Synthesis and Luminescent Properties of Silicon Nanocrystals

Nowadays, study of silicon-based visible light-emitting devices has increased due to large-scale microelectronic integration. Since then different physical and chemical processes have been performed to convert bulk silicon (Si) into a light-emitting material. From discovery of Photoluminescence (PL) in porous Silicon by Canham, a new field of research was opened in optical properties of the Si nanocrystals (Si-NCs) embedded in a dielectric matrix, such as SRO (silicon-rich oxide) and SRN (silicon-rich nitride). In this respect, SRO films obtained by sputtering technique have proved to be an option for light-emitting capacitors (LECs). For the synthesis of SRO films, growth parameters should be considered; Si-excess, growth temperature and annealing temperature. Such parameters affect generation of radiative defects, distribution of Si-NCs and luminescent properties. In this chapter, we report synthesis, structural and luminescent properties of SRO monolayers and SRO/SiO2 multilayers (MLs) obtained by sputtering technique modifying Si-excess, thickness and thermal treatments.


Introduction
The use of photonic signals instead of electrons to transmit information through an electronic circuit is an actual challenge. Unfortunately, it is well known that bulk silicon (Si) is an indirect bandgap semiconductor, making it an inefficient light emitter. Therefore, great efforts have been taken to obtain highly luminescent Si-based materials in order to get Si-based photonic devices, especially a light-emitting device [1][2][3]. Such circumstances have led to explore new options for converting silicon into a luminescent material. Si-NCs embedded in a dielectric material as silicon-rich oxide (SRO) or silicon-rich nitride (SRN) show a prominent photoluminescence (PL) emission in red and blue-green region, respectively [4,5]. Thus, SRO films have been considered as promising candidates for potential applications in Si-based optoelectronic devices, and their fully compatibility with the complementary metal-oxide-semiconductor (CMOS) processes. At present, different techniques or methods have been employed to produce SRO films or SRO/SiO 2 MLs [6][7][8][9][10][11][12][13]. In such structures, the SRO-thickness, Si-excess and annealing temperature are parameters that promote the formation of Si-NCs and radiative defects, affecting the optical and electrical properties of SRO monolayers or SRO/SiO 2 ML.
In this chapter, results about of Si-NCs embedded at SRO monolayer and SRO/SiO 2 MLs deposited by the RF Co-sputtering method as a function of Si-excess (5.2-14.3 at.%) and modulating the SRO-thickness layer (2.5-7.5 nm) of MLs are shown. Both, SRO monolayers and SRO/SiO 2 MLs have a broad emission band in the orange-red region (1.45-2.3 eV). Nevertheless, the SRO/SiO 2 MLs emit a stronger PL intensity compared with SRO monolayers. The most intense PL emission is observed in MLs when the SRO-thickness is 5 nm, and with the highest Si-excess (14.3 at.%), which is important for the design of electroluminescent devices with low threshold voltage. Although multilayer structures with good control in crystal size have been studied [14], in this chapter optical and structural properties are analyzed in detail. Furthermore a comprehensive study of synthesis of Si-NCs in SRO and SRO/SiO 2 MLs as a function of Si-excess (5.2-14.3 at.%) and annealing temperature is presented.

Experimental procedure
In this chapter, structural and optical properties of co-sputtered SRO monolayers and SRO/ SiO 2 MLs are obtained and studied as a function of the annealing temperature and the change in the Si-excess. Both structures were deposited by the co-sputtering of Si (2", purity of 99.999 %) and SiO 2 (2", purity of 99.99%) targets using a Torr International magnetron sputtering system (13.56 MHz). Both structures were deposited on p-type (100) Si wafers with resistivity of 2-5 Ω-cm. Before deposition, Si substrates were cleaned in ultrasonic bath with acetone, ethanol, and deionized water successively. After being dried with nitrogen, the substrates were immediately loaded into the chamber of the sputtering system. Once a base pressure of ~1 × 10 −6 Torr is achieved, Ar flow of 60 sccm is introduced into the chamber at a working pressure of 2.4 mTorr.
The SRO monolayers were deposited by the simultaneous co-sputtering of Si and fused quartz (SiO 2 ) targets. The Si-excess content in SRO monolayers was modified by changing the RF power applied to the Si-target (PSi) from 40 to 80 W maintaining constant the RF power applied to the SiO 2 target ( P ) was 5.09 W/cm 2 . For SRO/SiO 2 MLs, SRO interlayers with the same Si-excess that SRO monolayers were used. First, a SiO 2 layer was deposited onto the silicon substrate followed by a SRO film to obtain a SRO/SiO 2 bi-layer. Ten periods of SRO/SiO 2 bi-layers were deposited with an additional (upper) SiO 2 film (10 nm) to avoid oxidization during high temperature annealing. Each SiO 2 layer was about 6 nm in thick while the SRO layer thickness was modified from 2.5 to 7.5 nm (see Table 2).
All samples were deposited at 100°C. However, Si-NCs in SRO monolayers and SRO/SiO 2 MLs were produce by thermal annealing in a conventional tube furnace between 900 and 1200°C in N 2 environment for 2 h. Thickness of the films was measured by reflectance using a Filmetrics F20UV equipment. The chemical compositions and Si-excess of samples were analyzed by a Thermo Scientific X-ray photoelectron spectroscopy (XPS) Escalab 250Xi equipment. The Si oxide phase was studied by Fourier transform Infra-red (FTIR) spectroscopy using a Bruker Vector 22 spectrometer in the range 400-4000 cm −1 . The PL emission spectra were measured with a Horiba Fluoromax 3 system. The samples were excited using a 300 (4.13 eV) nm radiation and the PL emission signal was collected from 400 to 900 nm (1.37-3.1 eV) with a resolution of 1 nm.

Deposition rate Si and SiO 2 films
The first step to synthesize silicon nanocrystals embedded in SRO films by Co-sputtering is to consider the deposition rate of Si and SiO 2 films. Such procedure allows to obtain RF powers appropriate to Si and SiO 2 targets and silicon content suitable for SRO films. Some experimental studies report RF powers applied to the Si higher than SiO 2 target [15,16]. However, it is known that the deposition rate of SiO 2 is low (0.1Ǻ/sec) and high silicon content could be present in SRO films if the deposition rate of Si is much higher than the SiO 2 film. This effect is shown in Figure 1; where Si and SiO 2 films were individually deposited for 1200 s using Si and SiO 2 targets, respectively. Figure 1 shows the deposition rate and thickness of Si and SiO 2 films as function of the RF-power density (PD) applied to Si and SiO 2 targets. It can be observed that the deposition rate for SiO 2 is lower (0.15 Å/s) at 100 W while that Si films (0.12 Å/s) at 50 W, where similar thickness (~16-18 nm) is deposited with a PD Si = 2.47 W/cm 2 for a Si film and PD SiO 2 = 5.09 W/cm 2 for a SiO 2 film. Therefore, to ensure Si-excess similar to the reported by other techniques [17][18][19][20][21], SRO films were deposited considering RF-power densities between 1.97 and 3.94 W/cm 2 applied to the Si-target and a PD SiO 2 constant of 5.09 W/cm 2 . Selected the appropriate power densities, SRO monolayers and SRO/SiO 2 ML were deposited at 100°C and annealed at 1100°C.

SRO monolayers
The optimization of sub-stoichiometric silicon-rich oxide (SRO) from precise control of Si content in SRO films is an important parameter to tune the electrical and optical properties of luminescent and photovoltaic devices. In this regard, it has been reported that there might be an optimal Si-excess to find an efficient photoluminescence (PL) emission [18,19]. The PL quantum yield of SRO films (1.4 < x < 1.9) annealed at 1100°C, increase significantly for a Si content x = 1.8 (2.5 at.% of Si-excess) [18]. For this Si-excess, Si-NCs are not detected by HRTEM, however intense emission in red region was observed. It has been suggested that for a medium Si-excess (<9 at.%), oxygen-related defects, such as Si=O (nonbridged  oxygen passivation) and/or ultra-small oxidize Si-NCs (diameters <2 nm) are the most optically actives for promote an intense emission in red region (1.5 eV). Thus, a considerable Si-excess (>10 at.%) in SRO films is not necessary to obtain a maximum PL emission.
In this chapter Si-excess between 4.4 and 16.9 at.% are reported. In order to know the composition (x = O/Si) and Si-excess of SRO monolayers and SRO/SiO 2 multilayers, XPS measurements in-depth profile were performed. In Figure 2b, the analysis in-depth profile of an SRO film with 5.2 at.% of Si-excess compared to a SiO 2 film is shown. It can be seen that silicon content in SRO film increase respect to SiO 2 film. The atomic concentration of Si content in SRO film was ≈38.5 at.% respect to the SiO 2 film with an atomic concentration of ≈33.33 at.%. Such effect demonstrates Si-excess in SRO monolayers. For all SRO monolayers Si-excess increases from 4.4 to 16.9 at.% as the power density of Si-target rise from 1.97 to 3.94 W/cm 2 , respectively (see Figure 2a) The presence of Si-excess in SRO films is also detected by the oxidation states of Si obtained from Si2p-XPS data. This fact is suggested on the basis of the intermediate oxidation states detected in the Si2p-XPS spectra. According to Si-excess measured by XPS, five oxidation states were detected in samples, before and after annealing. For example for a SRO monolayer with 5.2 at.%, peaks located around 99.6 eV, 100.5, 101.5, 102.5 and 103.5 eV were found (Figure 3a). The aforementioned peaks correspond to the Si 0 (elemental Si), Si +1 (Si 2 O), Si +2 (SiO), Si +3 (Si 2 O 3 ) and Si +4 (SiO 2 ) oxidation states, respectively [22,23]. The contributions of the intermediate Si peaks (Si +1 , Si +2 , Si +3 ) corresponding to the suboxide phases become smaller when the SRO films are thermally annealed at 1100°C, while Si +4 peak for SiO 2 and Si 0 peak for elemental Si becomes dominant, especially the intensity of Si 0 peak increases (Figure 3b), explaining the Si-SiO 2 phase separation and Si-NCs formation after annealing, as shown by similar results found by Chen et al. [24].

SRO/SiO 2 MLs
For the design of SRO/SiO 2 ML structures, SRO layers with the same Si-excess that monolayers were used. In Figure 4a, the composition in-depth profile for an as-deposited SRO/SiO 2 MLs with 5 nm-thick SRO layer corresponding to 5.2 at.% and 10.2 at.% of Si-excess is shown. As we can see, different changes in atomic concentration (x) are observed corresponding to different SRO-SiO 2 interfaces as the etch time increases. A better contrast between SRO and SiO 2 stoichiometry is observed when the Si-excess and SRO-thickness is increased to 14.3 at.% and 7.5 nm, respectively (sample 3C, before and after annealing), as observed in Figure 4b. A densification in thickness is observed in this multilayer after thermal annealing and it is attributed to a microstructural reordering in SRO layers. The atomic concentrations (x) obtained in multilayers as-deposited (x = 1.37, x = 1.6) are near to SRO monolayers (x = 1.3, x = 1.5). Such difference or reduction of atomic concentration could be caused by the etch time performed by XPS. A reduction of the atomic concentration (x) in sample 3C by effect annealing is observed (Figure 4b), even this effect being more noticeable near the substrate (x = 1.3  x = 1.16). In this way, it is likely that a re-diffusion of Si atoms between SRO-SiO x interfaces as well as within SRO layers after thermal treatment is performed. Such annealing, promotes crystallization of Si clusters and Si-NCs formation between SRO-SiO x interfaces [25][26][27]. The Si-excess or Si clusters within of multiple SRO layers were detected from Si2p-XPS analysis. For this study, sample 3C was analyzed specifically in the etch time range of 520-560 s (see Figure 4b), specifically in critical points; SiOx layer (520 s), near the SRO-SiOx interfaces (544 s) and SRO layer (560 s) as is shown in Figure 5. It can be observed that for etch time at 560 (SRO) and 540 s (SRO-SiO x ), five oxidation states were detected, located around 99.6 eV, 100.5, 101.5, 102.5 and 103.5 eV (Figure 5b and c). For Si2p-XPS analysis in 520 s (SiO x ), the oxidation state Si 0 was not detected (Figure 5a), only oxidation states Si +3 and Si +4 were detected, so that Si atoms diffusion is unlikely in this region. Thereby, for an etch time at 544 and 560 s, five oxidation states were perfectly adjusted, even the Si 0 state is greater in 544 s, which could be tentatively edges SRO-SiO x interface. Therefore Si-diffusion is likely even in the SRO-SiO x interfaces and SRO nanolayers. This study evidences crystallization of silicon clusters after annealing and formation of Si-NCs in the SRO-SiO 2 interfaces and the SRO nanolayers by Si atoms diffusion. This effect was demonstrated before by HRTEM measurements [25,28].

SRO monolayer
The nature of the chemical bonds and the phase separation of Si and SiO 2 of SRO monolayers before and after thermal annealing are studied by FTIR. Figure 6a, shows FTIR spectrum of asdeposited SRO monolayers with 5.  observed at around 446-450, 800-872, and 1026-1040 cm −1 , respectively [29]. The Si-O-Si stretching band shifts toward low wavenumbers (1047-1027 cm −1 ) as the silicon content increases (from 5.2 to 14.3 at.%) in as-deposited monolayers. An additional IR band of low intensity is observed around 880 cm −1 . Such vibrational mode disappears when samples are thermally annealed, indicating a rearrangement of the SiO x network due to a diffusion of Si atoms, which promotes Si-SiO x phase separation and the formation of Si-NCs in SRO layers [25][26][27]. Figure 6b, shows the effect of annealing between 900 and 1100°C on SRO monolayer of 5.2 at.% of Si-excess, where the disappearance of band located around 880 cm −1 is observed. Likewise, a phase separation and Si-NCs formation in SRO monolayers is corroborated by the shift observed in the Si-O-Si stretching band position (1047-1080 cm −1 ) after thermal annealing [25,27]. It can be seen that for an annealing above about 1000°C, stretching peak appears at 1080 cm −1 . Such effect means that temperatures above 1000°C are sufficient for Si-SiO x phase separation and Si-NCs formation in the SRO monolayers. It is worth mentioning that an annealing temperature of 1100°C by 2 h provided better results in the structural and optical properties [25,28]. 7.5 nm (sample 3C), before and after thermal annealing (1100°C). For MLs as-deposited, the stretching peak position remains at 1055 cm −1 due to the same Si-excess, 14.3 at.%. However, intensity of the stretching peak is modified as thickness of SRO intermediate layers increase (2.5-7.5 nm), which is to be expected since the thickness of the MLs increases ( Table 2) and the number Si-O-Si bonds of these molecules increases. The shift in the Si-O-Si stretching band was also observed from 1055 to 1080 cm −1 after the thermal annealing. This effect, again, indicates a phase separation (Si-SiO x ) and therefore the Si-NCs formation within the SRO-bulk layers and possibly in the SRO-SiO 2 interfaces. By other hand, all SRO/SiO 2 MLs exhibited narrower Si-O-Si stretching bands respect to the SRO monolayers. Such effect at SRO/SiO 2 MLs can be related to a higher structural order in the SRO matrix due to a compressive stress in Si-NCs [30]. In MLs structures, the compressive stress exerted on the Si-NCs is higher unlike monolayers, where the structural order of the SRO matrix is the lowest resulting in a broad IR spectrum. Thus, it is likely that smaller nanocrystals (diameters <3 nm) must be confined in the SRO/SiO 2 MLs due to the high stress. Figure 8a shows the PL spectra of SRO monolayers thermally annealed at 1100°C for different Si-excess. It can be seen that when Si-excess increases of 4.4-16.9 at.%, the PL emission intensity is modified. The emission intensity of this band (red region) increases considerably for the SRO monolayer with 5.2 at.% thermally annealed at 1100°C. It has been reported that SRO films deposited by LPCVD or MBE with medium Si-excess (~2.5-5 at.%) produces an efficient PL emission, however Si-NCs have not been observed [17][18][19][20][21]. In our case, Si-NCs were observed from a Si-excess of 5.2 at.% (x = 1.54), as it was shown in images HRTEM [25]. The PL emission in SRO films with 14.3 and 16.9 at.% is still perceptible, however is weak due to the high silicon content, even close to SiO film (x = 1). In this case, PL intensity decreases due to an increase of Si-NCs size, larger to the Bohr radius or to annihilation of radiative centers.

SRO Monolayers
Regarding to the PL emission observed in sample whit 4.4 of Si-excess (PD Si = 1.97). A moderate emission is observed in the Figure 8a. According to the low concentration in this sample (x = 1.64), such emission is attributed to the formation of ultra-small oxidized Si-NCs as well as some point defects [18,19]. It has been suggested that for a medium Si-excess (<9 at.%), oxygen-related defects, such as Si=O (Nonbridged oxygen passivation) and/or ultra-small oxidized Si-NCs (diameters <2 nm) are the most optically active centers that promote an intense emission in red region (~1.5-1.7 eV). Accordingly, Si-NCs with diameters larger than 3 nm (Si-excess >10 at.%) are not direct light-emitting centers [18,19]. Thus, a considerable Si-excess (>10 at.%) in SRO films is not necessary to obtain a maximum PL emission.
The effect of the annealing temperature (900-1200°C) on the PL properties of the SRO monolayers also was analyzed (Figure 8b). The best PL emission intensity was obtained for the sample with Si-excess of 5.2 at.% (x = 1.54) and annealed at 1100°C. Thus, for the other annealing temperatures, PL intensity decreases, so a temperature of 1100° C is ideal for obtaining a maximum PL emission in SRO monolayers obtained by RF Co-Sputtering.

SRO/SiO 2 MLs
The maximum PL emission in the SRO monolayers is obtained with a Si-excess between 5.2 and 14.3 at.% and heat treated at 1100° C. In this way for the design of the structures MLs, Si-excess of the same order was used. Both, monolayers and multilayers, emit an intense and broad emission band in the red-orange region (1.45-2.1 eV). Figure 9b and a, shows, respectively, the PL spectra of thermally annealed SRO monolayers and SRO/SiO 2 MLs with a SROthickness of 5 nm with different Si-excesses [28]. In this case it was found that the MLs with a thickness of 5 nm in the SRO layer, has a maximum of PL emission independent of the Si-excess. For this, an analysis of PL intensities was made considering the ratio of maximum peaks of intensities of the SRO/SiO 2 MLs and SRO monolayer. Such analysis to determine the MLs with greater emission as function of Si-excess and SRO-thickness layer. The ratio of emission intensities was calculated taking the maximum PL intensity of each peak (~1.72, ~1.6 and ~1.52 eV) that form the PL spectrum of each SRO/SiO 2 ML (I PL ML) divided by the maximum PL intensity of SRO monolayers (I PL Mono) located at the same energy [28]. So, the most intense photoluminescence is achieved using SRO layers with 5 nm thick and 14 at.% of Si-excess. Such effect is important for the design of electroluminescent devices, particularly for supplying low voltages, since a large number of Si-NCs are generated as Si-excess increases.
The effect of confining Si-NCs in the SRO/SiO 2 MLs is to obtain nanocrystals sizes <3 nm due to the high stress, result in Si=O defects formation. Regarding the presence of Si-NCs and their PL properties, Wolkin et al. suggested that light emission in red-orange region originates from Si=O bonds, where oxidation can stabilize the energy position of the PL peak (~1.5-1.7 eV) from small Si-NCs (<3 nm) [31]. The model of Si=O bonds has been also used to explain the PL emission in connection to Si-NCs in a SiO 2 matrix [32]. It was suggested later that the lightemitting centers can be stabilized not only at Si-NCs surface but also in a disordered network [33], which could be promoted from the phase separation of Si and SiO 2 . The emission of the Si=O defect is around 1.5-2.1 eV and it strongly affect the PL intensity when the nanocrystal size is below 3 nm [31].

Si-NCs in SRO monolayer and SRO/SiO 2 MLs
The formation of Si-NCs in SRO monolayers and SRO/SiO 2 multilayers was demonstrated by HRTEM analysis [25,28]. A reduction in nanocrystal size in SRO/SiO 2 MLs respect to SRO monolayers is shown in Figure 10a. In both, SRO/SiO 2 MLs and SRO monolayers, an increase in the average nanocrystal size is observed as the Si-excess increases. However, in the ML structures, the Si-NCs tend to be confined due to the stress exerted by the SiO x matrix, so that the average nanocrystal size decreases in the MLs compared to the SRO monolayers with the same Si-excess. The increase in size of the Si-NCs according to the Si-excess is expected due to high Si content leads an enhanced in the agglomeration of Si clusters after thermal annealing. It can be deduced that a reduction in Si-NCs size below 3 nm promotes a high PL emission in both SRO monolayers and SRO/SiO 2 multilayers. Recent studies about SRO films confined by two SiO 2 layers have shown that Si atoms in excess from the SRO films mainly diffuse toward the center of the SRO layer, enhancing the Si atom aggregation and thus the Si-NCs formation [8]. Nevertheless, as observed in the XPS depth profile from this work, the Si-excess in the SRO films from SRO/SiO 2 MLs reduces as compared to SRO monolayers. This effect is related with the Si-diffusion from the SRO films toward the SiO 2 layer making them silicon rich instead of stoichiometric films. This Si-diffusion produces three important effects: (i) the Si-NCs formation near the edge of the SRO-SiO 2 interfaces, (ii) the reduction of the Si-NCs size as compared to SRO monolayers and (iii) and as a consequence, the increased Si-NCs density and improved PL intensity.
The phenomenon of light emission in SRO monolayers can be attributed to the presence and complete activation of Si=O defects especially in SRO films with 5.2 at.% where the average Si-NC size is about 1.68 ± 0.2 nm, as reported in other work [25]. In SRO/SiO 2 MLs, Si-NCs with diameters below 3 nm are necessary to obtain the activation of the Si=O luminescence centers. Therefore, SRO/SiO 2 MLs with SRO-5 nm where Si-NC sizes of about 1.7 ± 0.1, 2.1 ± 0.08 and 2.8 ± 0.09 nm were obtained with Si-excess of 5.2, 10.2 and 14.9 at.%, respectively [28]. The PL intensity for multilayers with SRO-thickness of 5 nm remains almost at the same order as the Si-excess increases, as shown in Figure 10b. However, the PL intensity of SRO/SiO 2 MLs with SRO-2.5 nm and SRO-7.5 nm and SRO monolayers decreases as the Si-excess increases. It was observed that the Si-NCs density in SRO/SiO 2 MLs is higher than SRO monolayers produced by a better control of the NCs formation in SRO nanolayers confined between SiO 2 layers. Moreover, Si-NCs with sizes below 3 nm and a large number of Si=O defects are created in SRO/SiO 2 MLs with SRO-thickness of 5 nm. These effects explain the enhanced PL intensity of multilayers making them suitable for the development of silicon-based light sources.

Conclusion
In the synthesis of Si-NCs by RF Co-sputtering in SRO monolayers or SRO/SiO 2 multilayers, it is important to consider parameters such as Si-excess, annealing temperature and SROthickness layer for SRO/SiO 2 MLs design. These parameters are important for the obtaining of a high emission either in SRO monolayers or SRO/SiO 2 MLs. It can be deduced that a high emission in both; SRO monolayers and SRO MLs is obtained for a medium Si-excess (5.2 at%). However, an important parameter to consider is the Si nanocrystal size. That is, while synthesis of Si-NCs is below 3 nm, a complete activation of Si=O defects are guaranteed. In this way it is possible to obtain a high emission in multilayers with a high Si content. For MLs where the intermediate SRO layer is 5 nm and Si-excess of 14.3 at%, a high emission is obtained due to the confinement of nanocrystals less than 3 nm. Such effect is important for the design of electroluminescent devices, particularly for supplying low voltages, since a large number of Si-NCs are generated as Si-excess increases.