Correlation between matrix structural order and compressive stress exerted on silicon nanocrystals embedded in silicon-rich silicon oxide
© Zatryb et al; licensee Springer. 2013
Received: 8 October 2012
Accepted: 19 November 2012
Published: 21 January 2013
Silicon nanocrystals embedded in a silicon oxide matrix were deposited by radio frequency reactive magnetron sputtering. By means of Raman spectroscopy, we have found that a compressive stress is exerted on the silicon nanocrystal cores. The stress varies as a function of silicon concentration in the silicon-rich silicon oxide layers varies, which can be attributed to changes of nanocrystal environment. By conducting the Fourier transform infrared absorption experiments, we have correlated the stresses exerted on the nanocrystal core to the degree of matrix structural order.
78.67.Bf, 78.67.Pt, 73.63.Bd, 78.47.D, 74.25.Nd
KeywordsSilicon Nanocrystals Stress Raman Phonon Confinement Order Disorder Matrix
Silicon nanocrystals (Si-NCs) embedded in a silicon-rich silicon oxide (SRSO) have been extensively studied due to their promising applications in the third generation tandem solar cells , light-emitting diodes , or silicon-based lasers . These SRSO structures may be successfully deposited by magnetron sputtering technique. After deposition, during annealing in a N2 atmosphere and 1,100°C temperature, the excess silicon in SRSO layer precipitates to form Si nanocrystals in nearly stoichiometric silicon dioxide matrix.
The structural quality of the matrix surrounding Si-NCs is very important since it influences the optical properties of Si-NCs . For example, it has been shown that various defects present in the matrix may quench the emission originated from Si-NCs due to non-radiative recombination . This is a serious problem from the point of view of applications, especially in the case of light-emitting devices. Besides the optical properties, due to differences in Si-NCs and SiO2 crystal structure, the matrix structural ordering may affect also the Si-NCs crystallinity and shape. It has been shown by first-principles calculations that the surrounding matrix always produces a strain on the nanocrystals, especially at the Si-NCs/SiO2 interface. According to theory, the amount of stress exerted on the nanocrystal is connected to the Si-NCs size  as well as to the number of oxygen per interface silicon . These structural parameters can be controlled during deposition process by varying the excess silicon concentration in the SRSO matrix . The structural properties of the Si-NCs may be then experimentally examined by means of the Raman spectroscopy, since the Si-Si bonding is Raman active. On the other hand, Si-O-Si bonds are active in the infrared (IR) region and therefore the matrix properties can be examined by means of the Fourier transform IR (FTIR) spectroscopy. In this work, we investigate the correlation between short-range structural order of the matrix and stress exerted on the Si-NCs by means of the Raman and FTIR spectroscopy. Our results indicate that there is a strong dependence of stress on the Si-NCs size and on the degree of short-range structural order of the matrix. We conclude that from the point of view of applications, a compromise has to be considered between good structural quality of the matrix and Si-NCs size.
The SRSO films with a nominal thickness of 500 nm used for this study were deposited onto the quartz substrates by radio frequency reactive magnetron sputtering. The incorporation of Si excess was monitored through the variation of the hydrogen rate rH = PH2 / (PAr + PH2). In this work we examined three samples deposited with rH value equal to 10%, 30%, and 50%. The films were deposited without any intentional heating of the substrates and with a power density of 0.75 W/cm2. More details on the process can be found elsewhere . All samples were subsequently annealed at 1,100°C for 1 h under N2 flux in order to favor the precipitation of Si excess and to induce Si-NCs formation.
The room temperature micro-Raman scattering was measured in vertical transmit, horizontal receive polarization, using single-stage spectrometer (T64000 Horiba Jobin Yvon, Longjumeau, France) equipped with a silicon charge-coupled device camera. An Ar+ laser (λ = 514.5 nm) was used as the excitation source. The lack of noticeable heating of the samples was assured by determination of the Stokes/anti-Stokes ratio. The FTIR spectra were collected using Nicolet iS10 spectrometer (Thermo Fisher Scientific Instruments, PA, USA). These measurements were conducted in attenuated total reflectance mode (ATR) using VariGATR accessory (Harrick Scientific Products Inc, NY, USA).
Results and discussion
In our previous papers [9, 10] we have reported results of structural investigations (including atomic force microscopy, X-ray diffraction, high-resolution electron microscopy or Rutherford backscattering) of SRSO films fabricated with the same technological parameters as the samples examined in the present study. The main conclusion of these investigations is that the deposition with rH = 10% favors the formation of well-crystallized Si-NCs with average size of about 3 nm, whereas deposition with rH = 50% favors formation of Si-NCs with size less than 2 nm. We have also shown that an increase of rH results in a drop of the crystalline fraction of nanoclusters.
The optical band gap ( E g ) and Urbach energy ( E U ) determined for the investigated samples
Eg(eV) (m= 1/2)
Inset in Figure 1 shows an example of the fit obtained for rH = 50% sample. It can be seen that the PC model accounts for the asymmetric shape of the Raman band of Si-NCs. This asymmetric shape is a result of a finite nanocrystals volume, which allows phonons away from the Brillouin zone center to contribute to the Raman scattering. Therefore, during the fitting procedure, we rely on two factors that directly depend on the Si-NCs size: the line-shape of the Raman band and the expected frequency of this band.
From the fit of Equation 4 to the Raman data, we obtained that the Si-NCs diameter d increases from about 2.4 nm for rH = 50% to about 2.7 nm for rH = 10% (the statistical error from the fitting procedure is less than 0.05 nm). The obtained results are in agreement with our expectations based on the structural data measured for similar samples. This result also confirms that the model given by Equation 1 can be used to estimate the Si-NCs size based on the Raman data.
The second important result obtained from the fit is ωc. For the unstressed Si crystal, this value equals to 520 cm−1. However, we obtained that ωc increases as a function of rH increases, from ωc = 523 cm−1 for rH = 10% to ωc = 525 cm−1 for rH = 50%. The obtained values strongly indicate that we deal with a compressive stress exerted on the Si-NCs which shifts the observed Raman lines towards higher wavenumbers . Similar effect has been observed for Si-NCs obtained by chemical vapor deposition technique and annealed at 1,250°C . Moreover, the observed rise of ωc indicates that the stress increases as a function of rH. Assuming that the hydrostatic pressure of about 1 GPa results in approximately 1.88 cm−1 shift of the Raman line , we may estimate the maximum stress to be about 2.6 GPa for rH = 50% sample. The obtained results also explain why we do not observe a clear downshift of the Raman frequency related to PC effect. Namely, the compressive stress increases as a function of rH and compensates for the downshift due to the finite crystallite size. It is worth to note that PC effect has been actually observed for Si-NCs synthesized in the form of free-standing powder . Therefore, the difficulties related to the observation of this effect in our case seem to be matrix-related. It should be also noted here that the obtained values of ωc do not strongly depend on the PC model selection. To check this, we fitted the HF Raman band with another PC model proposed by Campbell et al.  (with a Gaussian weighting function instead of sinc). Although this model predicted overestimated Si-NCs sizes (4 nm for rH = 50% and 5 nm for rH = 10%), the obtained values of ωc were similar (ωc = 523 cm−1 for rH = 10% and ωc = 524 cm−1 for rH = 50%). It should also be mentioned that both models are simplified since they do not take into account such effects as stress distribution or Si-NCs size distribution. Therefore, the estimated stress values should be treated as estimation.
where the intensities ISi-NC and IA are defined as integrals over ω of Equations 1 and 3, respectively. We prefer to calculate the relative contributions instead of the absolute amorphous and crystalline fractions since, as shown by Ossadnik et al. , the Raman-based estimates of the latter can be very inaccurate.
Figure 2b shows the integrated Raman intensities of Si-NCs and a-Si bands as a function of absorption coefficient (α). The absorption coefficient was determined at 4 eV (high-energy part of the absorption spectra). It can be seen that there is a linear correlation between α and the Raman intensity for Si-NCs as well as a-Si, with correlation coefficient equal to 0.98 and 0.97, respectively. Since both the Raman intensity and α depend linearly on the number of nanoparticles (e.g., Si-NCs), the obtained correlation indicates that the high-energy absorption is related to both: Si-NCs and a-Si. It should be also noted here that we obtained a strong correlation for the whole high-energy part of the absorption spectra (between 3 and 5 eV). Moreover, the correlation coefficient calculated for Si-NCs was always slightly higher than for a-Si. On the other hand, when energy drops approximately below 2.5 eV, the correlation coefficient also drops below 0.7. This result can be expected if we bear in mind that the estimated optical band gap exceeds 2.5 eV for all of the investigated structures (see Table 1). This result may also indicate that the low-energy part of the absorption spectra is (at least partially) related to some different structures e.g., defects in the matrix.
Figure 3 one can also see that the spectra of the samples deposited with excess silicon are much broader in comparison with the IR spectra of pure quartz. Moreover, the decrease of the rH used during deposition leads to a significant broadening of the IR spectra. This effect can be related to the lowering of the degree of matrix structural order. It should be emphasized here that we consider a short-range order since the matrix is non-crystalline. A more quantitative definition of the matrix structural disorder is discussed below.
where Δωi is the change of the ωi mode angular frequency, Δθ is the variation of the angle θ, γ is a constant, α is a bond force constant and m x denotes element mass. In this work we relate the structural disorder to a spread in θ and a wide distribution of n in the n-fold rings. This approach is clearly oversimplified since it does not account for the appearance of new modes induced by the disorder , which actually exist in an amorphous SiO2. Nevertheless, the above model enables us to understand the obtained results at least qualitatively and relate the observed broadening of the IR spectra to increase structural disorder of the matrix. This means that the siloxane rings structure is more diversified in the case of rH = 10% samples, with various ring orders n and a large spread in the intertetrahedral angle θ.
We would like to note that there is a correlation between the structural order of the matrix and the magnitude of the compressive stress exerted on Si-NCs. Namely, the stress is higher when the structural order of the matrix increases. Although several explanations of the compressive stress exerted on Si-NCs in SRSO matrix have been proposed [19, 28], we have not found any explanation which takes this effect into consideration. Here, we would like to suggest another possible origin of the compressive stress that accounts also for the observed correlation of the compressive stress magnitude on the structural order of the matrix.
Before we discuss this effect, we would like to note that after crystallization of a melted silicon nanoparticle, its volume increases by about 10% . This is rather not typical behavior, related to the fact that silicon has greater density in the liquid state than in the solid state. Therefore, the phase-transition from liquid to crystalline state should lead to a compressive stress, when Si-NCs are embedded in a SiO2 matrix, despite the different thermal expansion coefficients of Si and SiO2. This also means that the compressive stress observed in our experiment may be indicative of the crystallization process, which proceeds through melting. This kind of process has been suggested by Hirasawa et al.  for Si nanoparticles synthesized by pulsed laser ablation, where the determined crystallization temperatures were in the range of 800 to 1,300 K (depending on the nanoparticle size). These temperatures are far below the melting point of bulk Si (1,683 K). In our case, the annealing temperature of 1,373 K is also well below the melting point of bulk Si and only slightly below the melting point of a-Si (1,420 K for relaxed a-Si ). However, it is well known that the melting temperature of a nanoparticle decreases significantly with size, as a consequence of the additional free energy contribution of the surface to the overall Gibbs free energy . For example, it has been shown that free-standing Si nanoparticles with a size of 20 nm melt at around 1,000 K . On the other hand, nanoparticles embedded in a matrix can exhibit both melting-point depression and enhancement , and the actual melting behavior depends on the nature of the interface between the nanoparticle and the matrix. It has been found that when the interface between the nanoparticle and the matrix is coherent, the thermal vibration of the surface (interface) atoms of the nanoparticle is suppressed. This suppression may prevent the melting of the nanocrystals’ surface and lead to an increase of the melting temperature. This kind of behavior has been found for lead nanocrystals in an aluminum matrix and was attributed to the lattice structures of the two crystals ‘locking up’, suppressing the vibration of the nanoparticles’ surface atoms . Contrary to this, irregularly shaped and incoherent interfaces can be directly correlated with lowering of melting temperature of a nanoparticle .
In the investigated case, we expect that directly after deposition we deal with amorphous Si nanoparticles embedded in a disordered oxide matrix. Moreover, it is improbable that the sputtering technique allows deposit of coherent (epitaxial) interfaces between the amorphous nanoparticles and the matrix. Due to a large density gradient of the Si nanoparticles and the oxide host, when merged at their interface, the network topologies in either side deform in order to accommodate the transition . Therefore, we expect the interfaces between Si nanoparticles and the matrix to be incoherent. This can be further supported by the latest findings of molecular dynamics simulations which have shown that the interface structure between Si-NCs and the matrix is generally highly porous on the silica side, making the contact with the Si-NCs discontinuous . Taking this into account, we expect that the melting temperature of small, amorphous Si nanoparticles embedded in SRSO matrix might be depressed below the melting point of a-Si. If this is the case, melting of the nanoparticles may be possible at 1,100°C.
Having this in mind, we suggest the following origin of the compressive stress observed in our experiment. First of all, it should be emphasized that the matrix structure before and after annealing is not the same. Annealing at 1,100°C leads to phase separation on Si and SiO2 and the structural order of the matrix increases. Secondly, the crystallization of small a-Si nanoparticles takes place simultaneously to the matrix ordering. We suggest that for non-uniform structures obtained by sputtering, the crystallization may proceed through melting which in turn leads to volume expansion and compressive stress exerted on the Si-NC. Moreover, we may expect that the ability of Si-NCs to expand after crystallization should depend on the environment - particularly, on the degree of the structural order of the matrix (since expansion of the nanocrystal leads to matrix deformation). In other words, the matrix structure determines its ability to accommodate to the expanding Si-NCs. In this way, formation of a well-ordered matrix does not allow Si-NCs to expand freely, leading to a stronger compressive stress exerted on the Si-NCs. We deal with this situation for rH = 50%, where the compressive stress is the strongest and the FTIR spectra are quite narrow, suggesting a higher structural order of the matrix than for the other samples. On the other hand, for larger Si-NCs (rH = 10%), the structural order of the matrix is the lowest, resulting in a broad IR spectrum. This structural disorder indicates that the matrix can accommodate to the Si-NCs size/shape; therefore, compressive stress exerted on the Si-NCs is lowered.
Remarkably, the IR spectrum of pure quartz is much narrower than the spectra of the samples containing Si-NCs. It means that Si-NCs always introduce a large amount of the structural disorder to the matrix which may influence also the optical properties. This problem should be taken into account while designing structures for a particular application.
In conclusion, we have shown that compressive stress is exerted on Si-NCs in SRSO samples deposited by radio frequency reactive magnetron sputtering. This stress may completely compensate for the phonon quantum confinement effects, resulting in the lack of a clear dependence of the Si-NCs-originated Raman line on the Si-NCs size. The compressive stress increases with the increasing rH used during deposition. We relate the observed strong stress dependence on rH to the changes of structural order of the matrix surrounding Si-NCs induced by rH variation. The formation of an ordered matrix structure clearly competes with the formation of unstressed Si-NCs.
GZ would like to acknowledge for financial support to Program Iuventus Plus (no. IP2011 063471). In this work, the Raman spectra measurements were conducted as a part of the NLTK project (POIG. 02.02.00-00-003/08-00). This research was conducted as part of the Polonium program.
- Cho EC, Green MA, Conibeer G, Song D, Cho YH, Scardera G, Huang S, Park S, Hao XJ, Huang Y, Van Dao L: Silicon quantum dots in a dielectric matrix for all-silicon tandem solar cells. Adv Optoelectron 2007, 2007: 1–11.View Article
- Huh C, Kim K, Kim BK, Kim W, Ko H, Choi C, Sung GY: Enhancement in light emission efficiency of a silicon nanocrystal light emitting diode by multiple luminescent structures. Adv Mater 2010, 22: 5058–5062. 10.1002/adma.201001572View Article
- Pavesi L, Dal Negro L, Mazzoleni C, Franzo G, Priolo F: Optical gain in silicon nanocrystals. Nature 2000, 408: 440–444. 10.1038/35044012View Article
- Zatryb G, Podhorodecki A, Hao XJ, Misiewicz J, Shen YS, Green MA: Correlation between stress and carrier nonradiative recombination for silicon nanocrystals in an oxide matrix. Nanotechnology 2011, 22: 335703. 10.1088/0957-4484/22/33/335703View Article
- Zatryb G, Podhorodecki A, Hao XJ, Misiewicz J, Shen YS, Green MA: Quantitative evaluation of boron-induced disorder in multilayers containing silicon nanocrystals in an oxide matrix designed for photovoltaic applications. Opt Express 2010, 18: 22004–22009. 10.1364/OE.18.022004View Article
- Hadjisavvas G, Remediakis IN, Kelires PC: Shape and faceting of Si nanocrystals embedded in a-SiO2: a Monte Carlo study. Phys Rev B 2006, 74: 165419.View Article
- Guerra R, Degoli E, Ossicini S: Size, oxidation, and strain in small Si/SiO nanocrystals. Phys Rev B 2009, 80: 155332.View Article
- Podhorodecki A, Zatryb G, Misiewicz J, Wojcik J, Mascher P: Influence of the annealing temperature and silicon concentration on the absorption and emission properties of Si nanocrystals. J Appl Phys 2007, 102: 043104–043105. 10.1063/1.2772501View Article
- Ternon C, Gourbilleau F, Portier X, Voivenel P, Dufour C: An original approach for the fabrication of Si/SiO2 multilayers using reactive magnetron sputtering. Thin Sol Film 2002, 419: 5–10. 10.1016/S0040-6090(02)00294-8View Article
- Gourbilleau F, Levalois M, Dufour C, Vicens J, Rizk R: Optimized conditions for an enhanced coupling rate between Er ions and Si nanoclusters for an improved 1.54-μm emission. J Appl Phys 2004, 95: 3717–3722. 10.1063/1.1655680View Article
- Zatryb G, Podhorodecki A, Misiewicz J, Cardin J, Gourbilleau F: On the nature of the stretched exponential photoluminescence decay for silicon nanocrystals. Nanoscale Res Lett 2011, 6: 106. 10.1186/1556-276X-6-106View Article
- Podhorodecki A, Misiewicz J, Gourbilleau F, Rizk R: Absorption mechanisms of silicon nanocrystals in cosputtered silicon-rich-silicon oxide films. Electrochem Solid-State Lett 2008, 11: K31-K33. 10.1149/1.2828207View Article
- Khriachtchev L, Kilpelä O, Karirinne S, Keränen J, Lepisto T: Substrate-dependent crystallization and enhancement of visible photoluminescence in thermal annealing of Si/SiO2 superlattices. Appl Phys Lett 2001, 78: 323. 10.1063/1.1341227View Article
- Khriachtchev L, Räsänen M, Novikov S, Pavesi L: Systematic correlation between Raman spectra, photoluminescence intensity, and absorption coefficient of silica layers containing Si nanocrystals. Appl Phys Lett 2004, 85: 1511. 10.1063/1.1781733View Article
- Campbell IH, Fauchet PM: The effects of microcrystal size and shape on the one phonon Raman spectra of crystalline semiconductors. Solid State Commun 1986, 58: 739–741. 10.1016/0038-1098(86)90513-2View Article
- Paillard V, Puech P, Laguna MA, Carles R, Kohn B, Huisken F: Improved one-phonon confinement model for an accurate size determination of silicon nanocrystals. J Appl Phys 1999, 86: 1921–1924. 10.1063/1.370988View Article
- Zi J, Zhang K, Xie X: Comparison of models for Raman spectra of Si nanocrystals. Phys Rev B 1997, 55: 9263. 10.1103/PhysRevB.55.9263View Article
- Manotas S, Agulló-Rueda F, Moreno JD, Ben-Hander F, Martínez-Duart JM: Lattice-mismatch induced-stress in porous silicon films. Thin Sol Film 2001, 401: 306–309. 10.1016/S0040-6090(01)01641-8View Article
- Hernández S, Martínez A, Pellegrino P, Lebour Y, Garrido B, Jordana E, Fedeli JM: Silicon nanocluster crystallization in SiO x films studied by Raman scattering. J Appl Phys 2008, 104: 044304. 10.1063/1.2968244View Article
- Anastassakis E, Cantarero A, Cardona M: Piezo-Raman measurements and anharmonic parameters in silicon and diamond. Phys Rev B 1990, 41: 7529–7535. 10.1103/PhysRevB.41.7529View Article
- Hessel CM, Wei J, Reid D, Fujii H, Downer MC, Korgel BA: Raman spectroscopy of oxide-embedded and ligand-stabilized silicon nanocrystals. J Phys Chem Lett 2012, 3: 1089–1093. 10.1021/jz300309nView Article
- Ossadnik C, Vepřek S, Gregora I: Applicability of Raman scattering for the characterization of nanocrystalline silicon. Thin Sol Film 1999, 337: 148–151. 10.1016/S0040-6090(98)01175-4View Article
- Aguiar H, Serra J, González P, León B: Structural study of sol–gel silicate glasses by IR and Raman spectroscopies. J Non-Cryst Solids 2009, 355: 475–480. 10.1016/j.jnoncrysol.2009.01.010View Article
- Awazu K, Kawazoe H: Strained Si-O-Si bonds in amorphous SiO2 materials: a family member of active centers in radio, photo, and chemical responses. J Appl Phys 2003, 94: 6243–6262. 10.1063/1.1618351View Article
- Galeener FL, Thorpe MF: Rings in central-force network dynamics. Phys Rev B 1983, 28: 5802–5813. 10.1103/PhysRevB.28.5802View Article
- Galeener FL: Band limits and the vibrational spectra of tetrahedral glasses. Phys Rev B 1979, 19: 4292–4297. 10.1103/PhysRevB.19.4292View Article
- Kirk CT: Quantitative analysis of the effect of disorder-induced mode coupling on infrared absorption in silica. Phys Rev B 1988, 38: 1255. 10.1103/PhysRevB.38.1255View Article
- Crowe IF, Halsall MP, Hulko O, Knights AP, Gwilliam RM, Wojdak M, Kenyon AJ: Probing the phonon confinement in ultrasmall silicon nanocrystals reveals a size-dependent surface energy. J Appl Phys 2011, 109: 083534–083538. 10.1063/1.3575181View Article
- Ujihara T, Sazaki G, Fujiwara K, Usami N, Nakajima K: Physical model for the evaluation of solid–liquid interfacial tension in silicon. J Appl Phys 2001, 90: 750–755. 10.1063/1.1379349View Article
- Hirasawa M, Orii T, Seto T: Size-dependent crystallization of Si nanoparticles. Appl Phys Lett 2006, 88: 093119. 10.1063/1.2182018View Article
- Grimaldi MG, Baeri P, Malvezzi MA: Melting temperature of unrelaxed amorphous silicon. Phys Rev B 1991, 44: 1546–1553. 10.1103/PhysRevB.44.1546View Article
- Schierning G, Theissmann R, Wiggers H, Sudfeld D, Ebbers A, Franke D, Witusiewicz VT, Apel M: Microcrystalline silicon formation by silicon nanoparticles. J Appl Phys 2008, 103: 084305. 10.1063/1.2903908View Article
- Jiang Q, Zhang Z, Li J: Melting thermodynamics of nanocrystals embedded in a matrix. Acta Mater 2000, 48: 4791–4795. 10.1016/S1359-6454(00)00271-8View Article
- Xu Q, Sharp ID, Yuan CW, Yi DO, Liao CY, Glaeser AM, Minor AM, Beeman JW, Ridgway MC, Kluth P, Ager JW, Chrzan DC, Haller EE: Large melting-point hysteresis of Ge nanocrystals embedded in SiO2. Phys Rev Lett 2006, 97: 155701.View Article
- Singh A, Tsai AP: Melting behaviour of lead and bismuth nano-particles in quasicrystalline matrix - the role of interfaces. Sadhana 2003, 28: 63–80. 10.1007/BF02717126View Article
- Hadjisavvas G, Kelires PC: Structure and energetics of Si nanocrystals embedded in a-SiO2. Phys Rev Lett 2004, 93: 226104.View Article
- Soulairol R, Cleri F: Interface structure of silicon nanocrystals embedded in an amorphous silica matrix. Solid State Sci 2010, 12: 163–171. 10.1016/j.solidstatesciences.2009.05.004View Article
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.