Recently, the optical properties of Si nanostructures have been the subject of intensive study. The strong photoluminescence (PL) from porous Si, [1,2] Si nanocrystals, [2,3] and amorphous Si quantum structures [4±6] prepared by a variety of methods has been the subject of particular attention from both the fundamental and practical points of view. Although several emission mechanisms from Si, including quantum size effect, defects, and Si-based chemical species such as siloxane and polysilane, have been proposed, [7±10] quantum size effect has been most frequently proposed as the origin of the PL, and the size control of Si quantum structures is an important factor in Si-based optoelectronic devices. Most studies have been focused on quantum wells [11±13] and self-assembled Si quantum dots (QDs) grown on an insulating surface [14±16] for the size control of Si quantum structures. However, these self-assembled Si QDs have a large lateral size, and the quantum size effect is negligibly small in the lateral direction compared to that of sphere-like QDs. Therefore, quantum well and self-assembled Si QD structures grown on the substrate are not suitable for optoelectronic applications in the short wavelength region because quantum size effect in these two structures is very small due to only one-dimensional size effect in the growth direction.Meanwhile, amorphous Si (a-Si) has two important advantages compared with bulk crystalline Si. Firstly, the luminescence efficiency of bulk a-Si is higher than that of crystalline Si because of its structural disorder. [17] In addition, the bandgap energy of bulk a-Si (1.6 eV) is larger than that of bulk crystalline Si (1.1 eV), thus making it a good candidate for short-wavelength device applications such as a blue light-emitting diode. Therefore, it would be expected that these intrinsic advantages of a-Si and the quantum confinement effect in the a-Si QD in three directions would lead to extremely useful optical properties in short-wavelength devices. [18] In this study, a silicon nitride film was used as a matrix material to embed a-Si QDs. A silicon nitride film is expected to be better than silicon oxide in optical device applications because carriers can be easily transported to the a-Si QDs in the silicon nitride matrix due to the lower tunneling barriers for electrons and holes of silicon nitride than those of silicon oxide. [11] We report here on the growth and size control of a-Si QDs embedded in the silicon nitride film by plasma enhanced (PE) CVD. Figure 1 shows the Raman spectrum, that can be separated into two components. One peak in the vicinity of 460 cm ±1 is assigned to the asymmetric Si±N bond stretching mode in the silicon nitride matrix, and the other peak in the vicinity of 490 cm ±1 corresponds to scattering with localized optical phonons that exist in silicon QDs surrounded by nitrogen atoms. [19] Compared to those from bulk materials, these two peaks are also shifted to the highenergy side by about 10 cm ±1 , due to the compressive strain normally p...