5.3. Nano Silicon Grains
The presence of a population of silicon nanoparticles containing 5% of the total interstellar dust mass (with Si/H 6 ppm) in the ISM was proposed by Witt and his coworkers (Witt et al. 1998, Smith & Witt 2002) and Ledoux et al. (1998) to account for the ERE phenomenon. Witt et al. (1998) suggested that the formation of interstellar SNPs could occur as a result of the nucleation of SiO molecules in oxygen-rich stellar outflows, followed by annealing and phase separation into an elemental silicon phase in the core and a passivating mantle of SiO2. The SiO2 mantle is crucial for the SNP model since SNPs luminesce efficiently only when their surface dangling Si bonds are passivated.
The SNP model is recently receiving much attention because (1) as a consequence of quantum confinement (see Section 4), the SNP photoluminescence provides so far the best match to the observed ERE in both spectral profile (i.e. peak position and width) and the required extremely high quantum efficiency; (2) as shown by Smith & Witt (2002) in terms of a SNP photoionization and photofragmentation theory together with the experimentally established fact that photoionization quenches the PL of SNPs, this model successfully reproduces the observed dependences of the ERE intensity, the ERE quantum efficiency, and the ERE peak wavelength on the intensity and hardness of the UV/visible radiation fields of a wide variety of dusty environments; (3) as already mentioned in Section 2, other candidate materials all fail to simultaneously satisfy the ERE spectral characteristics and the high quantum efficiency requirement; it is worth noting that HAC, the long-thought ERE candidate, also falls in this category: HAC luminesces efficiently only in the near UV (quantum efficiencies as high as 10% have been reported; see Rusli, Robertson, & Amaratunga 1996 and references therein) when it has a higher degree of hydrogenation (indicating a larger band gap), dehydrogenation is expected to reduce the band gap and thus can redshift the PL peak into the wavelength range where the ERE is observed, but this results in an exponential drop in its PL efficiency by a factor of ~ 104.
As already mentioned in Section 2, although the SNP model seems very attractive, it also has some weakness: it has been shown by Li & Draine (2002a) that SNPs with oxide coatings, if they are free-flying in the ISM, would emit strongly in the 20 µm O-Si-O bending band. Existing COBE-DIRBE 25 µm photometry appears to already rule out such high abundances of SNPs. If SNPs are responsible for the ERE from the diffuse ISM, they must either be in a 50 Å clusters, or attached to larger "host" grains. This problem can not be resolved by invoking hydrogen passivation of SNPs because H-passivated SNPs luminesce at blue and near-UV wavelengths (Zhou, Brus, & Friesner 2003) while blue PL is not observed under interstellar conditions (Rush & Witt 1975; Vijh, Witt, & Gordon 2003). Very recently, Witt & Vijh (2004) argued that Fe- or C-passivated SNPs could overcome this problem. It would be very helpful to perform detailed calculations of the 11.3 µm Si-C emission feature of C-passivated SNPs and compare with the upcoming SIRTF (Space Infrared Telescope Facility) high resolution spectra of the diffuse ISM. Previously, Whittet, Duley, & Martin (1990) deduced that the abundance of Si in SiC dust is no more than ~ 5% of that in silicates, through an analysis of the 7.5-13.5 µm absorption spectra of 10 Galactic Center sources. It would also be very useful to investigate the IR emission properties of Fe-passivated SNPs which luminesce even more efficiently than O-passivated SNPs in the wavelength range over which the ERE is observed (Mavi et al. 2003).