3C-SiC-induced peak emission intensity in photoluminescence spectrum of SiC/SiO₂ core–shell nanowires using first-principles calculations

Since the use of one-dimensional (1D) silicon carbide (SiC) nanowires (NWs) in nanotechnology devices, SiC has attracted considerable interest in research.^1–5 The nanostructure of SiC has outstanding properties, such as high thermal conductivity, chemical stability, and mechanical strength, fast electron mobility, a wide bandgap, and cytocompatibility, which can be applied to electronic and optoelectronic nanoscale devices.^3–5 Due to the quantum confinement effect, the bandgap of the nanowire can be adjusted by controlling its diameter.⁶ SiC has stable luminescence and excellent chemical and thermal stability,^1,2 and it can be used in displays and LED components in harsh environments.² Bulk SiC has low luminous efficiency because of its indirect bandgap, but this problem can be solved by reducing its size to a few nanometers or tens of nanometers.^7,8 Many characteristics of SiCNWs are highly correlated with the small size of the crystallite, because of which it is important to examine their quantum effects. Rurali investigated the electronic properties of pure and hydrogen-passivated 3C-SiCNWs grown along the [110] direction. Pure NWs exhibited a metallic character because of surface reconstructions, and quantum confinement induces a gap-broadening effect in hydrogenated NWs.⁹ Oliveira et al. performed first-principles calculations to investigate the effects of different diameters on the mechanical and electronic properties of hydrogen-passivated [111]-oriented 3C- and [0001]-oriented 2H-, 4H-, and 6H-SiCNWs, and they found that homo-structures of the 3C/4H and 3C/6H SiCNWs presented a type-I band alignment.⁶ Yan et al. reported that small SiCNWs have a pseudo-direct bandgap that decreases monotonically as their diameter increases and gradually approaches that of bulk β-SiC (2.4 eV).¹⁰ Fu and Wang¹¹ and Wang et al.¹² claimed that quantum confinement is the source of the strong light, and Yu et al.¹³ indicated that it originates in nanosegments featuring a stacking fault. Zhang et al. synthesized 3C-SiC nanorods with a diameter larger than 100 nm, where the peak appeared at 378 nm in the photoluminescence (PL) spectrum. Due to the existence of stacking faults within 3C-SiC, the defects have different structures, such as 2H-, 4H-, and 6H-like nanoscale layers. In addition, 6H-SiC has been considered to be the source of the strongest blue-shift emission owing to a higher energy gap than 3C-SiC.¹⁴ 

Certain types of stacking faults in 4H- and 6H-SiC can create very clear quantum well-like structures.^15,16 Consider a 3C-like nano-scaled layer; there is a split-off band below the conduction band minimum (CBM), which is strongly localized around the stacking fault. Even though being strongly localized, it does not affect the electron dispersion parallel to the stacking fault plane. There are a large number of stacking faults in 3C-SiC, with 2H-- and 4H-like structures within the defects. Due to the offset in the conduction band between structures, the electrons are confined to the local lower conduction band near the thin cubic region. It eventually forms a quantum well and confines the electrons to layers of the stacking fault. These electrons bound around the stacking faults can move freely along the wells and can be considered to be two-dimensional (2D) electron gases.¹⁶ Therefore, each nanolayer of the stacking fault may act as a nanosegment of SiC polytypism.

This study constructs microstructures of SiCNWs based on their transmission electron microscope (TEM) micrographs and identifies the source of the strongest peak in the PL spectrum through simulations. The results of our calculations of the NWs show that the bandgap of 3C-SiCNWs with a diameter of 1.42 nm was 3.427 eV and this structure was related to the strongest emission. The interfacial calculations indicate that the charge will be transferred from the stacking faults (SFs) to the inside of the nano-segment as the energy increases.

Chemical vapor deposition (CVD) and TEM sample preparation were carried out according to work by Chen et al.¹⁷ The microstructure and composition of the samples were investigated using a transmission electron microscope (TEM) (JEOL ARM-200FTH) and a Cold Field Emission Gun (CFEG). The resolution of the information limit of the TEM was 0.10 nm.

Based on the experimental observations, the SiCNWs, based on crystal structure of 3C along the [111] direction, and 2H, 4H, and 6H along the [0001] direction, were constructed for the simulations. The diameters of the 3C-, 2H-, 4H-, and 6H-SiCNWs were 1.42, 9.40, 1.07, and 1.07 nm, respectively. Surface hydrogenation was required to deactivate the dangling bonds of Si and C.⁹ A periodic boundary was used such that the separation in the x–y plane between the NWs was greater than 1 nm. Figure 1(a) presents the ∼1.5 nm 3C, ∼1 nm 2H, ∼1 nm 4H, and ∼1.5 nm 6H structures of the SiCNWs for samples from CVD.¹⁷ We enlarged the frame of dashed yellow lines in Fig. 1(a) to obtain Figs. 1(b)–1(e).

Using the TEM, 2H/3C, and 2H/4H/3C interfaces were constructed. Figure 1(a) presents the stacking sequences, where the green and red boxes represent 2H/4H/3C and 2H/3C, respectively. The structures were constructed by two or three bulk SiC samples with different polytypisms; in other words, there was no limit to periodicity (or the periodic boundary).

Our calculations, based on the Heyd–Scuseria–Ernzerhof screened Coulomb hybrid density functional (HSE06),¹⁸ were performed using the code from the Vienna Ab Initio Simulation Package (VASP)^19,20 by using the projector augmented wave²¹ method. The exchange correlation interaction followed the generalized gradient approximations (GGAs) with the Perdew-Burke-Ernzerhof (PBE) functional.²² The conjugate gradient (CG) algorithm^23,24 was used to relax the ions into their instantaneous ground state, and all atomic positions were allowed to relax. The optimizations converged when the change in the total (free) energy was lower than 10⁻³ eV. The Brillouin zone (BZ) integral was based on the Gaussian smearing method,^25,26 where the width of the smearing was set to 0.05 eV, and a Γ-centered 1 × 1 × 5 k-mesh and a plane-wave cutoff energy of 500 eV were used.

The stacking sequences of 3C, 2H, 4H, and 6H were ABC, AB, ABAC, and ABCACB, respectively. The atomic arrangements were constructed by Visualization for Electronic and Structural Analysis (VESTA)^27,28 to examine the microstructures shown in Figs. 1(b)–1(e). If the proposed arrangements matched the atomic positions shown in Figs. 1(b)–1(e), the local structures of the NWs could be determined. According to the description in Fig. 1, the structures of the SiCNWs were constructed as shown in Fig. 2 for the simulations. The [111]-oriented 3C- and [0001]-oriented 2H-, 4H-, and 6H-SiCNWs were considered in the calculations. The 2H/4H/3C and 2H/3C interfacial structures were constructed from the green and red boxes in Fig. 1(a), as shown in Figs. 3(a) and 3(b), respectively. The NWs were calculated to reflect the quantum confinement effect, and the interfaces were permitted electronic transfers.

Density functional theory (DFT)^29,30 underestimates the bandgap of semiconductors and insulators. Therefore, our DFT calculations underestimated these values by about 30%. Thus, hybrid functional theory was used to avoid the semi-local problem.³¹ Figure 4 shows the band structure of 3C-, 2H-, 4H-, and 6H-SiC, for which the bandgaps were obtained by subtracting CBM from valence band maximum (VBM), and had values of 3.427, 4.475, 4.496, and 4.267 eV, respectively. The VBM and CBM of bulk SiC are located at the Γ-point and M-point, respectively. Since the [0001]-oriented NWs deconstruct the periodicity in the x and y directions, the K, M, and L points will disappear, and these disappeared points will be folded onto the NW axis along the plane parallel to the [0001] direction. For instance, K and M point will be folded onto the Γ-point as these three points are located at the same plane. Therefore, the Γ-point contains the information of K and M points in NW calculation. The bulk SiC had an indirect bandgap that became a direct bandgap when the structure was transformed into NWs (similar to that in the transformation of graphene into carbon nanotubes³² and silicon nanowires³³). In the 3C-SiC, CBM lay on the six equivalent X points within the primitive BZ. For NWs with the [111] orientation, it is clear that the bulk X and Γ points were on the plane perpendicular to the axis of the NWs.^33,34

Figure 4 also presents different CBMs of the 3C-, 2H-, 4H-, and 6H-SiCNWs that show similar quantum wells that hinder the normal carrier transport through stacking faults.¹⁶ 

Figures 5(b)–5(d) and 6(b)–6(d) show spheroid-like structures with different colors besides the atoms representing the wavefunctions of the electrons. The wave function shows the probability distribution of the electrons in space. The different colors represent the signs of wavefunctions describing its phase/direction. The atomic orbital along the positive axis had a positive sign, and that along the negative axial direction had a negative sign. Figures 5(a)–5(d) and 6(a)–6(d) show that as the energy was low (VBM), the electrons were present at the 2H/3C interface, but when the electrons transferred from VBM to CBM, they gathered inside the 3C. If the energy was higher, as in the exciton of the PL spectrometer (this event did not occur in the experiment on the PL spectrum), the electrons tended to move into 4H. When their energy was low (VBM), they formed a typical 2p orbital. As the electron transitioned to the CBM, its probability distribution formed a π orbital. If the exciton energy was high and the electron transitioned to a high-energy state, the s and p electrons formed an sp orbital. The bandgap in this case was 2.4 eV as the calculated interfacial structure was based on bulk SiC, which is different from NWs, and there was no quantum confinement effect.

The calculated VBM energy of the 2H/3C interface was lower than that of the bulk 3C, because of which the electrons tended to accumulate at the 2H/3C interface. When they were transported to CBM, the electrons were highly likely to be inside 3C. In other words, the transition from VBM to CBM was one where the electrons moved from the 2H/3C interface to 3C. The PL can be described as the condition in which an electron in an excited state with high energy (CBM) returned to a lower-energy state (VBM). According to the calculations of the interfaces, electrons are transferred from the 3C-like nanosegment back to the interface during the photon emission process of the two interfaces. Since the CBM of 3C–SiC is lower than that of 2H-, 4H- and 6H-SiC, electrons will be gathered inside the 3C. It can be seen that 3C plays an important role in the light-emitting process.

The highest peak in the PL spectrum (after decomposition) was at 392 nm (or 3.163 eV), as shown in Fig. 7, which is highly possible from 3C-like nanoscale layers within stacking faults from our simulations based on Fig. 1 of the TEM micrograph and Fig. 2. The bandgaps of bulk 2H-SiC (3.5 eV³⁵) and 4H-SiC (3.3 eV³⁵) were much higher than 3.163 eV. However, that of bulk 6H-SiC (3 eV³⁵) was very close to 3.163 eV. As shown in Fig. 1, 2H-, 4H-, and 6H-SiC were NWs with stacking faults, and the difference between their bandgaps and those of the bulk increased due to the quantum confinement effect.¹⁴ Therefore, the peak at 392 nm was highly likely to have been caused by 3C-like nanoscale layers, as shown in Fig. 7. The strongest experimental bandgap of nanoscale SiC layers was obtained through the decomposition of the PL spectrum and was about 3.163 eV (392 nm), as shown in Fig. 7. This is close to the value of the 1.42 nm 3C-SiCNWs. The bandgaps of the bulk 2H and 4H exceeded 3.163 eV, where the strong light was not emitted by 2H, 4H, or 6H in the simulations but was more likely to have originated in 3C-like nanoscale layers. Therefore, the case of 3C must be considered.

Figure 7 shows that the peak corresponding to the highest intensity of luminescence was obtained at 397 nm. No peak was observed in the bulk SiC in the range of 350–450 nm due to the existence of the NWs. The peaks of decomposition were observed at 392, 418, 445, and 496 nm, with sources of SiCNWs (392 nm), SiO₂ (418 nm),^36,37 O-vacancy (445 nm),^37,38 and bulk 3C-SiC (496 nm).¹⁴ 

The quantum confinement effect was observed when the diameter of the sample satisfied the relation d < 2R_Bohr,³⁹ where d is the diameter of NWs and R_Bohr is the Bohr radius of materials. The radius of 3C-SiC was 2.7 nm, as reported by Wu et al.⁴⁰ Hence, the quantum confinement effect was observed when the diameters of the nanowires were smaller than 5.4 nm. The diameter of the synthesized NWs of the SiC was about 40 nm each, and therefore, the peak at 496 nm had a low intensity of luminescence due to the indirect bandgap, which indicates that it originated from the bulk 3C-SiC, as shown in Fig. 7.

Because stacked fault nanolayers may act as nanosegments of SiC polytypism,¹⁶ which fulfills the quantum confinement condition, NWs in this study were formed by confining two directions and the related periodicities of bulk materials to modify the Brillouin zone (BZ) of the bulk material such that it was not 3D.³³ The BZ folding,⁴¹ due to the reduction in the number of dimensions of the system, led to the transition from an indirect (bulk) bandgap to a direct bandgap (NW).^32–34 Therefore, the NWs within the stacking faults could have created a peak with a high intensity of emission at 392 nm in the photoluminescent spectrum. Based on data in Fig. 7, the bandgap of 3C-SiCNWs with a diameter of 1.42 nm reached an intensity close to 3.163 eV (392 nm) in simulations, which suggests that 3C-like nanoscale layers are the source of the PL peak at a wavelength of 392 nm.

In this study, SiCNWs were created by chemical vapor deposition, and their TEM micrographs clearly presented structures of ∼1.5 nm 3C-SiC, ∼1 nm 2H-SiC, ∼1 nm 4H-SiC, and ∼1.5 nm 6H-SiC. First-principles calculations were used to study the bandgaps of each structure along the [111] orientation for the 3C- [0001], 2H-, 4H-, and 6H-SiCNWs to identify the sources of light of the peaks of decomposition in the PL spectrum shown in Fig. 7. When electrons were in a low-energy state, they gathered at the 2H/3C interface and moved to 3C with the transition from the valence band to the conduction band.

The bulk SiC had an indirect bandgap, but the SiCNWs had a direct bandgap based on Brillouin zone-folding, which created light with a high intensity of emission. The sources of the strong blue-shifted light were 6H-SiC-like nanoscale layers, as reported by Zhang et al.¹⁴ However, explanations in Fig. 7 based on the experiments and simulations show that the bandgap of the 3C-SiCNWs was close to 3.163 eV (392 nm), and the bandgap of 2H-, 4H-, and 6H-SiCNWs had errors higher than 30%. Therefore, the high blue-shifted peak at 392 nm had occurred due to the 3C-like nanoscale layers.

This work was financially supported by the Ministry of Science and Technology of Taiwan (Grant No. MOST-108WFA0410191).

The data that support the findings of this study are available from the corresponding author upon reasonable request.