Growth and Self-Assembly of Silicon–Silicon Carbide Nanoparticles into Hybrid Worm-Like Nanostructures at the Silicon Wafer Surface

This work describes the growth of silicon–silicon carbide nanoparticles (Si–SiC) and their self-assembly into worm-like 1D hybrid nanostructures at the interface of graphene oxide/silicon wafer (GO/Si) under Ar atmosphere at 1000 °C. Depending on GO film thickness, spread silicon nanoparticles apparently develop on GO layers, or GO-embedded Si–SiC nanoparticles self-assembled into some-micrometers-long worm-like nanowires. It was found that the nanoarrays show that carbon–silicon-based nanowires (CSNW) are standing on the Si wafer. It was assumed that Si nanoparticles originated from melted Si at the Si wafer surface and GO-induced nucleation. Additionally, a mechanism for the formation of CSNW is proposed.


Introduction
Nanostructures of organic and inorganic materials display singular physicochemical properties that depend on their chemical composition and morphology [1][2][3]. Recent advances on synthesis methods, and morphological and phase composition control [4,5] have led to the development of novel multifunctional nanomaterials for applications in medicine [6], energy conversion and storage [7], environmental remediation [8], and electronics [9].
The synthesis of the core-shell hybrid inorganic-organic nanostructures is intensively pursued because it combines both the inorganic core functionality and its stability under operation; moreover, the core itself may be designed to allow for an additional functionalization, expanding its technological applications [10,11]. In the last years, a number of graphene or graphene oxide (GO)-coated

Preparation of the CSNW
GO was prepared from natural graphite flakes by the modified Hummers' method, as previously reported by the authors [35]. A 1 × 1 cm 2 p-type silicon (Si) (100)-oriented wafer was used as the substrate. Before the GO film deposition, it was degreased by sequential sonication steps in hexane, acetone, and ethanol. GO films of different thickness were deposited by dripping~0.02 mg mL −1 or 2 mg mL −1 GO concentrated dispersions, to obtain samples named A and B, respectively. Finally, the samples were dried at 80 • C under nitrogen flux for 15 min to consolidate a film. The GO/Si samples were annealed in a quartz tube furnace at 1000 • C for 1 h under Ar atmosphere.

Raman and FT-IR Characterization
According to Raman analysis, thermal processing at 1000 • C under an Ar atmosphere produced the simultaneous formation of the silicon nanoparticles or carbon-silicon nanowires and GO reduction. Similar results regarding the reduction level and the silicon-silicon carbide nanoparticles formation were obtained for both samples, A and B. Figure 1 shows Raman spectrum (sample B) of (a) as-deposited GO and (b) one corresponding to GO-Si hybrid nanocomposite on the Si substrate. The as-deposited GO Raman spectrum exhibits the G and D bands at 1598 and 1357 cm −1 , respectively. As frequently reported, they originate from the in-plane vibrations of carbon sp 2 bonds (G band), and structural defects (D band) [36,37]. After the thermal process (Figure 1b), both bands changed in position and intensity, consistently with the GO reduction degree [38]. That is, the G and D peaks underwent a red-shift of about 13 and 10 cm −1 , respectively. The red shift of the G band suggests the reduction of graphene oxide and the recovery of the sp 2 domain, while the D band shift is associated with the size of the in-plane sp 2 domain [37,39]. Likewise, the increase in the I(D)/I(G) ratio from 0.91 to 1.21 indicates the formation of numerous aromatic domains of smaller overall size in graphene [40]. Note that Chan Lee et al. [41] reported the effective reduction of sprayed-GO films by Si.
Raman spectrum of rGO-Si nanocomposite (Figure 1b) displays the GO Raman bands and those belonging to silicon~515 and 950 cm −1 . The sharp band at 515 cm −1 could be assigned to TO vibration of Si-Si bond [42], the red shift from the assigned value of bulk silicon (~520 cm −1 ) could be attributed to the presence of nanocrystals in the nanowire [43]. The low intensity Raman band at 950 cm −1 appears to correspond to nanosized silicon, as reported by Meier et al. [44], however, the HR-TEM analysis of prepared material revealed that polycrystalline nanoparticles composed of Si and SiC were obtained.
On the other hand, the FT-IR characterization served to confirm the effective GO reduction as indicated by Raman spectroscopy. As-deposited GO (Figure 2a) shows absorption bands at 1117, 1201, 1427, 1622, and 3425 cm −1 , corresponding to C-O stretching vibrations, C-OH stretching vibrations, O-H deformation, C=C stretching (skeletal vibrations from unoxidized graphitic domains), and O-H stretching vibrations, respectively, all of them in good agreement with the wavenumber localization reported for GO prepared by the Hummers' method [45,46].
The FT-IR spectrum of GO-Si nanoparticle composite (Figure 2b) also suggests the reduction of thermally processed GO with the temperature, since all the absorption bands associated with the abovementioned oxygenated groups decreased in intensity. It was even observed that the thermal process led to the entire vanishing of some bands. In the FT-IR spectra, both of the silicon nanoparticles and of the carbon and silicon nanowires, the bands associated to Si-C or Si-O cannot be appreciated in Figure 2b, since the signals are screened, due to the core-shell structure [47]. to the presence of nanocrystals in the nanowire [43]. The low intensity Raman band at 950 cm −1 appears to correspond to nanosized silicon, as reported by Meier et al. [44], however, the HR-TEM analysis of prepared material revealed that polycrystalline nanoparticles composed of Si and SiC were obtained. On the other hand, the FT-IR characterization served to confirm the effective GO reduction as indicated by Raman spectroscopy. As-deposited GO (Figure 2a) shows absorption bands at 1117, 1201, 1427, 1622, and 3425 cm −1 , corresponding to C-O stretching vibrations, C-OH stretching vibrations, O-H deformation, C=C stretching (skeletal vibrations from unoxidized graphitic domains), and O-H stretching vibrations, respectively, all of them in good agreement with the wavenumber localization reported for GO prepared by the Hummers' method [45,46].
The FT-IR spectrum of GO-Si nanoparticle composite ( Figure 2b) also suggests the reduction of thermally processed GO with the temperature, since all the absorption bands associated with the abovementioned oxygenated groups decreased in intensity. It was even observed that the thermal process led to the entire vanishing of some bands. In the FT-IR spectra, both of the silicon nanoparticles and of the carbon and silicon nanowires, the bands associated to Si-C or Si-O cannot be appreciated in Figure 2b, since the signals are screened, due to the core-shell structure [47].   to the presence of nanocrystals in the nanowire [43]. The low intensity Raman band at 950 cm −1 appears to correspond to nanosized silicon, as reported by Meier et al. [44], however, the HR-TEM analysis of prepared material revealed that polycrystalline nanoparticles composed of Si and SiC were obtained. On the other hand, the FT-IR characterization served to confirm the effective GO reduction as indicated by Raman spectroscopy. As-deposited GO (Figure 2a) shows absorption bands at 1117, 1201, 1427, 1622, and 3425 cm −1 , corresponding to C-O stretching vibrations, C-OH stretching vibrations, O-H deformation, C=C stretching (skeletal vibrations from unoxidized graphitic domains), and O-H stretching vibrations, respectively, all of them in good agreement with the wavenumber localization reported for GO prepared by the Hummers' method [45,46].
The FT-IR spectrum of GO-Si nanoparticle composite (Figure 2b) also suggests the reduction of thermally processed GO with the temperature, since all the absorption bands associated with the abovementioned oxygenated groups decreased in intensity. It was even observed that the thermal process led to the entire vanishing of some bands. In the FT-IR spectra, both of the silicon nanoparticles and of the carbon and silicon nanowires, the bands associated to Si-C or Si-O cannot be appreciated in Figure 2b, since the signals are screened, due to the core-shell structure [47].

Morphology and Structure
After being prepared, the structure and morphology of samples A and B were examined by FE-SEM and HR-TEM. It was found that the thermal annealing led to the formation of silicon-derived nanoparticles at the GO/Si interface, and their morphology and mean particle size strongly depended on the GO film thickness. The HR-TEM analysis revealed that the nanoparticles are multiphase, comprising Si and SiC as dominant phases. The sample A (GO film thickness of around 100 nm) consists of Si-SiC nanoparticles of 20 nm mean size decorating the GO sheet surface (Figure 3a,b). On the other hand, in sample B (GO film thickness of 1 µm), Si-SiC nanoparticles (30 nm mean size) were self-assembled into worm-like hybrid nanostructures, as shown in the FE-SEM images at low-and high-magnification (Figure 3c,d, respectively). These hybrid rGO-Si-SiC worm-like nanostructures are mostly 10-100 nm in diameter with a length of several micrometers, covering a relatively large area of the Si wafer surface (note the magnitude scale of FE-SEM image in Figure 3d).
After being prepared, the structure and morphology of samples A and B were examined by FE-SEM and HR-TEM. It was found that the thermal annealing led to the formation of silicon-derived nanoparticles at the GO/Si interface, and their morphology and mean particle size strongly depended on the GO film thickness. The HR-TEM analysis revealed that the nanoparticles are multiphase, comprising Si and SiC as dominant phases. The sample A (GO film thickness of around 100 nm) consists of Si-SiC nanoparticles of 20 nm mean size decorating the GO sheet surface (Figure 3a,b). On the other hand, in sample B (GO film thickness of 1 µm), Si-SiC nanoparticles (30 nm mean size) were self-assembled into worm-like hybrid nanostructures, as shown in the FE-SEM images at low-and high-magnification (Figure 3c,d, respectively). These hybrid rGO-Si-SiC worm-like nanostructures are mostly 10-100 nm in diameter with a length of several micrometers, covering a relatively large area of the Si wafer surface (note the magnitude scale of FE-SEM image in Figure 3d).
The TEM and HRTEM images in Figure 4 display the morphological features of samples A and B. Figure 4a corroborates the FE-SEM characterization result in that Si-based nanoparticles decorate the rGO sheets in the sample A. On the other hand, Figure 4c,d displays the striking morphology of sample B, which consists of different-sized nanoparticles that self-assembled to develop the wormlike structures, as those shown in Figure 3d. TEM images, Figure 4c,d, suggest that silicon-derived nanoparticles are embedded into 1D self-assembled rGO sheets. In Figure 4d, a magnified image of the framed zone in Figure 4b is shown. It provides a more detailed view of the branched characteristics of the worm-like nanostructures, as they are formed by rGO-supported individual nanoparticles.  Figure 4b,e, respectively. It is seen that both are 30 nm diameter ball-shaped nanoparticles, apparently coated with an rGO shell. A careful determination of interplanar spacing revealed that various crystalline The TEM and HRTEM images in Figure 4 display the morphological features of samples A and B. Figure 4a corroborates the FE-SEM characterization result in that Si-based nanoparticles decorate the rGO sheets in the sample A. On the other hand, Figure 4c,d displays the striking morphology of sample B, which consists of different-sized nanoparticles that self-assembled to develop the worm-like structures, as those shown in Figure 3d. TEM images, Figure 4c,d, suggest that silicon-derived nanoparticles are embedded into 1D self-assembled rGO sheets. In Figure 4d, a magnified image of the framed zone in Figure 4b is shown. It provides a more detailed view of the branched characteristics of the worm-like nanostructures, as they are formed by rGO-supported individual nanoparticles.

The HR-TEM images of individual nanoparticles of samples A and B are shown in
The HR-TEM images of individual nanoparticles of samples A and B are shown in Figure 4b,e, respectively. It is seen that both are 30 nm diameter ball-shaped nanoparticles, apparently coated with an rGO shell. A careful determination of interplanar spacing revealed that various crystalline phases coexist in the formed nanoparticles, that is, they are multiphase polycrystalline entities; geometric shapes in Figure 4c were used to indicate the coexisting crystalline phases.

Mechanism Proposed for the CSNW Formation
The morphological characterization by FE-SEM and HR-TEM suggests that polycrystalline Si-SiC nanoparticles arise as the primary building blocks to develop more complex nanostructures. Both the mean particle size and morphology are strongly associated to the GO film thickness. Indeed, experiments carried out on silicon wafer GO-free surface were unsuccessful in growing silicon nanoparticles. After being formed, they interact, leading to a self-assembly process that produces the worm-like structures.
To explain the silicon nanoparticle formation, we have assumed that they originate from melted silicon at the silicon surface through a nucleation process that starts at the high energy edges of GO sheets. Since the Si melting point is around 1414 • C [48], and the hybrid materials were made at 1000 • C, we proposed that impurities and structural defects at the silicon wafer surface might promote the melting of silicon surface (pre-melting effect) at temperatures lower than the nominal silicon melting point. Also, it is quite probable that GO has a certain participation in the melting of silicon surface, because recent reports suggest that GO could act as a high temperature reactor [49]. Thus, we believe that the synergy between the pre-melting effect and the ability of GO to absorb IR radiation [50] promotes melting of the silicon surface. Additional support for the silicon surface pre-melting at temperatures lower than its melting point is provided by the roughness change observed in thermal oxidation at the silicon wafer surface, at temperatures of 900-1000 • C [51][52][53].
Regarding the phase composition of nanoparticles, it is known that the temperature of GO pyrolysis strongly depends on GO sheet size. It is quite probable that the combined participation of both GO pyrolysis and CO 2 emission by GO reduction supply elemental carbon to react with silicon and, in so doing, form the observed silicon carbide. The formed nanoparticles are stabilized by the rGO network.
Accordingly, we propose that the observed worm-like hybrid nanostructures develop as illustrated in Figure 5.

Mechanism Proposed for the CSNW Formation
The morphological characterization by FE-SEM and HR-TEM suggests that polycrystalline Si-SiC nanoparticles arise as the primary building blocks to develop more complex nanostructures. Both the mean particle size and morphology are strongly associated to the GO film thickness. Indeed, experiments carried out on silicon wafer GO-free surface were unsuccessful in growing silicon nanoparticles. After being formed, they interact, leading to a self-assembly process that produces the worm-like structures.
To explain the silicon nanoparticle formation, we have assumed that they originate from melted silicon at the silicon surface through a nucleation process that starts at the high energy edges of GO sheets. Since the Si melting point is around 1414 °C [48], and the hybrid materials were made at 1000 °C, we proposed that impurities and structural defects at the silicon wafer surface might promote the melting of silicon surface (pre-melting effect) at temperatures lower than the nominal silicon melting point. Also, it is quite probable that GO has a certain participation in the melting of silicon surface, because recent reports suggest that GO could act as a high temperature reactor [49]. Thus, we believe that the synergy between the pre-melting effect and the ability of GO to absorb IR radiation [50] promotes melting of the silicon surface. Additional support for the silicon surface pre-melting at temperatures lower than its melting point is provided by the roughness change observed in thermal oxidation at the silicon wafer surface, at temperatures of 900-1000 °C [51][52][53].
Regarding the phase composition of nanoparticles, it is known that the temperature of GO pyrolysis strongly depends on GO sheet size. It is quite probable that the combined participation of both GO pyrolysis and CO2 emission by GO reduction supply elemental carbon to react with silicon and, in so doing, form the observed silicon carbide. The formed nanoparticles are stabilized by the rGO network.
Accordingly, we propose that the observed worm-like hybrid nanostructures develop as illustrated in Figure 5.  At first, small Si-SiC nanoparticles develop (Figure 5a), then, they diffuse across the melted silicon surface to interact and, finally, they coalesce to form larger nanostructures (Figure 5b). The worm-like structure originates from the GO-promoted self-assembly of individual Si, SiC, and Si-SiC polycrystalline nanoparticles, with the latter formed by the coalescence among Si and SiC (see Figure 5c), as suggested by HR-TEM (Figure 4). The FE-SEM images (Figure 5d-f) were added to support the proposed mechanism for the nanoparticle self-assembly.

Conclusions
In summary, we have presented the development of silicon-silicon carbide (Si-SiC) nanoparticles at the graphene oxide/silicon interface. Depending on the GO film thickness, the obtained nanoparticles disperse on the silicon surface, or they self-assemble into a few-micrometer 1D worm-like hybrid nanostructure. It was demonstrated that dispersed silicon nanoparticles develop on graphene oxide layers, whilst 1D nanostructures comprise GO-embedded self-assembled Si-SiC nanoparticles. It was assumed that the silicon surface melts, and that Si and SiC nanoparticles originated through a graphene oxide-induced nucleation of melted silicon. In addition, a mechanism for the formation of the (CSNW) was proposed.