- Email: [email protected]
The influence of annealing temperature on the synthesis of silicon quantum dots embedded in hydrogenated amorphous Si-rich silicon carbide matrix
Journal of Non-Crystalline Solids 441 (2016) 10–15
Contents lists available at ScienceDirect
Journal of Non-Crystalline Solids journal homepage: www.elsevier.com/locate/jnoncrysol
The influence of annealing temperature on the synthesis of silicon quantum dots embedded in hydrogenated amorphous Si-rich silicon carbide matrix Guozhi Wen a,⁎, Xiangbin Zeng b, Xianghu Li a a b
School of Electronic and Electrical Engineering, Wuhan Polytechnic University, Wuhan, Hubei 430023, China School of Optical and Electronic Information, Huazhong University of Science and Technology, Wuhan, Hubei 430074, China
a r t i c l e
i n f o
Article history: Received 13 January 2016 Received in revised form 3 March 2016 Accepted 10 March 2016 Available online xxxx Keywords: Silicon Quantum dots Bonding configuration Temperature PECVD
a b s t r a c t Hydrogenated amorphous silicon carbide thin films (a-SiC:H) were prepared by plasma-enhanced chemical vapor deposition (PECVD) and thermal annealed at temperatures of 900, 1050, and 1200 °C, respectively. The influence of annealing temperature on the silicon quantum dot (QD) synthesis was investigated by Raman scattering spectroscopy, X-ray diffraction spectroscopy, and high-resolution transmission electron microscopy. The influence of annealing temperature on the chemical bonding configurations was revealed by Fourier transform infrared absorption microscopy. The element ratios of the as-deposited sample were deduced by X-ray photoelectron spectroscopy. Results reveal that the samples are in silicon-rich nature. Silicon in the as-deposited sample and the 900 °C annealed sample are amorphous. When the annealing temperature is increased to 1050 °C, crystal silicon QDs have come into being. The calculated number density is about 2.15 ± 0.03 × 1012 cm−2 and more than 80 ± 3% of the silicon QDs fall within a narrow size range of 2–3 nm. When the annealing temperature is increased to 1200 °C, the average size of crystal silicon QDs is tuned from 2.6 to 3.2 nm, while the crystallinity is enhanced from 56.7 ± 2.5 to 67.1 ± 1.5%. We attribute the influence of annealing temperature on the synthesis of silicon QDs to be dependent on the evolution of chemical bonding configurations and the agglomeration of silicon atoms from the host matrix. © 2016 Elsevier B.V. All rights reserved.
1. Introduction Silicon nanocrystals dispersed in amorphous dielectric matrix have been paid extensive attention for their promising applications in silicon-based optoelectronic devices [1–3] and solar cells [4,5]. It's known that when the sizes of silicon nanocrystals are less than the silicon Bohr radius (~5 nm in diameter), they would behave as quantum dots (QDs) due to the quantum confinement effects [4]. For silicon QD size in range of 3–8 nm, the optical energy bandgap could be tuned from 1.3 to 1.65 eV [6]. Silicon QD superlattice is proposed for allsilicon tandem cell, whose efficiency limit would increase to 42.5% for 2-cell and 47.5% for 3-cell [7,8]. Considerable work has been done on the growth and characterization of silicon QDs in silicon oxide (SiO2 or SiOx) [9,10], silicon nitrides (Si3N4 or SiNx) [11,12], and silicon oxytrides (SiOxNy) [13,14]. As a weaker polar dielectric matrix, silicon QDs in amorphous silicon carbide (SiCx) might exhibit much larger quantum confinement behavior. Taking into account the lower barrier height of SiCx (~2.5 eV), the barrier height of SiOx and SiNx are ~9 eV and ~5.3 eV, respectively, SiC is favorable for increasing the tunneling probability and in turn the carrier ⁎ Corresponding author. E-mail address: [email protected] (G. Wen).
http://dx.doi.org/10.1016/j.jnoncrysol.2016.03.006 0022-3093/© 2016 Elsevier B.V. All rights reserved.
transport could be easier [15]. However, few experimental studies have been carried out on the preparation of silicon QDs in SiC matrix [16]. A full understanding about the synthesis of silicon QDs has been less documented, and the discussion is still open [17–19]. In this work, hydrogenated amorphous silicon carbide thin films (a-SiC:H) thin films were deposited by decomposition of silane (SiH4) and methane (CH4) and annealed at 900, 1050, and 1200 °C, respectively. The influence of annealing temperature on the synthesis of silicon QDs and the evolution of chemical bonding configurations have been investigated. 2. Experimental details a-SiC:H thin films were deposited by 13.56 MHz capacitive-coupled plasma-enhanced chemical vapor deposition (PECVD) on (100) p-type crystalline silicon wafer and quartz plate simultaneously. The substrates were first cleaned with piranha cleaning solution (3:1 H2SO4:H2O2) and then rinsed in de-ionized water. A diluted HF acid (5%) dip was performed only for the silicon wafers to remove the native surface oxide for 180 s additionally. Prior to deposition, the chamber was evacuated down to a base pressure of ~ 5 × 10−5 Pa. Hydrogen-diluted 10% SiH4 of 50 sccm and pure CH4 of 10 sccm were then introduced to maintain a working pressure of 106.7 Pa. The substrate temperature, the power
G. Wen et al. / Journal of Non-Crystalline Solids 441 (2016) 10–15
supply and the deposition time were kept at 200 °C, 160 mW/cm2 and 90 min, respectively. After deposition, the samples were cut into four smaller parts. Three parts of them were annealed in a quartz tube furnace at 900, 1050, 1200 °C for 30 min in a flowing 99.999% N2 atmosphere, respectively, and then were cooled down naturally to room temperature within the quartz tube furnace. The synthesis of silicon QD was characterized by means of Raman scattering spectroscopy, X-ray diffraction spectroscopy, and transmission electron microscopy. Raman scattering measurements were carried out by HORIBA Jobin Yvon LabRAM Spectrometer (HR 800 UV) in backscattering configuration using the 514 nm line of Ar+ laser, where the incident power was kept low to avoid crystallization and distortion of the spectrum. X-ray diffraction measurements was carried out by Philips's X'Pert Pro (XRD, PANalytical PW3040/60) with Cu Kα radiation (λ = 1.540562 Å) at a voltage of 40 kV and a current of 40 mA. The glancing angle between the incident X-ray and sample surface was 1.00 with a resolution of 0.01°. Crystalline silicon QDs were characterized directly on the 1050 °C annealed sample by transmission electron microscopy (TEM, JEM-2100F) in plan-view at a working voltage of 200 kV. The sample for TEM was prepared transparently by conventional mechanical polishing and Ar+ thinning technology at room temperature. The element ratios were analyzed by X-ray photoelectron spectroscopy (XPS, VG Multilab 2000), using a monochromatic Al Kα (1486.5 eV) Xray source and a hemispherical energy analyzer. The X-ray source power, instrument resolution, working pressure, and analyzed area of the samples were kept at 300 W, 0.47 eV, 6.67 × 10− 8 Pa, and 0.36 mm2, respectively. Before detections, the samples were sputtered using a beam of 3 kV × 2 μA Ar+ bombardments for 270 s. The spectra were collected at 25 eV pass energy and the binding energy values were calibrated by using the contaminant carbon C1s = 284.6 eV. The narrow-scan peaks were fitted with Thermo Avantage® and a standard Smart background was used for fitting the spectra. Chemical bonding configuration behaviors were deduced from Fourier transform infrared absorption measurements (FTIR, VERTEX 70). The absorption of substrate was eliminated by a bare silicon wafer. Raman scattering, XRD, TEM, and XPS characterizations were performed on the samples deposited on the quartz plates. FTIR measurements were performed on the samples deposited on the crystalline silicon substrates.
11
3. Results and discussion Fig. 1 displays the Raman scattering measurements obtained from the as-deposited sample and the samples annealed at 900, 1050 and 1200 °C, respectively. Before measurement, the system was calibrated with a single crystal Si wafer, whose Raman shift peak is at about 521 cm−1. For the as-deposited sample, the spectrum exhibits one broad peak at about 472.8 cm−1 with the full width at half maximum (FWHM) of 93.5 ± 2.8 cm−1. The 900 °C annealed sample also exhibits one broad peak in the same behavior, but whose peak has shifted to about 477.5 cm− 1 and FWHM has decreased to 83.7 ± 0.13 cm−1. When the annealing temperature is further elevated to 1050 °C, the spectrum exhibits two peaks: one is sharper asymmetrical and the other is broad weak. The sharper asymmetrical peak has shifted to about 514.0 cm−1 and the FWHM has decreased to 16.1 ± 0.10 cm−1. The low-energy tail has extended down to 290 cm−1. The broad weak protuberance is located at about 952.1 cm−1. The 1200 °C annealed sample exhibits one significantly sharper asymmetrical peak and one enhanced broad weak protuberance in the same behavior as that of the 1050 °C annealed sample. The significantly sharper peak is also at about 514.0 cm−1, but the FWHM has decreased further to 8.3 ± 0.07 cm−1. The 472.8 cm−1 peak could be attributed to scattering by the transverse optical (TO) mode of Si\\C and/or Si\\Si vibrations in the amorphous matrix, suggesting that the as-deposited sample is in completely amorphous feature [20,21]. The shifts to 477.5 cm−1 could be ascribed to the presentation of ultra-small silicon particles, but whose number-density is quite small [22]. It is well known that Raman spectroscopy is sensitive to local atomic arrangements and lattice vibrations [23,24]. The grain size-related effects as well as compressive stress within the material usually result in a low-frequency asymmetric broadening and a high low-number shift of the first order Raman band from 521 cm−1 of bulk crystalline Si [25–27]. Thus, the asymmetrical low-frequency shift of the 1050 °C and 1200 °C annealed samples demonstrate the formation of crystalline silicon particles. The left axis in Fig. 2 shows the trend of FWHMs. It has been reported that FWHM of Si\\Si peak increases with decrease in order and decreases with increase in order [28]. Thus the continuous decrease of FWHM also reveals the increase in order or the synthesis of crystalline silicon particles in the samples.
Fig. 1. Raman scattering measurements of the as-deposited sample and the samples annealed at 900, 1050, and 1200 °C, respectively, in range of 100–1200 cm−1. The inset shows a typical deconvolution of the spectrum of the 1050 °C annealed sample in range of 300–580 cm−1.
12
G. Wen et al. / Journal of Non-Crystalline Solids 441 (2016) 10–15
Fig. 2. Full width at half maximum (left axis) and silicon crystallinities (right axis) of the as-grown sample and the samples annealed at temperatures of 900, 1050 and 1200 °C (solid dots). The solid lines are drawn as guides to the eyes.
Fig. 3 illustrates the grazing incident XRD patterns of the same samples as those characterized by Raman scattering measurements. Apart from one weak protuberance located at 33.9°, no other diffraction peaks
Fig. 3. Grazing incident XRD patterns of the as-deposited sample and the samples annealed at 900, 1050, and 1200 °C in range of 20°–80°, respectively. The vertical dashed lines are eye-guides for clarity.
could be observed in the patterns of the as-deposited sample and the 900 °C annealed sample. In the sample annealed at 1050 °C, three diffraction peaks located at 2θ = 28.4°, 47.4°, and 56.3°, respectively, could be observed clearly. The 33.9° protuberance remained the same in intensity nearly. Another new weak protuberance located at 63.3° could be observed. In the sample annealed at 1200 °C, the corresponding three diffraction peaks located at 2θ = 28.4°, 47.4°, and 56.3°, respectively, were a little stronger than those of the 1050 °C annealed sample. The 33.9° and the 63.3° protuberances nearly disappeared. The three diffraction peaks at 28.4°, 47.4°, 56.3° could be attributed to (111), (220) and (311) planes of crystalline silicon, respectively [29]. None of these three diffraction peaks could be observed in the as-deposited sample and the 900 °C annealed sample, indicating that silicon is in amorphous feature in both of them. The diffraction peaks support that silicon nanocrystals were synthesized in the 1050 °C annealed sample. The stronger peaks of the 1200 °C annealed sample reveal that bigger silicon nanocrystals were synthesized than that of the 1050 °C annealed sample. At the same time, one can notice that there is a preferential growth direction along the (111) plane. This may be due to the fact that the surface energy of the (111) crystal plane is the lowest in all the three crystal planes [30]. It is worthwhile to mention that the two weak protuberances at 33.9° and 63.3° are associated with SiC nanoparticles. These reveal that SiC nanoparticles were synthesized in the as-deposited sample and new particles were synthesized in the high temperature annealing process. But the weak intensities show that their sizes are very small and their number densities are very low. In order to get more insight into the influence of annealing temperature on the synthesis of silicon QD, element ratios were analyzed by XPS and chemical bonding configuration behaviors were revealed by FTIR. Fig. 4 displays the narrow-scan XPS investigations of the asdeposited sample. Besides the expected Si and C elements, O is detected. The Si2p spectrum (a) could be deconvoluted into three Gaussian components centered at binding energies of 99.2, 100.9 and 101.7 eV, which could be ascribed to Si\\Si, Si\\C and Si\\O bond, respectively. The C1s spectrum (b) could be deconvoluted into two components located at 282.4 and 284.4 eV, corresponding to C\\Si and C\\C bonding, respectively. The O1s spectrum (c) could be deconvoluted into three components locating at 530.8, 531.8, and 532.5 eV, respectively. The 530.8 eV bond and 531.8 eV band could be ascribed to O\\Si of
G. Wen et al. / Journal of Non-Crystalline Solids 441 (2016) 10–15
13
0.17, 0.205, and 0.63, respectively [36,37]. From the data presented, the relative ratios could be roughly estimated: (Si\\C)/(C\\Si) = 2.19, and (Si\\O)/(O\\Si) = 2.40. The relative stoichiometric element ratio Si/C of SiC and Si/O of SiO2 are 1 and 0.5, which are smaller than the estimated 2.19 and 2.40, respectively. Thus it is silicon-rich inside the as-deposited sample and, in turn, in the other three annealed samples. The appearance of Si\\Si peak around 99.7 eV demonstrates the silicon phase in amorphous SiC host matrix. Fig. 5 shows the FTIR measurements of the as-deposited sample and the samples annealed at 900, 1050, and 1200 °C, respectively. Two prominent absorption bands could be observed for the as-deposited sample. One is around 2085 cm− 1, the other extends from 500 to 1200 cm− 1. The 2085 cm− 1 could be assigned to SiHn (n = 1,2,3) and/or C\\SiH stretching vibration mode [38]. While the extending from 500 to 1200 cm−1 could be ascribed to superposition of the following four absorption components: SiHn (n = 1,2,3) rocking and wagging mode near 655 cm−1 [39], Si\\C non-hydrogen rocking and/or wagging mode or Si\\CH3 stretching vibration mode near 780 cm−1 [40], SiH2 bending vibration mode at 800–900 cm−1 [41], CHn (n = 1,2,3) wagging or bending vibration mode near 1000 cm−1 [42]. When the annealing temperature is elevated to 900 °C, the SiHn and/or C\\SiH mode around 2085 cm− 1 disappear completely. The SiHn mode near 655 cm−1 and CHn near 1000 cm−1 diminish and the Si\\C or Si\\CH3 mode near 780 cm−1 increase gradually. When the annealing temperature is further elevated to 1200 °C, the Si\\C or Si\\CH3 peak near 780 cm− 1 is shifted toward a higher wave number of 828 cm−1, the CHn peak near 1000 cm−1 is shifted toward 1078 cm−1, and both intensities are enhanced obviously. Atomic hydrogen would dissociate the precursor gases via electron impact reactions in PECVD. There are plenty of Si\\Si, SiHn, C\\SiH, Si\\C and Si\\CH3 radicals in the as-deposited sample, whose vibration modes are around 2085 cm−1 and from 500 to 1200 cm−1. The abundant hydrogenated silicon radicals lead to the formation of primary amorphous Si nuclei in the as-deposited sample and are characterized by the broad Raman peak at 472.8 cm−1. When the annealing temperature increases to 900 °C, Si\\H bonds are broken and hydrogen atoms are released from the SiHn and/or C\\SiH entities. This is consistent with the disappearance of the mode around 2085 cm−1 and the reduction of the mode near 655 cm−1. The non-stoichiometric SiC radicals would be dissociated as Si and SiC radicals: Si1−xCx → x(SiC) + (1 − 2×)Si [22], which is in consistent with the enhancement of the Si\\C vibration mode near 780 cm− 1. The abundant atomic Si with dangling bonds would precipitate to form new radicals. In turn, the high number
Fig. 4. (a) Si2p, (b) C1s, (c) O1s, narrow-scan XPS survey spectra of the as-grown nonstoichiometric hydrogenated amorphous SiC samples (open circles), together with the Gaussian curves fitted (solid dots).
SiOx (x b 2), while the 532.5 eV band could be ascribed to O\\Si of SiO2 [31–33]. Quantitative element ratios could be estimated from the integrated intensities under the Si2p and C1s narrow-scan peaks by the following equation: RSi/C = X/Y = AxSy/AySx, where X and Y represent element Si and C, respectively. Ax is the area under the narrow-scan peak of Si2p and Ay is that of the C1s. Sx is the sensitivity factor of element Si and Sy is that of C [34,35]. The sensitivity factors of Si, C, and O are
Fig. 5. Infrared absorption spectra for the as-grown sample and the samples annealed at temperatures of 900, 1050 and 1200 °C, respectively. The spectrum was recorded in range of 400–4000 cm−1 with resolution set at 4 cm−1.
14
G. Wen et al. / Journal of Non-Crystalline Solids 441 (2016) 10–15
Fig. 6. Typical plan-view TEM images of the a-SiC:H sample annealed at 1050 °C in N2 atmosphere for 30 min. (a) Low-magnification plan-view image of the sample, (b) Highmagnification image of one individual clearly-observed silicon QD.
density of new radicals together with primary clusters would absorb other silicon atoms, diffusing from the amorphous host matrix, to form amorphous silicon particles. The peak shifts to 477.5 cm− 1 and FWHM decreases to 83.7 ± 0.13 cm−1 of the Raman spectrum of the 900 °C annealed sample demonstrate their growth in the amorphous host matrix. When the annealing temperature is further increased to 1050 °C, the peak shift of Si\\C or Si\\CH3 toward higher wave numbers reveals that the dissociation of non-stoichiometric SiC radicals is enhanced. More silicon atoms are absorbed and amount of the amorphous silicon particles grow into crystalline silicon particles in randomly orientation [43–45]. The Raman peak shifts to 514.0 cm− 1 and the three XRD diffraction peaks at 28.4°, 47.4°, 56.3° of the 1050 °C annealed sample demonstrate the synthesis of crystalline silicon particles. Crystalline silicon would continue their growth inclination with annealing temperature increase to 1200 °C. According to the confinement model about the Raman shifts, the average size of crystalline silicon particles is correspondent to its special shift and could be extracted by using equation: Δω = ω(L) − ω0 = − A(a/L)γ, where ω0 is the frequency of the optical phonon at the zone center, ω(L) is the frequency of the Raman phonon in a crystalline silicon particles with size L, and a represents the lattice constant of crystalline silicon. Both A = 47.41 cm−1 and γ = 1.44 are used to describe the vibrational confinement [46]. Supposing that crystalline silicon particles have a spherical shape, the average size is about 2.6–3.2 nm for the 1050 °C annealed sample. The same 514.0 cm− 1 shift of the 1200 °C annealed sample as that of the 1050 °C annealed sample reveals that its average size doesn't get a clear increase. The average size of these silicon nanocrystals is b 5 nm in diameter, they could be ascribed to crystal silicon QDs. Furthermore, the crystallinity Xc could be deduced from the deconvolution of Raman spectrum by the expression: Xc = (Ic + Im)/ (Ic + Im + σIα), where Ic, Im, and Iα are the integrated intensities of the crystalline silicon band, the intermediate silicon band, and the amorphous Si or SiC band. σ is chosen to be the order of unity [44,47]. The inset in Fig. 1 shows representative deconvolution of the 1050 °C annealed sample in range of 300–580 cm−1 using the least-squares routine. Three independent Gaussian components could be obtained: a broad band near 468.4 ± 1.2 cm−1 corresponding to amorphous Si or Si\\C phase, a narrow band around 515.9 ± 0.1 cm−1 corresponding to crystal silicon phase, and an intermediate one in the vicinity of 504.2 ± 0.3 cm−1 corresponding to the amorphous silicon particles. It is found that the crystal fraction is about 56.7 ± 2.5% for the 1050 °C annealed sample and increases to 67.1 ± 1.5% for the 1200 °C annealed sample. The crystallinities are showed by the right axis in Fig. 2. This implies that high temperature could promote Si atom precipitation and more ordered structures in the amorphous phase.
At the same time, the diffusivity of Si atoms in the solid silicon carbide matrix is rather low, though the diffusion is enhanced [48]. Ostwald ripening effect shows that the growing of silicon particles is limited by the thickness of amorphous SiC matrix around them. Once the thickness, in another word, the thermodynamic barrier doesn't decrease down to the critical value for the penetration of silicon atoms, the penetration would be quite difficult for the diffusion of silicon atoms. A great number of silicon atoms might diffuse only in a short distance and create an additional density of smaller amorphous silicon particles [49]. Chang et al. have ascribed amorphous silicon particles to be amorphous silicon QDs [27]. It was reported that the average size of amorphous silicon QDs in the annealed samples was shifted toward smaller than those embedded in the as-deposited samples, and the number-density of smaller silicon QDs was higher [13]. Fig. 6 shows the direct physical evidence of crystal silicon QDs in the amorphous SiC matrix by high resolution TEM. Fig. 6(a) depicts the representative plan-view image of the sample annealed at 1050 °C. One can notice that there is a high density of black spots distributed throughout the homogenous host matrix, as well as adjacent coalescence of parts of them. Fig. 6(b) depicts the high resolution image of one perfect isolated crystalline QD, whose lattice fringes could be observed clearly. The lattice spacing measured equals to 3.17 Å, approximately, corresponding to the (111) crystal orientation of silicon [29]. The calculated number density is about 2.15 ± 0.03 × 1012 cm−2 for the 1050 °C annealed sample, only counting those could be clearly resolved silicon QDs. Fig. 7 gives the size distribution histogram of silicon QDs in the representative plan-view images of the same sample as that in Fig. 6. The size of silicon QDs were measured one by DigitalMicrograph®. More than 80 ± 3% of the QDs were in a narrow size range of 2–3 nm, qualitatively or even quantitatively in agreement with the Raman analysis. A broad weak protuberance located at 952.1 cm−1 has been reported in the Raman spectrum of the 1050 °C annealed sample. When it is deconvoluted in range of 850–1050 cm−1, there are two components centered at 934.2 ± 1.7 and 965.9 ± 5.3 cm−1, respectively. Considering that amorphous SiC vibration density of state in Raman measurement is up to 900 cm−1, the maximum optical phonon energy of crystalline SiC is up to 972 cm−1, the origin of this weak protuberance could be attributed to Si clusters (two-phonon process) with a small quantity of SiC particles [50]. Therefore, the increase of this protuberance could be explained in terms of Si\\C and Si\\Si bonding states increase from amorphous to crystalline. 4. Conclusions In this work, a-SiC:H thin films were deposited by decomposition of SiH4 and CH4. In order to synthesize silicon QDs, the samples were annealed at different high temperatures. The influence of annealing
G. Wen et al. / Journal of Non-Crystalline Solids 441 (2016) 10–15
Fig. 7. The size distribution histogram of the silicon QDs in the sample annealed at 1050 °C.
temperature on the silicon QD synthesis was investigated by Raman scattering spectroscopy, X-ray diffraction spectroscopy, and highresolution transmission electron microscopy, respectively. The element ratios of the as-deposited sample were deduced by XPS. The chemical bonding configurations were revealed by FTIR. Results reveal that the samples are in silicon-rich nature. Silicon in the as-deposited sample and the 900 °C annealed sample are amorphous. When the temperature is increased to 1050 °C, crystalline silicon QDs have come into being. The average size increases from 2.6 to 3.2 nm, the crystallinity enhances from 56.7 ± 2.5% to 67.1 ± 1.5% when the temperature increases from 1050 to 1200 °C. The calculated number density is about 2.15 ± 0.03 × 1012 cm−2 and more than 80 ± 3% of the silicon QDs fall in a narrow size range of 2–3 nm for the 1050 °C annealed sample. We attribute the influence of annealing temperature on the synthesis of silicon QDs to be dependent on the evolution of chemical bonding configurations and the agglomeration of silicon atoms from the host matrix. These analyses provide a better understanding of the synthesis of silicon QDs in a-SiC:H matrix. Acknowledgements This work is supported by projects of Wuhan Polytechnic University (Grant No. 2013y08 and 2015RZ18). The authors would like to thank Analytical and Testing Center of Huazhong University of Science and Technology. References [1] R.J. Walters, G.I. Bourianoff, H.A. Atwater, Nat. Matters 4 (2005) 143. [2] L. Pavesi, L. Dal Negro, C. Mazzoleni, G. Franzo, F. Priolo, Nature 408 (2000) 440. [3] B. Garrido Fernandez, M. Lόpez, C. García, A. Pérezi Rodríguez, J.R. Morante, C. Bonafos, M. Carrada, A. Claverie, J. Appl. Phys. 91 (2002) 798. [4] G. Conibeer, M. Green, R. Corkish, Y. Cho, E.-C. Cho, C. Jiang, T. Fangsuwannarak, E. Pink, Y. Huang, T. Puzzer, T. Trupke, B. Richards, A. Shalav, K. Lin, Thin Solid Films 511-512 (2006) 654.
15
[5] M.A. Green, E.-C. Cho, Y.-H. Cho, E. Pink, T. Trupke, K.-L. Lin, T. Fangsuwannarak, T. Puzzer, G. Conibeer, R. Corkish, Proceeding of 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June 6–10, 2005 3. [6] E.-C. Cho, M.A. Green, G. Conibeer, D. Song, Y.-H. Cho, G. Scardera, S. Huang, S. Park, X.J. Hao, Y. Huang, L.V. Dao, Adv. OptoElectron. 2007 (2007) 69578. [7] E.-C. Cho, M.A. Green, J. Xia, R. Corkish, P. Reece, M. Gal, Appl. Phys. Lett. 84 (2004) 2286. [8] J. Nelson, The Physics of Solar Cells, Imperial College Press, London, UK, 2003. [9] M. Zacharias, J. Heitmann, R. Scholz, U. Kahler, M. Schmidt, J. Bläsing, Appl. Phys. Lett. 80 (2002) 661. [10] T. Arguirov, T. Mchedlidze, M. Kittler, R. Rӧlver, B. Berghoff, M. Fӧrst, B. Spangenber, Appl. Phys. Lett. 89 (2006) 053111. [11] L.H. Jiang, X.B. Zeng, X. Zhang, J. Non-Cryst. Solids 375 (2011) 2187. [12] T.-W. Kim, C.-H. Cho, B.-H. Kim, S.-J. Park, Appl. Phys. Lett. 88 (2006) 123102. [13] A. Benami, G. Santana, A. Ortiz, A. Ponce, D. Romeu, J. Aguilar-Hernández, G. Contreras-Puente, J.C. Alonso, Nanotechnology 18 (2007) 155704. [14] K. Ma, J.Y. Feng, Z.J. Zhang, Nanotechnology 17 (2006) 4560. [15] C.-W. Jiang, M.A. Green, J. Appl. Phys. 99 (2006) 114902. [16] Y. Kurokawa, S. Miyajima, A. Yamada, M. Konagai, Jpn. J. Appl. Phys. 45 (2006) L1064 Part 2. [17] D.K. Basa, Phys. Status Solidi (a) 195 (2003) 87. [18] Z. Xia, S. Huang, Solid State Commun. 150 (2010) 914. [19] G. Wen, X. Zeng, W. Liao, C. Cao, Thin Solid Films 552 (2014) 18. [20] N.-M. Park, C.-J. Choi, T.-Y. Seong, S.-J. Park, Phys. Rev. Lett. 86 (2001) 1355. [21] Q.J. Cheng, E. Tam, S.Y. Xu, K. Ostrikov, Nanoscale 2 (2010) 594. [22] D.Y. Song, E.-C. Cho, G. Conibeer, Y. Huang, C. Flynn, M.A. Green, J. Appl. Phys. 103 (2008) 083544. [23] E.-C. Cho, S.-W. Park, X.J. Hao, D.Y. Song, G. Conibeer, S.-C. Park, M.A. Green, Nanotechnology 19 (2008) 245201. [24] Q.J. Cheng, S.Y. Xu, K. Ostrikov, Acta Mater. 58 (2010) 560. [25] U. Coscia, G. Ambrosone, S. Lettieri, P. Maddalena, S. Ferrero, Thin Solid Films 511512 (2006) 399. [26] A.K. Panchal, C.S. Solanki, J. Cryst. Growth 311 (2009) 2659. [27] G.R. Chang, F. Ma, D. Ma, K. Xu, Nanotechnology 21 (2010) 465605. [28] P.M. Voyles, J.E. Gerbi, M.M.J. Treacy, J.M. Gibson, J.R. Abelson, Phys. Rev. Lett. 86 (2001) 5514. [29] JCPDS-International Centre for Diffraction Data Card No. 27-1402 (2004). [30] Q.J. Cheng, S.Y. Xu, S.Y. Huang, K. Ostrikov, Cryst. Growth Des. 9 (6) (2009) 2863. [31] G. Hollinger, F.J. Himpsel, Appl. Phys. Lett. 44 (1984) 93. [32] S. Contarini, S.P. Howlett, C. Rizzo, B.A. De Angelis, Appl. Surf. Sci. 51 (1991) 177. [33] T.N. Taylor, J. Mater. Res. 4 (1989) 189. [34] C.D. Wagner, L.E. Davis, M.V. Zeller, J.A. Taylor, R.H. Raymond, L.H. Gale, Surf. Interface Anal. 3 (1981) 211. [35] W.K. Choi, T.Y. Ong, L.S. Tan, F.C. Loh, K.L. Tan, J. Appl. Phys. 83 (1998) 4968. [36] C.D. Wagner, W.M. Riggs, L.E. Davis, J.F. Moulder, G.E. Muilenberg, Handbook of Xray Photoelectron Spectroscopy, Perkin-Elmer, Eden Prairie, 1979. [37] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of X-ray Photoelectron Septrocopy, Perkin-Elmer, Eden Prairie, 1992. [38] G. Dellamea, F. Demichelis, C.F. Pirri, P. Rava, V. Rigato, T. Stapinski, E. Tresso, J. NonCryst. Solids 137 (1991) 95. [39] A. Kole, P. Chaudhuri, Thin Solid Films 522 (2012) 45. [40] G. Ambrosone, U. Coscia, S. Lettieri, P. Maddalena, C. Privato, S. Ferrero, Thin Solid Films 403-404 (2002) 349. [41] F. Giorgis, C.F. Pirri, E. Tresso, P. Rava, Diam. Relat. Mater. 6 (1997) 1606. [42] M.H. Brodsky, M. Cardona, J.J. Cuomo, Phys. Rev. B 16 (1977) 3556. [43] D. Nesheva, C. Raptis, A. Perakis, I. Bineva, Z. Aneva, Z. Levi, S. Alexandrova, H. Hofmeister, J. Appl. Phys. 92 (2002) 4678. [44] Q.J. Cheng, S.Y. Xu, K. Ostrikov, Nanotechnology 20 (2009) 215606. [45] D.K. Basa, G. Ambrosone, U. Coscia, Nanotechnology 19 (2008) 415706. [46] J. Zi, H. Büscher, C. Falter, W. Ludwig, K. Zhang, X. Xie, Appl. Phys. Lett. 69 (1996) 200. [47] N. Budini, P.A. Rinaldi, J.A. Schmidt, R.D. Arce, R.H. Buitrago, Thin Solid Films 518 (2010) 5349. [48] K. Boer, Survey of Semiconductor Physics, van Nostrand Reinhold, 1990. [49] C. Das, A. Dasgupta, S.C. Saha, S. Ray, J. Appl. Phys. 91 (2002) 9401. [50] J. Wasyluk, T.S. Perova, S.A. Kukushkin, A.V. Osipov, N.A. Feoktistov, S.A. Grudinkin, Mater. Sci. Forum 645–648 (2010) 359.
Contents lists available at ScienceDirect
Journal of Non-Crystalline Solids journal homepage: www.elsevier.com/locate/jnoncrysol
The influence of annealing temperature on the synthesis of silicon quantum dots embedded in hydrogenated amorphous Si-rich silicon carbide matrix Guozhi Wen a,⁎, Xiangbin Zeng b, Xianghu Li a a b
School of Electronic and Electrical Engineering, Wuhan Polytechnic University, Wuhan, Hubei 430023, China School of Optical and Electronic Information, Huazhong University of Science and Technology, Wuhan, Hubei 430074, China
a r t i c l e
i n f o
Article history: Received 13 January 2016 Received in revised form 3 March 2016 Accepted 10 March 2016 Available online xxxx Keywords: Silicon Quantum dots Bonding configuration Temperature PECVD
a b s t r a c t Hydrogenated amorphous silicon carbide thin films (a-SiC:H) were prepared by plasma-enhanced chemical vapor deposition (PECVD) and thermal annealed at temperatures of 900, 1050, and 1200 °C, respectively. The influence of annealing temperature on the silicon quantum dot (QD) synthesis was investigated by Raman scattering spectroscopy, X-ray diffraction spectroscopy, and high-resolution transmission electron microscopy. The influence of annealing temperature on the chemical bonding configurations was revealed by Fourier transform infrared absorption microscopy. The element ratios of the as-deposited sample were deduced by X-ray photoelectron spectroscopy. Results reveal that the samples are in silicon-rich nature. Silicon in the as-deposited sample and the 900 °C annealed sample are amorphous. When the annealing temperature is increased to 1050 °C, crystal silicon QDs have come into being. The calculated number density is about 2.15 ± 0.03 × 1012 cm−2 and more than 80 ± 3% of the silicon QDs fall within a narrow size range of 2–3 nm. When the annealing temperature is increased to 1200 °C, the average size of crystal silicon QDs is tuned from 2.6 to 3.2 nm, while the crystallinity is enhanced from 56.7 ± 2.5 to 67.1 ± 1.5%. We attribute the influence of annealing temperature on the synthesis of silicon QDs to be dependent on the evolution of chemical bonding configurations and the agglomeration of silicon atoms from the host matrix. © 2016 Elsevier B.V. All rights reserved.
1. Introduction Silicon nanocrystals dispersed in amorphous dielectric matrix have been paid extensive attention for their promising applications in silicon-based optoelectronic devices [1–3] and solar cells [4,5]. It's known that when the sizes of silicon nanocrystals are less than the silicon Bohr radius (~5 nm in diameter), they would behave as quantum dots (QDs) due to the quantum confinement effects [4]. For silicon QD size in range of 3–8 nm, the optical energy bandgap could be tuned from 1.3 to 1.65 eV [6]. Silicon QD superlattice is proposed for allsilicon tandem cell, whose efficiency limit would increase to 42.5% for 2-cell and 47.5% for 3-cell [7,8]. Considerable work has been done on the growth and characterization of silicon QDs in silicon oxide (SiO2 or SiOx) [9,10], silicon nitrides (Si3N4 or SiNx) [11,12], and silicon oxytrides (SiOxNy) [13,14]. As a weaker polar dielectric matrix, silicon QDs in amorphous silicon carbide (SiCx) might exhibit much larger quantum confinement behavior. Taking into account the lower barrier height of SiCx (~2.5 eV), the barrier height of SiOx and SiNx are ~9 eV and ~5.3 eV, respectively, SiC is favorable for increasing the tunneling probability and in turn the carrier ⁎ Corresponding author. E-mail address: [email protected] (G. Wen).
http://dx.doi.org/10.1016/j.jnoncrysol.2016.03.006 0022-3093/© 2016 Elsevier B.V. All rights reserved.
transport could be easier [15]. However, few experimental studies have been carried out on the preparation of silicon QDs in SiC matrix [16]. A full understanding about the synthesis of silicon QDs has been less documented, and the discussion is still open [17–19]. In this work, hydrogenated amorphous silicon carbide thin films (a-SiC:H) thin films were deposited by decomposition of silane (SiH4) and methane (CH4) and annealed at 900, 1050, and 1200 °C, respectively. The influence of annealing temperature on the synthesis of silicon QDs and the evolution of chemical bonding configurations have been investigated. 2. Experimental details a-SiC:H thin films were deposited by 13.56 MHz capacitive-coupled plasma-enhanced chemical vapor deposition (PECVD) on (100) p-type crystalline silicon wafer and quartz plate simultaneously. The substrates were first cleaned with piranha cleaning solution (3:1 H2SO4:H2O2) and then rinsed in de-ionized water. A diluted HF acid (5%) dip was performed only for the silicon wafers to remove the native surface oxide for 180 s additionally. Prior to deposition, the chamber was evacuated down to a base pressure of ~ 5 × 10−5 Pa. Hydrogen-diluted 10% SiH4 of 50 sccm and pure CH4 of 10 sccm were then introduced to maintain a working pressure of 106.7 Pa. The substrate temperature, the power
G. Wen et al. / Journal of Non-Crystalline Solids 441 (2016) 10–15
supply and the deposition time were kept at 200 °C, 160 mW/cm2 and 90 min, respectively. After deposition, the samples were cut into four smaller parts. Three parts of them were annealed in a quartz tube furnace at 900, 1050, 1200 °C for 30 min in a flowing 99.999% N2 atmosphere, respectively, and then were cooled down naturally to room temperature within the quartz tube furnace. The synthesis of silicon QD was characterized by means of Raman scattering spectroscopy, X-ray diffraction spectroscopy, and transmission electron microscopy. Raman scattering measurements were carried out by HORIBA Jobin Yvon LabRAM Spectrometer (HR 800 UV) in backscattering configuration using the 514 nm line of Ar+ laser, where the incident power was kept low to avoid crystallization and distortion of the spectrum. X-ray diffraction measurements was carried out by Philips's X'Pert Pro (XRD, PANalytical PW3040/60) with Cu Kα radiation (λ = 1.540562 Å) at a voltage of 40 kV and a current of 40 mA. The glancing angle between the incident X-ray and sample surface was 1.00 with a resolution of 0.01°. Crystalline silicon QDs were characterized directly on the 1050 °C annealed sample by transmission electron microscopy (TEM, JEM-2100F) in plan-view at a working voltage of 200 kV. The sample for TEM was prepared transparently by conventional mechanical polishing and Ar+ thinning technology at room temperature. The element ratios were analyzed by X-ray photoelectron spectroscopy (XPS, VG Multilab 2000), using a monochromatic Al Kα (1486.5 eV) Xray source and a hemispherical energy analyzer. The X-ray source power, instrument resolution, working pressure, and analyzed area of the samples were kept at 300 W, 0.47 eV, 6.67 × 10− 8 Pa, and 0.36 mm2, respectively. Before detections, the samples were sputtered using a beam of 3 kV × 2 μA Ar+ bombardments for 270 s. The spectra were collected at 25 eV pass energy and the binding energy values were calibrated by using the contaminant carbon C1s = 284.6 eV. The narrow-scan peaks were fitted with Thermo Avantage® and a standard Smart background was used for fitting the spectra. Chemical bonding configuration behaviors were deduced from Fourier transform infrared absorption measurements (FTIR, VERTEX 70). The absorption of substrate was eliminated by a bare silicon wafer. Raman scattering, XRD, TEM, and XPS characterizations were performed on the samples deposited on the quartz plates. FTIR measurements were performed on the samples deposited on the crystalline silicon substrates.
11
3. Results and discussion Fig. 1 displays the Raman scattering measurements obtained from the as-deposited sample and the samples annealed at 900, 1050 and 1200 °C, respectively. Before measurement, the system was calibrated with a single crystal Si wafer, whose Raman shift peak is at about 521 cm−1. For the as-deposited sample, the spectrum exhibits one broad peak at about 472.8 cm−1 with the full width at half maximum (FWHM) of 93.5 ± 2.8 cm−1. The 900 °C annealed sample also exhibits one broad peak in the same behavior, but whose peak has shifted to about 477.5 cm− 1 and FWHM has decreased to 83.7 ± 0.13 cm−1. When the annealing temperature is further elevated to 1050 °C, the spectrum exhibits two peaks: one is sharper asymmetrical and the other is broad weak. The sharper asymmetrical peak has shifted to about 514.0 cm−1 and the FWHM has decreased to 16.1 ± 0.10 cm−1. The low-energy tail has extended down to 290 cm−1. The broad weak protuberance is located at about 952.1 cm−1. The 1200 °C annealed sample exhibits one significantly sharper asymmetrical peak and one enhanced broad weak protuberance in the same behavior as that of the 1050 °C annealed sample. The significantly sharper peak is also at about 514.0 cm−1, but the FWHM has decreased further to 8.3 ± 0.07 cm−1. The 472.8 cm−1 peak could be attributed to scattering by the transverse optical (TO) mode of Si\\C and/or Si\\Si vibrations in the amorphous matrix, suggesting that the as-deposited sample is in completely amorphous feature [20,21]. The shifts to 477.5 cm−1 could be ascribed to the presentation of ultra-small silicon particles, but whose number-density is quite small [22]. It is well known that Raman spectroscopy is sensitive to local atomic arrangements and lattice vibrations [23,24]. The grain size-related effects as well as compressive stress within the material usually result in a low-frequency asymmetric broadening and a high low-number shift of the first order Raman band from 521 cm−1 of bulk crystalline Si [25–27]. Thus, the asymmetrical low-frequency shift of the 1050 °C and 1200 °C annealed samples demonstrate the formation of crystalline silicon particles. The left axis in Fig. 2 shows the trend of FWHMs. It has been reported that FWHM of Si\\Si peak increases with decrease in order and decreases with increase in order [28]. Thus the continuous decrease of FWHM also reveals the increase in order or the synthesis of crystalline silicon particles in the samples.
Fig. 1. Raman scattering measurements of the as-deposited sample and the samples annealed at 900, 1050, and 1200 °C, respectively, in range of 100–1200 cm−1. The inset shows a typical deconvolution of the spectrum of the 1050 °C annealed sample in range of 300–580 cm−1.
12
G. Wen et al. / Journal of Non-Crystalline Solids 441 (2016) 10–15
Fig. 2. Full width at half maximum (left axis) and silicon crystallinities (right axis) of the as-grown sample and the samples annealed at temperatures of 900, 1050 and 1200 °C (solid dots). The solid lines are drawn as guides to the eyes.
Fig. 3 illustrates the grazing incident XRD patterns of the same samples as those characterized by Raman scattering measurements. Apart from one weak protuberance located at 33.9°, no other diffraction peaks
Fig. 3. Grazing incident XRD patterns of the as-deposited sample and the samples annealed at 900, 1050, and 1200 °C in range of 20°–80°, respectively. The vertical dashed lines are eye-guides for clarity.
could be observed in the patterns of the as-deposited sample and the 900 °C annealed sample. In the sample annealed at 1050 °C, three diffraction peaks located at 2θ = 28.4°, 47.4°, and 56.3°, respectively, could be observed clearly. The 33.9° protuberance remained the same in intensity nearly. Another new weak protuberance located at 63.3° could be observed. In the sample annealed at 1200 °C, the corresponding three diffraction peaks located at 2θ = 28.4°, 47.4°, and 56.3°, respectively, were a little stronger than those of the 1050 °C annealed sample. The 33.9° and the 63.3° protuberances nearly disappeared. The three diffraction peaks at 28.4°, 47.4°, 56.3° could be attributed to (111), (220) and (311) planes of crystalline silicon, respectively [29]. None of these three diffraction peaks could be observed in the as-deposited sample and the 900 °C annealed sample, indicating that silicon is in amorphous feature in both of them. The diffraction peaks support that silicon nanocrystals were synthesized in the 1050 °C annealed sample. The stronger peaks of the 1200 °C annealed sample reveal that bigger silicon nanocrystals were synthesized than that of the 1050 °C annealed sample. At the same time, one can notice that there is a preferential growth direction along the (111) plane. This may be due to the fact that the surface energy of the (111) crystal plane is the lowest in all the three crystal planes [30]. It is worthwhile to mention that the two weak protuberances at 33.9° and 63.3° are associated with SiC nanoparticles. These reveal that SiC nanoparticles were synthesized in the as-deposited sample and new particles were synthesized in the high temperature annealing process. But the weak intensities show that their sizes are very small and their number densities are very low. In order to get more insight into the influence of annealing temperature on the synthesis of silicon QD, element ratios were analyzed by XPS and chemical bonding configuration behaviors were revealed by FTIR. Fig. 4 displays the narrow-scan XPS investigations of the asdeposited sample. Besides the expected Si and C elements, O is detected. The Si2p spectrum (a) could be deconvoluted into three Gaussian components centered at binding energies of 99.2, 100.9 and 101.7 eV, which could be ascribed to Si\\Si, Si\\C and Si\\O bond, respectively. The C1s spectrum (b) could be deconvoluted into two components located at 282.4 and 284.4 eV, corresponding to C\\Si and C\\C bonding, respectively. The O1s spectrum (c) could be deconvoluted into three components locating at 530.8, 531.8, and 532.5 eV, respectively. The 530.8 eV bond and 531.8 eV band could be ascribed to O\\Si of
G. Wen et al. / Journal of Non-Crystalline Solids 441 (2016) 10–15
13
0.17, 0.205, and 0.63, respectively [36,37]. From the data presented, the relative ratios could be roughly estimated: (Si\\C)/(C\\Si) = 2.19, and (Si\\O)/(O\\Si) = 2.40. The relative stoichiometric element ratio Si/C of SiC and Si/O of SiO2 are 1 and 0.5, which are smaller than the estimated 2.19 and 2.40, respectively. Thus it is silicon-rich inside the as-deposited sample and, in turn, in the other three annealed samples. The appearance of Si\\Si peak around 99.7 eV demonstrates the silicon phase in amorphous SiC host matrix. Fig. 5 shows the FTIR measurements of the as-deposited sample and the samples annealed at 900, 1050, and 1200 °C, respectively. Two prominent absorption bands could be observed for the as-deposited sample. One is around 2085 cm− 1, the other extends from 500 to 1200 cm− 1. The 2085 cm− 1 could be assigned to SiHn (n = 1,2,3) and/or C\\SiH stretching vibration mode [38]. While the extending from 500 to 1200 cm−1 could be ascribed to superposition of the following four absorption components: SiHn (n = 1,2,3) rocking and wagging mode near 655 cm−1 [39], Si\\C non-hydrogen rocking and/or wagging mode or Si\\CH3 stretching vibration mode near 780 cm−1 [40], SiH2 bending vibration mode at 800–900 cm−1 [41], CHn (n = 1,2,3) wagging or bending vibration mode near 1000 cm−1 [42]. When the annealing temperature is elevated to 900 °C, the SiHn and/or C\\SiH mode around 2085 cm− 1 disappear completely. The SiHn mode near 655 cm−1 and CHn near 1000 cm−1 diminish and the Si\\C or Si\\CH3 mode near 780 cm−1 increase gradually. When the annealing temperature is further elevated to 1200 °C, the Si\\C or Si\\CH3 peak near 780 cm− 1 is shifted toward a higher wave number of 828 cm−1, the CHn peak near 1000 cm−1 is shifted toward 1078 cm−1, and both intensities are enhanced obviously. Atomic hydrogen would dissociate the precursor gases via electron impact reactions in PECVD. There are plenty of Si\\Si, SiHn, C\\SiH, Si\\C and Si\\CH3 radicals in the as-deposited sample, whose vibration modes are around 2085 cm−1 and from 500 to 1200 cm−1. The abundant hydrogenated silicon radicals lead to the formation of primary amorphous Si nuclei in the as-deposited sample and are characterized by the broad Raman peak at 472.8 cm−1. When the annealing temperature increases to 900 °C, Si\\H bonds are broken and hydrogen atoms are released from the SiHn and/or C\\SiH entities. This is consistent with the disappearance of the mode around 2085 cm−1 and the reduction of the mode near 655 cm−1. The non-stoichiometric SiC radicals would be dissociated as Si and SiC radicals: Si1−xCx → x(SiC) + (1 − 2×)Si [22], which is in consistent with the enhancement of the Si\\C vibration mode near 780 cm− 1. The abundant atomic Si with dangling bonds would precipitate to form new radicals. In turn, the high number
Fig. 4. (a) Si2p, (b) C1s, (c) O1s, narrow-scan XPS survey spectra of the as-grown nonstoichiometric hydrogenated amorphous SiC samples (open circles), together with the Gaussian curves fitted (solid dots).
SiOx (x b 2), while the 532.5 eV band could be ascribed to O\\Si of SiO2 [31–33]. Quantitative element ratios could be estimated from the integrated intensities under the Si2p and C1s narrow-scan peaks by the following equation: RSi/C = X/Y = AxSy/AySx, where X and Y represent element Si and C, respectively. Ax is the area under the narrow-scan peak of Si2p and Ay is that of the C1s. Sx is the sensitivity factor of element Si and Sy is that of C [34,35]. The sensitivity factors of Si, C, and O are
Fig. 5. Infrared absorption spectra for the as-grown sample and the samples annealed at temperatures of 900, 1050 and 1200 °C, respectively. The spectrum was recorded in range of 400–4000 cm−1 with resolution set at 4 cm−1.
14
G. Wen et al. / Journal of Non-Crystalline Solids 441 (2016) 10–15
Fig. 6. Typical plan-view TEM images of the a-SiC:H sample annealed at 1050 °C in N2 atmosphere for 30 min. (a) Low-magnification plan-view image of the sample, (b) Highmagnification image of one individual clearly-observed silicon QD.
density of new radicals together with primary clusters would absorb other silicon atoms, diffusing from the amorphous host matrix, to form amorphous silicon particles. The peak shifts to 477.5 cm− 1 and FWHM decreases to 83.7 ± 0.13 cm−1 of the Raman spectrum of the 900 °C annealed sample demonstrate their growth in the amorphous host matrix. When the annealing temperature is further increased to 1050 °C, the peak shift of Si\\C or Si\\CH3 toward higher wave numbers reveals that the dissociation of non-stoichiometric SiC radicals is enhanced. More silicon atoms are absorbed and amount of the amorphous silicon particles grow into crystalline silicon particles in randomly orientation [43–45]. The Raman peak shifts to 514.0 cm− 1 and the three XRD diffraction peaks at 28.4°, 47.4°, 56.3° of the 1050 °C annealed sample demonstrate the synthesis of crystalline silicon particles. Crystalline silicon would continue their growth inclination with annealing temperature increase to 1200 °C. According to the confinement model about the Raman shifts, the average size of crystalline silicon particles is correspondent to its special shift and could be extracted by using equation: Δω = ω(L) − ω0 = − A(a/L)γ, where ω0 is the frequency of the optical phonon at the zone center, ω(L) is the frequency of the Raman phonon in a crystalline silicon particles with size L, and a represents the lattice constant of crystalline silicon. Both A = 47.41 cm−1 and γ = 1.44 are used to describe the vibrational confinement [46]. Supposing that crystalline silicon particles have a spherical shape, the average size is about 2.6–3.2 nm for the 1050 °C annealed sample. The same 514.0 cm− 1 shift of the 1200 °C annealed sample as that of the 1050 °C annealed sample reveals that its average size doesn't get a clear increase. The average size of these silicon nanocrystals is b 5 nm in diameter, they could be ascribed to crystal silicon QDs. Furthermore, the crystallinity Xc could be deduced from the deconvolution of Raman spectrum by the expression: Xc = (Ic + Im)/ (Ic + Im + σIα), where Ic, Im, and Iα are the integrated intensities of the crystalline silicon band, the intermediate silicon band, and the amorphous Si or SiC band. σ is chosen to be the order of unity [44,47]. The inset in Fig. 1 shows representative deconvolution of the 1050 °C annealed sample in range of 300–580 cm−1 using the least-squares routine. Three independent Gaussian components could be obtained: a broad band near 468.4 ± 1.2 cm−1 corresponding to amorphous Si or Si\\C phase, a narrow band around 515.9 ± 0.1 cm−1 corresponding to crystal silicon phase, and an intermediate one in the vicinity of 504.2 ± 0.3 cm−1 corresponding to the amorphous silicon particles. It is found that the crystal fraction is about 56.7 ± 2.5% for the 1050 °C annealed sample and increases to 67.1 ± 1.5% for the 1200 °C annealed sample. The crystallinities are showed by the right axis in Fig. 2. This implies that high temperature could promote Si atom precipitation and more ordered structures in the amorphous phase.
At the same time, the diffusivity of Si atoms in the solid silicon carbide matrix is rather low, though the diffusion is enhanced [48]. Ostwald ripening effect shows that the growing of silicon particles is limited by the thickness of amorphous SiC matrix around them. Once the thickness, in another word, the thermodynamic barrier doesn't decrease down to the critical value for the penetration of silicon atoms, the penetration would be quite difficult for the diffusion of silicon atoms. A great number of silicon atoms might diffuse only in a short distance and create an additional density of smaller amorphous silicon particles [49]. Chang et al. have ascribed amorphous silicon particles to be amorphous silicon QDs [27]. It was reported that the average size of amorphous silicon QDs in the annealed samples was shifted toward smaller than those embedded in the as-deposited samples, and the number-density of smaller silicon QDs was higher [13]. Fig. 6 shows the direct physical evidence of crystal silicon QDs in the amorphous SiC matrix by high resolution TEM. Fig. 6(a) depicts the representative plan-view image of the sample annealed at 1050 °C. One can notice that there is a high density of black spots distributed throughout the homogenous host matrix, as well as adjacent coalescence of parts of them. Fig. 6(b) depicts the high resolution image of one perfect isolated crystalline QD, whose lattice fringes could be observed clearly. The lattice spacing measured equals to 3.17 Å, approximately, corresponding to the (111) crystal orientation of silicon [29]. The calculated number density is about 2.15 ± 0.03 × 1012 cm−2 for the 1050 °C annealed sample, only counting those could be clearly resolved silicon QDs. Fig. 7 gives the size distribution histogram of silicon QDs in the representative plan-view images of the same sample as that in Fig. 6. The size of silicon QDs were measured one by DigitalMicrograph®. More than 80 ± 3% of the QDs were in a narrow size range of 2–3 nm, qualitatively or even quantitatively in agreement with the Raman analysis. A broad weak protuberance located at 952.1 cm−1 has been reported in the Raman spectrum of the 1050 °C annealed sample. When it is deconvoluted in range of 850–1050 cm−1, there are two components centered at 934.2 ± 1.7 and 965.9 ± 5.3 cm−1, respectively. Considering that amorphous SiC vibration density of state in Raman measurement is up to 900 cm−1, the maximum optical phonon energy of crystalline SiC is up to 972 cm−1, the origin of this weak protuberance could be attributed to Si clusters (two-phonon process) with a small quantity of SiC particles [50]. Therefore, the increase of this protuberance could be explained in terms of Si\\C and Si\\Si bonding states increase from amorphous to crystalline. 4. Conclusions In this work, a-SiC:H thin films were deposited by decomposition of SiH4 and CH4. In order to synthesize silicon QDs, the samples were annealed at different high temperatures. The influence of annealing
G. Wen et al. / Journal of Non-Crystalline Solids 441 (2016) 10–15
Fig. 7. The size distribution histogram of the silicon QDs in the sample annealed at 1050 °C.
temperature on the silicon QD synthesis was investigated by Raman scattering spectroscopy, X-ray diffraction spectroscopy, and highresolution transmission electron microscopy, respectively. The element ratios of the as-deposited sample were deduced by XPS. The chemical bonding configurations were revealed by FTIR. Results reveal that the samples are in silicon-rich nature. Silicon in the as-deposited sample and the 900 °C annealed sample are amorphous. When the temperature is increased to 1050 °C, crystalline silicon QDs have come into being. The average size increases from 2.6 to 3.2 nm, the crystallinity enhances from 56.7 ± 2.5% to 67.1 ± 1.5% when the temperature increases from 1050 to 1200 °C. The calculated number density is about 2.15 ± 0.03 × 1012 cm−2 and more than 80 ± 3% of the silicon QDs fall in a narrow size range of 2–3 nm for the 1050 °C annealed sample. We attribute the influence of annealing temperature on the synthesis of silicon QDs to be dependent on the evolution of chemical bonding configurations and the agglomeration of silicon atoms from the host matrix. These analyses provide a better understanding of the synthesis of silicon QDs in a-SiC:H matrix. Acknowledgements This work is supported by projects of Wuhan Polytechnic University (Grant No. 2013y08 and 2015RZ18). The authors would like to thank Analytical and Testing Center of Huazhong University of Science and Technology. References [1] R.J. Walters, G.I. Bourianoff, H.A. Atwater, Nat. Matters 4 (2005) 143. [2] L. Pavesi, L. Dal Negro, C. Mazzoleni, G. Franzo, F. Priolo, Nature 408 (2000) 440. [3] B. Garrido Fernandez, M. Lόpez, C. García, A. Pérezi Rodríguez, J.R. Morante, C. Bonafos, M. Carrada, A. Claverie, J. Appl. Phys. 91 (2002) 798. [4] G. Conibeer, M. Green, R. Corkish, Y. Cho, E.-C. Cho, C. Jiang, T. Fangsuwannarak, E. Pink, Y. Huang, T. Puzzer, T. Trupke, B. Richards, A. Shalav, K. Lin, Thin Solid Films 511-512 (2006) 654.
15
[5] M.A. Green, E.-C. Cho, Y.-H. Cho, E. Pink, T. Trupke, K.-L. Lin, T. Fangsuwannarak, T. Puzzer, G. Conibeer, R. Corkish, Proceeding of 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June 6–10, 2005 3. [6] E.-C. Cho, M.A. Green, G. Conibeer, D. Song, Y.-H. Cho, G. Scardera, S. Huang, S. Park, X.J. Hao, Y. Huang, L.V. Dao, Adv. OptoElectron. 2007 (2007) 69578. [7] E.-C. Cho, M.A. Green, J. Xia, R. Corkish, P. Reece, M. Gal, Appl. Phys. Lett. 84 (2004) 2286. [8] J. Nelson, The Physics of Solar Cells, Imperial College Press, London, UK, 2003. [9] M. Zacharias, J. Heitmann, R. Scholz, U. Kahler, M. Schmidt, J. Bläsing, Appl. Phys. Lett. 80 (2002) 661. [10] T. Arguirov, T. Mchedlidze, M. Kittler, R. Rӧlver, B. Berghoff, M. Fӧrst, B. Spangenber, Appl. Phys. Lett. 89 (2006) 053111. [11] L.H. Jiang, X.B. Zeng, X. Zhang, J. Non-Cryst. Solids 375 (2011) 2187. [12] T.-W. Kim, C.-H. Cho, B.-H. Kim, S.-J. Park, Appl. Phys. Lett. 88 (2006) 123102. [13] A. Benami, G. Santana, A. Ortiz, A. Ponce, D. Romeu, J. Aguilar-Hernández, G. Contreras-Puente, J.C. Alonso, Nanotechnology 18 (2007) 155704. [14] K. Ma, J.Y. Feng, Z.J. Zhang, Nanotechnology 17 (2006) 4560. [15] C.-W. Jiang, M.A. Green, J. Appl. Phys. 99 (2006) 114902. [16] Y. Kurokawa, S. Miyajima, A. Yamada, M. Konagai, Jpn. J. Appl. Phys. 45 (2006) L1064 Part 2. [17] D.K. Basa, Phys. Status Solidi (a) 195 (2003) 87. [18] Z. Xia, S. Huang, Solid State Commun. 150 (2010) 914. [19] G. Wen, X. Zeng, W. Liao, C. Cao, Thin Solid Films 552 (2014) 18. [20] N.-M. Park, C.-J. Choi, T.-Y. Seong, S.-J. Park, Phys. Rev. Lett. 86 (2001) 1355. [21] Q.J. Cheng, E. Tam, S.Y. Xu, K. Ostrikov, Nanoscale 2 (2010) 594. [22] D.Y. Song, E.-C. Cho, G. Conibeer, Y. Huang, C. Flynn, M.A. Green, J. Appl. Phys. 103 (2008) 083544. [23] E.-C. Cho, S.-W. Park, X.J. Hao, D.Y. Song, G. Conibeer, S.-C. Park, M.A. Green, Nanotechnology 19 (2008) 245201. [24] Q.J. Cheng, S.Y. Xu, K. Ostrikov, Acta Mater. 58 (2010) 560. [25] U. Coscia, G. Ambrosone, S. Lettieri, P. Maddalena, S. Ferrero, Thin Solid Films 511512 (2006) 399. [26] A.K. Panchal, C.S. Solanki, J. Cryst. Growth 311 (2009) 2659. [27] G.R. Chang, F. Ma, D. Ma, K. Xu, Nanotechnology 21 (2010) 465605. [28] P.M. Voyles, J.E. Gerbi, M.M.J. Treacy, J.M. Gibson, J.R. Abelson, Phys. Rev. Lett. 86 (2001) 5514. [29] JCPDS-International Centre for Diffraction Data Card No. 27-1402 (2004). [30] Q.J. Cheng, S.Y. Xu, S.Y. Huang, K. Ostrikov, Cryst. Growth Des. 9 (6) (2009) 2863. [31] G. Hollinger, F.J. Himpsel, Appl. Phys. Lett. 44 (1984) 93. [32] S. Contarini, S.P. Howlett, C. Rizzo, B.A. De Angelis, Appl. Surf. Sci. 51 (1991) 177. [33] T.N. Taylor, J. Mater. Res. 4 (1989) 189. [34] C.D. Wagner, L.E. Davis, M.V. Zeller, J.A. Taylor, R.H. Raymond, L.H. Gale, Surf. Interface Anal. 3 (1981) 211. [35] W.K. Choi, T.Y. Ong, L.S. Tan, F.C. Loh, K.L. Tan, J. Appl. Phys. 83 (1998) 4968. [36] C.D. Wagner, W.M. Riggs, L.E. Davis, J.F. Moulder, G.E. Muilenberg, Handbook of Xray Photoelectron Spectroscopy, Perkin-Elmer, Eden Prairie, 1979. [37] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of X-ray Photoelectron Septrocopy, Perkin-Elmer, Eden Prairie, 1992. [38] G. Dellamea, F. Demichelis, C.F. Pirri, P. Rava, V. Rigato, T. Stapinski, E. Tresso, J. NonCryst. Solids 137 (1991) 95. [39] A. Kole, P. Chaudhuri, Thin Solid Films 522 (2012) 45. [40] G. Ambrosone, U. Coscia, S. Lettieri, P. Maddalena, C. Privato, S. Ferrero, Thin Solid Films 403-404 (2002) 349. [41] F. Giorgis, C.F. Pirri, E. Tresso, P. Rava, Diam. Relat. Mater. 6 (1997) 1606. [42] M.H. Brodsky, M. Cardona, J.J. Cuomo, Phys. Rev. B 16 (1977) 3556. [43] D. Nesheva, C. Raptis, A. Perakis, I. Bineva, Z. Aneva, Z. Levi, S. Alexandrova, H. Hofmeister, J. Appl. Phys. 92 (2002) 4678. [44] Q.J. Cheng, S.Y. Xu, K. Ostrikov, Nanotechnology 20 (2009) 215606. [45] D.K. Basa, G. Ambrosone, U. Coscia, Nanotechnology 19 (2008) 415706. [46] J. Zi, H. Büscher, C. Falter, W. Ludwig, K. Zhang, X. Xie, Appl. Phys. Lett. 69 (1996) 200. [47] N. Budini, P.A. Rinaldi, J.A. Schmidt, R.D. Arce, R.H. Buitrago, Thin Solid Films 518 (2010) 5349. [48] K. Boer, Survey of Semiconductor Physics, van Nostrand Reinhold, 1990. [49] C. Das, A. Dasgupta, S.C. Saha, S. Ray, J. Appl. Phys. 91 (2002) 9401. [50] J. Wasyluk, T.S. Perova, S.A. Kukushkin, A.V. Osipov, N.A. Feoktistov, S.A. Grudinkin, Mater. Sci. Forum 645–648 (2010) 359.