On the 2H- to 3C-Type Transformation and Growth Mechanism of SiC Nanowires upon Carbothermal Reduction of Rice Straws.

SiC nanowires (NWs) and nanoparticles (NPs) fabricated by carbothermal reduction of rice straws with/without FeSi catalysts were characterized by transmission electron microscopy to study the catalyst-facilitated vapor-liquid-solid (VLS) growth against the oriented attachment of the crystals, which underwent 2H- to 3C-type transformation. The cotectic melt of the FeSi catalyst in the Fe-Si-C-O system turned out to promote the VLS growth to form straight and occasionally tapered NWs in contrast to the zigzag ones via the (hkl)-specific coalescence of the faceted NPs. The SiC NWs showed [0001]2H-directed growth more or less stacked with {111}3C interlayers following the optimum crystallographic relationship (0001)2H//{111̅}3C; [21̅1̅0]2H//⟨101⟩3C with zigzag {111}3C lateral steps and polysynthetic twins/faults near the (0001)2H/(111)3C interface. The FeSi-assisted VLS growth and twinning/stacking fault-coupled 2H to 3C phase change may be extended to novel green manufacturing and design of sustainable resources for other semiconductor NWs.

Introduction

The silicon carbide nanowire (SiC NW) is a promising material in various applications including light-emitting diodes,[1] functional ceramics,[2,3] and composite reinforcement[3,4] due to the exceptional properties of wide band gap, chemical inertness, good thermal conductivity, high Yang’s modulus and hardness,[5] and high temperature strength.[2,3] All of the excellent physical and chemical properties of SiC NWs depend on the grain size and the crystal structure, inducing the research communities to still focus on the different synthetic routes for the nanostructured SiC.[2,3,6−11]

SiC NWs and whiskers have been synthesized by chemical vapor deposition,[12] thermal pyrolysis of organic precursors,[13] electrospinning,[14] and reaction of silicon- and carbon-containing compounds.[15] The formation mechanisms behind the growth of SiC NWs through vapor–liquid–solid (VLS) and vapor–solid (VS) methods have also drawn much attention.[3,16] However, the methods with a long processing time and costly raw materials lead to high energy consumption and are not environmentally friendly.

Recently, natural resources have received much attention since they are environmentally friendly, less harmful, and low cost for the green synthesis of nanomaterials. For example, SiC particulates and nanoparticles (NPs) could be synthesized from rice husks (RHs) by carbothermal[17,18] and magesiothermic reduction,[19−24] respectively. The SiC NWs and whiskers could also be produced from electronic waste and coir fibers coupled with silicon slurry waste using carbothermal reduction[25] and the spark plasma process.[26]

Rice straw (RS), a major agricultural waste product, is a byproduct of rice production at harvest. The ratio of straw to paddy ranges from 0.7 to 1.4 depending on the variety and growth. Globally, roughly 800–1000 million tons per year of rice straw is produced, with about 600–800 million tons per year produced in Asia.[27] However, most RSs are generally disposed of by burning or land filling, resulting in waste of energy, air pollution, and greenhouse gas emissions.

RSs mainly contain lignin, cellulose, hemicellulose, and hydrated silica, and the content of the hydrated silica is about 10–20 wt % depending on the species, origin, climate, and geographic location.[28]

Since the content of hydrated silica is less in RHs (15–28 wt %),[29] the research on extracting silicon and synthetic silicon oxide/carbide/nitride has always focused on RHs rather than on RSs. However, the huge volume percentage of RSs rather than that of RHs is the main byproduct of rice production. According to the International Rice Research Institute (IRRI), the amount of straw is roughly 0.7–1.4 kg for every kg of milled rice.[27] Therefore, the research on RSs should be a crucial issue for ecofriendly and recycling economics. Here, we report the VLS synthetic process of 3C-type SiC nanowires from RSs by carbothermal reduction. Not only three possible formation mechanisms of SiC NWs are proposed, but also the 2H to 3C phase transformation is also addressed. Additionally, the FeSi-assisted VLS growth by the addition of FeCl3 with the high yield rate and the high NW/NP ratio is extensively investigated and discussed.

Results

The color changes from the starter RSs to the NW and NP products are shown in the photos shown in Figure . After the carbonization process at 600 °C for 3h, the khaki-colored RSs transformed to the black CRSs. Then, greenish SiC NWs occurred on the top of the products in the crucible after 3C-type SiC growth at 1400 °C for 4h. Note that xylene was used to separate the hydrophilic SiC NWs and the hydrophobic SiC NPs. The high-quality yellow-green SiC NPs were obtained by subsequent thermal treatment at 600 °C for 12 h to remove the residual carbonized relatives.

Figure 1

Flow chart with photos of the SiC NW and NP production by carbothermal reduction.

The Raman spectra provided in Figure a show the NWs produced by carbothermal reduction from 1100 to 1400 °C with a 400 sccm Ar flow rate compared to the commercial SiC NPs. The Raman bands for 3C-type SiC are 790 cm–1 (transverse optical mode, TO) and 968 cm–1 (longitudinal optical mode, LO), and their second-order bands are identified by the commercial SiC NPs. Only the graphitical D and G bands (1410 and 1590 cm–1) and amorphous silica (300–500 cm–1) occur when the NWs were produced at 1100 and 1200 °C; then, the obvious 790 cm–1 (TO) and 968 cm–1 (LO) Raman bands arose when the NWs were produced at 1300 and 1400 °C. In fact, the sample almost completely transformed from amorphous silica and the carbon source to form 3C-type and minor 2H-type and also relic amorphous SiC upon heating to 1400 °C. The magnified Raman spectra for the NWs produced at 1300 °C provided in Figure b show a broad band with 780, 890, and 937 cm–1, in which 780 cm–1 is attributed to 796 cm–1 (TO) for 3C-type SiC and 764 cm–1 (TO) and 799 cm–1 (TO) for 2H-type SiC,[30] 890 cm–1 is attributed to the amorphous phase of SiC,[31] and 937 cm–1 is attributed to 972 cm–1 (LO) for 3C-type SiC and 964 cm–1 (LO) for 2H-type SiC.[30] Apparently, the nucleation and growth of the SiC NWs and NPs with 2H-type phases should be within the temperature range of 1200–1300 °C. Similarly, the photoluminescence spectra (Figure S3) also confirm the typical emissions for 3C-type SiC for the NWs produced at 1300 °C. Therefore, the following transmission electron microscopy (TEM) characterization was settled at 1300 °C to investigate the VLS growth and the 2H to 3C phase transformation in SiC NWs. Additionally, the Raman bands showed a significant blue shift that can be attributed to the high density of stacking faults of the SiC NWs.[32]

Figure 2

(a) Raman spectrum of the NWs produced by carbothermal reduction from 1100 to 1400 °C with a 400 sccm Ar flow rate compared to the commercial SiC NPs. (b) Magnified Raman spectra of the SiC NWs produced by carbothermal reduction at 1300 °C with a 400 sccm Ar flow rate in (a).

The X-ray diffraction (XRD) data provided in Figure a show the typical 3C-type SiC in the F4̅3m space group with (111), (200), (220), (311), and (222) diffractions and a broad band of amorphous silica from 20 to 50° after carbothermal reduction at 1400 °C with 100–600 sccm Ar flow rates. Apparently, the highest crystallinity and the lowest background are when the Ar flow rate is 400 sccm, which was settled as the standard parameter for further experiments. It should be noted that the two small peaks denoted as 2H-type SiC near the strongest (111) peaks of 3C-type SiC are attributed to the relic of 2H to 3C transformation during the growth process of 3C-type SiC, the details of which will be addressed later. Except for the strong (101̅0) and (101̅1) diffractions, there is still a weak (101̅3) diffraction for 2H-type SiC denoted by the red arrow in Figure a. The scanning electron microscopy (SEM) image in Figure b illustrates the straight SiC NWs with the nanometer-sized diameter and the ultralong length of about 50–150 μm. The SiC particulates with a size of 100 nm to 2 μm in the NW forest play the role of a self-catalyst for further growth of the NWs.

Figure 3

(a) XRD trace of the SiC NWs produced by carbothermal reduction at 1400 °C with 100–600 sccm Ar flow rates. (b) SEM image of the SiC NWs produced by carbothermal reduction at 1400 °C with a 400 sccm Ar flow rate.

The TEM bright field image (BFI) shown in Figure a illustrates the SiC NPs coalescing with each other to form a polycrystalline SiC NW that has a zigzag form, and the size of the coalesced nanoparticles is less than 20 nm. The TEM BFI and the corresponding selected area electron diffraction (SAED) shown in Figure b,c indicate that the polycrystalline SiC NW has a preferred orientation as it tilts in the [101] zone axis, and the (hkl)-specific (i.e., {111}3C, {100}3C and {110}3C) coalescence of the faceted NPs even still has extra diffraction spots due to the tiny coalescence interfaces, 2H-type NPs, and double diffractions. This could be attributed to the oriented attachment of the nanoparticles, which was intensively investigated in several kinds of nanoparticle systems.[33,34]

Figure 4

(a, b) TEM BFI of the SiC NW without/with tilting in the [101] zone axis, respectively, and (c) corresponding SAED of the SiC NW in (b).

The TEM BFI of a self-catalyzed SiC NW is illustrated in Figure a, in which the SiC catalyst on the top with a diameter of 150 nm and the body of the NW with the particulates coalesced with each other in the core, and the amorphous phase showed lower contrast in the shell. The corresponding SAED of the top catalyst is illustrated in Figure b, indicating that the 2H- and 3C-type SiC coexists in the catalyst particulate, and these two phases have a crystallographic relationship (0001)2H //{112}3C; ⟨1̅010⟩2H //⟨1̅1̅1⟩3C. Figure c,d shows the lattice image of the 2H- and 3C-type SiC area indicated as the top red and bottom black arrows in Figure a, respectively. It is still an open question how the 2H- and/or 3C-type SiC catalysts would affect the self-catalyst VLS growth. The energy dispersive X-ray (EDX) linescan of the green line AB and red line CD in Figure a shows that the top catalyst only has Si and C signals; however, the body of the NW has the Si and C signals in the core and the Si and O signals in the shell. Apparently, this SiC NW rather than the NW produced by typical VLS growth has a polycrystalline core via oriented attachment and a SiO2 shell via further carbothermal reduction.

Figure 5

(a, b) TEM BFI and the corresponding SAED of the SiC NW via self-catalyzed VLS growth; (c, d) lattice images of the top and bottom local area in the SiC catalyst indicated by red and black arrows, respectively; and (e, f) EDX linescan of the blue line AB and orange line CD in (a), respectively.

The TEM BFI reveals a self-catalyzed NW (the catalyst not shown) with a distinct neck in the middle in Figure a, in which the growth direction of the NW is from the top right to the bottom left as indicated by the black arrow. As the NW tilted to an exact zone axis in Figure b, the 2H-type SiC with an amorphous SiO2 shell developed outward to form a neck and then transformed into 3C-type SiC. Figure c,d shows the corresponding SAED of the 2H- and 3C-type SiC area shown in Figure a, indicating that these two phases have a crystallographic relationship, (0001)2H//{111̅}3C; [21̅1̅0]2H//⟨101⟩3C. As a result, the 2H-type SiC rather than the 3C-type SiC occurred first, and then, the 3C-type SiC nucleated on 2H-type SiC with the specific crystallographic relationship in self-catalyzed VLS growth. It is worth noting that the neck has a high density of stacking faults, which are confirmed by the striking along [0001]2H in Figure c, and they are attributed to the 2H- to 3C-type SiC phase transformation.

Figure 6

(a, b) TEM BFI of the SiC NW without/with tilting in the exact zone axis, respectivly, and (c, d) corresponding SAED of the 2H- and 3C-type SiC regions in (b), respectively.

The TEM BFI and the corresponding dark field image (DFI) at g = [111̅] in Figure a,b illustrate the polysynthetic twins and stacking faults in the NW, respectively. In the DFI, the twin plane can be distinguished at the interface between the bright and dark stripes denoted by the white arrow, and the stacking fault can be distinguished by the stripe contrast in the bright or dark area denoted by the red arrow. The corresponding SAED in Figure c shows the symmetric twin spots denoted by the white arrows in the [101] zone axis. The lattice image of the NW in Figure d illustrates the mirror-symmetrical twin plane denoted by the white arrow, and the lattice discontinuous stacking fault is denoted by the red arrow. In fact, the SiC nanowires usually have a high density of stacking faults; however, uniformly and periodically twinned structures are rarely found in SiC nanowires, although they are observed in many nanowire materials.[35,36] Additionally, the formation of periodic twins in the SiC nanowires could be attributed to a minimum surface energy and strain energy argument.[36] The EDX spectra show that the chemical constitution of the NW is mainly Si and C contents, with minor O from the SiO2 shell.

Figure 7

(a, b) TEM BFI and DFI in g = (111̅) of the SiC NW with multiple stacking faults and twins indicated by red and white arrows; (c) corresponding SAED of the SiC NW in (a) with the [101] zone axis; (d, e) lattice image of the local stacking faults and twins; and (d) EDX spectrum of the whole NW.

The growth mechanisms of the SiC NWs without any additives were shown in the TEM BFI as can be seen in Figure . The oriented attachment in Figure a reveals that the SiC NPs coalesced by specific surfaces via Brownian-type motion to lower the surface energy. Therefore, the NW produced via oriented attachment usually has a zigzag form, smaller size, and random orientation compared to the NW produced via VLS growth. The VLS growth in Figure b, which promotes seeding and oriented growth by introducing a catalytic liquid alloy phase that can rapidly adsorb vapor to supersaturation levels and from which crystal growth can subsequently occur from nucleated seeds at the liquid–solid interface, has achieved the most success in producing various semiconductor nanowires. In this case, without any additives, the seed for VLS growth is the SiC itself, and the body of the NW should be single-crystalline SiC. The special case of VLS growth and oriented attachment is illustrated by TEM BFIs in Figure c, in which the NW still develops via VLS growth, but the body of the NW is polycrystalline, which could be explained by the oriented attachment of the SiC nanoparticles. Note that all the NWs are clad by the shell of SiO2.

Figure 8

Growth mechanisms of the SiC NWs produced from rice straws by carbothermal reduction illustrated by TEM BFIs: (a) orientated attachment; (b) VLS growth; and (c) VLS growth and orientated attachment.

The XRD diffraction of the SiC NWs produced by carbothermal reduction with the additive 3 wt % FeCl3 as the precursor of the FeSi catalyst at 1400 °C with a 400 sccm Ar flow rate as shown in Figure a reveals that extra FeSi (210) and (211) diffractions occur rather than the XRD without any additives in Figure a. It should be noted that 3 wt % FeCl3 is the optimum content for NW growth since the 4 and 5 wt % FeCl3 additives (Figure S4) are too much to promote the yielding rate of the NWs. In addition, the selected additive FeCl3 was transformed into the catalyst FeSi during the carbothermal reduction. The SEM image in Figure b shows the main straight SiC NWs with minor curved NWs. In fact, the additive FeCl3 as the precursor of the FeSi catalyst not only increased the production of NWs rather than that of NPs but also increased the ratio of the straight SiC NWs rather than zigzag NWs via oriented attachment (Tables and 2).

Figure 9

(a, b) XRD trace and SEM image of the SiC NWs produced by carbothermal reduction with the additive 3 wt % FeCl3 as the precursor of the catalyst at 1400 °C with a 400 sccm Ar flow rate.

Table 1

Yield rate of the total SiC products, NWs, and NPs produced by carbothermal reduction at 1400 °C with 100–600 sccm Ar flow rates

	total yield %	NWs yield %	NPs yield %
600 sccm Ar	70.28	13.64	56.64
400 sccm Ar	73.57	16.27	57.30
300 sccm Ar	71.66	14.95	56.71
200 sccm Ar	66.53	8.65	57.88
100 sccm Ar	61.46	7.42	54.01

Table 2

Yield rate of the total SiC products, NWs, and NPs produced by carbothermal reduction at 1400 °C and 400 sccm Ar flow rates with 1, 3, 4, and 5 wt % FeCl3 additives

(wt %)	total yield %	NWs yield %	NPs yield %
FeCl3, 1	77.25	14.76	62.49
FeCl3, 3	78.52	21.22	57.30
FeCl3, 4	78.25	22.13	56.12
FeCl3, 5	78.01	21.33	56.68

The TEM BFI of the SiC NW produced by FeSi-catalyzed VLS growth in Figure a illustrates both the top FeSi catalyst and the NW cladding underneath with vague contrast of the SiO2 shell. The corresponding SAEDs of the top FeSi catalyst and the NW underneath in Figure b,c show the single crystalline diffractions in the [101] and [21̅1̅0] zone axes. Apparently, the top FeSi catalyst and the NW underneath did not have the crystallographic orientation that is crucial evidence for FeSi as a catalyst rather than a nucleator for NWs via the VLS growth. The lattice image of the interface between the FeSi catalyst and the NW in Figure S5 also illustrates that the faceted FeSi particle is nonepitaxy with respect to the tip of the SiC NW due to the amorphous Fe-Si-O-C layer mediation. The lattice image of the top FeSi catalyst in Figure (d) illustrates the well-developed (111̅) and (101̅) facets in the [101] zone axis. The EDX spectra of the top catalyst (Figure e) showed the main Fe and Si contents and minor O, C, Ni, and Cr impurities, while the Ni and Cr counts should be from the TEM holder. In addition, the further EDX linescan in Figure S6 also reveals that the FeSi catalyst has a FeSi core and SiO2 shell. The elemental concentrations of the NW produced by carbothermal reduction at 1400 °C and 400 sccm Ar flow rate without/with FeCl3 additives by TEM EDX are summarized in Tables S1 and S2, respectively. The abundant carbon concentration of 57.27–75.49 atom % without/with FeCl3 is due to the carbon-coated collodion film on TEM copper grids. The little amount of oxygen concentration is 3.00–3.39 atom % from the amorphous silica shell, and it can be expected that the SiO2 shell is about 4.50–5.09 atom % of the single NW. Note that other than the FeSi catalyst, only the constitution of carbon, oxygen, and silicon is in the NW.

Figure 10

(a) TEM BFI of the NW produced by FeSi-catalyzed VLS growth; (b, c) corresponding SAED of the upper FeSi catalyst and the lower 2H-type SiC NW, respectively; and (d, e) lattice image and the EDX spectrum of the FeSi catalyst, respectively.

The TEM BFI and the corresponding SAED of the 2H-type SiC NW in Figure a,b produced by the FeSi-catalyzed VLS growth illustrate the ledge growth on the NW along the ⟨0001⟩ growth direction. The lattice images of the ledges in Figure c,d denoted by the red arrows in Figure a show the ledge growth accompanying the formation of the stacking faults.

Figure 11

(a, b) TEM BFI and the corresponding SAED of the 2H-type SiC NW produced by FeSi-catalyzed VLS growth and (c, d) lattice image of the ledge in the middle and the tip of the 2H-type SiC NW along the growth direction indicated by the red arrows in (a).

Figure 12

(a) Schematic drawing of 2H→3C phase transformation and (b, c) lattice image and the magnified image in the yellow rectangular area of the interface between 2H-type SiC and 3C-type SiC.

Discussion

Oriented Attachment Growth and the Shape of SiC NWs

The size-dependent oriented attachment of the relatively small-sized SiC nanoparticles can be rationalized by a Brownian-type rotation of the particles in terms of anchorage release (i.e., debonding) at the SiC interface analogous to the high-temperature dynamics of Y-PSZ particles in Co1–O grains and SnO2 particles in TiO2 grains via reactive sintering.[37,38] The diffusion coefficient (D) of spherical particles confined in a grain was formulated to decrease exponentially with the increase in the number of atoms in anchorage (i.e., atoms in good coherency) at the interface, whereas it increases exponentially with ΔT above a critical temperature (T0) for anchorage release[37]where πd/a is the number of atoms in anchorage (assuming crystallite as a sphere of diameter d), a is the periodic distance of an atom, ΔG0 (=ΔH0 −TΔS0) for viscous motion in terms of atom diffusion along the interface, Δhm is the latent enthalpy to untie an atom from the interface, h is the Planck constant, and other symbols have their usual meanings.

In fact, T0 and the free energy are a function of the particle size due to the nanosize effect. Therefore, both the small number of atoms in anchorage at the interface and a lower T0 could make the nanoparticle have a higher diffusivity during the Brownian-type rotation. The anchorage release for SiC nanoparticles at the homogeneous SiC/SiC interface depends on the critical temperature T0, which is a moderate homologous temperature of 0.60–0.67 Tm. As a result, the T0 should be lower than 1300 °C to activate the Brownian motion of the connected nanoparticles in the solid state. Additionally, the nanometer-sized SiC produced from the gas phase via the carbothermal reaction would promote Brownian rotation of the nanoparticles until the energetically favored epitaxial state was reached.

The (hkl)-specific coalescence of 2H- and 3C-type NPs condensed directly from vapor without FeSi catalysis in Figure a with the crystallographic relationship (0001)2H//{112}3C; ⟨1̅010⟩2H//⟨1̅1̅1⟩3C is relevant to the special boundaries induced by the coalescence of the diamond-type NPs.[39] The diamond-type particles, i.e., C-overdoped Si, 3C–Si1+C, and Si-overdoped C, formed sintered polycrystals and individual particles with well-developed ∼{111} and ∼{110} vicinal surfaces for (hkl)-specific coalescence to form a {111} twin boundary, (110) 70.5° twist boundary, and [111̅](123)/(011) tilt boundary.[39] As a result, both the phase and the chemical constitution of the Si-C system would dominate the surface energy and the energetically favored special boundaries via the (hkl)-specific coalescence.

Additionally, the periodic bond chain (PBC) model can explain the Wulff shape of a crystal, in which the shape possesses the lowest surface energy for a fixed volume and hence represents the ideal shape that the crystal would take in the absence of other constraints.[40−42] For 3C-type SiC, the only flat (F) face {111} has three PBCs, while the step (S) face {110} and the kink (K) face {100} contain one and no PBCs, respectively. Consequently, the morphology of the 3C-type SiC is expected to be an octahedron with narrow {110} facets. However, in the present case of 3C-type SiC NWs, the morphology reveals straight and even zigzag features to show {110} and {100} facets. This nanosize effect accounts for the prevailing S rather than the F face for the (hkl)-specific coalescence of 3C-type SiC, despite the VLS growth that typically leads the NW formed via vapor to the solid phase through liquidous catalysis.

Self-Catalyzed VLS Growth of SiC NWs

The possible formation mechanism of SiC NWs by carbothermal reduction has already been investigated by several researchers.[25,43−46] The SiO gas is generated as the following two reactions and R2The produced SiO vapor would react with solid C or vapor CO to form solid SiC with vapor CO or solid SiC with vapor CO2, respectively, as shown by the following two reactions and R4. It is worth noting that the ΔG0(R3) was lower than ΔG0(R4) when the temperature settled at 1300 °C. Therefore, the nuclei of SiC NPs as the catalyst for self-catalyzed VLS growth occurred when the vapor SiO deposited and reacted on the solid carbon via reaction . Meanwhile, the SiC content accumulated on the nuclei surface to form the NW as the saturated vapor CO reacted with the vapor SiO via reaction .The vapor CO2 produced by R2 and R4 would react with solid carbon again to form vapor CO via reaction . As a result, it not only keeps vapor CO in saturation but also promotes the growth of the SiC NWs.At the end of the growth process, when the temperature cools from 1400 °C to room temperature, the residue vapor SiO would deposit on the SiC NW to form an amorphous SiO2 shell. It is difficult if not impossible to remove this amorphous SiO2 shell from the SiC NW by optimum control of the SiO2 constitutions from the starting materials or increasing the percentage of the reducing gas at the end of the growth process.

The yield rate of the total SiC products, NWs, and NPs produced by carbothermal reduction at 1400 °C with 100–600 sccm Ar flow rate is shown in Table . The calculated SiC yield is based on the number of moles of SiC present in the sample (nSiC) and the number of moles of silica initially present ((nSiO2)) as expressed by the following equation[47]Note that the weight of the silica initially present in the sample can be measured by TGA of CRS beyond the temperature of 800 °C (Figure S2). The yield rate of the total SiC products is 73.57%, including the NWs of 16.27% and the NPs of 57.30% under the 400 sccm Ar flow without any additives.

FeSi-Assisted VLS Growth of SiC NWs

The additive FeCl3 is transformed into FeSi in high-temperature chemical reactions, which is confirmed by the TEM observations and is illustrated as the formation mechanisms by the following two reactions and R7.[48] The solid Si in reaction could be generated by the process of boiling in HCl (0.5 N) for 1 h to remove metal impurities. Both reactions could generate FeSi catalysts for the further VLS growth of SiC NWs.The FeSi-assisted VLS growth not only increased the ratio of NWs rather than NPs (Table ) but also promoted the long straight NWs via VLS growth rather than the zigzag NWs via oriented attachment. This can be proved by the ratio of NWs/NPs, which is 37.03% with 3 wt % FeCl3 additives compared to the ratio of NWs/NPs which is 28.39% without any additives. As the NW grows via VLS growth at the beginning, the nearby nanoparticles attracted by the coulomb force would coalesce and reorient to the energetically favored epitaxial state. Therefore, the body of the NW is polycrystalline rather than single-crystalline via typical VLS growth. It should be noted that the special VLS growth and oriented attachment are still suppressed by FeSi-assisted VLS growth.

It is important to consider the role of the cotectic FeSi melt core and the cladding oxygen/amorphous silica shell during the high-temperature VLS growth of the NW. Regarding the FeSi catalyst for the gas-phase reaction, the melting point of the FeSi catalyst in the Fe-Si-C-O system would be around 1200–1300 °C, while the typical melting point of ε-type FeSi is about 1407 °C according to the Fe-Si phase diagram.[49] Besides, a large depression of melting temperature with decreasing size is expected, as a larger fraction of the total number of atoms is on the surface.[50] Thus, the FeSi catalyst-based Fe-Si-C-O droplet would remain as a liquid phase for beneficial mass transportation from the Si-C-O-enriched vapor phases. By contrast, Fe diffusion from the liquid droplet to the solid NW was rather limited through the crystalline SiC core and amorphous Si-C-O shell when heated up to 1400 °C according to the TEM EDX spectrum. Therefore, the FeSi-based droplets have effectively directed the linear growth of the NWs, analogous to the Au-assisted VLS growth of Si NWs with/without oxygen.[51,52]

2H/3C Phase Transformation of SiC NWs

The common-occurring polymorphs of SiC are 3C, 2H, 4H, and 6H, in which 3C-type SiC has the zinc blende structure with a...ABCABC... sequence, and the other three polytypes have the hexagonal structure with...ABAB... (2H),...ABCBABCB... (4H), and...ABCACBABCACB... (6H) sequences. The synthesis and properties of the 3C-, 4H-, and 6H-type SiC polymorphs are intensively investigated; however, little research has focused on the synthesis of 2H-type SiC since the free energy between different polytypes is quite small so that the 2H-type SiC is easily transformed to other polytypes of SiC and thus is difficult to stabilize.[53,54]

In this research, we distinguished the polytypes of SiC by electron diffraction and lattice images, respectively. The 2H- and 3C-type SiC could be easily identified by SAED due to the different symmetries in the cubic and hexagonal systems, which are shown in Figure c,d. However, the 4H, 6H, 15R, and 21R types with similar symmetry and d-spacings give rise to the difficulty of identifying these polytypes of SiC by XRD and Raman scattering. The lattice parameter of the c-axis for hexagonal and rhombohedral (in the hexagonal axis) SiC is 0.503 nm for 2H (JCPDS 21-1126), 1.006 nm for 4H (JCPDS 29-1127), 1.508 nm for 6H (JCPDS 29-1131), 3.770 nm for 15R (JCPDS 39-1196), and 5.278 nm for 21R (JCPDS 89-2219). Therefore, as we tilted the crystal to the direction perpendicular to the c-axis like [21̅1̅0] or [101̅0] in TEM, (0002) diffraction occurred and its d-spacing could be taken to distinguish the polytypes of SiC. For example, the d-spacing of (0002) diffractions in Figures c, 10c, 11b is about 0.252 nm, which is half the c-axis for 2H-type SiC. Additionally, regarding the spacing of (0002)/{111} fringes in lattice images Figures c,d and 12b, it has also been confirmed that only 2H/3C-type SiC exist in this system.

The 2H to 3C transformation-induced stacking fault is illustrated by a schematic drawing and the lattice image in Figure . The stacking fault (denoted by an orange dashed line) always dwells on the {0001} plane and is bound with partial dislocations (denoted by a red upside-down T) at the edge of the NW where they form a closed dislocation loop around the NW. The stacking sequence of 2H- to 3C-type SiC on the stacking fault is ...ABABABACABCABC... and is confirmed by the experimental lattice image in Figure b and the magnified image in Figure c. The macroscopic interface plane including ledges or zigzag features could be attributed to the dominating fast surface diffusion, in which the atomic layer by layer growth is obviously seen in Figure b. As the processing temperature increased to 1400 °C, the 2H-type SiC vanished, and most of the NW had a 3C-type SiC structure accompanying the high density of stacking faults, which is a relic of the 2H to 3C phase transformation.

The individual NWs with a 2H-type hot thin tip near the molten FeSi catalyst (Figures and 10) and 2H/3C polytypes at the cool thicker root (Figures and 11) indicate capillarity-hindered yet cooling stress-induced formation of diamond-like 3C-type SiC. The 2H-type SiC nucleates first on the SiC or FeSi surface at 1200–1300 °C and then develops as a wire by atomic ledge growth, which was confirmed by TEM observations in Figure . In previous work, 2H-type SiC was formed above 1200–1250 °C, in which it was grown using carbothermal reduction, pulsed laser ablation, and hot pressed sintering.[53,55,56] As the temperature decreased, the tip of the 2H-type SiC NWs transformed to the more stable 3C-type SiC as reported before.[53,57] Simultaneously, the diameter of the tip would decrease so that the surface tension could be reduced. Apparently, the phase selection of NWs rather than NPs was mainly dominated by cooling stress but not the capillarity effect. Since the temperature decreased quickly (down to 1000 °C within 2 min), the cooling stress is high enough as a driving force for the 2H to 3C transformation. This could be confirmed by the cooling from 1300 °C with 2H/3C polytypes compared to 1400 °C with almost complete transformation to the 3C-type SiC.

Conclusions

SiC NWs produced from rice straws by carbothermal reduction with/without FeSi catalysts show that there are three kinds of growth mechanisms by oriented attachment, VLS growth, and VLS growth and oriented attachment. The NWs produced by oriented attachment with a zig-zig shape are formed by Brownian-type rotation of the nanoparticles until an energetically favored epitaxial state was reached. In contrast, the NWs produced by VLS growth with a long and straight shape are formed by the nucleation on the SiC and/or FeSi surfaces and then grow as a wire by dominated rapid surface diffusion. Such VLS growth combined with oriented attachment shows that the NW has a SiC and/or FeSi catalyst with a polycrystalline body. Actually, the yield rate of NWs/NPs is 37.03 and 28.39 wt % with/without FeSi catalysts, respectively, indicating that the FeSi-assisted VLS growth not only increased the ratio of NWs rather than NPs but also promoted the long straight NWs via VLS growth rather than the zigzag NWs via oriented attachment. As the processing temperature increases to 1200–1300 °C, the 2H/3C-type SiC NWs follow a preferred crystallographic relationship of (0001)2H//{111̅}2H; [21̅1̅0]2H//⟨101⟩3C, in which the resultant {111}2H/{0001}3C interface induced a significant stacking fault by {111}2H to {0001}3C transformation. As the processing temperature increased to 1400 °C, the 2H-type SiC vanished and most of the NW had a 3C-type SiC structure, accompanying the high density of stacking faults, which is a relic of the 2H to 3C phase transformation.

This research not only provides an effective route for the synthesis of SiC NWs but also gives an ecofriendly way to avoid the waste of energy, air pollution, and greenhouse gas emissions by traditional burning and land filling of RSs. Besides, RSs have the advantages of being among the cheapest raw materials, convenient loading and transporting, easy granulation and pretreatment, and low waste generation and energy consumption compared to other natural or artificial waste sources. Therefore, this FeSi-assisted VLS growth and stacking fault-induced 2H to 3C phase change may be extended to novel green manufacturing and design of sustainable resources for other semiconductor NWs.

Experimental Section

The powders which are ground and dried from RSs were obtained from the Green Technology Institute of the CPC Corporation, Taiwan. The RS powders were thoroughly washed several times with distilled water to remove soil and impurities until the filtrated water changed from muddy to clean. The filtrated RS powders were further boiled in HCl (0.5 N) for 1 h to remove metal impurities, were washed again several times with distilled water until the pH value approached neutral, and then were dried overnight at a temperature of 80 °C. The dried RS powders were heat-treated in a tube furnace at 600 °C for 3 h at a heating rate of 5 °C/min under an Ar atmosphere for carbonization and removal of small organic molecules.

The carbonized rice straw (CRS) powders were further ground into tiny particles and boiled in HCl (0.5 N) for 1 h to remove metal impurities and were washed several times with distilled water again until the pH value approached neutral. They were then dried overnight at a temperature of 80 °C. The dried RS powders were heat-treated in a tube furnace at 1400 °C for 4 h at a heating rate of 5 °C/min in an Ar atmosphere for 3C-type SiC crystal growth. The selected additive FeCl3 was added to the RS powders before the 1400 °C heat treatment. It is transformed into FeSi in high-temperature chemical reactions, and then, the product FeSi is used as a catalyst for the further 3C-type SiC crystal growth. It should be noted that no Si precursor such as silica or other catalysts were added in this experiment except for the selected additive FeCl3. Both the products without/with the FeCl3 additive were added to the DI water and then treated using a disperser (IKA T25 digital ULTRA-TURRAX) at 5000 rpm/min. Then, we separated the SiC NPs and NWs using xylene solution due to the hydrophilic feature of NWs and the hydrophobic feature of NPs. Finally, the separated NWs and NPs were dried overnight at a temperature of 80 °C. The organized flow chart of the SiC NW and NP production by carbothermal reduction is shown in the Supporting Information, Figure S1. Furthermore, thermogravimetric analysis (TGA) of the RSs and CRSs from room temperature to 900 °C (Figure S2) revealed that the hemicellulose, cellulose, and lignin decomposed at 258, 312, and 437 °C, respectively.

The phases and crystallinity of NWs were characterized by X-ray diffraction (Bruker D2 phaser). Scanning electron microscopy (SEM, JEOL JSM-6701F) coupled with energy dispersive X-ray (EDX) analysis was carried out to reveal the size, shape, and chemical constituents of the NWs. The NWs were also characterized by micro-Raman/photoluminescence (PL) spectroscopy (Horiba Jobin Yvon Labram HR800) under 325 or 532 nm excitation with 1 μm spatial resolution.

The NWs were collected on copper grids overlaid with a carbon-coated collodion film for composition and crystal structure/shape characterizations using transmission electron microscopy (TEM, FEI Tecnai G2 F20 and JEOL JEM-2100F Cs STEM) coupled with a bright field image (BFI), selected area electron diffraction (SAED), and point-count energy dispersive X-ray (EDX) analysis at a beam size of 1 nm. The lattice image coupled with two-dimensional (2D) Fourier transform spectroscopy of the NWs further showed the shape, defects, and interfaces due to particle impingement, VLS growth, and phase transformation.