Group-IV-semiconductor quantum-dots in thermal SiO2 layer fabricated by hot-ion implantation technique: different wavelength photon emissions

Tomohisa Mizuno; Rikito Kanazawa; Kazuhiro Yamamoto; Kohki Murakawa; Kazuma Yoshimizu; Midori Tanaka; Takashi Aoki; Toshiyuki Sameshima

doi:10.35848/1347-4065/abdb80

1. Introduction

SiC semiconductors have been widely studied to evaluate the quantum phenomena of low-dimensional SiC structures as well as for realizing SiC power devices.¹⁾ Although three-dimensional SiC is an indirect-bandgap structure, SiC can emit photoluminescent (PL) photons which are attributable to the free exciton recombination of electrons excited by photons in SiC.^2–4) The peak-PL photon energy (E_PH) of SiC is equal to the exciton energy gap (E_GX), which is approximately 0.1 eV lower than the bandgap energy (E_G).^2–4) Moreover, there are many diverse polytypes in SiC for which the physical properties including E_GX, strongly depend on the polytype,^2,3) so it is possible that the peak-PL wavelength (λ_PL) of SiC photonic devices can be controlled by the polytype. Because the E_G of SiC also depends on the diameter of SiC,²⁾ SiC nanostructures,²⁾—such as a porous-SiC,^2,5–7) 2D-SiC,^2,8,9) SiC-nanowires,^2,10,11) and SiC-dots,^2,12,13)—are also candidates not only for materials science, including quantum effects, but also for photonic devices with various emission wavelengths.

Using the self-clustering effects of C atoms in a Si layer via hot-C⁺ ion implantation into Si, which was evaluated by atom probe tomography (ATP),^14–16) SiC nano-dots (dot-diameter Φ ≈ 2 nm) can be easily formed in various Si crystal structures from amorphous (a-Si) to crystal Si (c-Si) by a hot-C⁺-ion implantation technique performed in the wide ranges of Si-substrate temperature T and C⁺ ion-dose D_C, that is, 500 ≤ T ≤ 1000 °C and 5 × 10¹² ≤ D_C ≤ 7 × 10¹⁶ cm⁻²,^16–21) to evaluate the quantum mechanical phenomena in SiC-dots as well as to realize Si-based photonic devices.^22–24) The self-clustering effects of ion-implanted C atoms in Si leads to the local condensation of C-atoms with the diameter of several nm in Si layer, resulting in the local formation of SiC nano-dots in Si layers.^16,19,21) The hot-C⁺-ion implantation process can reduce the ion-implantation-induced damage to the Si layer, which is one of the advantageous characteristics of the hot-C⁺-ion implantation process.¹⁸⁾ Moreover, the partial formation of SiC-polytypes with different E_GX values for cubic- (3C-SiC) and hexagonal-SiC (H-SiC) nano-dots were also confirmed both at the oxide/Si interface and in the Si layer, using corrector-spherical aberration transmission electron microscopy (CSTEM), high-angle annular-dark-field (HAADF) scanning transmission electron microscopy (STEM), and electron diffraction (ED) patterns that were obtained by fast Fourier transform (FFT) analysis of the lattice spots of CSTEM data.¹⁶⁾ As a result, we demonstrated a broad photoluminescence (PL) spectrum with very strong emission intensity (I_PL) from the visible to the near-UV regions (>400 nm) even from indirect-bandgap SiC-dots, for which the I_PL is two orders of magnitude larger than that of 2D-Si.^25–26)

Because the SiC dots that have a larger E_G (>2.4 eV) embed in the Si layer that has a smaller E_G (≈1.1 eV),²⁷⁾ the SiC-dots in the Si layer are not quantum-dot (QD), resulting in a very small PL quantum efficiency for visible Si-based photonic devices. Thus, we experimentally realized SiC-QDs embedded in SiO₂ with a large E_G of 9 eV,²⁷⁾ using the simple processes of implanting double hot-Si⁺/C⁺-ion into a SiO₂ layer and post N₂ annealing at 1000 °C.²⁸⁾ HAADF-STEM observation showed that the SiC-QD diameter and density were approximately 2 nm and 1.5 × 10¹² cm⁻², respectively, and the clear lattice spots of some SiC-QDs were also verified by CSTEM.²⁸⁾ Moreover, after N₂ annealing, the PL intensity of SiC-QDs rapidly increased, and as a result, we successfully confirmed that the PL quantum efficiency of SiC-QDs was approximately 2.5 times greater than that of SiC-dots in the Si layer because of the increased life time of excited electrons, which are quantum mechanically confined in SiC-QDs.²⁸⁾ Thus, the post N₂ annealing is also a key process for forming SiC-QDs. In addition, to realize different wavelength photonic devices from IR to UV ranges, QD structures with various E_G values, such as Si-and C-QDs as well as SiC-QDs, are also needed. Si-^29,30) and C-QDs³¹⁾ have been widely studied, but they have not yet been realized via the easy and simple processes of hot-ion implantation and the post N₂ annealing techniques.

In this work, we experimentally studied the group-IV-semiconductor QDs (IV-QDs) embedded in the SiO₂ layer—Si-, SiC-, and C-QDs—using very simple processes of both hot-ion implantation into the SiO₂ layer and post N₂ annealing.³²⁾ Many QD formations of Si- and C-atoms in addition to SiC in SiO₂ layer were successfully confirmed by HAADF-STEM, and some IV-QDs also showed clear lattice spots when observed by CSTEM. We successfully demonstrated very strong PL emissions with different peak photon energies E_PH values from Si-QDs (near-IR) fabricated by Si⁺ hot-implantation, SiC-QDs (visible range) fabricated by double Si⁺/C⁺ hot-implantation,²⁸⁾ and C-QDs (near-UV) fabricated by C⁺ hot-implantation.

2. Experimental procedures

Using the simple fabrication processes of the hot-ion implantation into a thermal surface-SiO₂ layer (SOX) before post N₂ annealing,²⁸⁾ we realized three types of IV-QDs (Si-, SiC-, and C-QDs) in a SiO₂ layer. Figure 1 shows the fabrication steps for IV-QDs, and Table I shows the process conditions for hot-ion implantation and temperature T for each IV-QD. Figure 1(b) shows that IV-QDs were fabricated by hot-ion implantation conditions shown in Table I into the 140 nm thick SiO₂ layer (SOX) on the (100) bulk-Si substrate at T, after the SOX was formed via the dry O₂ oxidation of (100)-Si, as shown in Fig. 1(a). Table I shows that Si-, SiC- and C-QDs were fabricated by single-Si⁺, double-Si⁺/C⁺, and single-C⁺ hot-ion implantations, respectively, where 200 ≤ T ≤ 900 °C, and the hot-ion dose conditions of Si⁺ (D_S) and C⁺-ion doses (D_C) were varied from 4 × 10¹⁶ to 1 × 10¹⁷ cm⁻² with the ion projection range of the middle of SiO₂ layer to realize a higher PL intensity of IV-QDs. The acceleration energies of the Si⁺ and C⁺ ions were 60 and 25 keV, respectively, for which the projection range was the middle of the SiO₂ layer. For SiC-QDs, the D_S /D_C ratio dependence of the PL properties was also studied, although our previous study²⁸⁾ showed that the optimum D_C condition for realizing higher I_PL of SiC-QDs was 4 × 10¹⁶ cm⁻². Figure 1(c) shows that the post N₂ annealing was carried out at an annealing temperature of T_N = 1000 °C for various annealing times t_N (0 ≤ t_N ≤ 120 min) to recover the crystal quality of the IV-QDs.

**Fig. 1.** (Color online) IV-QD fabrication steps via hot-ion-implantation into SiO₂ layer. After (a) dry-oxidation process of bulk-Si substrate at 1000 °C (oxide thickness T_OX = 140 nm), (b) hot-ions were implanted into SiO₂ layer at T. (c) Post N₂ annealing was carried out at T_N of 1000 °C for t_N. Process conditions for (b) are shown in Table I.
Download figure:
Standard image High-resolution image

Table I. Process conditions of hot-ion temperature T and ion doses of Si (D_S) and C (D_C) for Si-, SiC- and C-QDs.

QD-type	Hot-ions	D_S (×10¹⁶ cm⁻²)	D_C (×10¹⁶ cm⁻²)	T (°C)
Si	Si⁺	6–10		600
SiC	Si⁺/C⁺	4–8	4	200–900
C	C⁺		4–10	400–600

The PL and Raman properties of IV-QDs were measured at room temperature, at which the excitation laser energy, power, and diameter were 3.8 eV, 0.6 mW, and 1 μm, respectively. The broad-wavelength (λ_PL) PL spectrum from the near-UV to near-IR regions was calibrated using a standard illuminant.

3. Results and discussion

3.1. Material structures of IV-QDs

The depth profiles of implanted Si at D_S = 6 × 10¹⁶ cm⁻² and C-atom concentrations at D_C = 4 × 10¹⁶ cm⁻² in IV-QDs, after N₂ annealing, were evaluated by the Si2p and C1s spectra of X-ray photoelectron spectroscopy (XPS), respectively, as shown in Fig. 2(a). The XPS accuracies for the content and depth positions were ±1 at% and ±2 nm, respectively, and the X-ray beam diameter was 100 μm.²¹⁾ The peak Si- and C-concentrations in the SiO₂ layer were approximately 6 × 10²¹ and 4 × 10²¹ cm⁻³, respectively. Because the Si- and C-peak concentrations are proportional to D_S and D_C, the maximum Si at D_S = 1 × 10¹⁷ cm⁻² and C concentrations at D_C = 1 × 10¹⁷ cm⁻² in this work were estimated to be the same as 1 × 10²² cm⁻³ in SiO₂. Moreover, Fig. 2(b) shows the depth profiles of the C-contents of the Si−C and C–C bonds in SiC-QDs at T = 200 °C (solid line) and 600 °C (dashed line), and the C-contents are nearly independent of T despite a changing 400 °C of T. Thus, Fig. 2(b) suggests that the effect of T on the C-contents of Si−C and C–C bonds is very small within a large area of 100 μm in diameter, which is similar to the results for the SiC-dots in the Si layer.²¹⁾ However, the influence of T on the SiC-QD properties will be discussed in Fig. 5(c).

**Fig. 2.** (Color online) (a) Concentration depth-profiles of implanted Si (solid line) at D_S = 6 × 10¹⁶ cm⁻² or C (dashed line) atoms at D_C = 4 × 10¹⁶ cm⁻² in SiO₂ layer, which was evaluated by Si2p and C1s spectra of XPS, respectively, where T = 600 °C, T_N = 1000 °C, and t_N = 30 min. (b) T dependence of C-content depth profiles of Si–C (blue lines) and C–C bonds (red lines) in double Si⁺/C⁺ ion implanted SiO₂ layer at t_N = 30 min, where solid and dashed lines show the data at T = 200 °C and 600 °C, respectively. Concentration accuracy evaluated by XPS was estimated to be ±1 at%, depth error bar was approximately ±2 nm, and X-ray beam diameter of XPS was 100 μm. Figure 2(a) shows that the peak concentrations of Si and C atoms were approximately 6 × 10²¹ and 4 × 10²¹ cm⁻³, respectively, in the middle of SiO₂ layer. Figure 2(b) shows that the C-contents of Si–C and C–C bonds are almost independent of T.
Download figure:
Standard image High-resolution image

Next, we discuss the IV-QD structures as evaluated using electron microscopes. We experimentally confirmed many Si- and C-QDs, in addition to SiC-QDs, using HAADF-STEM observations, as shown in Figs. 3(a)−3(d). Figures 3(a) and 3(b) shows the whole SOX area and the SOX middle area of SiC-QDs, respectively, where D_S = 6 × 10¹⁶ cm⁻², T = 600 °C, and t_N = 60 min. Si-QDs were uniformly formed in the middle SOX area with an 80 nm width of higher Si concentration region [>2.5 × 10²¹ cm⁻³ in Fig. 2(a)], but Fig. 3(a) shows that in the surface and bottom areas of SOX with lower Si concentration, few Si-QDs could be observed. This may be attributable to the detection limitation of HAADF-STEM, that is, it is possible that the Φ of the Si-QDs in the lower Si concentration region is too small to be detected by HAADF-STEM. We also confirmed the similar depth distributions of other SiC- and C-QDs with higher dopant regions. Figures 3(c) and 3(d) also shows that other SiC- at D_S = 6 × 10¹⁶ cm⁻², D_C = 4 × 10¹⁶ cm⁻², and T = 400 °C as well as C-QDs at D_C = 1 × 10¹⁷ cm⁻² and T = 400 °C were uniformly formed in the middle areas of the SOX layer with higher dopant concentrations, respectively. In addition, the diameter Φ and surface density N of the IV-QDs also depend on the type of IV-QDs.

Figures 4(a) and 4(b) shows the Φ histogram of Si-QDs and the normal probability plot of Φ as those used in Fig. 3(b), respectively. The average Φ and standard deviation of Φ (σ_Φ) of Si-QDs were 2.37 and 0.28 nm, respectively. Moreover, Fig. 4(b) shows that the Φ distribution can be explained by the Gaussian function (solid line). The Φ distribution of other IV-QDs was also confirmed by the Gaussian function, which indicates that the Φ of the IV-QDs randomly fluctuates. As a result, the σ_Φ of Si-QDs (≈0.28 nm) was much smaller than the σ_Φ of SiC- and C-QDs (≈1.0 nm), which may be caused by the diffusion difference between Si and C atoms in SiO₂. We will discuss the effect of N₂ annealing on the Φ of Si-QDs. Because the Si-diffusivity d_S in SiO₂ is reported to be d_S = 1.38 exp (−4.74 eV/kT) cm² s⁻¹,³³⁾ where k denotes the Boltzmann constant and the diffusion length (L_D ≡ $2\sqrt{{d}_{S}{t}_{N}}$ )²⁷⁾ of Si is estimated to be approximately 0.4 nm at T_N = 1000 °C and t_N = 30 min. Therefore, the L_D/Φ of Si-QDs is approximately 17% and is in almost the same order as σ_Φ. This L_D increase after N₂ annealing could not be observed using HAADF-STEM, because the Si-QD image at t_N = 0, which is considered to be small, could not be observed under the detection limitations of HAADF-STEM.²⁸⁾

**Fig. 4.** (Color online) (a) Histogram of Φ of Si-QDs in 5000 nm² area, and (b) normal probability plot of Φ of Fig. 4(a) (circles). Process conditions of IV-QDs are the same as those in Fig. 3. Figure 4(a) shows that average Φ and σ_Φ of Si-QDs are approximately 2.37 nm and 0.28 nm, respectively. Figure 4(b) shows that Φ distribution of (a) can be explained by the Gaussian function (solid line).
Download figure:
Standard image High-resolution image

Here, we summarize the Φ and N for each type of IV-QDs, using the data in Fig. 3. Figure 5(a) shows the average Φ and N for each of the three types of IV-QDs evaluated by the HAADF-STEM data shown in Fig. 3. N was determined by the number of QDs (n) over the cross section area (S_CS) shown in Fig. 3, that is, N = n/S_CS. The average Φ of the IV-QDs varied from approximately 2–4 nm, and the average Φ of the Si-QDs was the smallest among the three IV-QDs. Moreover, the N of the Si-QDs was the highest of the three IV-QDs, and the IV-QD surface densities were approximately 2 × 10¹² cm⁻² with an error bar of 20%. The error bars of Φ and N in Fig. 5 showed the σ_Φ and the statistical deviation of N (δN), respectively, where δN/N is given by the statistical deviation of n, that is, δN/N = $1/\sqrt{n},$ assuming that n randomly fluctuates.

**Fig. 5.** (Color online) (a) Average Φ (circles) and N (triangles) and (b) total QD area S_QD in a unit area obtained by S_QD = Nπ(Φ/2)² as a function of the type of IV-QDs of Si-, SiC-, and C-QDs, where the process conditions are the same as those in Fig. 4(c). (c) T dependence of average Φ and N in SiC-QDs, where D_S = 6 × 10¹⁶ cm⁻², D_C = 4 × 10¹⁶ cm⁻², and t_N = 60 min. Figure 5(a) shows that the Φ and N strongly depend on the type of IV-QD. Φ varies from 2 to 4 nm in IV-QDs, and σ_Φ of Si-QDs, shown as an error bar, is small (≈0.28 nm), but σ_Φ of SiC- and C-QDs is relatively large (≈1.0 nm). N of IV-QDs was approximately 2 × 10¹² cm⁻² with standard deviation of approximately 20%, as shown in error bars. In addition, the N increases with decreasing Φ. Figure 5(b) shows that the S_QD of IV-QDs varies from approximately 0.1–0.2, and maximum variation of δS_QD/S_QD calculated by Eq. (2) is approximately ±31% in SiC-QDs. Figure 5(c) shows that with decreasing T, the Φ of SiC-QDs decreases, but N increases.
Download figure:
Standard image High-resolution image

We discuss the total QD area (S_QD) of IV-QDs in the SiO₂ layer, because the PL intensity (I_PL) of the QDs is proportional to S_QD, as shown in Eq. (1),²⁸⁾ where S_QD = Nπ(Φ/2)², assuming that the QDs are spheres.

$\begin{eqnarray}&&{I}_{{\rm{P}}{\rm{L}}}=\eta {I}_{0}{S}_{{\rm{Q}}{\rm{D}}},\end{eqnarray} \tag{ 1 }$

where η and I₀ denote the PL emission coefficient of IV-QDs and the excited laser flux at the surface SiO₂. The penetration length of laser photons with 3.8 eV in the SiO₂ layer can be assumed to be infinite, because the E_G of SiO₂ (9 eV)²⁸⁾ is much higher than the laser photon energy of 3.8 eV. Moreover, the S_QD variation δS_QD can be given by the following equation:

$\begin{eqnarray}&&\displaystyle \frac{\delta {S}_{{\rm{Q}}{\rm{D}}}}{{S}_{{\rm{Q}}{\rm{D}}}}=\sqrt{{\left(\displaystyle \frac{2{\sigma }_{{\rm{\Phi }}}}{{\rm{\Phi }}}\right)}^{2}+{\left(\displaystyle \frac{\delta N}{N}\right)}^{2}}=\sqrt{{\left(\displaystyle \frac{2{\sigma }_{{\rm{\Phi }}}}{{\rm{\Phi }}}\right)}^{2}+{\left(\displaystyle \frac{1}{\sqrt{n}}\right)}^{2}}.\end{eqnarray} \tag{ 2 }$

Figure 5(b) shows the estimated S_QD per unit area and δS_QD/S_QD of the three types of IV-QDs, respectively, using the data in Fig. 5(a). The S_QD of IV-QDs varies from approximately 0.1–0.2, which slightly depends on the type of IV-QDs. Equation (1) indicates that the S_QD is considered to affect the PL properties, as discussed in Sect. 3.3. The maximum δS_QD/S_QD is approximately ±31% in SiC-QDs, because of the lower N of SiC-QDs shown in Fig. 5(a).

Figure 5(c) shows the T dependence of the average Φ and N in SiC-QDs, where D_S = 6 × 10¹⁶ cm⁻², D_C = 4 × 10¹⁶ cm⁻², and t_N = 60 min. With increasing T, the average Φ increases, but N decreases, which is the similar to the results for the SiC-dots in the Si layer.²¹⁾ Therefore, even after N₂ annealing, we confirmed the influence of T on the SiC-QD formation, which may be attributable to the SiC-QD growth by gathering small SiC-QDs during the high-T hot-ion implantation process.

Next, we discuss the crystal structures of the IV-QDs, using the lattice spots of IV-QDs evaluated by CSTEM. Figures 6(a-1), 6(b-1), 6(c-1), and 6(d) shows the CSTEM images of the cross section of Si-, H-SiC-, 3C-SiC- and C-QDs in SiO₂ (encircled areas) under the same process conditions as those in Fig. 3. All three IV-QDs show clear lattice spots, confirming that some Si- and C-QDs also consist of crystal structures. Figures 6(a-2), 6(b-2), and 6(c-2) shows the ED patterns of [111]Si with lattice spaces d = 0.314 nm, $\left[10\overline{1}3\right]$ H-SiC with d = 0.237 nm, and (111) 3C-SiC with d = 0.251 nm, as evaluated by FFT analysis of the lattice spots of Figs. 6(a-1), 6(b-1), and 6(c-1), respectively. As a result, it was found that SiC-QDs also consist of 3C- and H-SiC polytypes, although it has already been confirmed^16,19,21) that SiC-dots in the Si layer consist of 3C- and H-SiC polytypes. In contrast, the lattice distance of the C-atoms in Fig. 6(d) was approximately 0.36 nm, which is nearly equal to the layer distance of graphite (≈0.335 nm).³⁴⁾ Thus, some C-QDs consist of nano-graphite with a diameter of approximately 2 nm, as indicated by the G- and D-bands of the UV-Raman spectroscopy shown in Fig. 7(b). However, all IV-QDs, as confirmed by the HAADF-STEM images in Figs. 3(a)−3(c), showed no clear lattice spots. For example, only 1/4 of the Si-QDs, confirmed by HAADF-STEM, shows clear lattice images via CSTEM. Therefore, the crystal quality of some IV-QDs is poor, which suggests that some part of C-QDs consist of a-C. Consequently, IV-QDs consist of both amorphous and crystal structures.

**Fig. 6.** (Color online) CSTEM lattice images of cross sections of (a-1) Si-QD, (b-1) H-SiC-QD, (c-1) 3C-SiC-QD, (d) C-QD, and ED-patterns of (a-2) [111]Si-QD, (b-2) H-SiC-QD, and (c-2) 3C-SiC-QD evaluated by FFT analyses of (a-1), (b-1), and (c-1), in which the process conditions are the same as Fig. 3. Lattice spaces of 0.237 nm of (b-2) and 0.251 nm of (c-2) indicate that polytypes of SiC-QD of (b-1) and (c-1) are $\left[10\bar{1}3\right]$ H-SiC and (111) 3C-SiC, respectively. Lattice distance of C-QD in (d) was approximately 0.36 nm, which is nearly equal to the layer distance of graphite (≈0.335 nm).³⁴⁾
Download figure:
Standard image High-resolution image

**Fig. 6.** (Color online) CSTEM lattice images of cross sections of (a-1) Si-QD, (b-1) H-SiC-QD, (c-1) 3C-SiC-QD, (d) C-QD, and ED-patterns of (a-2) [111]Si-QD, (b-2) H-SiC-QD, and (c-2) 3C-SiC-QD evaluated by FFT analyses of (a-1), (b-1), and (c-1), in which the process conditions are the same as Fig. 3. Lattice spaces of 0.237 nm of (b-2) and 0.251 nm of (c-2) indicate that polytypes of SiC-QD of (b-1) and (c-1) are $\left[10\bar{1}3\right]$ H-SiC and (111) 3C-SiC, respectively. Lattice distance of C-QD in (d) was approximately 0.36 nm, which is nearly equal to the layer distance of graphite (≈0.335 nm).³⁴⁾
Download figure:
Standard image High-resolution image

**Fig. 7.** (Color online) t_N dependence of UV-Raman spectrum of (a) Si-QD at D_S = 6 × 10¹⁶ cm⁻², and (b) C-QD at D_C = 1 × 10¹⁷ cm⁻², where T = 600 °C. Arrows in (a) show the peak Raman positions of c-Si (520 cm⁻¹), including Raman intensity from Si-substrate under SOX layer and a-Si in Si-QDs (480 cm⁻¹). After N₂ annealing, peak Raman intensity of c-Si increases, but the peak Raman intensity of a-Si decreases. Arrows in (b) show T (a-C), D, and G bands of graphite, and after N₂ annealing, all peak-Raman intensities of graphite also increase.
Download figure:
Standard image High-resolution image

As an example of a Si-QD for which the crystal structure is clear, we estimate the Si atom number of Si-QDs observed by HAADF-STEM data in Figs. 5(a)−5(b), compared with that vis the XPS data shown in Fig. 2(a). First, we estimate the Si atom number of the Si-QDs N_QD using Eq. (3).

$\begin{eqnarray}&&{N}_{{\rm{Q}}{\rm{D}}}={N}_{{\rm{S}}{\rm{i}}}{S}_{{\rm{Q}}{\rm{D}}}W,\end{eqnarray} \tag{ 3 }$

where N_Si is the Si volume density (≡5 × 10²² cm⁻³),²⁷⁾ S_QD ≈ 0.1 in Fig. 5(b), and W denotes the depth width of the Si-QD formation area (≈80 nm) shown in Fig. 3(a), resulting in N_SD ≈ 4 × 10¹⁶ cm⁻².

In contrast, the Si atom number via the XPS data in Fig. 2(a); N_XPS was obtained by integrating the depth profile of the Si atom profile with more than 2.5 × 10²¹ cm⁻³ of the Si-QD formation area shown in Fig. 3(a), resulting in N_XPS ≈ 3.6 × 10¹⁶ ± 0.5 × 10¹⁶ cm⁻². Thus, N_SD ≈ N_XPS within an XPS accuracy of 1 at%. Consequently, we confirmed that approximately 100% of the Si atoms evaluated by XPS data form Si-QDs, as observed by HAADF-STEM in Fig. 3(a).

3.2. UV-Raman properties of IV-QDs

In this subsection, we discuss the material properties of the IV-QDs evaluated by UV-Raman spectroscopy. Figures 7(a) and 7(b) shows the t_N dependence of the UV-Raman spectrum of Si- at D_S = 6 × 10¹⁶ cm⁻² and C-QDs at D_C = 1 × 10¹⁷ cm⁻², respectively, where T = 600 °C. The arrows in Fig. 7(a) show the peak Raman shifts of c-Si (520 cm⁻¹), including the Raman intensity I_R0 from the Si-substrate under the SOX layer, in addition to a-Si in Si-QDs (480 cm⁻¹). The peak Raman intensity of c-Si increases and the a-Si Raman peak decreases after N₂ annealing, where the measured c-Si Raman intensity minus I_R0 (≡ΔI_R) shows the net Si-Raman intensity of Si-QDs in SiO₂. Thus, the crystal quality of Si-QDs can be improved via N₂ annealing. Moreover, the arrows in Fig. 7(b) show the T-, D-, and G-bands of C–C vibrations in C-QDs. After N₂ annealing, the peak Raman intensities of the G- and D-bands increased, but the N₂ annealing effect on the T-band intensity was small. Thus, the graphite component of C-QDs, which was already confirmed via CSTEM, as shown in Fig. 6(c), also increases after N₂ annealing. Our previous work²⁸⁾ on the t_N dependence of UV-Raman analysis for SiC-QDs also demonstrated that the peak-Raman intensities of the D-, T-, and TO-modes of Si−C vibration increase after a brief N₂ annealing, which leads to the improvement of the quality of the SiC-QDs.

Figure 8(a) shows the UV-Raman spectra of SiC-QDs as a function of the D_S/D_C dose ratio at a fixed D_C of 4 × 10¹⁶ cm⁻², where T = 200 °C and t_N = 0. The arrows in Fig. 8(a) show the G-, D-, and T-bands, which are attributable to the separated C atom areas in SiO₂,²⁸⁾ and the TO mode of the Si−C vibration. The Raman spectrum line shape strongly depends on D_S/D_C even at the same D_C. The G band can be observed at D_S/D_C ≤ 1.5. Moreover, Fig. 8(b) shows the peak Raman intensities of the D- and TO-bands versus D_S/D_C at the same data used in Fig. 8(a). The D-band intensity increases with decreasing D_S/D_C, which suggests that the separated C atoms in SiO₂ increase at low D_S conditions. The TO intensity is reduced only at D_S/D_C = 2. Thus, the Si−C bond formation shown by the TO-mode increases with decreasing D_S, so this study indicates that D_S/D_C ≤ 1.5 is optimum for forming SiC in this study.

3.3. PL properties

3.3.1. QD type dependence

In this subsection, we discuss the PL properties as a function of the type of IV-QDs. Figure 9(a) shows the PL spectrum comparison among three types of IV-QDs: Si-QDs (solid line: D_S = 6 × 10¹⁶ cm⁻², T = 600 °C, t_N = 1.5 h), SiC-QDs (dashed line: D_S = 6 × 10¹⁶ cm⁻², D_C = 4 × 10¹⁶ cm⁻², T = 200 °C, t_N = 30 min), and C-QDs (dotted line: D_C = 1 × 10¹⁷ cm⁻², T = 600 °C, t_N = 30 min) after N₂ annealing, in which the PL intensity of SiC-DQ is reduced to 1/3 of measured PL data. The lower and upper axes show the PL photon energy and wavelength λ_PL, respectively. We experimentally demonstrated the PL emissions from three types of IV-QDs with different peak-PL energy E_PH (peak-λ_PL) emissions from Si-QDs (near-IR), SiC-QDs (visible region), and C-QDs (near-UV). Thus, the PL spectrum line shape and I_PL strongly depend on the type of IV-QDs, and the E_PH (peak-λ_PL) of Si-, and SiC-, and C-QDs were approximately 1.56 eV (800 nm), 2.42 eV (500 nm), and 3.28 eV (380 nm), respectively. The I_PL of the SiC-QDs was the largest and was approximately 2.6 and 6.6 times greater than that of Si- and C-QDs, respectively. Moreover, the Si-QDs showed a very sharp PL spectrum with a FWHM of 0.36 eV, compared with the broad FWHM of SiC- (0.85 eV) and C-QDs (0.89 eV). The broad PL spectrum of SiC-QDs can be explained by the PL components of cubic and hexagonal SiC-polytypes with different E_GX, as discussed in Fig. 15. Consequently, it is easy for IV-QDs to control PL emission wavelength from near-IR to near-UV by changing the type of atoms implanted into the SiO₂ layer. The PL emission mechanisms for SiC- and Si-QDs are attributable to the sum of the emissions from various SiC polytypes as shown in Fig. 15, and the quantum-mechanically induced E_G-expansion of Si-QDs which are indicated in Fig. 11(b), respectively. However, the physical mechanism for C-QD emission is not currently understood, but it may be caused by the photon emissions from a-C and nano-graphite.^31,34) Moreover, Fig. 9(b) shows the T dependence of the maximum I_PL (I_MAX) of SiC-QDs (squares) and C-QDs (triangles) at the same ion dose conditions in Fig. 9(a). The I_MAX of SiC-QDs drastically increases with decreasing T, but the I_MAX of C-QDs slightly increases with increasing T. As a result, the optimum T conditions for realizing the strongest I_MAX of SiC- and C-QDs are 200 and 600 °C, respectively. This T dependence of SiC-QDs could be attributable to the N increase in the SiC-QDs at lower T conditions.²⁸⁾ Therefore, we will discuss the PL data of SiC- and C-QDs under the optimum T conditions shown in Fig. 9(b).

**Fig. 9.** (Color online) (a) PL spectrum comparison among three IV-QDs of Si-QD (solid line: D_S = 6 × 10¹⁶ cm⁻², T = 600 °C, t_N = 1.5 h), SiC-QD (dashed line: D_S = 6 × 10¹⁶ cm⁻², D_C = 4 × 10¹⁶ cm⁻², T = 200 °C, t_N = 30 min), and C-QD (dotted line: D_C = 1 × 10¹⁷ cm⁻², T = 600 °C, t_N = 30 min). Lower and upper axes show the PL photon energy and wavelength, respectively. The PL intensity of SiC-DQ is reduced to 1/3 of measured PL data. PL spectrum strongly depends on type of IV-QDs, and very different E_PH (peak-λ_PL) values were realized among three IV-QDs. E_PH (peak-λ_PL) of Si-, and SiC-, and C-QDs were approximately 1.56 eV (800 nm), 2.42 eV (500 nm), and 3.28 eV (380 nm), respectively. (b) T dependence of I_MAX of SiC-QDs (squares) and C-QDs (triangles) under the same ion dose conditions as those shown in Fig. 9(a). Figure 9(b) shows that optimum T conditions for increasing I_MAX of SiC- and C-QDs are 200 °C and 600 °C after N₂ annealing, respectively.
Download figure:
Standard image High-resolution image

Next, we address the effect of N₂ annealing on the PL properties of IV-QDs. Figures 10(a)–10(c) shows the t_N dependence of the PL spectra of the Si-, SiC-, and C-QDs, respectively, where the process conditions of IV-QDs are the same as those in Fig. 9. The PL spectrum line shape of each IV-QD after N₂ annealing is mostly independent of t_N. However, the I_PL drastically increases with increasing t_N, but the I_PL increase factors of N₂ annealing are also affected by the type of IV-QDs. The I_PL of the three IV-QDs rapidly increases after short N₂ annealing, and in particular, the t_N dependence of the I_PL of Si-QDs continues to grow larger with increasing t_N even for long t_N value.

**Fig. 10.** (Color online) t_N dependence of PL spectrum of (a) Si-, (b) SiC-, and (c) C-QDs, where process conditions of IV-QDs are the same as those seen in Fig. 9. PL spectrum line shape of each IV-QD is nearly independent of t_N, but PL intensity of each QDs increases with increasing t_N.
Download figure:
Standard image High-resolution image

We summarize the t_N dependence of the PL properties of IV-QDs. Figures 11(a) and 11(b) shows the t_N dependence of the I_MAX and E_PH (peak-λ_PL) under the same process conditions in Fig. 9, respectively, where circles, squares, and triangles show the data of Si-, SiC-, and C-QD, respectively, and the right vertical axis in Fig. 11(b) shows the peak PL wavelength (peak-λ_PL). The rhombi in Fig. 11(a) show the data of SiC dots in the Si layer at D_C = 4 × 10¹⁶ cm⁻² and T = 600 °C. Both the I_MAX and E_PH (peak-λ_PL) of IV-QDs strongly depend on the type of IV-QDs. The I_MAX of IV-QDs rapidly increased after a brief N₂ annealing, which could be attributable to the improved crystal quality of IV-QDs, as shown by the Raman data of IV-QDs in Figs. 7(a) and 7(b). The maximum I_MAX enhancement factors of Si-, SiC-, and C-QDs, compared to I_MAX at t_N = 0, reached approximately 59, 23, and 3.3, respectively, which indicates that post N₂ annealing is also a key process in realizing a higher I_PL. Thus, the I_MAX enhancement factor of Si-QDs is at its maximum in the three IV-QDs. In addition, even during high-T_N annealing, the I_MAX of IV-QDs continues to increase with increasing t_N. However, Fig. 11(a) shows that SiC-dots in the Si layer show a drastic decrease in I_MAX with increasing t_N, that is, ${I}_{{\rm{MAX}}}\left({t}_{N}\right)\propto \exp \left(-{t}_{N}/{t}_{D}\right),$ where t_D is a decay scaling time of approximately 21 min.²¹⁾ Thus, the SiC-dot structures in the Si layer are thermally unstable at high T_N conditions, which may be caused by the decomposition of SiC during high-T_N annealing.²¹⁾ Consequently, IV-QD structures have thermal stability even at a higher T_N, which is an advantageous characteristic of IV-QDs in the SiO₂ layer.

**Fig. 11.** (Color online) t_N dependence of (a) I_MAX, and (b) E_PH (left vertical-axis) and peak-λ_PL (right vertical-axis) at the same data of Fig. 9 [Si-QD (circles), SiC-QD (squares), and C-QD (triangles)]. Rhombi in Fig. 11(a) show the data of SiC dots in c-Si layer at D_C = 4 × 10¹⁶ cm⁻² and T = 600 °C. I_MAX of three IV-QDs rapidly increases after a brief N₂ annealing and continues increasing with increasing t_N. However, only SiC dots in c-Si layer decrease with increasing t_N under t_N ≥ 5 min. As shown in Fig. 11(b), only SiC-QDs shows the rapid increase of E_PH after a brief N₂ annealing, because 3C-SiC, which have an E_GX = 2.39 eV, was formed. However, the E_PH of other IV-QDs is nearly independent of t_N. Figures 11(a) and 11(b) show that both the I_MAX and E_PH (peak-λ_PL) of IV-QDs strongly depend on the type of IV-QDs. The I_MAX of SiC-QDs is approximately 2.6 and 6.6 times greater than that of Si- and C-QDs, respectively. Consequently, the E_PH of IV-QDs varies from 1.56 to 3.28 eV, that is, peak-λ_PL of IV-QDs varies from 380 to 800 nm.
Download figure:
Standard image High-resolution image

Figure 11(b) shows that different peak-λ_PL values for a near-IR of 800 nm to a near-UV of 380 nm can be obtained only by changing the type of IV-QDs. Moreover, only SiC-QDs showed strong t_N dependence on E_PH, because the E_PH increase from 2.0 eV of Si–C alloy at t_N = 0 to 2.3 eV at t_N = 5 min could be attributable to the 3C-SiC (E_GX = 2.39 eV) QD-formation that occurs by binding the Si–C atoms during high-T_N annealing for t_N = 5 min.²⁸⁾ Because Fig. 5(a) shows that the Φ of SiC-QDs is approximately 4 nm and relatively large, and in addition, the quantum-mechanics effects on E_G expansion (E_G ∝ Φ⁻²) in SiC-QDs occurs in Φ < 3 nm,²⁾ the quantum-mechanical effects on E_G expansion of SiC-QDs in this study was very small. On the other hand, Fig. 11(b) shows that the E_PH of Si-QDs with Φ ≈ 2.5 nm [as shown in Fig. 5(a)] was 1.56 eV, which is nearly equal to the reported experimental results reached in E_G via quantum E_G-expansion of Si-QDs at Φ ≈ 3 nm.³⁵⁾ In addition, considering the long tailing of the PL spectrum of Si-QDs at the half-width-tenth-maximum (HWTM), shown in Figs. 9 and 10(a), the HWTM of E_PH in Si-QDs (δE_PH) was approximately 0.28 eV. Assuming that the δE_PH is attributable to the quantum E_G-expansion of E_G ∝ Φ⁻², this δE_PH is caused by the σ_Φ shown in Fig. 5(a), so δE_PH/E_PH can be given by Eq. (4)

$\begin{eqnarray}&&\displaystyle \frac{\delta {E}_{{\rm{P}}{\rm{H}}}}{{E}_{{\rm{P}}{\rm{H}}}}=2\displaystyle \frac{{\sigma }_{{\rm{\Phi }}}}{{\rm{\Phi }}}.\end{eqnarray} \tag{ 4 }$

Therefore, the experimental δE_PH/E_PH seen in Fig. 9 was approximately 0.18, and the δE_PH/E_PH obtained by substituting the experimental σ_Φ/Φ of 0.12 in Fig. 5(a) into Eq. (4) was 0.24, which is neatly equal to the experimental δE_PH/E_PH. As a result, Eq. (4) is nearly valid in Si-QDs, so the HWTM of the PL spectrum in Si-QDs is also attributable to the statistical variation of the Si-QD diameter. Thus, for Si-QDs, we experimentally confirmed the quantum E_G-expansion of Si-QDs in this study.

We summarize the QD type dependence of the PL properties. Figure 12(a) shows the QD type dependence of I_MAX (circles) and E_PH (squares) for the same data as that used in Figs. 9 and 11. The I_MAX of SiC-QDs is approximately 2.6 and 6.6 times greater than that of the Si- and C-QDs, respectively. Moreover, the E_PH of IV-QDs can be controlled by the type of IV-QDs in the range of 1.5 < E_PH < 3.3 eV (380 < peak-λ_PL < 800 nm). Assuming that the PL emission coefficient η of Eq. (1) is independent of the type of IV-QDs, the universal I_MAX ∝ S_QD of Eq. (1) should be valid in the three IV-QDs. Figure 12(b) shows the I_MAX as a function of the S_QD [Fig. 5(b)] of Si- (circle), SiC- (square), and C-QDs (triangle). We verified that the I_MAX of Si- and SiC-QDs obeyed Eq. (1) [dashed line in Fig. 12(b)], which indicates that the η of Si-QDs is almost equal to that of SiC-QDs. However, the I_MAX of the C-QDs deviates from the dashed line in Eq. (1). We estimate the η of QDs by η = I_MAX/I₀ S_QD of Eq. (1), where the η deviation (δη) can be expressed by δη/η = δS_QD /S_QD and δS_QD /S_QD is given by Eq. (2). Figure 12(c) shows the η of the IV-QDs as a function of the type of IV-QD. The η of the Si-QDs is equal to that of the SiC-QDs, as shown in Fig. 12(b), but the η of the C-QDs is approximately 1/4 of those of the Si- and SiC-QDs. The physical mechanism for the low η in the C-QDs is not currently understood, but it is possible that the fabrication conditions for C-QDs, such as the T of hot-ion implantation, were not optimized in this work, resulting in poor quality C-QDs.

**Fig. 12.** (Color online) (a) Summary of QD type dependence of I_MAX (circles) and E_PH (squares) using the same data in Fig. 9, (b) total QD area dependence of I_MAX, and (c) QD type dependence of η (circles) calculated by substituting I_MAX and S_QD into Eq. (1) using the same data in Fig. 12(a). Circle, square, and triangle in Fig. 12(b) show the data for the Si-, SiC-, and C-QDs, respectively. Dashed line in Fig. 12(b) shows the I_MAX ∝ S_QD relationship of Eq. (1), and the data for Si- and SiC-QDs obey Eq. (1) of I_MAX ∝ S_QD, but the I_MAX of C-QDs deviates from the dashed line of Eq. (1). Figure 12(c) shows that the η of Si-QDs is equal to that of SiC-QDs, but the η of C-QDs is approximately 1/4 of those of Si- and SiC-QDs.
Download figure:
Standard image High-resolution image

3.3.2. Process dependence

Figure 13(a) shows the ion dose dependence of I_MAX and E_PH of Si- (circles) and C-QDs (triangles) after N₂ annealing, respectively, in which T = 600 °C. With an increasing ion dose, the I_MAX of Si-QDs decreases, but the I_MAX of C-QDs increases. The E_PH of Si-QDs is nearly independent of D_S, but the E_PH of the C-QDs increases with increasing D_C. It has already been reported³⁶⁾ that an integrated Raman intensity ratio—G-band to D-band of C–C vibration (S_G /S_D)—shown in Eq. (5) is an indicator of the quality of the graphite. In the following equation

$\begin{eqnarray}&&\displaystyle \frac{{S}_{G}}{{S}_{D}}=\displaystyle \frac{\displaystyle \int {I}_{G}\left(\omega \right)d\omega }{\displaystyle \int {I}_{D}\left(\omega \right)d\omega },\end{eqnarray} \tag{ 5 }$

where ω is the wavenumber, and I_G and I_D denote the fitting Raman spectra of the G- and D-bands for the measured data, respectively. Figure 13(b) shows the ion dose dependence of the peak Si-Raman intensity I_R (circles) of c-Si (520 cm⁻¹) of Si-QDs and the S_G /S_D (triangles) of C-QDs, when the process conditions are the same as those used in Fig. 13(a). The I_R of Si-QDs increases with decreasing D_S, which indicates that the Si-QD quality improved at a lower D_S, because of the ion-implantation damage reduction of Si-QDs at a lower D_S. In contract, the S_G /S_D of C-QDs increases with increasing D_C, which suggests that the graphite quality improved at a higher D_C. Consequently, the I_MAX and E_PH of Si- and C-QDs strongly depend on the ion dose, which is attributable to the ion dose dependence of the QD crystal quality.

**Fig. 13.** (Color online) Ion dose dependence of (a) I_MAX (dashed lines) and E_PH (solid lines) of Si-QD (circles) and C-QD (triangles), and (b) Raman intensity I_R at 520 cm⁻¹ of Si-QDs (circles) and DG-band-integrated intensity ratio S_G /S_D of C-QDs (triangles) of Eq. (5), where T = 600 °C and t_N = 30 min. Figure 13(a) shows that the I_MAX of Si-QDs increases with decreasing D_S, but the I_MAX of C-QDs increases with increasing D_C. The E_PH of Si-QDs slightly increases with decreasing D_S, but the E_PH of C-QDs increases with increasing D_C. Figure 13(b) shows that I_R of Si-QDs increases with decreasing D_S, but the S_G /S_D of C-QDs increases with increasing D_C.
Download figure:
Standard image High-resolution image

Next, the PL spectrum of SiC-QDs strongly depends on the D_S/D_C ratio even at a fixed D_C condition of 4 × 10¹⁶ cm⁻², as shown in Fig. 14(a), where T = 200 °C and t_N = 5 min. The arrows in Fig. 14(a) show the E_PH-positions. Figure 14(a) shows that the PL spectrum line shape strongly depends on the D_S/D_C. The I_PL at D_S/D_C = 1.5 is almost the same as that at D_S/D_C = 1, and the I_PL drastically decreases at D_S/D_C = 2, but the E_PH increases with decreasing D_S/D_C. In summary, Fig. 14(b) shows the I_MAX and E_PH of SiC-QDs as a function of D_S/D_C under the same process conditions as those used in Fig. 14(a). With decreasing D_S/D_C, both the I_MAX and E_PH of the SiC-QDs continue to increase even at the same D_C. In other words, after N₂ annealing, the I_MAX of the SiC-QDs is nearly constant under D_S/D_C ≤ 1.5, and rapidly decreases at D_S/D_C = 2, because of the reduced Si–C bonding shown as the TO intensity reduction in Fig. 8(b). Thus, the condition of D_S/D_C ≤ 1.5 is optimum for realizing a higher I_PL in SiC-QDs. In addition, the large D_S/D_C dependence of E_PH indicates that the peak-λ_PL of SiC-QDs can be controlled by D_S/D_C.

Our previous work¹⁹⁾ showed that the broad PL spectrum of SiC-dots in a-Si can be explained by five PL components of different SiC-polytypes of cubic- and hexagonal-SiC as well as an additional Si–C alloy, whereas the broad PL spectrum of SiC-dots in c-Si is attributable to the sum of only four PL components of different SiC-polytypes, which excludes the Si–C alloy.²¹⁾ This Si–C alloy component, which has an E_GX < 2.39 eV (3C-SiC), suggests an imperfect SiC structure in SiO₂.¹⁹⁾ Actually, Fig. 15(a) shows that the measured PL spectrum at D_S/D_C = 1 (bold line) can be well fitted via five PL component fitting [Gaussian curve (dashed lines)] even in SiC-QDs, where D_C is fixed to 4 × 10¹⁶ cm⁻², t_N = 5 min, and T = 200 °C. Thus, the PL spectrum of SiC-QDs in SiO₂ can be also explained by the sum of the PL emissions from four different SiC-polytypes: 3C-SiC, which has an E_GX of 2.39 eV (3C: red); 8H-SiC, which has an E_GX of 2.73 eV (8H: green); 6H-SiC, which has an E_GX of 3.05 eV (6H: blue); 4H-SiC, which has an E_GX of 3.32 eV (4H: purple); and an additional one-component of Si–C alloy (SC: brown), which has an E_GX of lower than 2 eV.¹⁹⁾ We can estimate the total photon emission number from each PL component, which is an indicator for each polytype ratio for the SiC and Si–C alloy.¹⁹⁾ The integrated PL component ratio P_I of the one PL component I_I to the total I_PL emission (subscript I is from SC to 4H), as shown in Fig. 15(a), was calculated using Eq. (6).¹⁹⁾

$\begin{eqnarray}&&{P}_{I}={\int }^{}{I}_{I}(E)dE/{\int }^{}{I}_{{\rm{P}}{\rm{L}}}(E)dE,\end{eqnarray} \tag{ 6 }$

where E is the photon energy. Figure 15(b) shows the integrated intensity ratio of the five PL components¹⁹⁾ of SiC-QDs as a function of D_S/D_C, and the P_I ratio strongly depends on D_S/D_C. The 3C-SiC PL component ratio of approximately 44% is nearly independent of D_S/D_C. However, with decreasing D_S/D_C, the hexagonal-SiC PL component ratio increases, and the SC component rapidly decreases. Thus, the SC component indicates that the imperfect SiC formation of the Si–C alloy in SiO₂ increases with increasing D_S/D_C. Consequently, even at the same D_C, the SiC-QD formation is strongly affected by D_S, and the H-SiC polytype ratio increases with an increasing D_C ratio. These results regarding the D_S/D_C dependence of PL components can indicate that the peak-λ_PL (E_PH) can also be controlled by D_S/D_C, as shown in Figs. 14(b) and 15(b).

**Fig. 15.** (Color online) (a) Five PL component fittings [Gaussian curve (dashed lines)] for measured PL spectrum at D_S/D_C = 1 (solid line) and (b) integrated PL intensity ratios of five PL components of SiC-QD calculated by Eq. (6) as a function of D_S/D_C, where D_C is fixed at 4 × 10¹⁶ cm⁻², t_N = 5 min, and T = 200 °C. Figure 15(a) shows that the PL spectrum of SiC-QDs can also be explained by the sum of five PL components¹⁹⁾ from four different SiC-polytypes of 3C-SiC with E_GX of 2.39 eV (I_3C: red), 8H-SiC with E_GX of 2.73 eV (I_8H: green), 6H-SiC with E_GX of 3.05 eV (I_6H: blue), 4H-SiC with E_GX of 3.32 eV (I_4H: purple), and an additional one-component of Si–C alloy I_SC (brown) with lower E_G of 2 eV. Figure 15(b) shows that five PL component ratio of Eq. (6) strongly depends on D_S/D_C. Hexagonal PL component ratio increases with decreasing D_S/D_C, but the 3C-SiC PL component is almost independent of D_S/D_C.
Download figure:
Standard image High-resolution image

According to the different PL peak wavelengths versus the type of IV-QDs shown in Figs. 9, 11(b), 12(a), and 14(b), the peak-λ_PL of IV-QDs in SiO₂ can be controlled by the combination of D_S and D_C. Figure 16 shows the peak-λ_PL contour map of IV-QDs in various D_S and D_C conditions obtained from the data in Figs. 9, 12(a), 14(b), and 15(b). A longer λ_PL can be obtained by increasing the D_S at a lower D_C. In contrast, a shorter λ_PL can be realized by increasing the D_C at a lower D_S. Consequently, the peak-λ_PL of IV-QDs from near-UV to near-IR regions can be easily designed only by the combination of D_S and D_C conditions.

**Fig. 16.** (Color online) Peak-λ_PL contour map in various D_S and D_C conditions for designing peak emission wavelength of IV-QDs. This contour map of peak-λ_PL was obtained using the data from Figs. 9, 12, and 14. A longer λ_PL can be obtained by increasing D_S at a lower D_C. In contrast, a shorter λ_PL can be realized by increasing D_C at a lower D_S.
Download figure:
Standard image High-resolution image

4. Conclusion

In this work, we experimentally studied the group-IV semiconductor-QDs of Si, SiC, and C in a SiO₂ layer fabricated by hot-ion implantation at T and post N₂ annealing at a T_N of 1000 °C. Si- and C-QDs, in addition to SiC-QDs, can be successfully formed around the middle of the SiO₂ layer with a relatively higher dopant concentration using HAADF-STEM. CSTEM also shows clear lattice spots in some areas of IV-QDs, as confirmed by HAADF-STEM, which indicates that some areas of IV-QDs are crystallized. The IV-QD diameter Φ varied from approximately 2–4 nm, and the statistical distribution of Φ can be explained by the Gaussian function. The QD surface-density N was approximately 2 × 10¹² cm⁻². The Φ and N of the SiC-QDs were affected by T.

The UV-Raman data for the Si- and C-QDs show that the QD crystal quality was improved after N₂ annealing because the c-Si Raman intensity of the Si-QDs and the G-band intensity of the C-QDs increased. Moreover, the UV-Raman spectra of the SiC-QDs strongly depend on D_S/D_C, and the Raman intensities of the TO and D-bands increase with decreasing D_S even at a fixed D_C, which suggests that the SiC and graphite formations in SiC-QDs increase at low D_S conditions.

We experimentally demonstrated strong PL emissions in a wide range of E_PH (peak-λ_PL) values, even from Si- and C-QDs, in addition to SiC-QDs. The PL spectrum line shape strongly depends on the type of IV-QDs, and different E_PH (peak-λ_PL) values can be obtained from 1.56 eV (800 nm) of Si-QDs to 3.3 eV (380 nm) of C-QDs, simply by changing the type of IV-QDs. Moreover, the PL intensity of the IV-QDs drastically increased after a short N₂ annealing, which indicates that post N₂ annealing is a key process for realizing strong PL intensity by improving the crystal quality of the IV-QDs. The PL intensity of the SiC-dots in the Si layer gradually decreased after a long N₂ annealing, which could be attributable to the decomposition of the Si−C bond during long high-temperature annealing. However, the PL intensity of the IV-QDs in SiO₂ layer continues to increase with increasing t_N, which is advantageous for the thermal stability of IV-QDs. We experimentally confirmed that the PL emission efficiency of Si-QDs is equal to that of SiC-QDs, but the PL emission efficiency of C-QDs was approximately 1/4 that of Si- and SiC-QDs. In addition, for Si-QDs with an average Φ of 2.4 nm, the E_PH of 1.56 eV can be explained by the quantum E_G-expansion via the Φ reduction. In addition, the HWTM of the PL spectrum of Si-QDs is also attributable to the Si-QD Φ variation effect on quantum E_G-expansion in this study. It was also found that the I_PL and E_PH of Si- and C-QDs depend on the ion doses.

The PL spectrum of SiC-QDs also depends on the D_S /D_C eve at a fixed D_C, and the I_PL and E_PH increase with decreasing D_S /D_C. These results suggest that the optimum D_S /D_C condition exists to realize higher PL intensity and to design a desired E_PH (peak-λ_PL). Moreover, the PL spectrum of SiC-QDs can be also explained by five PL components of different cubic/hexagonal SiC-polytypes and Si–C alloys. With increasing D_S /D_C, the PL component of 3C-SiC is almost constant, but the hexagonal-SiC component decreases.

Consequently, this work showed that the peak-λ_PL of IV-QDs can be controlled using only two parameters, D_S and D_C. A longer λ_PL can be obtained by increasing D_S at a lower D_C, and a shorter λ_PL can be realized by increasing the D_C at a lower D_S.

In summary, we demonstrated very strong PL emissions from IV-QDs with different λ_PL values for near-IR in Si-QDs, the visible range in SiC-QDs, and near-UV in C-QDs, showing that it is easy to design a desired peak-λ_PL simply by controlling the combination of ion doses of Si⁺ and C⁺ that are implanted into the SiO₂ layer.

Acknowledgments

This work was partially supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (17K06359).

Group-IV-semiconductor quantum-dots in thermal SiO₂ layer fabricated by hot-ion implantation technique: different wavelength photon emissions

Article metrics

Permissions

Author e-mails

Author affiliations

Dates

Abstract

1. Introduction

2. Experimental procedures