Two-dimensional (2D) materials have attracted immense attention and interest from the scientific community and industry. By definition, 2D materials refer to laterally infinite and ordered structures with a thickness within single or several atoms. Differing from their three-dimensional (3D) bulk counterparts, materials with 2D layered, anisotropic structures have many compelling physical and chemical properties such as dimensional confinement of charge carriers, anisotropic electronic and magnetic response, non-linear optical properties, superior mechanical strength and flexibility, and facet-dependent catalytic behaviors^[1-5]. With these features, 2D materials have shown great promise in a number of applications, such as energy storage^[6-9], optoelectronics^[10,11], and artificial neural devices^[12-15].

Graphene, the carbon allotrope in a monolayer with a planar structure and has undoubtedly been the center of 2D material research in the last decade. Other Group 14 elements, such as silicon and germanium, have also gained considerable attention because of the similarity of their electronic structure compared to graphene based on theoretical calculations^[16]. It is also believed that 2D silicon (2DSi) may be one of the material candidates to substitute crystalline silicon (c-Si) for the next-generation electronics. To date, the materials properties of 2DSis have been widely investigated through both experimental and computational studies^[16-25]. The corresponding optoelectronic features of 2DSis have been described in simulation reports, such as size-tunable electronic structures and direct-bandgap-like electronic structures^[26,27]. These remarkable properties enable 2DSis to be considered as candidates for active materials in various optoelectronic devices, such as light-emitting diodes (LEDs), and photodetectors^[28-32].

To date, despite a vast number of reports about the simulated properties of 2DSi nanostructures, there have only been a few experimental results focusing on the synthetic approaches and the material characterizations of 2DSis. One of the major reasons for this gap is the relatively complex procedures for the preparation of 2DSis. Unlike graphene, which can be prepared using numerous chemical and physical approaches (e.g., mechanical exfoliation, electrochemical synthesis, hydrothermal assembly, etc.), there are only a few synthetic routes to make 2DSis with clear structures^[33-35]. Meanwhile, inconsistent results about the physical and chemical properties of 2DSis are commonly found in the literature, including experimental studies and theoretical predictions. It is therefore important to summarize the results from both experimental and theoretical studies for deepen the understanding of the chemical and physical properties of 2DSi. Therefore, in this review article, we aim to provide a comprehensive summary of the recent development in the research field of 2DSi nanomaterials, especially focusing on the progress of synthesis, surface functionalization, and the corresponding optoelectronic applications.

The contents of this review are divided into four major sections. The discussion starts with a description of the various types of 2DSi materials (the first section) and the general synthetic protocols (the second section). In the third section, we describe several surface functionalization approaches to chemically modify the surface of 2DSis, which improve the chemical and physical properties of the intrinsic layered silicon backbones. The discussion in the fourth section focuses on the electronic properties and the state-of-the-art optoelectronics based on 2DSi materials. We sought to connect the structural properties and the surface bonding environments of various types of 2DSis with their optoelectronic behaviors and device performance. We finally provide some viewpoints on the future directions of the field of 2DSi-based optoelectronics.

There may be a misconception that 2DSi is a new member of the 2D material family, although the immense discussion about 2DSi only arises after the discovery of graphene^[36]. The finding of silicon with layered structures was first reported by Wöhler et al. in 1863^[37]. It was noticed that yellow-green-colored sheet-like powders could be obtained when calcium silicide (CaSi₂) is mixed with concentrated HCl solution. It was later confirmed that the product was partially hydroxylated silicane (Si₆H₃(OH)₃)^[38,39]. According to the definition, there are three types of 2DSis whose terms have been widely used in literature: silicene, silicane, and multilayered silicon nanosheets. Because these structures are highly relevant and may cause potential confusion, we summarize their structural features inFig. 1.

Silicene is a two-dimensional allotrope of silicon whose structure is similar to that of graphene. In contrast to the graphene-like flat 2D structure, silicene has a monolayered hexagonal structure but buckled to tiny degrees (Fig. 1). This happens because the Si atoms in silicene tend to adopt mixed sp^x hybridization, where 2 < x < 3. Based on the results from computational studies, the energy difference of mixed hybridized (i.e., sp²-sp³) silicene with a buckle-like structure is much more stable than the planar counterpart^[16]. The in-plane lattice constant of 6.4 Å was measured experimentally from the silicene sample grown on an Ag(111) structure under ultrahigh vacuum (UHV) conditions^[29]. Because silicon atoms favor the sp³ hybridization to construct the Si-Si-bond-based network, it is not unexpected that silicene is not stable against air or moisture; indeed, there is no report about the successful preparation of freestanding silicene under those conditions.

The partially conjugated structures enable silicene to have a similar electronic structure to that of graphene. For instance, it has been predicted that silicene also has a Dirac cone, and the charge carriers within may behave like Dirac fermions at the K point^[16]. Silicene also shows linear electronic dispersion at the regimes close to the Dirac points with a calculated Fermi velocity value of 5.1 × 10⁵ m/s^[17]. Meanwhile, the slightly buckled structure enables silicene to have a more substantial quantum spin Hall effect than graphene. Based on the first-principle calculations, the Dirac point in silicene is expected to be opened at low temperatures due to the strong spin-orbit coupling^[19]. Because the transition temperature of 18 K is much higher than that of graphene (below ~0.01 K)^[40], the bandgap of silicene can be further tuned by the application of an external inhomogeneous electric field (the critical electric field is around 17 mV/Å)^[41]. These features suggest that silicene can be applied to spintronic devices under more realistic operational conditions.

Although it is commonly misnamed as “silicene” in the literature^[42], silicane is another class of 2DSi in which silicon atoms are fully sp³ hybridized (Fig. 1). Unlike the sp²-based framework of silicene, silicon atoms on silicane are passivated by adatoms or organic groups to maintain sp³ hybridization. Because of the additional surface passivating groups, silicane cannot be considered as the allotrope of silicon. Various names have been given for silicane structures with different surface passivants (e.g., siloxene (Si₆H₃(OH)₃), layered polysilane (Si₆H₆)) to distinguish them from their linear polysilane counterparts, which are mostly synthesized from the dehydrogenative coupling or Wurtz-type reductive dehalogenation reactions^[43].

Regardless of different passivating ligands, the buckled silicon backbone within silicane is similar to the (111) facet of crystalline silicon (c-Si). The Si–Si bond length is ~2.35 Å^[39], which is the same as that in c-Si. The lateral distance between two adjacent silicane sheets varies from 0.59 nm up to 3 nm depending on the type of passivating ligand^[39,44,45]. However, the electronic structure is insensitive to the length of the ligands^[45]. This happens because the valance band (VB) maximum energetic states and the conduction band (CB) minimum energetic states are dominated by the contribution of the silicon atoms^[45]. Ryan et al. investigated the band structure of silicane with OH- and Cl-terminating groups and showed similar band structures and projected density of states (pDOS) compared to previous reports^[27]. The density functional theory (DFT) simulation results also confirmed the “direct-like” bandgap behavior of silicane (direct bandgap: 2.26 eV, indirect bandgap: 2.16 eV). It was also predicted that the surface-terminated elements with larger electronegativity values may cause the downshifting of the bandedge compared with that of H-terminated silicane^[27].

Multilayered silicon nanosheets (MSNs) are representative 2DSi nanostructures with a thickness greater than one but less than tens of silicon atomic layers. MSNs with various structures have been investigated through computational studies^[22]. However, only two general structures have been experimentally determined: (I) multilayered silicene and (II) ultrathin diamond-like c-Si^[46]. These two structures can be distinguished through spectroscopic and microscopic techniques such as Raman spectroscopy^[47], energy dispersive grazing incidence X-ray diffraction (ED-GIXRD), and scanning tunneling microscopy (STM)^[48]. While the former is only found from the sample prepared by the epitaxial growth approaches^[49], the 2D structure with intrinsic c-Si scaffold can be prepared by multiple synthetic methods such as solution-based wet chemistry methods or vapor-deposition-based techniques^[50].

It has been theoretically predicted that multilayered silicene has linearly dispersed Dirac bands^[51], similar to that present in the bilayered graphene^[22]. The presence of Dirac fermions has been experimentally confirmed by spectroscopic techniques such as angle-resolved photoelectron spectroscopy and scanning tunneling spectroscopy (STS)^[52,53]. Computational results also reveal that the band structures and the doping properties of ultrathin c-Si nanosheets with double-layered diamond structures are highly tunable and dependent on the surface terminating groups^[18].

Due to the incorporation of mixed sp²/sp³ hybridization of silicon atoms within the constitutive multilayered backbone, the structural transfer between the multilayered silicene and the ultrathin c-Si can occur under relatively high temperatures (e.g., 200–300 °C)^[46,47], which can be monitored by STS and Raman spectroscopy. It has also been noticed that the ultrathin c-Si structure (i.e., cubic structured silicon nanosheet) experiences an irreversible phase transition (cubic → tetragonal → hexagonal) during the increase of pressure, and finally forms one-dimensional (1D) c-Si nanowires^[54]. The corresponding computational studies have further indicated the reduction of thermal conductivity in such 1D structures, which may allow the application of these ultrathin structures in thermoelectric devices.

Despite the existence of numerous 2D models in theoretical studies, only a few structures of 2DSi have been successfully synthesized with comprehensive analyses of the crystal structures. There are some excellent reviews that have summarized detailed experimental conditions to prepare 2D silicon materials^[50,55,56]. In this review, we prefer the description of the formation mechanisms within those synthetic protocols.

As for graphene, the typical 2D structures of elemental carbon can be directly synthesized by the exfoliation of graphite^[36]. However, 2DSi cannot be obtained from the physical exfoliation of c-Si because of the strong covalent backbone of the diamond structure. Nowadays, there are two general approaches to preparing 2D silicon nanomaterials: epitaxial growth and chemical exfoliation (Fig. 2).

The preparation of silicene is achieved vias the epitaxial growth approach. Generally, freestanding silicon atoms are first generated by the pyrolysis of silicon precursors (e.g., silanes) or the ionization of bulk silicon under high current (e.g., > 90 A)^[57,58], which are subsequently deposited on specific substrates (e.g., Ag(111)^[29,59], Ir(111)^[60], and ZrB₂(0001))^[61]. The high crystalline coherence between the graphene-like honeycomb lattice structure of silicon and the predesigned inorganic phase of the substrate allows the alignment of silicon atoms and forms planar silicene via heterointerface with the substrate. It is important to note that silicene is extremely reactive with oxidants due to the unstable sp²-hybridization of silicon. Consequently, the preparation and characterizations of silicene are required to performed under UHV conditions. Combined with the computational structural analysis, atomically resolved STM is a powerful tool to reveal the epitaxially grown silicene structures^[23,59]. Feng et al. reported the experimental results of silicene sheets grown on Ag(111) surfaces. The line-scan profiles analysis revealed a height difference of 0.2 Å between two adjacent layers of silicon atoms, indicating Si sp²-sp³ hybridization, which aligns well with the theoretical prediction^[29].

Chemical exfoliation is a “top-down” strategy that has been widely used in the preparation of materials with layered structures^[62-64]. Chemical exfoliation is also the first recorded method to prepare 2DSi. In 1863, Wöhler et al. mixed calcium silicide (CaSi₂) with concentrated HCl aqueous solution and obtained yellow-green colored sheet-like powders, which were later confirmed to be partially hydroxy-terminated silicane (Si₆H₃(OH)₃)^[37-39]. It is widely accepted that metal silicide precursors (e.g., CaSi₂, MgSi₂) with layered structures are key to preparing silicane sheets with various terminating groups. These metal silicides share Si₆ cyclohexane-like subunits interconnected in each plane separated by a bivalent cation layer (Fig. 2(b)). The liberation of a single layer of silicon can be achieved by the deintercalation of bivalent metal cations during the hydrogenation reaction^[65].

The reaction parameters are crucial to the preparation of silicane. Yamanaka et al. reported that low synthetic temperatures (e.g., below –30 °C) could effectively inhibit the surface oxidation of as-formed Si–H surfaces, yielding hydride-terminated silicane^[65]. Once the temperature increases to 0 °C, the oxidation becomes evident, yielding silicanes with mixed hydride/hydroxide termination^[65]. The size of silicane can be further tailored by the reaction time. Mixed hydride/hydroxide-terminated silicane with a size up to ~0.25 µm² can be formed from a prolonged reaction (e.g., up to 10 days) at room temperature^[66].

Both epitaxial growth and chemical redox reactions can be utilized to prepare multilayered silicon nanosheets. Despite the difference in thickness and morphology, silicon nanosheets share a similar cubic diamond crystal structure and thus can be readily prepared using the epitaxial growth approach. Dong et al. reported the formation of freestanding silicon nanosheets by depositing silicon atoms on a water-cooled wall of the evaporation chamber at high discharge voltage under a slightly reducing atmosphere^[57]. The redox reactions of metal silicides can also form a multilayered structure. In this case, metal silicide precursors may not contain the closely packed layer of silicon atoms. For instance, lithium silicide (Li₁₃Si₄) with an orthorhombic structure can be converted to layered silicon nanosheets while leaching in ethyl alcohol and thermal annealing at 550 ^oC^[67].

The optimization of synthetic parameters enables the structural conversion of 2DSi. Grazianetti et al. demonstrated that both multilayered, stacked silicene and diamond-structured silicon nanosheets were formed by the epitaxial growth of silicon on an Ag(111) substrate at temperature regimes^[47]. Single-layered silicene can be used as the template or precursor for the growth of the successive layer with a thickness of up to 24 layers of silicon atoms. In situ STM analysis and Raman spectroscopic results indicated the formation of multilayered silicene and diamond-structured silicon at intermediate (200–230 °C) and high (350 °C) temperature regimes, respectively. The sp²-hybridized silicon atoms in silicene can be converted to partially sp³-hybridization under hydrogenation, yielding a “half-silicane” structure whose sublattice is fully hydrogen-saturated and the other lattice is intact due to the direct contact with the Ag(111) substrate^[68,69].Table 1 summarizes recent progress in the preparation of 2DSi materials and the corresponding synthetic features.

Similar to other silicon-based nanostructures, the as-formed 2DSi materials are prone to oxidation^[70,71]. As for silicene, the backbone is constructed by the sp²-sp³ network of silicon with partially dangling bonds. Therefore, the surface silicon atoms on silicene are supposed to be “naked”, and the silicene's pseudo-planar structure is only stable under UHV conditions^[72]. The post-synthetic passivation of the silicon atoms with sp²-hybridization or dangling bond structures causes the reconstruction of silicene, yielding half- or full-silicane with various terminating groups^[68,69]. In contrast, surface modification does not change the hybridization of silicon atoms in silicane and multilayered silicon nanosheets. Consequently, the discussion in this section only focuses on sp³-hybridized 2DSi structures.

The as-formed silicane and multilayered silicon nanosheets are mostly terminated by either hydride or hydroxyl groups, regardless of the synthetic process used. Because of the weaker binding energy compared to that of the Si–O bond (Si-H: 318 kJ/mol, Si–O: 452 kJ/mol)^[73], the surface Si–H bonds are inevitably oxidized when 2DSi materials are exposed to air or moisture atmosphere. The non-controllable oxidation^[27] cleaves the Si–Si bonds and forms Si–O–Si connections^[74]. The reactions involving the adjacent layers of silicanes further destroys the original stable layered structure, and turn into an irregular amorphous silicon structure^[8].

It is highly possible that the oxidation of silicon introduces surface trap states and impair the optical properties of silicane. Consequently, it is crucial to apply surface modification approaches to stabilize the 2DSis and inhibit the formation of undesired electrical/optical traps. The general reactivity of Si–H bonds on 2DSi is similar to those on the molecular silanes, bulk c-Si, and Si nanocrystals, although the bond strength may slightly vary due to the difference in steric hindrance on these structures. Various procedures have been developed for passivating silicon surfaces which have been demonstrated to functionalize 2DSis. This section reviews several strategies to modify the surfaces of hydride-terminated 2DSi with robust chemical bonds such as Si–O, Si–C, and Si–N. Some recently reported surface modification approaches and related benefits are summarized inFig. 3.

Organoamines can directly react with hydride-terminated silicon surfaces due to their nucleophilic nature, yielding stable Si-N covalent bonds^[75]. The observation of amine-passivation on 2DSi was first reported by Okamoto et al. and gave well-defined layered structural organoamine-functionalized surfaces(Fig. 4)^[44]. Hydride-terminated silicane prepared at low temperature reacts with decylamine at 60 °C for 12 h under an inert atmosphere, yielding stacked, amine-functionalized silicane. Similar to silicon nanocrystals (SiNCs), the interaction between silicane with nitrogen source leads to size-independent blue photoluminescence (PL) at 450 nm, which was assigned to the trap-induced radiative recombination^[75]. The surface modification of SiNCs using amine sources is usually incomplete and introduces a substantial amount of oxide species^[75]. However, the oxidation was not found from the amine-passivated silicane as indicated by spectroscopic techniques such as Fourier-transform infrared (FT-IR) spectroscopy and X-ray absorption near-edge structure (XANES) analysis^[44].

Ohshita et al. further demonstrated that the single-step amine-functionalization approach could effectively link silicane with aromatic substituents^[76].Three types of aromatic organoamines (benzylamine, 1-naphthylmethylamine, and (9-Butyl-9H-carbazol-3-yl)-methanamine) can directly substitute the original Si-H bonds, yielding arylmethylamine modified 2DSi. The additional absorption features and excitation-dependent PL features suggest that arylmethylamine-ligands with expanded conjugation affect the optical properties of 2DSi.

Hydrosilylation is a class of reactions of Si-H bonds with unsaturated organic groups (e.g., alkenes, alkynes) to form robust Si-C bonds. Because hydrosilylation strategies provide effective passivation of silicon atoms, they have been widely applied to stabilize and modify silicon surfaces. Linford and Chidsey reported the first example of hydrosilylation on the surface of bulk silicon in 1993^[78]. Since then, several proposed mechanisms have been proposed, such as homolytic cleavage of the Si–H bond^[79-82], exciton mediation^[83], radical-driven Si–H cleavage^[84], and photoemission^[85]. Many of these hydrosilylation approaches have been applied to modify Si–H bonds on 2DSis, with very few exceptions. For instance, Helbich et al. observed that the 2D features and the PL of silicon nanosheets disappear after the UV irradiation^[77]. This result indicates that the 2DSi material is sensitive to the light irradiation and thus photoinitiated hydrosilylation reactions (e.g., photoinitiated Si–H cleavage, exciton mediation) should not be applied on 2DSis.

Homolytic cleavage of Si–H bonds on 2DSi and the subsequent hydrosilylation reactions can be driven by heat or radical initiators under mild conditions. Thermal hydrosilylation is one of the most commonly used hydrosilylation approaches to functionalize silicon surfaces (Fig. 5). It is efficient, catalyst-free, and morphologically independent, and has been examined to be effective to functionalize hydride-terminated silicane with functional alkenes and alkynes^[86,87]. Commercially available radical initiators (e.g., decylbenzene diazonium tetrafluoroborate (DDB) and azobisisobutyronitrile (AIBN))^[77,86] have been proven to effectively initiate the hydrosilylation reactions on 2DSi at relatively low temperatures, yielding functionalized 2DSis and even 2DSi/polymer hybrid materials (Fig. 4(f))^[77,88,89].

Lewis acids are capable of initiating hydrosilylation reactions on molecular silanes^[90], bulk c-Si^[91], and colloidal SiNCs^[92]. Similar approaches have also been applied on 2DSis to prepare inorganic/organic hybrid thin films^[93]. Helbich et al. reported a series of organoboranes that can effectively cleave the Si–H bonds on 2DSis and enable subsequent functionalization with unsaturated organic molecules (Fig. 5)^[93].

The Grignard-based protocol to the surface modification of silicon nanostructures is drawn from its molecular silane analogues. A two-step approach to passivate surface Si–H bonds has been proposed: (i) halogenation to convert surface Si–H bonds to Si–X bonds (X = halide groups, such as Cl, Br); and (ii) subsequent reaction with Grignard reagents to form alkyl-terminated surfaces. Following the idea of Grignard reactions, Sugiyama et al. proposed to functionalize silicane using phenyl magnesium bromide as the Grignard reagent^[94]. Interestingly, the modification was even achieved by a single-step reaction between hydride-terminated 2DSi and the Grignard reagent, indicating that the Si–H bonds on silicane have higher reactivity than those on bulk c-Si or silicon nanocrystals. However, due to the large steric hindrance of the phenyl ring, the substitution from hydrogen to phenyl group was not complete, yielding partially phenyl-passivated silicane with intense blue emission (Fig. 6(a)).

Ohashi et al. also demonstrated a direct surface modification approach on 2DSi using benzyl bromide as the passivant and the solvent^[95]. The nucleophilic reaction between calcium silicide and benzyl bromide can occur at 150 °C in a sealed Teflon reactor, in which the benzyl-passivation on Si atoms proceeded in parallel with the cleavage of CaSi₂. The introduction of the surface benzyl group leads to the shrinkage of the bandgap, resulting in the red-shift of the absorption edge of the functionalized 2DSi (Fig. 6(b)).

As a new member in the 2D material family and a silicon allotrope in ultrathin structure, 2DSi may pave the way for silicon in diverse applications such as lithium-ion battery anodes^[42,96], sensors^[97,98], photocathode^[99], and catalysis^[100]. 2DSi have also been considered for use as active materials in electronic devices due to their unique fundamental features such as indirect-bandgap-like band structures^[27], quantum spin Hall effect^[19], chiral superconductivity^[101], and exotic field-dependent states^[102]. In the following section, we focus on the recent computational and experimental advances in 2DSi-based optoelectronics.

Theoretical studies have predicted that the charge carriers on sp²-hybridized silicene would have massless particles due to the linear Fermi–Dirac dispersion from the half-filled p_z orbitals^[16]. Unlike graphene, silicene has a smaller bandgap (around 1.55 eV) opening at the Dirac point^[19], and the Fermi velocity extracted from the Dirac cone is around 5.8 × 10⁷ cm/s^[103]. It has alsobeen predicted that silicene will exhibit extremely high electron mobility. The calculation results by Kim et al. indicate that the intrinsic mobility of silicene is ~1200 cm²/(V·s) with a saturation velocity of 3.9 × 10⁶ cm/s s at 300 K (Fig. 7(a))^[103]. The bandgap and carrier mobility can be further tuned by adjusting dopant elements, surface modification, and defects. Ding et al. investigated the bandgap structures of hydride- and fluoride-passivated silicene (H-silicene and F-silicene) and showed that the surface adatoms could significantly change the band structure (H-silicene: indirect bandgap, E_g = 2.19 eV; F-silicene: direct bandgap, E_g = 0.70 eV)^[104]. The bandgap modification can also be achieved by the attachment of organic ligands such as alkyl-, alkoxyl-, and amine-groups (Figs. 7(b)–7(g))^[45,105]. The high tunability of band structure makes 2DSi an attractive, active material for transistor applications. In 2012, Ni et al. investigated the functions of 2DSi-based FETs and showed that dual-gated silicene has a low on-off ratio of 4.2 when applying a vertical electric field of 1 V/Å at 300 K^[106]. Quhe et al. noted that sodium adatoms can effectively increase the bandgap of 2DSi, which substantially improves the on-off ratio of 2DSi-based FETs up to 4 × 10^8[26].

The light-induced photocurrent of silicane was first analyzed by Sugiyama et al. in the report about phenyl passivation on silicane via the Grignard reaction approach^[94]. The phenyl-capped silicane (Si₆H₄Ph₂) sample was drop-cast on a SnO₂:F (FTO) substrate and used as the photoelectrode in a photoelectrochemical cell. The photocurrent was detected when the electrode was directly irradiated using a xenon lamp and disappeared when the irradiation was discontinued, indicating that the photocurrent forms via a bandgap transition.

The photoexcitation wavelength depends on the absorption bandedge; therefore, it is crucial to tailor the absorption range of 2DSi. Ohshita et al. observed that the capping ligands with extended conjugating features (e.g., naphthylmethyl and carbazolylmethyl groups) could introduce additional absorption features and facilitate the photocurrent generation^[76]. This phenomenon was found from the benzyl-passivated silicane sample as the bandgap of silicane reduced from 2.31 to 2.13 eV^[95]. The significant red-shift of the absorption bandedge of silicane from ~520 to ~800 nm enabled the generation of photocurrent response under visible light irradiation (> 420 nm). Because of the anchoring of the densely packed benzyl group, the stability of the 2DSi structure was substantially enhanced. No decomposition of the silicane electrode was observed even after a 7-day immersion in the electrolyte solution.

The application of 2DSi/polymer homogenous hybrids as active materials for optoelectronic sensors has also been investigated^[32]. Polymerizable molecules, such as styrene and tert-butyl methacrylate (^tBMA), have been applied to functionalized hydride-terminated silicane via radical-induced hydrosilylation, and were subsequently deposited on the metal electrodes via an airbrush-coating system (Fig. 8). An evident light-dependent change in drain current was observed as the photo-induced current increased exponentially toward higher and lower drain voltages once the absolute value of the drain voltages rose over 2 V.

Many experimental studies have also been conducted on the fabrication of 2DSi-based transistors. In 2015, Tao et al. fabricated the first proof-of-concept silicene-based FETs^[31]. Silicene was epitaxially grown on Ag(111)/mica substrate under UHV condition, encapsulated by Al₂O₃, and transferred onto the SiO₂/Si substrate for contact deposition (Fig. 9). The manufacturing processes successfully stabilized silicene, so that the Raman and electrical signatures can last for ~2 min in air. The carrier mobility of ~100 cm²/(V·s) (µ_h = 99–129 cm²/(V·s), µ_e = 58–86 cm²/(V·s)) was obtained from the non-degraded devices, which was believed to be attributed to the acoustic phonon-limited transport and grain boundary scattering. The differences in the charge carrier transport behaviors between single-layered and multilayered silicene were investigated by Grazianetti et al.^[47]. A similar encapsulation-free delamination method^[31] was applied on 2DSi samples with various layers of silicene (10 and 24 layers) to fabricate FETs. The multilayer silicene sample exhibited distinguished ambipolar behavior of the drain current at both sides of the branch bias. The measured charge neutrality point (i.e., the Dirac point V_D) ranged between 6 and 8 V with the current ON/OFF ratio up to 1.5×. The room-temperature field-effect mobilities of 10 layers and 24 layers of silicene were 66–200 and 49–63 cm²/(V·s), respectively, with a residual carrier density V_D of ~10¹² cm^–2, which were two orders of magnitude higher than that of the FETs using single-layer silicene as the active material. More importantly, these multilayered silicene FETs showed higher reproducible ambipolar response and improved stability (stable up to two days) than the single-layered silicene devices.

In addition to silicene, both silicane and MSNs have been used to fabricate thin-film transistors (TFTs). Helbich et al. functionalized hydride-terminated silicane with various unsaturated organic groups using hydrosilylation catalyzed by Lewis acids^[93]. The functionalized silicanes exhibited improved materials stability and solution-processability and were capable of being blended with commercial semiconducting polymer poly(3-hexylthiophene-2,5-diyl) (P3HT) to construct solution-gated FETs (SGFETs). The devices operated at room temperature under ambient conditions with transconductances (g_m) of 7.45 × 10⁵ S. Compared to the control P3HT-only sample, devices with functionalized silicanes showed enhanced sensitivity and increased device transconductance by more than 100%. Lyuleeva et al. further optimized the silicane/polymer hybrids and directly functionalized silicane with thiophene-based ligands^[87]. The direct functionalization with thiophene ligands improves the electronic transportation from the surface-modified silicane to P3HT via the polaron hopping processes from the P3HT polymer chain to the thiophene group. These devices had relatively low OFF currents and high ON currents, leading to an improved ON/OFF ratio of 8.3 × 10².

Nakano et al. recently developed phototransistors using the silicane/graphene hybrids as the active materials^[107]. Hydride-terminated silicane was functionalized via dehydrogenation with arylmethylamine derivatives such as 4-(aminomethyl)quinoline (AM-Qu), 4-(aminomethyl)pyridine (AM-Py), and p-(aminomethyl)triphenylamine (AM-TPA). Computational studies confirmed that the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of passivated silicane were mainly contributed on the anchored organoamines and the silicon atoms, respectively. These features were believed to facilitate the photo-induced charge transfer from the ligands to the silicane backbone. The V_D of the TPA-Si/graphene hybrid phototransistors was 7 V, which was significantly smaller than those of pristine graphene (V_D > 50 V) and polymer-coated graphene devices (V_D = 20 V). Although the hole mobility wass smaller than that of the silicene-based devices^[31], the phototransistors with amine-passivated silicanes show improved stability, which was attributed to the surface protection provided by the organic ligands.

Differing from other quantum-confined materials (e.g., quantum dots) whose PL properties can be tailored through precise particle size control, the PL tuning of silicane can be achieved through surface modification. The green PL of hydride- or hydroxyl-terminated silicane may originate from the radiative recombination at the direct-bandgap-like bandedge^[27]. In contrast, the blue PL from the amine-functionalized silicane was attributed to the trap states created by the interaction with organoamines or oxide species^[56]. The PL quantum yield (PLQY) values of the existent silicanes are relatively low (usually below 10%) compared to their SiNC counterparts^[108]. For instance, Karar et al. reported that the PLQY of green-emitting quasi-two-dimensional silicon nanosheets (i.e., stacked, partially oxide-crosslinked silicanes) is 9%^[109]. A similar PLQY value was obtained from the hydride-/hydroxyl-terminated silicane^[27].

Single-layered silicene has garnered much interest due to its unique electrical and optical properties; however, the instability of the intrinsic materials under operating conditions precludes its use as the active materials for optoelectronics in many circumstances. However, the freestanding bilayer silicene (BSi) has been proven experimentally to have better stability and has been considered as an ideal material candidate for superconductors and topological insulators^{[28,29,101,110]}. The potentials of BSi as active materials for optoelectronics were investigated using DFT calculations by Huang et al.^[111] The results indicate that the naked BSi has an indirect bandgap, which may not be suitable for optoelectronic applications, whereas the partially hydride-terminated BSi nanoclusters with the numbers of Si atoms below 28 show dipole-allowed direct (or quasi-direct) bandgaps with gaps between 1.01 and 2.85 eV. Three BSi structures (Si₈H₄, Si₁₆H₁₂, and Si₁₂H₁₀) have been predicted to have direct (or quasi direct) with a range of the full visible regime, which can be used as the active materials for white-light LEDs.

The first 2DSi-based LED was reported by Kim et al. in 2014^[30]. Single-crystalline MSNs were grown through chemical vapor deposition using SiCl₄ as the silicon precursor under the H₂ atmosphere. By controlling the growth times ranging from 1.5 to 20 min, MSNs with various sizes were formed with PL across the visible to near-infrared region (from 385 to 765 nm). White-light emissions were also obtained by mixing the MSNs prepared with different growth times. The authors further fabricated solution-processable LEDs with an MSN-based emissive layer (Fig. 10). The resulting white-light and blue-light electroluminescence (EL) spectra were consistent with the PL features of the corresponding MSNs, indicating that the EL originates from the radiative recombination of the charge carriers at the bandedge of MSNs.

This review has outlined the recent progress relative to 2DSi nanomaterials, surface modification approaches, and the corresponding optoelectronic applications. As one of the important members in the family of nanomaterials, 2DSis have demonstrated encouraging electronic and optical properties, and the corresponding devices have advanced during the last decade. The intrinsic features of silicon, such as elemental abundance and high biocompatibility, further make 2DSi amenable to be applied in energy storage and conversion, biotechnology, and photonics.

Despite the recent success in 2DSi research, several critical issues must be addressed. The first issue concerns the scale-up synthesis of 2DSi materials with high crystalline quality. Although preparation approaches, such as epitaxial growth and chemical exfoliation, have been optimized for various types of 2DSi, we do not yet understand the growth mechanisms of 2DSi materials under certain reaction conditions. Thus, it is crucial to revisit and further optimize the synthesis routes of 2DSi, such as:

(1) To gain a better understanding of the growth mechanism behind the synthesis route, to determine the step reactions and the roles of silicon precursors, solvents, environments, reaction temperature and time;

(2) To precisely control the structure, size, and morphology of the product 2DSi and to analyze the physical and chemical properties of the product(s) and byproduct(s);

(3) To design and realize the potential internal transition of 2DSi materials, uncover the fine-tuning conditions such as reaction temperature and time, substrate, carrier gas or solvent, precursor ratio, and the type of catalyst.

Second, surface analysis and ligand engineering are crucial to 2DSi materials. The covalent nature of the silicon backbones in 2DSis differs from those of many other ionic nanomaterials. The “naked” sp²-hybridized silicon atoms on silicene and the as-formed hydride-terminated surfaces of silicane and MSNs are both prone to oxidation, which introduces trap states that may promote additional recombination pathways of charge carriers and thus impair the electronic properties. Although conventional surface modification approaches (such as hydrosilylation and Grignard reactions) have been widely used for functionalizing c-Si, differences exist when comparing the bond strength and steric hindrance of the silicon atoms on the surfaces on the bulky and nanoscaled silicon materials. Consequently, new functionalization methods that can circumvent the limitation of these existent methods are urgently needed to passivate 2DSis to gain better stability, more vigorous PL intensity, and high charge carrier mobilities.

Third, although considerable progress has been made on the understanding of the properties of 2DSis, the fabrication and optimization of 2DSi-based devices remain a significant challenge. Although a few prototype devices (e.g., transistors, sensors, LEDs) have been reported, most of them suffer from materials and device instability and relatively low performance. Additional efforts have to be made to further boost the device performance and improve the device stability under the operational conditions. Coupling the materials with polymers^[32,93] and the search for proper device architectures^[96] would be effective routes towards those goals. Meanwhile, the computational studies on the properties of new 2DSis (e.g., doping, alloy)^[112,113] would accelerate the searching of new material candidates with better optoelectronic properties in the 2DSi family.

Although 2DSi materials have been found and studied since the nineteenth century, they still behave like newcomers in the fields of silicon-based science and engineering. Therefore, future studies on the synthesis and surface modification of 2DSis are crucial because they pave the way to deepen the understanding of the properties of these reduced-dimensional materials and to boost device performance.