Enhanced photoluminescence from porous silicon nanowire arrays

The enhanced room-temperature photoluminescence of porous Si nanowire arrays and its mechanism are investigated. Over 4 orders of magnitude enhancement of light intensity is observed by tuning their nanostructures and surface modification. It is concluded that the localized states related to Si-O bonds and self-trapped excitations in the nanoporous structures are attributed to the strong light emission.


Background
The past decade has seen intense interest in nanoscale structures as these materials exhibit significantly different optical and electrical properties from their bulk materials [1][2][3][4]. Si, as one of the most conventional semiconductor materials, plays an important role in microelectronics [5][6][7]. Its application in integrated circuits has drastically changed the way we live. However, due to its indirect bandgap structure, the weak light emission from Si limits its application for future on-chip optical interconnection. Much effort has been devoted to overcome this limitation [8][9][10], and nanostructured materials are believed to be a candidate for light-emitting devices. There have been many reports discussing light emission and its mechanism from porous Si [11][12][13], Si sphere [14], and nanowire [3,[15][16][17][18][19][20] structures. Several perspectives, such as quantum size effects [2], interfacial state [11,14], and radiative defects in SiO x [19,21] are used to explain their contribution on the strong photoluminescence (PL). However, there are only limited investigations on the enhancement of light emission. In this letter, we will discuss the ways to improve the PL properties of porous Si nanowire arrays. Over 4 orders of magnitude enhancement of PL intensity is observed at room temperature by engineering their nanostructures and chemically modifying their surfaces.

Methods
Si nanowire arrays (Si NWAs) were prepared by metalassisted chemical etching on p-Si(100) with the resistivity of 0.02 Ω cm. The Si wafers were firstly cleaned in acetone, ethanol, and diluted hydrofluoric acid (HF) solution to remove the organic contaminants and the native SiO 2 layer. Ag particles were then formed in the solution of AgNO 3 (0.06 M) and HF (5 M) for 10 min followed by the chemical etching of Si NWAs in the solution of HF (5 M) and H 2 O 2 for 15 min. Ag catalysts were finally removed in concentrated HNO 3 . Si NWAs with different surface morphology were obtained by tuning the H 2 O 2 concentration at 0.2, 0.5, 2, and 5 M. Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were utilized to investigate the surface morphology and the crystallinity of the Si nanowires. PL measurements were performed to investigate their optical property with LabRam HR 800 Raman instrumentation (Horiba Jobin Yvon) within the range of 500 to 1,000 nm using the 488-nm line of an Ar + laser at a laser power of 2 mW. This trend is consistent with that found in the PL intensity in Figure 1a, and it indicates that the PL enhancement is related to the surface nanostructures of Si NWAs.

Results and discussion
All the PL emissions in Figure 1a exhibit similar broad peaks centered around 750 nm with a short-wavelength shoulder. They can be deconvoluted to two bands centered at 752 and 688 nm as shown in Figure 1b. The former (p1) is consistent with reports before [3], and it is believed to arise from the silicon nanostructure coated with a thin oxide layer. However, the weak PL peak located at 688 nm has not been discussed yet. It is 8 nm longer than that observed in [19,20]. This red shift may be due to the relatively big skeleton size (approximately 20 nm) of the porous NWA as shown in Figure 2d or from other emission mechanisms.
To investigate the enhancement mechanism of light emission from the porous Si NWAs and confirm their emission origins, these samples are divided into two groups and processed with further treatment. For group 1, oxidization was performed at 1,000°C for 5 min to passivate the surface with Si-O bonds; in group 2, the Si NWAs were rinsed in diluted HF to remove the Si-O bonds on the surface. Figure 3 shows the PL spectra of pristine and treated NWA samples. Interestingly, for the samples with low porosity (those obtained at 0.2, 0.5, and 2 M H 2 O 2 concentrations), oxidization treatments are always helpful to improve the PL intensity, and over 30 times enhancement is observed compared to their pristine ones. This is easily understood as the intense SiO 2 surface can greatly reduce the nonradiative recombination and help the light emission. The maximum PL intensity comes from the oxidized Si NWAs prepared at 2 M H 2 O 2 concentration, and a 2.5 × 10 4 times enhancement is observed compared to that from Si NWAs prepared at 0.2 M (solid line in the inset of Figure 1a). However, for the NWAs obtained at 5 M H 2 O 2 concentration, an opposite trend is observed. After oxidization, the PL intensity has a twofold decrease, and we attribute this to the reduction of effective lightemitting centers or interface state as the small-sized silicon skeleton is fully oxidized into SiO 2 . Even proper thermal oxidization helps the light emission from the Si NWAs; compared with the 4 orders of magnitude enhancement for the pristine samples as shown in Figure 1a, only 2 orders of magnitude enhancement is observed with the increase of H 2 O 2 concentration for all oxidized Si NWAs. In our experiment, we find that the best PL intensity comes from the thermal treatment at 1,000°C for 5 min for the Si NWA sample prepared at 2M H 2 O 2 concentration. Long-time thermal oxidization will induce the reduction of PL intensity as the small nanostructures are oxidized.
To our surprise, after oxidization, the PL peaks have a red shift for all the samples. The shift increases with the porosity of NWAs, and a maximum shift of 50 nm from 750 to 800 nm was observed for the sample prepared at 5 M H 2 O 2 concentration. This phenomenon cannot be explained by the quantum confinement (QC) effect. According to QC theory, the bandgap should increase with the size decrease of the nanostructure by oxidization and lead to a blue shift. Moreover, their temperature-dependent PL spectrum also indicates that the light emission did not originate from the QC effect. As shown in Figure 3d, the intensity of PL increases with decreasing temperature, while the peak position remains stable. Apparently, the emission mechanism is also contradictive with the well-known Varshni formula in the QC that it will induce a blueshift with decreasing temperature. At the same time, the emission linewidth decreases with increasing temperature in porous Si NW arrays. This abnormal phenomenon has been explained by a multilevel model for light emission as discussed before [18].
Simultaneously, HF treatment on the Si NWAs always arouses the great decrease of intensity. We know that HF treatment removes the Si-O layer and introduces the Si-H bonds on the surface, which will impede the forma- are the main origins of the light emission. With the increase of the porosity of Si NWAs at high H 2 O 2 concentration, it offers more light-emitting centers and the PL intensity is greatly enhanced. From Figure 3a,b,c, it is found that the small shoulder in the short wavelength corresponding to the p2 peak disappears, and it agrees well with the discussion in [19].

Conclusion
Si NWAs on Si substrates with different morphology were prepared by two-step metal-assisted chemical etching. With the increase of porosity, the light emission intensity increases. Surface treatment affects the intensity significantly, and oxidization substantially strengthens the intensity. The origin of the strong emission of Si NWAs is concluded to be from the localized state related to Si-O bonds and self-trapped excitations in the nanoporous structures.