Optical characterization of luminescent silicon nanocrystals embedded in glass matrices

Interstellar dust in nebulae and in the Diffuse Interstellar Medium (DISM) of galaxies contains a component which exhibits efficient visible-near infrared luminescence ranging from 600 to 1000 nm, known as Extended Red Emission (ERE). The ERE, which was first detected by Cohen et al.1 in the spectrum of the Red Rectangle nebula more than 30 years ago, generally appears as a broad structureless emission band (120 to 210-nm-wide) peaking between 600 and 850 nm (see Sec. Interestingly, the bandwidth of the ERE in-creases with the position of the maximum of the band (Fig. 2.2). The ERE has been observed in many different dusty environments that are illuminated by ultraviolet photons. Therefore, this visible-near infrared emission is probably due to the interaction of ultraviolet photons with dust grains through photoluminescence (PL). The maximal photon conversion efficiency of the PL process leading to ERE approaches 10% (Fig. 2.3). The ERE carrier is a component of interstellar dust that must consist of the most abundant refractive elements in the cosmos (C, O, Si, Fe and Mg), and, once formed, the carriers must survive under a wide range of as-trophysical conditions. Silicon nanocrystals (nc Si) are discussed as possible carriers of the ERE.2, 3 Indeed, nc Si can emit intense PL in a wide spectral range from 600 to 1100 nm. Interestingly, the emission energy depends on the nanoparticle size. This effect has been explained by the quantum confinement model proposed first by Canham.4, 5 Ledoux et al.2 have shown that the properties of the PL of free oxygen-passivated nc Si could satisfy the observational con-straints of the ERE in terms of spectral variability and luminescence yield. However, the re-cent model calculations of Li and Draine6 revealed that heated (~100 K) free-flying O-passivated nc Si should give rise to an emission feature around 20 m, which in fact is not observed in ERE regions. A possibility to overcome this objection is to assume that the O-passivated nc Si are not free nanoparticles, but instead attached to or embedded in larger grains of some other material, as for instance, silicates which have already been identified in space.7, 8, 9 Another way to overcome the objection of Li and Draine is to consider other forms of passivation. As has been shown by several authors,10, 11, 12, 13, 14 nc Si embedded in SiO2 (nc Si/SiO2) emit strong PL. Ion implantation is a promising technique to produce nanoparti-cles in solid matrices. We employed the accelerator facilities of the Institute of Solid State Physics of the University of Jena to implant Si ions into fused silica windows. An excess concentration of silicon atoms is thus produced in the host SiO2 matrix which, by applying an annealing at 1100 °C, condensates to silicon nanoparticles and crystallizes. Although the condensation and crystallization occur after an annealing of one minute,10, 15 the samples were annealed during one hour in order to well-passivate the nc Si, that means, to reduce effectively the number of Si-dangling bonds at the nc Si surface that are efficient non-radiative recombination centers.10, 16 Upon excitation with UV light, most of our nc Si/SiO2 samples revealed strong PL. In this thesis, we compared the PL of free O-passivated nc Si and silicon nanocrystals embedded in SiO2 (see Chap. 4). The free O-passivated nc Si were synthesized by laser pyro-lysis of silane. We demonstrated that the emission energy depends on the size of the nanopar-ticles which supports the fact that, in our samples, the PL is due the recombination of exci-tons confined within the nanocrystal volume. However, our measurements revealed that the size dependence of the PL differ obviously whether the nanocrystals are free or embedded in a host SiO2 matrix. Indeed, PL spectra measured by different groups (Fig. 2.22), including ours (Fig. 4.6 and Fig. 4.14), show that, for the same peak position, the PL originated either from nc Si/SiO2 systems or from free silicon nanoparticles but which were about two times larger. We assume that the SiO2-embedding strongly affects the PL energy of nc Si. As far as nc Si/SiO2 systems are concerned, quantitative measurements revealed a broad (150 to 250-nm-wide) PL band peaking between 705 and 1050 nm (see Sec 4.2.17.). However, taking into account the fact that the size distributions of nc Si in fused silica win-dows are rather broad, and the fact that cooling the samples down to temperatures below 20 K results in a blue shift of about 30 nm (Fig. 4.17), the entire spectral range observed for the ERE can be covered. We noticed that the bandwidth of the PL spectra of our nc Si/SiO2 systems increases with the peak position (Fig 4.13). Free and embedded silicon nanoparticles exhibit an efficient luminescence. The maximum quantum yield of embedded nc Si (60%) reported by Walters et al.160 shows that the SiO2-embedding may enhance the PL efficiency since it is two times higher than the maximum PL yield of free oxide-covered nc Si reported by Ledoux et al.72 However, the PL efficiency is even higher at low temperature, since we observed an increase of the PL intensity by a factor of ~1.4 when we cooled one of our nc Si/SiO2 samples down to 50 K (Fig. 4.17). Therefore, it can be concluded that nc Si em-bedded in SiO2 matrices do fulfil the requirements of an ERE carrier as far as the spectros-copy (spectral peak position, bandwidth, and PL yield) is concerned. In order to simulate the conditions of silicates containing nc Si, we have post-implanted into strongly luminescent nc Si/SiO2 systems various atomic elements, for instance magnesium and calcium, which form silicates if their oxide is combined with SiO2. To recover the PL of the nc Si which disappeared completely upon the Mg+ or Ca+ ion irradiation, it was necessary to apply a post-annealing. We demonstrated that the quenching of the PL observed after a post-annealing at 900 °C is due to the presence of residual defects and that a post-annealing at 1100 °C is needed to recover the PL completely (Fig.5.6). Nevertheless, since Kachurin et al.44 observed a strong decrease of the PL by incorporating boron into their nc Si/SiO2 systems with an ion fluence of 1 × 1016 ions/cm2, we assume that, above concen-trations of about 2 at.% in fused silica, the PL originating from nc Si should be strongly quenched even after a post-annealing at 1100 °C. Recently, red PL peaking at 750 nm have been observed in crystalline Al2O3 samples (sapphire) where the presence of nc Si has been clearly demonstrated.50 However, this PL would arise from luminescent defect centers in the host matrix. Furthermore, since no PL has been observed for nc Si embedded in hexagonal silicon carbide,161 it seems that the embed-ding in crystalline matrices prevent the recombination of excitons in quantum confined nc Si. Therefore, one can question the existence of luminescent nc Si embedded in crystalline sili-cates. In order to understand the effect of the incorporation of foreign atoms on the PL properties of our nc Si/SiO2 systems, we proceeded to similar experiments with Er and Ge. As has been demonstrated by several authors,17, 18 the presence of nc Si in a glass matrix enhances considerably the emission of Er3+ ions at 1.536 µm. At the same time, the PL of nc Si is considerably quenched. Since the solubility of Er in crystalline silicon is about 2 orders of magnitude lower than in SiO2,19 the optically active Er3+ ions are believed to be localized outside the nc Si core, demonstrating that ions present in the host SiO2 matrix influence the PL properties of embedded nc Si. We also incorporated Ge into our luminescent nc Si/SiO2 systems. We observed many strong resemblances with previous studies on the optical properties of nc Si1 xGex/SiO2 sys-tems (PL,191 time-resolved PL,125 temperature dependence of PL,191 and Raman spectros-copy125, 193) which indicates that we also synthesized nc Si1 xGex embedded in SiO2. PL origi-nating from the recombination of excitons confined within the nanocrystal volume has been previously observed for Si1 xGex alloy nanocrystals (nc Si1 xGex) produced by a cosputtering technique. We observed a red shift of the PL spectra with increasing Ge content x in the nc Si1 xGex/SiO2 samples. The red shift could be due to the alloying, since it is expected that the band gap energy of nc Si1 xGex changes continuously from the band gap of nc Si to the band gap of nc Ge that is comparatively lower in energy.5, 144 We also observed a shortening of the PL lifetimes with increasing Ge content which has been predicted by calculations189, 192 However, it is reasonable to assume that the PL decays are also strongly affected by an in-crease of non-radiative recombinations caused by defects. This picture is supported by the decrease of the PL intensity observed as the Ge concentration increases, whereas the calcula-tions have predicted the contrary.189 Toshikiyo et al.,190 have demonstrated that Ge atoms in the alloy nanocrystals provide new recombination channels induced by the presence of Ge dangling bonds at the interface between the nanocrystal and the host matrix which are re-sponsible for the PL quenching. Finally, for the first time, luminescent alloy nc Si1 xGex have been produced by ion implantation into fused silica and subsequent annealing at high tem-perature. Taking into account this result, one can reasonably put forward the assumption that not only pure nc Si can be related to the ERE. Interestingly, we showed that the PL of our nc Si/SiO2 systems was seriously quenched when the samples were implanted with high-energy ions (see Chap. 5). We demon-strated that even at low ion fluences (~1013 ions/cm2), the amount of ions traversing the nanoparticle induces enough structural defects within the nc Si to explain a complete quench-ing of the PL (see Sec. 5.1.5). In the same way, Barratta et al.,194 who studied the optical properties of porous silicon thin films subjected to He irradiation, found an apparent correla-tion between the PL yield and the amorphization state evaluated by Raman spectroscopy. The authors questioned the possibility that crystalline silicon nanoparticles could maintain their initially high PL efficiency when they are subjected to intense ion beam irradiation like that expected in some astrophysical environments where the ERE has been observed. On another hand, Kanemitsu195 demonstrated that amorphous silicon (a-Si) nanostructures in the same size range of 2-5 nm exhibit broad Vis-NIR PL similar to that one which emanates from nc Si. Although the PL efficiency of a Si nanoparticles is one order of magnitude lower than that of nc Si at room temperature, the PL intensity of a Si nanoparticles is significantly in-creased below 100 °K, and the two PL yields of a Si nanoparticles and nc Si become compa-rable. Therefore, it can be assume that under astrophysical conditions a Si nanoparticles (free, embedded or attached to larger grains) are competitive candidates for the ERE. Furthermore, this material is not concerned by the constraint expressed by Li and Draine6 which was about crystalline Si nanoparticles. Koike et al.196 observed a broad PL band peaking at 1.9 eV emitted by forsterite sin-gle crystals that were irradiated by γ-ray. The authors197 proposed that forsterite could be responsible for the ERE since their laboratory spectra were similar to astronomical observation of ERE in the Red Rectangle.22 It should be remembered that forsterite has been already detected in space. Interestingly, we observed an identical PL band from our fused silica win-dows subjected to He irradiation (Fig. 4.11). The occurrence of this luminescence is due to the formation of defect centers in glassy SiO2 (non bridging oxygen hole centers, see Sec. However, it has been demonstrated that this defect PL disappears completely at low temperature (~6 K).198 Therefore, if we consider irradiated silicates as possible carriers of the ERE, they should be present in larger grains. Another formation of luminescent defects should be evoked,199 since we observed an identical 1.9-eV PL band from windows that were subjected to an intense UV radiation (not shown in this work).


Citation style:
Could not load citation form.