Wet chemistry method

The first attempt at making an atomic layer of silicon was via chemical exfoliation of calcium disilicide ^{Reference Nakano, Ishii and Nakamura4,Reference Nakano, Mitsuoka, Harada, Horibuchi, Nozaki, Takahashi, Nonaka, Seno and Nakamura5} ( Figure 1 ). Chemical exfoliation is an attractive fabrication technique for nanostructures since it is a relatively inexpensive and low-technology technique. The basic idea is to use foreign atoms or ions to assist in splitting bonds (primarily via steric and kinetic attacks), leading to free layers. This, therefore, requires that the starting material have a layered structure (i.e., certain weaker bonds that can be preferentially broken while preserving other bonds within the layers). Thus, this technique is very effective for making graphene from graphite due to the weak interlayer van der Waals bonds. One common side effect of chemical exfoliation is a resulting functionalization of the prepared materials, particularly from the foreign atoms. However, if functionalization is desired, then this is an added advantage.

Figure 1.

(a) Model of CaSi₂; (b–c) top and side view of Mg-doped silicon sheet capped with oxygen; (d) TEM image of nanosheet. Reprinted with permission from Reference 5. © 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

A number of possible candidates exist for making silicene via chemical exfoliation. Thus, calcium disilicide (CaSi₂) has been used to prepare siloxene (a flat form of silicon with attached OH groups), which can then be exfoliated to produce siloxene nanosheets, ^{Reference Nakano, Ishii and Nakamura4} a form of functionalized silicene. Mg-doped silicene sheets capped with oxygen were obtained when Mg-doped CaSi₂ was directly exfoliated using propylamine hydrochloride; Mg doping was used because the doping reduced the charges on the Ca and Si and, therefore, aided in the exfoliation. On the other hand, polysilane could be exfoliated in an organic solvent to give silicon nanosheets covered with organic groups instead of oxygen. ^{Reference Okamoto, Kumai, Sugiyama, Mitsuoka, Nakanishi, Ohta, Nozaki, Yamaguchi, Shirai and Nakano8,Reference Sugiyama, Okamoto, Mitsuoka, Morikawa, Nakanishi, Ohta and Nakano9}

UHV deposition

The first breakthrough in experimental silicene research was the report of the fabrication of silicene nanoribbons on a silver substrate oriented in the (110) direction. The technique used was an ultrahigh vacuum (UHV) deposition of silicon atoms onto a metallic substrate. Deposition is one approach to prevent 3D island formation, as bulk Si is more stable than silicene by 1.56 eV/atom. ^{Reference Zhang and Yan10} Currently, successful fabrication of silicene sheets has been reported using an Ag (111) substrate held at 150–300°C and a slow deposition rate of Si below 0.1 monolayer per minute ^{Reference Vogt, De Padova, Quaresima, Avila, Frantzeskakis, Asensio, Resta, Ealet and Le Lay11–Reference Chiappe, Grazianetti, Tallarida, Fanciulli and Molle14} ( Figure 2 ). Silver has turned out to be an ideal substrate because of the low reactivity between Si and Ag and because of the compatibility of the crystal structures and lattice constants.

Figure 2.

Scanning tunneling microscopic image of silicene on Ag (111). Reprinted with permission from Reference 11. © 2012 American Physical Society.

There are still many unanswered questions about the growth of silicene on Ag. Thus, different superstructures of silicene on Ag have been reported, whether at one substrate temperature or at different temperatures, ^{Reference Lin, Arafune, Kawahara, Tsukahara, Minamitani, Kim, Takagi and Kawai12,Reference Jamgotchian, Colignon, Hamzaoui, Ealet, Hoarau, Aufray and Bibérian15} and there is not yet a conclusive explanation for all of these results. There appears to be a correlation with substrate temperature; ^{Reference Jamgotchian, Colignon, Hamzaoui, Ealet, Hoarau, Aufray and Bibérian15} however, theoretical modeling of structures (e.g., using molecular dynamics) is not at the level of predicting the superstructures. Two other results still under dispute concern the level of interaction between Si and Ag electronic states, and the presence or not of the Dirac electrons in silicene. These are related questions since they both concern what happens to the electrons in silicene in the presence of the Ag substrate; we will discuss this further later in the article. Silicene has also been reported to have been grown on ZrB₂ ^{Reference Fleurence, Friedlein, Ozaki, Kawai, Wang and Yamada-Takamura16} and Ir (111). ^{Reference Meng, Wang, Zhang, Du, Wu, Li, Zhang, Li, Zhou, Hofer and Gao17}

Structure

Freestanding silicene can be viewed as a 2D material. However, it is not completely flat, as is graphene. The most commonly reported structure has been with alternate Si atoms lying in planes that are separated from each other in the direction perpendicular to the planes by 0.45 Å ( Figure 3 a). Nevertheless, it makes sense to still refer to these planes as forming a buckled 2D sheet. In fact, such a sheet is very similar to the (111) plane of Si.

Figure 3.

(a–d) Structures of a freestanding monoatomic layer of silicon. Each figure represents a minimum energy configuration found, with the top drawing being a top view and the bottom drawing being a side view of the structure. Reproduced, in part, from Reference 20. © 2013 PCCP Owner Societies.

A hypothetical flat silicene has been shown to be metastable via calculation of the phonon modes. ^{Reference Cahangirov, Topsakal, Aktürk, Şahin and Ciraci18} The buckled form is also lower in energy by 30 meV/atom ^{Reference Durgun, Tongay and Ciraci19} and has a binding energy of 4.9 eV/atom, which is about 0.6 eV/atom lower than for bulk Si. ^{Reference Cahangirov, Topsakal, Aktürk, Şahin and Ciraci18} More recent work now indicates that the honeycomb structure of Si might not be the lowest energy one; ^{Reference Kaltsas and Tsetseris20} rather, other slightly more stable structures were obtained by allowing the Si (111) plane to reconstruct. In fact, the other structures, the so-called √3×√3, 5×5, and 7×7, (where the expressions label the geometrical commensurability with the basic unit cell) were found to be more stable than the silicene structure (Figure 3b–d). These examples were chosen because they are present in the surface reconstruction of bulk Si. Nevertheless, it is expected that the structure that is obtained in any given growth would be related to the growth conditions and substrate.

Thus, the structure of silicene on Ag is much more complex than previously thought, and a variety of superstructures have been reported from both experiments and theory. ^{Reference Vogt, De Padova, Quaresima, Avila, Frantzeskakis, Asensio, Resta, Ealet and Le Lay11,Reference Feng, Ding, Meng, Yao, He, Cheng, Chen and Wu13,Reference Chiappe, Grazianetti, Tallarida, Fanciulli and Molle14,Reference Lalmi, Oughaddou, Enriquez, Kara, Vizzini, Ealet and Aufray21–Reference Huang, Kang and Yang26} Growth conditions, particularly substrate temperature, ^{Reference Huang, Kang and Yang26} determine which superstructure is obtained. Some ambiguity remains due in part to experimental characterization (e.g., scanning tunneling microscopy) not associating a unique and well-defined structure to a given image.

Elastic properties

Two-dimensional materials require a new definition of their elastic constants (e.g., the 3D bulk modulus is defined in terms of isotropic pressure and volume change). One could define in-plane stiffness,

(1) $$C = h{C_{11}}\left[ {1 - {{\left( {{{{C_{11}}} \over {C_{12}^{}}}} \right)}^2}} \right]$$

in terms of the 3D elastic constants C ₁₁ and C ₁₂, and an effective thickness h. ^{Reference Şahin, Cahangirov, Topsakal, Bekaroglu, Aktürk, Senger and Ciraci27} One can also introduce 2D elastic constants via ^{Reference Wang28}

(2) $${\gamma _{{\rm{ij}}}}\, = \,{{\rm{C}}_{{\rm{ij}}}} \times {{\rm{c}}_{\rm{0}}},$$

where c₀ is the interlayer spacing of a hypothetical 3D supercell. Overall, it has been found that silicene is less rigid than graphene. This is easily understood in terms of the buckled structure of silicene compared to the flat structure of graphene.

With uniaxial strain, it is possible to open a bandgap, though the dependence has been found to be nonlinear. ^{Reference Zhao29} The largest gap was found to be about 80 meV for a strain of about 8%. Instability of the structure was obtained for a strain of 14–18%. On the other hand, a biaxial tensile strain converted silicene into a metallic state (for strain higher than 7.5%) due to the lowering of the conduction band at the Γ point (the point in the Brillouin zone with zero wave vector). ^{Reference Liu, Wu, Ouyang and Xu30,Reference Qin, Wang, Zhu and Zhang31} The latter result is different from graphene, which remains as a zero-gap material.

Electronic properties

The fascination with silicene is similar to that for graphene—freestanding silicene has been predicted to have Dirac cones just like graphene. It should be repeated that this result, at first sight, is not obvious due to the lower symmetry of silicene compared to graphene. Nevertheless, it has been shown rigorously that the degree of buckling does not affect the existence of the zero bandgap in the absence of spin–orbit coupling, from symmetry arguments. ^{Reference Guzmán-Verri and Yan Voon6} A similar result derived from ab initio calculations is given in Figure 4 .

Figure 4.

Band structure of flat and buckled silicene. Reprinted with permission from Reference 19. © 2005 American Physical Society. TDOS, total density of states.

Figure 5 depicts the so-called band structure, which is a plot of the electron energy versus its wave vector (related to its momentum). The point at the intersection of the Fermi level (which basically represents the separation between filled and unfilled electron states), given here at zero energy, and the point K shows the absence of the bandgap and the linear crossing of curves, the Dirac cone (when the line is rotated). The band structure can be recovered experimentally, with the most direct technique currently in use being the so-called ARPES (angle-resolved photoelectron spectroscopy) method. A photon of energy E is used to eject an electron of wave vector k; thus, simultaneous measurements of the two quantities are possible.

Figure 5.

(a) Angle-resolved photoelectron spectroscopy of a clean silver surface (left) and with silicene coverage (right); (b) Brillouin zone showing the location of various points. Reprinted with permission from Reference 11. © 2012 American Physical Society.

An example of an ARPES result is shown in Figure 5, which appears fairly convincing in showing a linear dispersion curve (the lighter region in the second panel on the left figure). Nevertheless, this result and another similar result ^{Reference Chen, Liu, Feng, He, Cheng, Ding, Meng, Yao and Wu32} have recently been questioned. ^{Reference Chen, Li, Feng, Ding, Qiu, Cheng, Wu and Meng33–Reference Wang and Cheng36} Recent experiments have come to the conclusion that the linear dispersion actually becomes quadratic near the Dirac point. ^{Reference Arafune, Lin, Kawahara, Tsukahara, Minamitani, Kim, Takagi and Kawai25} Moreover, a search for massless Dirac electrons via the generation of the quantum Hall effect turned out negative. ^{Reference Lin, Arafune, Kawahara, Kanno, Tsukahara, Minamitani, Kim, Kawai and Takagi35} The latter experimental results have also received support from theoretical calculations, revealing a strong hybridization between the Si and Ag electronic states ^{Reference Lin, Arafune, Kawahara, Kanno, Tsukahara, Minamitani, Kim, Kawai and Takagi35} and also others attaching the linear dispersion to an sp band origination from the silver. Indeed, the linearity of the silicene electron is unlikely down to 3 eV, as implied in Figure 5 (see the calculation of Wang and Cheng ^{Reference Wang and Cheng36} ). Still, defense of the Dirac cones persists, ranging from renewed identification of a linear tail to the silicene electronic dispersion even in the presence of a small gap ^{Reference Avila, de Padova, Cho, Colambo, Lorcy, Quaresima, Vogt, Resta, Le Lay and Asensio37} to associating Dirac cones to only certain superstructures. ^{Reference Chen, Li, Feng, Ding, Qiu, Cheng, Wu and Meng33} Clearly, the study of silicene on silver remains complex and rich.

The electronic properties of silicene in the presence of a vertical electric field (i.e., perpendicular to the plane) are important for possible application as field-effect transistors. In contrast to graphene, a bandgap opens for silicene due to the lower symmetry of the buckled structure. In effect, the two atoms in the unit cell feel different electric potentials since they are at different heights. Two early calculations both obtained a fairly linear dependence of the gap on the electric field; ^{Reference Ni, Liu, Tang, Zheng, Zhou, Qin, Gao, Yu and Lu38,Reference Drummond, Zólyomi and Fal’ko39} however, they obtained rates that differ by a factor of two.

It was noted previously that the bandgap is zero in the absence of spin–orbit coupling, which can split the bands and lead to a small gap opening. There are two contributions to the magnitude of the spin–orbit gap. One is the magnitude of the atomic spin–orbit coupling; the other relates to symmetry reduction due to buckling. It turns out that both effects contribute to a larger spin–orbit gap for silicene compared to graphene. It is, in fact, negligible for graphene, while it is about 1.55 meV for silicene. ^{Reference Liu, Feng and Yao40} While still small, it is sufficient to help demonstrate the quantum spin Hall effect; this effect arises from the existence of a bulk gapped state and gapless conducting edge states at the boundaries, an example of a topological insulator. An interesting proposal is to control the bandgap using an external electric field, ^{Reference Ezawa41–Reference Dyrdal and Barnaś43} transforming silicene from a topological insulator into a band insulator. Indeed, silicene has been predicted to have an extremely rich phase diagram of topological states with unique quantum states of matter such as a hybrid quantum Hall-quantum anomalous Hall state (the anomalous effect being the well-known quantum Hall effect in the absence of an external magnetic field) and a so-called valley-polarized metal (resulting from electron transfer from a conduction valley to a different hole valley), leading to the new field of spin valleytronics. ^{Reference Ezawa44–Reference Tabert and Nicol46}

Optical

In any study of silicon-based materials, optical properties are an area of interest since one of the “holy grails” of materials science is silicon-integrated photonics. Bulk Si is known to be a poor light emitter due to its indirect bandgap. There have been few studies on the optical properties of silicene to date due in part to the zero gap for the freestanding sheet and to the increased difficulty in computing optical properties using density functional theory (DFT). The latter is known to be deficient in computing excited-state properties, necessary to describe optical properties. Furthermore, the semi-metallic nature of gapless silicene reduces the possible contribution due to electron screening, complicating the theory, particularly for single-particle ones. Experimentally, no results have yet been reported, due to the difficulty in peeling silicene from the Ag substrate.

Early calculations on hydrogenated silicon nanosheets (the effect of hydrogenation on the electronic properties is covered in more detail in the next section) of different orientations have revealed the possibility of obtaining direct energy gaps of the order of 2–3 eV. ^{Reference Lu, Yang and Zhang47} We note, however, that detailed structures of the nanosheets were not provided, and they appear to be flat rather than buckled as we understand them to be now. Furthermore, those calculations were done using the generalized gradient approximation to DFT, which is known to underestimate the bandgap. Calculations that fix the previous problems ^{Reference Pulci, Gori, Marsili, Garbuio, Del Sole and Bechstedt48} show that there can be a blueshift of the transition by about 1 eV upon exciton formation and also an enhanced fundamental oscillator strength, which is about 500 times stronger than for graphane (the hydrogenated form of graphene); the latter has been attributed to the different nature of the lowest conduction states (being mostly localized on silicon atoms for silicane (the hydrogenated form of silicene) and mostly on hydrogen atoms for graphane).

One would expect the infrared properties to be very similar to those of graphene since those are due to the zero gap, linear energy dispersion, and the 2D nature. Indeed, the universality of the optical absorption (proportional to the universal fine structure constant) based upon a single-particle picture has been predicted. ^{Reference Bechstedt, Matthes, Gori and Pulci49}

Another aspect missing from the above is a study of excitons (the collective behavior of electrons and holes in optically excited materials) in silicene. It should be pointed out that even the excitonic spectrum of graphene is not fully understood. Recent calculations ^{Reference Wei, Dai, Huang and Jacob50} indicate that silicene also has resonant excitons like graphene.

Functionalization

In spite of the interest in pure silicene, there are a number of reasons why functionalization is of some importance. First and foremost, growth processes might naturally lead to functionalized sheets (e.g., via wet chemical methods). Second, the additional functionality introduced by adding various other atoms enlarges the variety of properties attainable. Four general types of functionalizations have been studied for silicene: hydrogenation, halogenation, by metals, and by organic groups.

Hydrogenation is, by far, the most studied. ^{Reference Lu, Yang and Zhang47,Reference Lew Yan Voon, Sandberg, Aga and Farajian51–Reference Ding and Ni55} Hydrogenation of graphene was studied early on as a way of opening an energy gap; hence, it is natural to consider it for silicene as well. Also, experimentally, it is often possible to obtain hydrogenated samples. Theoretically, incorporation of hydrogen atoms is the standard approach to saturating dangling bonds. Finally, from an application perspective, one can also envision hydrogen storage as an important goal of silicon (and other material) research. The basic results are that full hydrogenation, with the resulting material being called silicane, does lead to a bandgap opening. ^{Reference Lew Yan Voon, Sandberg, Aga and Farajian51} Boat-like and chair-like configurations are found to be the most stable ones but with very different bandgaps (2.9 eV direct and 3.8 eV indirect, respectively, using many-body perturbation theory ^{Reference Houssa, Scalise, Sankaran, Pourtois, Afanas’ev and Stesmans53} ). More interestingly, half-hydrogenation is found to lead to a direct-gap insulator with a gap of 1.74–1.79 eV for the chair configuration and a ferromagnetic state. ^{Reference Wang, Li and Wang56,Reference Zhang, Li, Hu, Wu and Zu57} The latter has been attributed to the unpaired 3p electrons on the unhydrogenated silicon sites. On the basis of mean-field approximation arguments, the Curie temperature was estimated to be 122–300 K. ^{Reference Wang, Li and Wang56,Reference Zhang, Li, Hu, Wu and Zu57} One approach to obtaining half-hydrogenation is to first fully hydrogenate silicene and then apply a vertical electric field. ^{Reference Gang, Liu and Li58}

Fluorination performs a similar function as hydrogenation. It has been found that fluorinated silicene is more stable than silicane and with a smaller bandgap. ^{Reference Garcia, de Lima, Assali and Justo52,Reference Ding and Ni55} Bromine was found to lead to even more interesting behavior. ^{Reference Zheng and Zhang59} Half-brominated silicene is an antiferromagnetic half-metal, whereby one spin channel is metallic, and the other is semiconducting with a 1.73 eV energy gap.

Metal adatoms display various states and properties, as one can expect from their variety. Even for graphene, metal adatoms have led to a variety of behaviors such as superconductivity with lithium and catalytic properties with transition metals. We refer the reader to the paper by Lin and Ni for an extensive study. ^{Reference Lin and Ni60} A general conclusion is that metal adatoms bind much more strongly to silicene than to graphene. The bonding is covalent in all cases except for alkali metal atoms.

Other properties

In addition to the previous extensive studies, we would like to mention some other work on the properties of silicene. A few papers have been published on various thermal properties. ^{Reference Li and Zhang61,Reference Hu, Zhang and Poulikakos62} In particular, the in-plane thermal conductivity of silicene is found to be about an order of magnitude lower than for bulk Si. Vacancy defects lead to further reduction due to phonon-defect scattering. Furthermore, the thermal conductivity of silicene is found to initially increase with tensile strain, ^{Reference Hu, Zhang and Poulikakos62} opposite to what is observed for graphene. The different behavior was traced to the fact that all the vibrational modes for graphene softened with tensile strain, whereas this is true for the longitudinal and transverse modes for silicene but not for the flexural modes (perpendicular to the layer). The latter competition between phonon softening and stiffening for silicene leads to a nonmonotonic behavior of the thermal conductivity as a function of strain.