- Email: [email protected]
Structural, optoelectronic and mechanical properties of PECVD Si-C-N films: An effect of substrate bias
Materials Science in Semiconductor Processing 88 (2018) 65–72
Contents lists available at ScienceDirect
Materials Science in Semiconductor Processing journal homepage: www.elsevier.com/locate/mssp
Structural, optoelectronic and mechanical properties of PECVD Si-C-N films: An effect of substrate bias
T
⁎
A.O. Kozaka, , V.I. Ivashchenkoa, O.K. Poradaa, L.A. Ivashchenkoa, T.V. Tomilaa, V.S. Manjarab, G.V. Klishevychb a b
Institute for Problems of Materials Sciences NAS of Ukraine, 3, Krzhyzhanovsky Str., 03142 Kyiv, Ukraine Institute of Physics NAS of Ukraine, 46, Nauki ave., 03028 Kyiv, Ukraine
A R T I C LE I N FO
A B S T R A C T
Keywords: Amorphous Si-C-N films Plasma-enhanced chemical vapor deposition Photoluminescence Photoluminescence excitation Chemical bonding Nanoindentaion
Structural, optoelectronic and mechanical properties of amorphous silicon carbon nitride (Si-C-N) thin films produced by plasma enhanced chemical vapor deposition (PECVD) at different negative substrate biases (Ud) are studied. The films are characterized by X-ray diffraction, atomic force microscopy, Fourier transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), Raman spectroscopy, optical transmittance spectroscopy, nanoindentation as well using the results of the measurements of photoluminescence (PL) and PL excitation (PLE) spectra. All deposited films are found to be X-ray amorphous. An increase in Ud leads to: a smoothing of the film surface; a decrease of the transparency; an increase of refractive index from 1.69 to 1.92; a decrease of the energy gap from 4.15 to 2.38 eV; an increase in nanohardness and elastic modulus from 14 GPa to 24 GPa and from 147 GPa to 190 GPa, respectively. These results were explained in terms of the bonding configuration from XPS and FTIR measurements. The PL spectra of the films deposited at lower negative substrate bias have one PL band in the region between 530 and 540 nm, whereas the PL spectra of the films deposited at higher negative substrate bias show two PL bands at 530–570 nm and 640–650 nm. On the basis of the PLE data, it was shown that these PL bands are related to the electronic recombination between the conduction band and the valence bands and their tails within the amorphous N-rich and C-rich Si-C-N-O-H networks.
1. Introduction Up to date, thin silicon carbon nitride (Si-C-N) films are of interest due to the unique optoelectronic properties as a variable band gap in the range from 0.96 to 5.6 eV and refractive index between 1.44 and 2.2, high transparency in the visible range of spectra, on the one hand, and the excellent mechanical properties, namely, high hardness (up to 38 GPa), high structural and thermal stability until 1350 °C, low roughness, strong adhesion to a substrate, and good abrasive wear resistance [1–7]. The combination of these properties has enabled the efficient application of Si-C-N films in semiconductor devices and as wear-resistant coatings. In particular, they can be used as thin films with variable optical characteristics in ultraviolet light sensors, and as sensitive layers in gas sensors, as well as protective and wear-resistant coatings on optoelectronic devices and metal surfaces [7–10]. Different chemical vapor deposition (CVD) or physical vapor deposition (PVD) methods, such as plasma enhanced CVD (PECVD) [2,5,8,11,12], thermo-CVD (high temperature heating) [3], hot-wire CVD [9,10], magnetron sputtering [5], pulsed laser deposition [13] are ⁎
often used for the preparation of Si-C-N films. It is well established that film properties are sensitive to deposition parameters and precursors used. The main PECVD parameters are substrate temperature, discharge power and negative substrate bias (Ud). During deposition, two processes take place: i) film deposition due to the chemical reactions between adsorbed species on the growing film surface, and ii) the sputtering of the deposited film. The sputtering occurs due to the breaking of the weak bonds on the film surface, which promotes the densification of the films, and an increase of their quality [2,6]. These two processes are controlled by deposition conditions. The recent review of the properties of Si-C-N films deposited by using different methods was done in Ref. [14]. We note that, despite a huge experimental material accumulated on Si-C-N films, so far, the comprehensive studies of the Si-C-N films that could combine good optoelectronic and mechanical properties are in an infant stage. Such films might be used for the production of semiconductor devices for the exploitation under extreme conditions. In this work, Si-C-N films were produced by PECVD at different substrate biases using the liquid precursor hexamethyldisilazane
Corresponding author. E-mail address: [email protected] (A.O. Kozak).
https://doi.org/10.1016/j.mssp.2018.07.023 Received 5 April 2018; Received in revised form 13 June 2018; Accepted 15 July 2018 1369-8001/ © 2018 Elsevier Ltd. All rights reserved.
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
Table 1 Deposition parameters. Ud (V) FN2 (sccm) FH+HMDS (sccm) PC (Тоrr) TS (°С) PW (W/сm2)
− 5, − 50, − 100, − 150, − 250 0.5 12 0.2 450 0.2
Ud – substrate bias voltage, FN2 – nitrogen flow rate, FH+HMDS – hydrogen flow rate through a bubbler with the HMDS vapor, PC – working pressure, TS – substrate temperature, PW – discharge power density.
(HMDS) [5]. The optoelectronic, mechanical and structural properties of the deposited Si-C-N films were investigated using X-ray diffraction (XRD), atomic force microscopy (AFM), Fourier transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), Raman spectroscopy, optical transmittance spectroscopy, nanoindentation, as well the results of the measurements of photoluminescence (PL) and PL excitation (PLE) spectra. Below, for simplicity, the expression “increase (decrease) of substrate bias” will mean the change of its absolute value.
Fig. 1. X-ray patterns of Si-C-N films deposited at Ud = − 5 V (a) and Ud = − 250 V (b). The peaks at 2Θ ~ 33°, 62° and 69° correspond to the silicon substrate.
triangular pyramid diamond tip. Nanohardness (H) and elastic modulus (E) were determined according to the Oliver and Pharr method [16].
2. Experimental details Si-C-N films were produced using the HMDS vapor and nitrogen by means of PECVD. The negative bias was applied to a substrate by an additional radio-frequency (5.27 MHz) generator. The HMDS vapor was delivered into the reactor chamber by hydrogen from the thermostatcontrolled bubbler heated up to 40 °C. All deposition parameters are listed in Table 1. The films were deposited on both the (100) oriented silicon wafers and transparent quartz substrates. The surface oxide on the silicon wafer surfaces were cleaned by dipping into the 10% hydrofluoric acid for 3 min and the final etching of the substrates was carried out by treating with the hydrogen plasma into the reaction chamber. The film thickness was estimated by an optical interference profilometer. The film surface was analysed by AFM microscope “NanoScope IIIa Dimension 3000TM”. The structure of the film was investigated by Xray diffraction using a diffractometer “DRON-3M” in Cu Kα- radiation. FTIR measurements were carried out by using a spectrometer “FSM 1202” LLC “Infraspek” in the range of the wave numbers of 400–4000 cm−1. XPS core-level spectra were measured by a UHVAnalysis-System, SPECS, under the Mg Kα radiation (E = 1253.6 eV). XPS spectra were measured after argon etching during 3 min. The Raman spectra were measured in the range of 100–3200 cm−1 with the help of the Via Renishaw Raman microscope equipped with the He-Ne laser excitation at 632.8 nm. Measurements of the UV–VIS transmission spectra were carried out by using a two-beam optical spectrometer SPECORD-M40 in the range of wavelengths of 200–900 nm. Photoluminescence spectra (PL) were investigated at room temperature by using an experimental installation for the recording of PL spectra. The main unit of this installation is the automated monochromator SPM-2 (Carl Zeiss, Jena). The installation enables one to measure the intensity of PL spectra at a certain wavelength in the visible range. Photoluminescence is excited by a low powerful LED laser (~ 25 µW/ cm2, λ = 405 nm). Our analysis shows that this power is enough to excite PL from bulk states [15]. This laser does not heat films during optical measurements, which, in contrast to Hg lamps, does not lead to lowering the accuracy of measurements. The photoluminescence excitation (PLE) spectra of the deposited Si-C-N films were recorded using the HITACHI MPF-4 spectrofluorometer equipped with a xenon lamp. The conversion of its analogue output signals into a digital form was carried out using a digital converter. The excitation wavelengths were in the range of 320–525 nm (3.88–2.36 eV, respectively). Nanoindentaion of the films was carried out by the NanoIndenter G200 device (Agilent Technologies) using continuous stiffness measurement (CSM) mode. The indentations were produced with a Berkovich
3. Results and discussion 3.1. Film structure We carried out XRD measurements of the deposited films to analyze their microstructure. X-ray patterns of the Si-C-N films deposited on silicon substrates are shown in Fig. 1. There are no any features that could be assigned to any crystallites in the films. It follows that the deposited films are X-ray amorphous. 3.2. Film surface morphology and thickness Fig. 2(a) and (b) show the AFM images of the surface of the films deposited at Ud = − 5 V and Ud = − 250 V, respectively. The scanning area was 1 µm × 1 µm. From Fig. 2 one can observe that the films are very smooth and the film surfaces exhibit an excellent morphological homogeneity with low roughness. The values of the RMS roughness (Rq) and average roughness (Ra) of the films deposited at various substrate biases were Rq = 0.33 nm and Ra = 0.27 nm (Ud = − 5 V), Rq = 0.27 nm and Ra = 0.22 nm (Ud = − 250 V). We see that an increase in negative substrate bias leads to the insignificant smoothing of the film surface: substrate bias accelerates ions and charged particles, thereby promoting the etching of weak bonds and possibly the densification of the films. As a result, the roughness of the film surface slightly decreases and the film thickness increases from 200 nm to 700 nm with increasing Ud from − 5 V to − 250 V, respectively. 3.3. Film composition The chemical composition of the films was estimated using the XPS data. The Si, C, N, O contents in these films are approximately 36, 36, 14 and 14 at% and 37, 41, 13 and 9 at% for the films deposited at Ud = − 5 V and − 250 V, respectively. An increase of the concentration of carbon in the film is accompanied by decreasing the oxygen content, whereas the concentration of silicon and nitrogen remains unchanged. The oxygen in the films can be due to the residual oxygen absorbed on the reactor chamber walls. Since the reduction of the film roughness is accompanied with lowering the oxygen content some oxygen may come from the ambient air after deposition [17]. We could not estimate the hydrogen content in the films because the XPS analysis did not give such a possibility. However, in the FTIR spectra the hydrogen bonds are clearly seen. Also, it was found that hydrogen bands decrease with increasing substrate bias which indicates that the hydrogen content 66
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
Fig. 2. AFM images of the surface of the Si-C-N films deposited at Ud = − 5 V (a) and Ud = − 250 V (b).
1250–3500 cm−1. The band at 1549–1553 cm−1 can be assigned to C-C and/or C-N stretching modes. Three bands located in the range of 2160–2220 cm−1, 2865–2890 cm−1 and 3370–3385 cm−1can be caused by the vibration of the hydrogen bonds C≡N/Si–H, C–H and N–H bonds, respectively (cf. Table 2). We see that the small band related to the vibration of N–H bonds is present in all the spectra shown in Fig. 3a. This means that the reduction of the intensity of the band in the range of 1157–1170 cm−1 (cf. Fig. 3b) occurs due to decreasing a number of C–N bonds. The measured XPS core-level spectra of the deposited films are shown in Fig. 4. The intensity of the Si2p and N1s spectra weakly depend on substrate bias, whereas the intensity of the C1s spectra increases and the intensity of the O1s spectra decreases with increasing Ud. The peaks in the Si2p, C1s, N1s and O1s spectra of the films prepared at Ud = – 5 V centered at 102.3 eV, 284.8 eV, 398.7 eV and 532.9 eV, respectively, can be assigned to Si–O and/or Si-N [26], C–C [27] and/or C–N [28], N–C [29] and O–Si [26] bonds, respectively. As substrate bias rises, all the measured spectra shift toward low binding energies. For the film deposited at − 250 V, the Si2p peak at 101.5 eV is assigned to Si–N bonds [26,27], the C1s peak at 284.2 eV can be caused by C–C bonds [30], the N1s peak at 398 eV is associated with N–Si bonds [13] and the O1s peak at 532.4 eV is assigned to O–Si bonds [26]. The observed asymmetry of the XPS spectra points to that they are not formed by a single chemical bonding state.
decreases as substrate bias increases. Here it should be noted that the hydrogen was detected in the hydrogenated Si-C-N films using elastic recoil detection method [18]. 3.4. Chemical bonding configurations An analysis of the chemical bonding structure was performed by the FTIR, XPS and Raman spectroscopy measurements of the films deposited on Si substrates. Fig. 3(a) illustrates the FTIR spectra of the deposited films. Table 2 summarizes the identified FTIR absorption bands of the films deposited under different conditions. For peak identification, the literature data [11,13,19–26] is also presented. The wide absorption band ranging from 550 to 1250 cm−1 inherent to Si-CN films [5,12] can be related to the vibrations of Si–C, Si–N, Si–O, Si–H, C–N and N–H bonds (cf. Table 2). Fig. 3(a) shows that this band changes with increasing Ud. To gain more insight into the origin of this band, we decomposed it into several Gaussians, as shown in Fig. 3(b). The concentration of Si–O, Si–H, C–N/N–H bonds, estimated from the area of a corresponding Gaussian, decreases, and the concentration of Si–C bonds increases with increasing substrate bias. Such a re-distribution of bonding configuration could be caused by etching weak bonds by increasing ion bombardment. The concentration of Si–N bonds remains the highest independently of Ud. Other small absorption bands are observed in the range of
Fig. 3. (a) FTIR spectra of the films deposited at different Ud. Numbers − 5 to − 250 denote Ud in V. (b) FTIR deconvoluted spectra in the range of 550–1250 cm−1. The experimental FTIR spectra marked by open circles and their deconvoluted counterparts marked in red color. The inversion of the bands was done. 67
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
C–N [29–31], C–C [30,32] and/or C–H [33,34] and C–Si [35] bonds, respectively. With increasing negative substrate bias to −250 V, the area of the C-N and C-C/C-H components decreases, and the area of the C-Si component increases. The N1s spectra consist of two Gaussians at 398.07 eV and 399.03–399.18 eV indicating the existence of N–Si and N–C/N–H bonds [34], respectively. The N-C/N-H component decreases as Ud increases. The Si-N peak shifts towards high binding energies relatively the peak of the Si–N bonds in a-Si3N4 (397.64 eV) [36] that could indicate the formation of the Si-C-N network in the deposited films. Finally, the O1s spectrum of the film deposited at Ud = – 5 V was fitted by one Gaussian curve centered at 531.0 eV. It is assigned to Si–O bonds. A future increase of Ud resulted in increasing the asymmetry of this peak and, therefore, it was presented by two Gaussian curves centered at 530.9 eV and 533.0 eV. These curves are suggested to be formed by Si–O and C–O bonds, respectively. So, an increase in substrate bias gives rise to an appearance of C–O bonds. We note that the chemical binding states of the deposited Si-C-N films gained from the XPS measurements are consistent with those derived from the FTIR measurements. The results indicate the formation Si-C-N-O-H network in which Si–C and Si–N bonds are predominant. The ion bombardment induced by Ud promotes both the formation of strong Si–C bonds instead of weak C–N bonds and the breaking of Si–H bonds. Such changes in chemical bonding can cause the densification of the films [6]. The Raman spectra of the deposited films are shown in Fig. 6. These spectra are very useful for an analysis of the bonding configurations of carbon and silicon in carbon-based materials. It is well known that there are Si-rich and C-rich regions that coexist together with the Si-C clusters in Si-C-N films, and can be observed by the Raman spectroscopy [37]. In the Raman spectra, shown in Fig. 6, only one peak at 520 cm−1 related to the silicon substrate was detected. The absence of the carbon related peaks D (near 1360 cm−1), corresponding to the disorder-induced vibration modes of sp2 atoms in rings, and G (near 1560 cm−1) related to an in-plane bond stretching of sp2-hybridized carbon atoms in rings and in chains, is probably due to the fact that the intensive luminescence of the samples in the investigated area overshadows the non-intensive Raman peaks [38]. To reduce the intensity of the luminescence, the laser power was reduced to a minimum, but the luminescent bands remains very intense. This luminescence is typical for disordered polymeric carbons which contain a relatively large amount of nitrogen and hydrogen [39]. Obviously, strong luminescence caused by the presence of a large number of nitrogen and hydrogen in the films makes the identification of any peculiarities in the Raman spectra impossible [40]. Moreover, the Raman spectra, shown in Fig. 6, indicate the disorder of the Si-C-N network implying the absence of free amorphous carbon clusters. However, the results of the FTIR and XPS measurements indicate the presence of the C–C bonds in the films. We suppose that these bonds can be due to the single C–C bonds in the surrounding of the Si and N atoms [12,41].
Table 2 Peak positions in the FTIR spectra shown in Fig. 3 and their identification based on literature data. Observed FTIR band (cm−1)
Bonding type
Wavenumber (cm−1) from early publications [Ref.]
690–708a 796–799a 908–928a 1023–1035a
Si–H Si–C Si–N Si–O
690–710 [19] 770–810 [20] 900–940 [19] 1000–1030 [19] 1030–1050 [13]
1157–1170a
C–N N–H
1156–1198 [21] 1156–1198 [21] 1130–1160 [19]
1549–1553
C=C
1500–1850 1500–1600 1500–1850 1500–1600 1530–1725
[22] [23] [22] [23] [24]
2100–2150 2123–2202 2100–2300 2105–2173
[19] [24] [11] [25]
C=N
2160–2220
C≡N Si–H
2865–2890
C–Hn
2857–2890 [25]
3370–3385
N–Hn
3350–3400 [26] 3342–3378 [25]
a
Peak position in the deconvoluted spectra.
Fig. 4. Core level XPS spectra of the deposited films. The spectra were measured after argon etching during 3 min. Numbers − 5 and − 250 denote Ud in V.
3.5. Optical properties An optical energy gap (Eg) and refractive index (n) were estimated from the ultraviolet-visible transmission spectra. The optical transmission spectra of the Si-C-N films deposited on quartz substrates at different negative subtract biases are shown in Fig. 7a. All the films exhibit a sufficiently high optical transparency (up to 90%) in the range of 400–900 nm. An increase in negative substrate bias from − 5 V to − 250 V leads to darkening the film (Fig. 7b), which, most likely, is due to a decrease in the band gap and to an increase of film thickness. Besides, an increase of the carbon content can cause the shift of the edge of the optical absorption band toward high wavelengths [7]. The interference fringes, observed in the region with a low level of absorption (cf. Fig. 7a) is caused by multiple reflections between two interfaces: film/air and film/substrate. The interference
To gain the additional information on the bonding picture of the films we decomposed the XPS spectra onto separate Gaussians. The deconvoluted XPS spectra are shown in Fig. 5. The Si 2p spectra of each film are composed of three Gaussians. The deconvoluted peaks at 100.7–100.75 eV, 101.85–102 eV and 103 eV are attributed to Si–С [26,29], Si–N and Si–O [26] bonds, respectively. From Fig. 5 we see that the peak related to Si–N bonds is predominant over other peaks. Fig. 5 shows that the Si-N component decreases, whereas and the Si-C component increases with substrate bias in agreement with the results of the FTIR measurements (cf. Fig. 3). In the C1s spectra, three peaks centered at 285.9, 284.8–284.85 and 283.9 eV can be assigned to the 68
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
Fig. 5. Deconvolution of the XPS spectra of the films deposited at Ud= − 5V and Ud = − 250 V. The experimental XPS spectra marked by open circles and their deconvoluted counterparts marked in red color. Numbers − 5 and − 250 denote Ud in V.
Fig. 6. Raman spectra of the films deposited at different negative substrate bias (Normalized by band maximum).
fringes can appear in the transmittance spectra of a film provided it has: i) uniform thickness; ii) the smooth and homogenous film surface; iii) high structural perfection of the interfaces [42,43]. Interference fringes in transmittance spectra can be used to determine the refractive index, n. Estimation of the refractive index was done by using the method suggested by Swanepoel [42]. This method is based on drawing envelope curves through the extremes of the interference fringes in the transmission spectrum. The refractive index as a function of wavelength is calculated using the expression:
n=
Fig. 7. a) Optical transmission spectra; b) The images of the Si-C-N film deposited on quartz substrate (from left to right): uncoated substrate, film deposited at Ud = − 5 V, Ud = − 50 V, Ud = − 100 V, Ud = − 150 V, Ud = − 250 V.
N+ N 2 − ns2 ,
where
N = 2n s
Fig. 8 displays the refractive index as a function of wavelength for the films deposited under various Ud. An increase in the reflective index with increasing negative substrate bias is observed. The refractive index, evaluated at 632 nm (inset in Fig. 8), increases from 1.69 to 1.92 as the Ud increases from − 5 V to − 250 V, respectively. Taking into account the values of the refractive indexes of SiO2 (n = 1.47), Si3N4
n2+1 TM − Tm + s . TMTm 2
ns = 1.51, is the refractive index of a quartz substrate (transmittance of uncoated substrate is 92%), TM and Tm are maximum and minimum transmittances within the envelope curves, respectively. 69
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
Fig. 10. PL spectra of the Si-C-N films deposited at different negative substrate biases, Ud. (Excitation wavelength is 405 nm). Fig. 8. Refractive index vs. wavelength of the Si-C-N films, and the refractive index, evaluated at 632 nm (inset).
with increasing Ud. Such a behavior of the band gap depending on negative substrate bias can be explained in the terms of chemical bonding using the results of the FTIR and XPS measurements. The films prepared at low Ud have the largest value of Eg among other films due to the significant content of Si–N and Si–O bonds. An increase in negative substrate bias was shown to result in increasing Si–C bonds and reducing Si–O bonds. These changes in the bonding configuration of the films caused by increasing Ud will promote the narrowing of the band gap, since the energy gap in silicon carbide is lower than in silicon nitride and silicon oxide. Here, it should be noted also that high substrate biases give rise to intensive ion bombardment of the growing film surface causing its randomization and breaking of Si–H bonds. As a result, a number of gap states will rise, and the band gap will narrow. The photoluminescence (PL) spectra of the films deposited on silicon substrates under different Ud are presented in Fig. 10. We see that the PL spectra are very sensitive to substrate bias. The one-peak structure of the PL spectra is observed for the films deposited at Ud = − 5 V and Ud= − 50 V with the maximum at 530 nm, whereas the photoemission of the films deposited at highest substrate biases have two PL bands centered at 535 nm and 645 nm (Ud = − 150 V) and 570 nm and 640 nm (Ud = − 250 V). To gain more insight into the origin of the PL peaks let us analyze the photoluminescence excitation, PLE, spectra of the films deposited at Ud = − 5 V and − 150 V shown in Fig. 11. In the PLE measurements, the PL intensity at a fixed wavelength is determined as a function of wavelength of irradiation. Each PLE peak corresponds to a specific light absorption process. For intelligible interpretation, it would be appropriate to consider first the PLE spectra of the high-biased film. The PLE spectra of the film deposited at Ud = − 150 V was monitoring at 540 nm and 650 nm. There are three PLE peaks at 360 nm, 410 nm and 490 nm for this sample (cf. Fig. 11). The positions of these peaks practically coincide in both the PLE spectra under two monitoring wavelengths. This means that the peaks are caused by absorption process within one network formed trough the film. To identify these peaks we used literature data [47–50]. The PLE peak at 360 nm represents a characteristic light absorption line of cubic SiC [47], whereas the peak at 410 nm could be caused by the oxygen defects and the formed Si–C–O network within the film [48]. We note also that these two peaks are in the range of 340–400 nm that is characteristic to the light absorption in Si3N4 [49]. The peak position of 490 nm is very close to that of 500 nm in the PLE spectra of porous SiC [50]. The PL excitation spectrum of the film prepared at Ud = − 5 V was measured for the fixed PL wavelength of 530 nm. There is only one broad asymmetric peak in the range of at 330–480 nm in the PLE
(n = 2.0), CNx (n = 1.4–1.7) [24] and SiC (n = 2.4–2.8) films, the refractive indexes of 1.69–1.92 can indicate the availability of the mixed bonds Si–C–N–O in the deposited Si-C-N films It is well known that the value of the refractive index depends on film density, film morphology and film chemical stoichiometry [8,20,44,45]. We have shown above that an increase in substrate bias led to: densification of the films; lowering of the oxygen content; enhancement of the Si-C network. We suppose that it is these changes in the structure of the Si-C-N films that are responsible for an increase of the reflective index with negative subtract bias (cf. Fig. 8). The optical band gap, Eg, of the films is estimated from the Tauc plot [46]: (αhν)1/2 = A(hν − Eg), where α is the absorption coefficient, A is constant, hν is the incident photon energy. The Tauc plots for the films deposited at different Ud and Eg as a function of substrate bias are shown in Fig. 9. It is clearly seen that the optical band gap decreases
Fig. 9. The dependence of (αhν)1/2 on the photon energy hν, and dependence of the Eg on the negative substrate bias (inset). 70
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
Fig. 11. PLE spectra of the Si-C-N films deposited at different negative substrate bias. Numbers − 5 and − 150 denote Ud in V. Excitation wavelengths were in the range of 275–525 nm.
spectrum. Since for the low-biased film there is the large number of Si–N bonds, we could suppose that the Si-N and Si-N-O networks are responsible for the photoemission of this film. However, here we observed only one intensive broad peak instead of three narrow ones, as in the case of the high-biased film. The high disorder of the amorphous structure and a larger number of Si–N bonds in the low-biased film can give rise to the intensive broad PLE peak shown in Fig. 11. An appearance of the additional peaks at 650 nm in the PL spectrum and at 490 nm in the PLE spectrum of the high-biased film that are absent in the spectra of the low-biased film can be assigned to the Si-C and Si-C-N networks, in agreement with the results of the FTIR and XPS measurements. So, the results presented in Figs. 10 and 11 enable us to assume that the one-peak and two-peak structures of the PL spectra of the films are caused by the N-rich and C-rich Si-C-N-O-H networks, respectively.
Fig. 12. (a) Nanohardness and (b) elastic modulus as functions of indentation depth (L) of the films deposited at various substrate biases, Ud.
3.6. Mechanical properties study The dependence of nanohardness and elastic modulus values of the films deposited at different Ud as functions of indentation depth are shown in Fig. 12. The nanohardness and elastic modulus increase from 14.5 GPa to 24 GPa and from 147 GPa to 195 GPa, respectively, as the negative subtract bias increases from − 5 V to − 250 V. The values of H and E measured at the indentation depth of 60 nm as functions of Ud are shown in Fig. 13. Negative substrate bias was found to strongly influence on the mechanical properties of the Si-C-N films: the values of H and E increase with Ud. It is known that the nanohardness of amorphous Si-C-N films depends on the concentrations of Si–N and Si–C bonds [5,6]. In our case, an increase in Ud enhances the ion bombardment that promotes the cleaning of the grooving film surface from weakly bonded C-N, Si-H and Si-O clusters, and the released silicon and carbon atoms form new strong Si–C bonds which is confirmed by both FTIR and XPS measurements. It follows that the observed increase in hardness is due to strengthening Si-C network. The densification of the high-biased films [6,30] can also lead to the observed strength enhancement.
Fig. 13. (a) Nanohardness and (b) elastic modulus as functions of substrate bias, Ud.
low surface roughness. The films were characterized by FTIR, XPS and Raman spectroscopy. It was found that Si–C, Si–N and Si–O bonds are predominant. Hydrogen Si–H, C–H and N–H bonds were also detected in the films. These bonds form the amorphous hydrogenated Si-N-C-O-H network in the films. An increase in Ud enhances the ion bombardment that promotes the cleaning of the grooving film surface from weakly bonded C–N, Si–H and Si–O clusters, and the released silicon and carbon atoms form new strong Si–C bonds. The observed increase in hardness is supposed to be due to strengthening Si-C network. The deposited films display good transparency up to 90% in visible optic region. An increase of the refractive index with Ud is a sequence of the narrowing of the band gap caused by reducing Si–O and Si–H bonds and increasing Si–C bonds. All the films demonstrate the visible photoluminescence at room temperature. The photoluminescence spectra of the films prepared at Ud = − 5 V and − 50 V, have the maximum at 530 nm caused by the N-rich Si-C-N-O-H network. A further increase in substrate bias leads to shifting this maximum towards high wavelengths and to appearing an additional PL peak at 650 nm. These changes in the
4. Conclusion Hydrogenated Si-C-N films have been deposited by using PECVD from hexamethyldisilazane at different negative substrate biases ranged from − 5 to − 250 V. The deposited films were amorphous with ultra71
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
PL spectra are caused by strengthening Si–C bonds in Si-C-N-O-H network. The strengthening of Si–C bonds is also responsive for strength enhancement: the high-biased films demonstrate nanohardess above 24 GPa. The results of this investigation show that the Si-C-N films, prepared by the PECVD from HMDS due to their good optical properties, high hardness and low surface roughness can be recommended for their use in semiconductor devices that operate under extreme conditions.
(2011) 369–373. [21] I. Blaszczyk-Lezak, A.M. Wrobel, M.P.M. Kivitorma, I.J. Vayrynen, A. Tracz, Silicon carbonitride by remote microwave plasma CVD from organosilicon precursor: growth mechanism and structure of resulting Si:C:N films, Appl. Surf. Sci. 253 (2007) 7211–7218. [22] Y. Awad, M.A. El Khakani, M. Scarlete, C. Aktik, R. Smirani, N. Camiré, M. Lessard, J. Mouine, Structural analysis of silicon carbon nitride films prepared by vapor transport-chemical vapor deposition, J. Appl. Phys. 107 (2010) 033517. [23] A.C. Ferrari, S.E. Rodil, J. Robertson, Interpretation of infrared and Raman spectra of amorphous carbon nitrides, Phys. Rev. B 67 (2003) 155306. [24] M. Balaceanu, E. Grigore, G. Pavelescu, F. Ionescu, J.P. Roger, Optical characteristics of carbon nitride films prepared by hollow cathode discharge, J. Optoelectron. Adv. Mater. 2 (2000) 351–355. [25] S. Peter, S. Bernütz, S. Berg, F. Richter, S. Peter, FTIR analysis of a-SiCN:H films deposited by PECVD, Vacuum 98 (2013) 81–87. [26] W.F.A. Besling, A. Goossens, B. Meester, J. Schoonman, Laser-induced chemical vapor deposition of nanostructured silicon carbonitride thin films, J. Appl. Phys. 83 (1998) 544–553. [27] X.B. Yan, B.K. Tay, G. Chen, S.R. Yang, Synthesis of silicon carbide nitride nanocomposite films by a simple electrochemical method, Electrochem. Commun. 8 (2006) 737–740. [28] Z. Ma, J. Z, Z. Chen, E. Xie, Luminescence properties of terbium-doped SiCN thin films by rf magnetron reactive sputtering, Diam. Relat. Mater. 20 (2011) 475–479. [29] X. Xiao, Y. Li, L. Song, X. Peng, X. Hu, Structural analysis and microstructural observation of SiCxNy films prepared by reactive sputtering of SiC in N2 and Ar, Appl. Surf. Sci. 156 (2000) 155–160. [30] Y. Awad, M.A. El Khakani, C. Aktik, J. Mouine, N. Camiré, M. Lessard, M. Scarlete, H.A. Al-Abadleh, R. Smirani, Structural and mechanical properties of amorphous silicon carbonitride films prepared by vapor-transport chemical vapor deposition, Surf. Coat. Technol. 204 (2009) 539–545. [31] P. Gao, J. Xu, Y. Piao, W. Ding, D. Wang, X. Deng, C. Dong, Deposition of silicon carbon nitride thin films by microwave ECR plasma enhanced unbalance magnetron sputtering, Surf. Coat. Technol. 201 (2007) 5298–5301. [32] A. Majumdar, G. Das, N. Patel, P. Mishra, Debabrata Ghose, Rainer Hippler, Microstructural and chemical evolution of –CH3-incorporated (low-k) SiCO(H) films prepared by dielectric barrier discharge plasma, J. Electrochem. Soc. 155 (2008) D22–D26. [33] L. Colin, A. Cassuto, J.J. Ehrhardt, M.F. Ruiz-Lopez, D. Jamois, Adsorption and decomposition of hexamethyldisiloxane on platinum: an XPS, UPS and TDS study, Appl. Surf. Sci. 99 (1996) 245–254. [34] L. Štrbková, A. Manakhov, L. Zajíčková, A. Stoica, P. Veselý, R. Chmelík, The adhesion of normal human dermal fibroblasts to the cyclopropylamine plasma polymers studied by holographic microscopy, Surf. Coat. Technol. 295 (2016) 70–77. [35] M. Ahmed, J. Byrne, J. McLaughlin, W. Ahmed, Study of human serum albumin adsorption and conformational change on DLC and silicon doped DLC using XPS and FTIR spectroscopy, Biomater. Nanobiotechnol. 4 (2013) 194–203. [36] H.-L. Zhu, F.-D. Han, J.-Q. Bi, Y.-J. Bai, Y.-X. Qi, L.-L. Pang, C.-G. Wang, S.-J. Li, Facile synthesis of Si3N4 nanocrystals via an organic–inorganic reaction route, J. Am. Ceram. Soc. 92 (2009) 535–538. [37] V. Kulikovsky, R. Ctvrtlik, V. Vorlicek, V. Zelezny, P. Bohac, L. Jastrabik, Effect of air annealing on mechanical properties and structure of SiCxNy magnetron sputtered films, Surf. Coat. Technol. 240 (2014) 76–85. [38] A.C. Ferrari, J. Robertson, Resonant Raman spectroscopy of disordered, amorphous, and diamondlike carbon, Phys. Rev. B 64 (2000) 075414. [39] G. Ferini, G.A. Baratta, M.E. Palumbo, A Raman study of ion irradiated icy mixtures, Astron. Astrophys. 414 (2004) 757–766. [40] F. Neri, P. Tripodi, S. Trusso, G. Faggio, Optical and structural properties of silicon carbon nitride thin films deposited by reactive pulsed laser ablation, Radiat. Eff. Defects Solids: Inc. Plasma Sci. Plasma Technol. 165 (2010) 754–759. [41] V.I. Ivashchenko, V.I. Shevchenko, Effects of short-range disorder upon electronic properties of a-SiC alloys, Appl. Surf. Sci. 184 (1–4) (2001) 137–143. [42] R. Swanepoel, Determination of the thickness and optical constants of amorphous silicon, J. Phys. E: Sci. Instrum. 16 (1983) 1214–1222. [43] J. Ruiz-Perez, E. Marquez, D. Minkov, J. Reyes, J.B. Ramirez-Malo, P. Villares, R. Jimenez-Garay, Computation of the optical constants of thermally-evaporated thin films of GeSe2 chalcogenide glass from their reflection spectra, Phys. Scr. 53 (1996) 76–82. [44] E. Vassallo, A. Cremona, F. Ghezzi, F. Dellera, L. Laguardia, G. Ambrosone, U. Coscia, Structural and optical properties of amorphous hydrogenated silicon carbonitride films produced by PECVD, Appl. Surf. Sci. 252 (2006) 7993–8000. [45] P.S. Hoffmann, N.I. Fainer, O. Baake, M.L. Kosinova, Y.M. Rumyantsev, V.A. Trunova, A. Klein, B. Pollakowski, B. Beckhoff, W. Ensinger, Silicon carbonitride nanolayers—synthesis and chemical characterization, Thin Solid Films 520 (2012) 5906–5913. [46] J. Tauc, Absorption edge and internal electric fields in amorphous semiconductors, Mater. Res. Bull. 5 (1970) 721–729. [47] J. Fan, H. Li, J. Wang, M. Xiao, Fabrication and photoluminescence of SiC quantum dots stemming from 3C, 6H, and 4H polytypes of bulk SiC, Appl. Phys. Lett. 101 (2012) 131906. [48] Z. Chen, J. Zhou, X. Song, X. Xu, Y. Liu, J. Song, Z. Ma, E. Xie, The effect of annealing temperatures on morphologies and photoluminescence properties of terbium-doped SiCN films, Opt. Mater. 32 (2010) 1077–1084. [49] D. Hiller, A. Zelenina, S. Gutsch, S.A. Dyakov, L. Lopez-Conesa, J. Lopez-Vidrier, S. Estrade, F. Peiro, B. Garrido, J. Valenta, M. Korınek, F. Trojanek, P. Maly, M. Schnabel, C. Weiss, S. Janz, M. Zacharias, Absence of quantum confinement effects in the photoluminescence of Si3N4–embedded Si nanocrystals, J. Appl. Phys. 115 (2014) 204301. [50] K.-H. Lee, Y.-L. Du, T.-H. Lee, Photoluminescence and photoluminescence excitation from porous silicon carbide, Bull. Korean Chem. Soc. 21 (2000) 769–773.
Acknowledgements The authors thank the head of international laboratory “Surface Enhanced Spectroscopy” at the Institute of Physics NAS of Ukraine dr. Fesenko O.M. for the help in Raman studies; dr. Dub. S.M. (Institute for Superhard Material NASU) for help with nanoindentation measurements. References [1] I.V. Afanasyev-Charkin, M. Nastasi, Hard Si–N–C films with a tunable bandgap produced by pulsed glow discharge deposition, Surf. Coat. Technol. 199 (2005) 38–42. [2] W.-Y. Chang, C.-Y. Chang, J. Leu, Optical properties of plasma-enhanced chemical vapor deposited SiCxNy films by using silazane precursors, Thin Solid Films 636 (2017) 671–679. [3] A. Bendeddouche, R. Berjoan, E. Bêche, R. Hillel, Hardness and stiffness of amorphous SiCxNy chemical vapor deposited coatings, Surf. Coat. Technol. 111 (1999) 184–190. [4] H. Hoche, C. Pusch, R. Riedel, C. Fasel, A. Klein, Properties of SiCN coatings for high temperature applications—comparison of RF-, DC- and HPPMS-sputtering, Surf. Coat. Technol. 205 (2010) S21–S27. [5] A.O. Kozak, O.K. Porada, V.I. Ivashchenko, L.A. Ivashchenko, P.L. Scrynskyy, T.V. Tomila, V.S. Manzhara, Comparative investigation of Si-C-N Films prepared by plasma enhanced chemical vapour deposition and magnetron sputtering, Appl. Surf. Sci. 425 (2017) 646–653. [6] I. Blaszczyk-Lezak, A.M. Wrobel, D.M. Bielinski, Remote hydrogen microwave plasma chemical vapor deposition of silicon carbonitride films from a (dimethylamino)dimethylsilane precursor: compositional and structural dependencies of film properties, Diam. Relat. Mater. 15 (2006) 1650–1658. [7] C.W. Chen, C.C. Huang, Y.Y. Lin, L.C. Chen, K.H. Chen, W.F. Su, Optical properties and photoconductivity of amorphous silicon carbon nitride thin film and its application for UV detection, Diam. Relat. Mater. 14 (2005) 1010–1013. [8] S. Saloum, B. Alkhaled, Growth rate and sensing properties of plasmadeposited silicon organic thin films from hexamethyldisilazane compound, Acta Phys. Pol. A 117 (2010) 484–489. [9] T. Harada, H. Nakanishi, T. Ogata, Y. Kadotani, A. Izumi, Evaluation of corrosion resistance of SiCN-coated metals deposited on an NH3-radical-treated substrate, Thin Solid Films 519 (2011) 4487–4490. [10] H. Nakayama, M. Ito, Super H2O-barrier film using Cat-CVD (HWCVD)-grown SiCN for film-based electronics, Thin Solid Films 519 (2011) 4483–4486. [11] F. Kraus, S. Cruz, J. Müller, Plasmapolymerized silicon organic thin films from HMDSN for capacitive humidity sensors, Sens. Actuators B 88 (2003) 300–311. [12] V.I. Ivashchenko, A.O. Kozak, O.K. Porada, L.A. Ivashchenko, O.K. Sinelnichenko, O.S. Lytvyn, T.V. Tomila, V.J. Malakhov, Characterization of SiCN thin films: experimental and theoretical investigations, Thin Solid Films 569 (2014) 57–63. [13] N.-M. Park, S.H. Kim, G.Y. Sung, Band gap engineering of SiCN film grown by pulsed laser deposition, J. Appl. Phys. 94 (2003) 2725–2728. [14] P. Hoffmann, N. Fainer, M. Kosinova, O. Baake, W. Ensinger, Compilation on synthesis, characterization and properties of silicon and boron carbonitride films, Open access peer-reviewed Edited volume, Silicon Carbiden Materials, Processing and Applications in Electronic Devices, 2011 InTech, 2015, pp. 487–546, , https:// doi.org/10.5772/24330 (Chapter 21) Available from: https://www.intechopen. com/books/silicon-carbide-materials-processing-and-applications-in-electronicdevices/compilation-on-synthesis-characterization-and-properties-of-silicon-andboron-carbonitride-films. [15] O.K. Porada, V.S. Manzhara, A.O. Kozak, V.I. Ivashchenko, L.A. Ivashchenko, Photoluminescence properties of PECVD Si-C-N films, J. Nano Electron. Phys. 9 (2017) 02022. [16] W.C. Oliver, G.M. Pharr, An improved technique for determining hardness andelastic modulus using load and displacement sensing indentation experiments, J. Mater. Res. 1 (1992) 1564–1583. [17] S. Bulou, L. Brizoual, P. Miska, L. Poucques, R. Hugon, M. Belmahi, J. Bougdira, The influence of CH4 addition on composition, structure and optical characteristics of SiCN thin films deposited in a CH4/N2/Ar/hexamethyldisilazane microwave plasma, Thin Solid Films 520 (2011) 245–250. [18] Z. Khatami, C. Nowikow, J. Wojcik, P. Mascher, Annealing of silicon carbonitride nanostructured thin films: interdependency of hydrogen content, optical, and structural properties, J. Mater. Sci. 53 (2018) 1497–1513. [19] I. Ferreira, E. Fortunato, P. Vilarinho, A.S. Viana, A.R. Ramos, E. Alves, R. Martins, Hydrogenated silicon carbon nitride films obtained by HWCVDPA-HWCVD and PECVD technique, J. Non-Cryst. Solids 352 (2006) 1361–1366. [20] S. Saloum, B. Alkhaled, Structural, optical and electrical properties of plasmadeposited thin films from hexamethyldisilazane compound, Acta Phys. Pol. A119
72
Contents lists available at ScienceDirect
Materials Science in Semiconductor Processing journal homepage: www.elsevier.com/locate/mssp
Structural, optoelectronic and mechanical properties of PECVD Si-C-N films: An effect of substrate bias
T
⁎
A.O. Kozaka, , V.I. Ivashchenkoa, O.K. Poradaa, L.A. Ivashchenkoa, T.V. Tomilaa, V.S. Manjarab, G.V. Klishevychb a b
Institute for Problems of Materials Sciences NAS of Ukraine, 3, Krzhyzhanovsky Str., 03142 Kyiv, Ukraine Institute of Physics NAS of Ukraine, 46, Nauki ave., 03028 Kyiv, Ukraine
A R T I C LE I N FO
A B S T R A C T
Keywords: Amorphous Si-C-N films Plasma-enhanced chemical vapor deposition Photoluminescence Photoluminescence excitation Chemical bonding Nanoindentaion
Structural, optoelectronic and mechanical properties of amorphous silicon carbon nitride (Si-C-N) thin films produced by plasma enhanced chemical vapor deposition (PECVD) at different negative substrate biases (Ud) are studied. The films are characterized by X-ray diffraction, atomic force microscopy, Fourier transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), Raman spectroscopy, optical transmittance spectroscopy, nanoindentation as well using the results of the measurements of photoluminescence (PL) and PL excitation (PLE) spectra. All deposited films are found to be X-ray amorphous. An increase in Ud leads to: a smoothing of the film surface; a decrease of the transparency; an increase of refractive index from 1.69 to 1.92; a decrease of the energy gap from 4.15 to 2.38 eV; an increase in nanohardness and elastic modulus from 14 GPa to 24 GPa and from 147 GPa to 190 GPa, respectively. These results were explained in terms of the bonding configuration from XPS and FTIR measurements. The PL spectra of the films deposited at lower negative substrate bias have one PL band in the region between 530 and 540 nm, whereas the PL spectra of the films deposited at higher negative substrate bias show two PL bands at 530–570 nm and 640–650 nm. On the basis of the PLE data, it was shown that these PL bands are related to the electronic recombination between the conduction band and the valence bands and their tails within the amorphous N-rich and C-rich Si-C-N-O-H networks.
1. Introduction Up to date, thin silicon carbon nitride (Si-C-N) films are of interest due to the unique optoelectronic properties as a variable band gap in the range from 0.96 to 5.6 eV and refractive index between 1.44 and 2.2, high transparency in the visible range of spectra, on the one hand, and the excellent mechanical properties, namely, high hardness (up to 38 GPa), high structural and thermal stability until 1350 °C, low roughness, strong adhesion to a substrate, and good abrasive wear resistance [1–7]. The combination of these properties has enabled the efficient application of Si-C-N films in semiconductor devices and as wear-resistant coatings. In particular, they can be used as thin films with variable optical characteristics in ultraviolet light sensors, and as sensitive layers in gas sensors, as well as protective and wear-resistant coatings on optoelectronic devices and metal surfaces [7–10]. Different chemical vapor deposition (CVD) or physical vapor deposition (PVD) methods, such as plasma enhanced CVD (PECVD) [2,5,8,11,12], thermo-CVD (high temperature heating) [3], hot-wire CVD [9,10], magnetron sputtering [5], pulsed laser deposition [13] are ⁎
often used for the preparation of Si-C-N films. It is well established that film properties are sensitive to deposition parameters and precursors used. The main PECVD parameters are substrate temperature, discharge power and negative substrate bias (Ud). During deposition, two processes take place: i) film deposition due to the chemical reactions between adsorbed species on the growing film surface, and ii) the sputtering of the deposited film. The sputtering occurs due to the breaking of the weak bonds on the film surface, which promotes the densification of the films, and an increase of their quality [2,6]. These two processes are controlled by deposition conditions. The recent review of the properties of Si-C-N films deposited by using different methods was done in Ref. [14]. We note that, despite a huge experimental material accumulated on Si-C-N films, so far, the comprehensive studies of the Si-C-N films that could combine good optoelectronic and mechanical properties are in an infant stage. Such films might be used for the production of semiconductor devices for the exploitation under extreme conditions. In this work, Si-C-N films were produced by PECVD at different substrate biases using the liquid precursor hexamethyldisilazane
Corresponding author. E-mail address: [email protected] (A.O. Kozak).
https://doi.org/10.1016/j.mssp.2018.07.023 Received 5 April 2018; Received in revised form 13 June 2018; Accepted 15 July 2018 1369-8001/ © 2018 Elsevier Ltd. All rights reserved.
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
Table 1 Deposition parameters. Ud (V) FN2 (sccm) FH+HMDS (sccm) PC (Тоrr) TS (°С) PW (W/сm2)
− 5, − 50, − 100, − 150, − 250 0.5 12 0.2 450 0.2
Ud – substrate bias voltage, FN2 – nitrogen flow rate, FH+HMDS – hydrogen flow rate through a bubbler with the HMDS vapor, PC – working pressure, TS – substrate temperature, PW – discharge power density.
(HMDS) [5]. The optoelectronic, mechanical and structural properties of the deposited Si-C-N films were investigated using X-ray diffraction (XRD), atomic force microscopy (AFM), Fourier transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), Raman spectroscopy, optical transmittance spectroscopy, nanoindentation, as well the results of the measurements of photoluminescence (PL) and PL excitation (PLE) spectra. Below, for simplicity, the expression “increase (decrease) of substrate bias” will mean the change of its absolute value.
Fig. 1. X-ray patterns of Si-C-N films deposited at Ud = − 5 V (a) and Ud = − 250 V (b). The peaks at 2Θ ~ 33°, 62° and 69° correspond to the silicon substrate.
triangular pyramid diamond tip. Nanohardness (H) and elastic modulus (E) were determined according to the Oliver and Pharr method [16].
2. Experimental details Si-C-N films were produced using the HMDS vapor and nitrogen by means of PECVD. The negative bias was applied to a substrate by an additional radio-frequency (5.27 MHz) generator. The HMDS vapor was delivered into the reactor chamber by hydrogen from the thermostatcontrolled bubbler heated up to 40 °C. All deposition parameters are listed in Table 1. The films were deposited on both the (100) oriented silicon wafers and transparent quartz substrates. The surface oxide on the silicon wafer surfaces were cleaned by dipping into the 10% hydrofluoric acid for 3 min and the final etching of the substrates was carried out by treating with the hydrogen plasma into the reaction chamber. The film thickness was estimated by an optical interference profilometer. The film surface was analysed by AFM microscope “NanoScope IIIa Dimension 3000TM”. The structure of the film was investigated by Xray diffraction using a diffractometer “DRON-3M” in Cu Kα- radiation. FTIR measurements were carried out by using a spectrometer “FSM 1202” LLC “Infraspek” in the range of the wave numbers of 400–4000 cm−1. XPS core-level spectra were measured by a UHVAnalysis-System, SPECS, under the Mg Kα radiation (E = 1253.6 eV). XPS spectra were measured after argon etching during 3 min. The Raman spectra were measured in the range of 100–3200 cm−1 with the help of the Via Renishaw Raman microscope equipped with the He-Ne laser excitation at 632.8 nm. Measurements of the UV–VIS transmission spectra were carried out by using a two-beam optical spectrometer SPECORD-M40 in the range of wavelengths of 200–900 nm. Photoluminescence spectra (PL) were investigated at room temperature by using an experimental installation for the recording of PL spectra. The main unit of this installation is the automated monochromator SPM-2 (Carl Zeiss, Jena). The installation enables one to measure the intensity of PL spectra at a certain wavelength in the visible range. Photoluminescence is excited by a low powerful LED laser (~ 25 µW/ cm2, λ = 405 nm). Our analysis shows that this power is enough to excite PL from bulk states [15]. This laser does not heat films during optical measurements, which, in contrast to Hg lamps, does not lead to lowering the accuracy of measurements. The photoluminescence excitation (PLE) spectra of the deposited Si-C-N films were recorded using the HITACHI MPF-4 spectrofluorometer equipped with a xenon lamp. The conversion of its analogue output signals into a digital form was carried out using a digital converter. The excitation wavelengths were in the range of 320–525 nm (3.88–2.36 eV, respectively). Nanoindentaion of the films was carried out by the NanoIndenter G200 device (Agilent Technologies) using continuous stiffness measurement (CSM) mode. The indentations were produced with a Berkovich
3. Results and discussion 3.1. Film structure We carried out XRD measurements of the deposited films to analyze their microstructure. X-ray patterns of the Si-C-N films deposited on silicon substrates are shown in Fig. 1. There are no any features that could be assigned to any crystallites in the films. It follows that the deposited films are X-ray amorphous. 3.2. Film surface morphology and thickness Fig. 2(a) and (b) show the AFM images of the surface of the films deposited at Ud = − 5 V and Ud = − 250 V, respectively. The scanning area was 1 µm × 1 µm. From Fig. 2 one can observe that the films are very smooth and the film surfaces exhibit an excellent morphological homogeneity with low roughness. The values of the RMS roughness (Rq) and average roughness (Ra) of the films deposited at various substrate biases were Rq = 0.33 nm and Ra = 0.27 nm (Ud = − 5 V), Rq = 0.27 nm and Ra = 0.22 nm (Ud = − 250 V). We see that an increase in negative substrate bias leads to the insignificant smoothing of the film surface: substrate bias accelerates ions and charged particles, thereby promoting the etching of weak bonds and possibly the densification of the films. As a result, the roughness of the film surface slightly decreases and the film thickness increases from 200 nm to 700 nm with increasing Ud from − 5 V to − 250 V, respectively. 3.3. Film composition The chemical composition of the films was estimated using the XPS data. The Si, C, N, O contents in these films are approximately 36, 36, 14 and 14 at% and 37, 41, 13 and 9 at% for the films deposited at Ud = − 5 V and − 250 V, respectively. An increase of the concentration of carbon in the film is accompanied by decreasing the oxygen content, whereas the concentration of silicon and nitrogen remains unchanged. The oxygen in the films can be due to the residual oxygen absorbed on the reactor chamber walls. Since the reduction of the film roughness is accompanied with lowering the oxygen content some oxygen may come from the ambient air after deposition [17]. We could not estimate the hydrogen content in the films because the XPS analysis did not give such a possibility. However, in the FTIR spectra the hydrogen bonds are clearly seen. Also, it was found that hydrogen bands decrease with increasing substrate bias which indicates that the hydrogen content 66
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
Fig. 2. AFM images of the surface of the Si-C-N films deposited at Ud = − 5 V (a) and Ud = − 250 V (b).
1250–3500 cm−1. The band at 1549–1553 cm−1 can be assigned to C-C and/or C-N stretching modes. Three bands located in the range of 2160–2220 cm−1, 2865–2890 cm−1 and 3370–3385 cm−1can be caused by the vibration of the hydrogen bonds C≡N/Si–H, C–H and N–H bonds, respectively (cf. Table 2). We see that the small band related to the vibration of N–H bonds is present in all the spectra shown in Fig. 3a. This means that the reduction of the intensity of the band in the range of 1157–1170 cm−1 (cf. Fig. 3b) occurs due to decreasing a number of C–N bonds. The measured XPS core-level spectra of the deposited films are shown in Fig. 4. The intensity of the Si2p and N1s spectra weakly depend on substrate bias, whereas the intensity of the C1s spectra increases and the intensity of the O1s spectra decreases with increasing Ud. The peaks in the Si2p, C1s, N1s and O1s spectra of the films prepared at Ud = – 5 V centered at 102.3 eV, 284.8 eV, 398.7 eV and 532.9 eV, respectively, can be assigned to Si–O and/or Si-N [26], C–C [27] and/or C–N [28], N–C [29] and O–Si [26] bonds, respectively. As substrate bias rises, all the measured spectra shift toward low binding energies. For the film deposited at − 250 V, the Si2p peak at 101.5 eV is assigned to Si–N bonds [26,27], the C1s peak at 284.2 eV can be caused by C–C bonds [30], the N1s peak at 398 eV is associated with N–Si bonds [13] and the O1s peak at 532.4 eV is assigned to O–Si bonds [26]. The observed asymmetry of the XPS spectra points to that they are not formed by a single chemical bonding state.
decreases as substrate bias increases. Here it should be noted that the hydrogen was detected in the hydrogenated Si-C-N films using elastic recoil detection method [18]. 3.4. Chemical bonding configurations An analysis of the chemical bonding structure was performed by the FTIR, XPS and Raman spectroscopy measurements of the films deposited on Si substrates. Fig. 3(a) illustrates the FTIR spectra of the deposited films. Table 2 summarizes the identified FTIR absorption bands of the films deposited under different conditions. For peak identification, the literature data [11,13,19–26] is also presented. The wide absorption band ranging from 550 to 1250 cm−1 inherent to Si-CN films [5,12] can be related to the vibrations of Si–C, Si–N, Si–O, Si–H, C–N and N–H bonds (cf. Table 2). Fig. 3(a) shows that this band changes with increasing Ud. To gain more insight into the origin of this band, we decomposed it into several Gaussians, as shown in Fig. 3(b). The concentration of Si–O, Si–H, C–N/N–H bonds, estimated from the area of a corresponding Gaussian, decreases, and the concentration of Si–C bonds increases with increasing substrate bias. Such a re-distribution of bonding configuration could be caused by etching weak bonds by increasing ion bombardment. The concentration of Si–N bonds remains the highest independently of Ud. Other small absorption bands are observed in the range of
Fig. 3. (a) FTIR spectra of the films deposited at different Ud. Numbers − 5 to − 250 denote Ud in V. (b) FTIR deconvoluted spectra in the range of 550–1250 cm−1. The experimental FTIR spectra marked by open circles and their deconvoluted counterparts marked in red color. The inversion of the bands was done. 67
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
C–N [29–31], C–C [30,32] and/or C–H [33,34] and C–Si [35] bonds, respectively. With increasing negative substrate bias to −250 V, the area of the C-N and C-C/C-H components decreases, and the area of the C-Si component increases. The N1s spectra consist of two Gaussians at 398.07 eV and 399.03–399.18 eV indicating the existence of N–Si and N–C/N–H bonds [34], respectively. The N-C/N-H component decreases as Ud increases. The Si-N peak shifts towards high binding energies relatively the peak of the Si–N bonds in a-Si3N4 (397.64 eV) [36] that could indicate the formation of the Si-C-N network in the deposited films. Finally, the O1s spectrum of the film deposited at Ud = – 5 V was fitted by one Gaussian curve centered at 531.0 eV. It is assigned to Si–O bonds. A future increase of Ud resulted in increasing the asymmetry of this peak and, therefore, it was presented by two Gaussian curves centered at 530.9 eV and 533.0 eV. These curves are suggested to be formed by Si–O and C–O bonds, respectively. So, an increase in substrate bias gives rise to an appearance of C–O bonds. We note that the chemical binding states of the deposited Si-C-N films gained from the XPS measurements are consistent with those derived from the FTIR measurements. The results indicate the formation Si-C-N-O-H network in which Si–C and Si–N bonds are predominant. The ion bombardment induced by Ud promotes both the formation of strong Si–C bonds instead of weak C–N bonds and the breaking of Si–H bonds. Such changes in chemical bonding can cause the densification of the films [6]. The Raman spectra of the deposited films are shown in Fig. 6. These spectra are very useful for an analysis of the bonding configurations of carbon and silicon in carbon-based materials. It is well known that there are Si-rich and C-rich regions that coexist together with the Si-C clusters in Si-C-N films, and can be observed by the Raman spectroscopy [37]. In the Raman spectra, shown in Fig. 6, only one peak at 520 cm−1 related to the silicon substrate was detected. The absence of the carbon related peaks D (near 1360 cm−1), corresponding to the disorder-induced vibration modes of sp2 atoms in rings, and G (near 1560 cm−1) related to an in-plane bond stretching of sp2-hybridized carbon atoms in rings and in chains, is probably due to the fact that the intensive luminescence of the samples in the investigated area overshadows the non-intensive Raman peaks [38]. To reduce the intensity of the luminescence, the laser power was reduced to a minimum, but the luminescent bands remains very intense. This luminescence is typical for disordered polymeric carbons which contain a relatively large amount of nitrogen and hydrogen [39]. Obviously, strong luminescence caused by the presence of a large number of nitrogen and hydrogen in the films makes the identification of any peculiarities in the Raman spectra impossible [40]. Moreover, the Raman spectra, shown in Fig. 6, indicate the disorder of the Si-C-N network implying the absence of free amorphous carbon clusters. However, the results of the FTIR and XPS measurements indicate the presence of the C–C bonds in the films. We suppose that these bonds can be due to the single C–C bonds in the surrounding of the Si and N atoms [12,41].
Table 2 Peak positions in the FTIR spectra shown in Fig. 3 and their identification based on literature data. Observed FTIR band (cm−1)
Bonding type
Wavenumber (cm−1) from early publications [Ref.]
690–708a 796–799a 908–928a 1023–1035a
Si–H Si–C Si–N Si–O
690–710 [19] 770–810 [20] 900–940 [19] 1000–1030 [19] 1030–1050 [13]
1157–1170a
C–N N–H
1156–1198 [21] 1156–1198 [21] 1130–1160 [19]
1549–1553
C=C
1500–1850 1500–1600 1500–1850 1500–1600 1530–1725
[22] [23] [22] [23] [24]
2100–2150 2123–2202 2100–2300 2105–2173
[19] [24] [11] [25]
C=N
2160–2220
C≡N Si–H
2865–2890
C–Hn
2857–2890 [25]
3370–3385
N–Hn
3350–3400 [26] 3342–3378 [25]
a
Peak position in the deconvoluted spectra.
Fig. 4. Core level XPS spectra of the deposited films. The spectra were measured after argon etching during 3 min. Numbers − 5 and − 250 denote Ud in V.
3.5. Optical properties An optical energy gap (Eg) and refractive index (n) were estimated from the ultraviolet-visible transmission spectra. The optical transmission spectra of the Si-C-N films deposited on quartz substrates at different negative subtract biases are shown in Fig. 7a. All the films exhibit a sufficiently high optical transparency (up to 90%) in the range of 400–900 nm. An increase in negative substrate bias from − 5 V to − 250 V leads to darkening the film (Fig. 7b), which, most likely, is due to a decrease in the band gap and to an increase of film thickness. Besides, an increase of the carbon content can cause the shift of the edge of the optical absorption band toward high wavelengths [7]. The interference fringes, observed in the region with a low level of absorption (cf. Fig. 7a) is caused by multiple reflections between two interfaces: film/air and film/substrate. The interference
To gain the additional information on the bonding picture of the films we decomposed the XPS spectra onto separate Gaussians. The deconvoluted XPS spectra are shown in Fig. 5. The Si 2p spectra of each film are composed of three Gaussians. The deconvoluted peaks at 100.7–100.75 eV, 101.85–102 eV and 103 eV are attributed to Si–С [26,29], Si–N and Si–O [26] bonds, respectively. From Fig. 5 we see that the peak related to Si–N bonds is predominant over other peaks. Fig. 5 shows that the Si-N component decreases, whereas and the Si-C component increases with substrate bias in agreement with the results of the FTIR measurements (cf. Fig. 3). In the C1s spectra, three peaks centered at 285.9, 284.8–284.85 and 283.9 eV can be assigned to the 68
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
Fig. 5. Deconvolution of the XPS spectra of the films deposited at Ud= − 5V and Ud = − 250 V. The experimental XPS spectra marked by open circles and their deconvoluted counterparts marked in red color. Numbers − 5 and − 250 denote Ud in V.
Fig. 6. Raman spectra of the films deposited at different negative substrate bias (Normalized by band maximum).
fringes can appear in the transmittance spectra of a film provided it has: i) uniform thickness; ii) the smooth and homogenous film surface; iii) high structural perfection of the interfaces [42,43]. Interference fringes in transmittance spectra can be used to determine the refractive index, n. Estimation of the refractive index was done by using the method suggested by Swanepoel [42]. This method is based on drawing envelope curves through the extremes of the interference fringes in the transmission spectrum. The refractive index as a function of wavelength is calculated using the expression:
n=
Fig. 7. a) Optical transmission spectra; b) The images of the Si-C-N film deposited on quartz substrate (from left to right): uncoated substrate, film deposited at Ud = − 5 V, Ud = − 50 V, Ud = − 100 V, Ud = − 150 V, Ud = − 250 V.
N+ N 2 − ns2 ,
where
N = 2n s
Fig. 8 displays the refractive index as a function of wavelength for the films deposited under various Ud. An increase in the reflective index with increasing negative substrate bias is observed. The refractive index, evaluated at 632 nm (inset in Fig. 8), increases from 1.69 to 1.92 as the Ud increases from − 5 V to − 250 V, respectively. Taking into account the values of the refractive indexes of SiO2 (n = 1.47), Si3N4
n2+1 TM − Tm + s . TMTm 2
ns = 1.51, is the refractive index of a quartz substrate (transmittance of uncoated substrate is 92%), TM and Tm are maximum and minimum transmittances within the envelope curves, respectively. 69
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
Fig. 10. PL spectra of the Si-C-N films deposited at different negative substrate biases, Ud. (Excitation wavelength is 405 nm). Fig. 8. Refractive index vs. wavelength of the Si-C-N films, and the refractive index, evaluated at 632 nm (inset).
with increasing Ud. Such a behavior of the band gap depending on negative substrate bias can be explained in the terms of chemical bonding using the results of the FTIR and XPS measurements. The films prepared at low Ud have the largest value of Eg among other films due to the significant content of Si–N and Si–O bonds. An increase in negative substrate bias was shown to result in increasing Si–C bonds and reducing Si–O bonds. These changes in the bonding configuration of the films caused by increasing Ud will promote the narrowing of the band gap, since the energy gap in silicon carbide is lower than in silicon nitride and silicon oxide. Here, it should be noted also that high substrate biases give rise to intensive ion bombardment of the growing film surface causing its randomization and breaking of Si–H bonds. As a result, a number of gap states will rise, and the band gap will narrow. The photoluminescence (PL) spectra of the films deposited on silicon substrates under different Ud are presented in Fig. 10. We see that the PL spectra are very sensitive to substrate bias. The one-peak structure of the PL spectra is observed for the films deposited at Ud = − 5 V and Ud= − 50 V with the maximum at 530 nm, whereas the photoemission of the films deposited at highest substrate biases have two PL bands centered at 535 nm and 645 nm (Ud = − 150 V) and 570 nm and 640 nm (Ud = − 250 V). To gain more insight into the origin of the PL peaks let us analyze the photoluminescence excitation, PLE, spectra of the films deposited at Ud = − 5 V and − 150 V shown in Fig. 11. In the PLE measurements, the PL intensity at a fixed wavelength is determined as a function of wavelength of irradiation. Each PLE peak corresponds to a specific light absorption process. For intelligible interpretation, it would be appropriate to consider first the PLE spectra of the high-biased film. The PLE spectra of the film deposited at Ud = − 150 V was monitoring at 540 nm and 650 nm. There are three PLE peaks at 360 nm, 410 nm and 490 nm for this sample (cf. Fig. 11). The positions of these peaks practically coincide in both the PLE spectra under two monitoring wavelengths. This means that the peaks are caused by absorption process within one network formed trough the film. To identify these peaks we used literature data [47–50]. The PLE peak at 360 nm represents a characteristic light absorption line of cubic SiC [47], whereas the peak at 410 nm could be caused by the oxygen defects and the formed Si–C–O network within the film [48]. We note also that these two peaks are in the range of 340–400 nm that is characteristic to the light absorption in Si3N4 [49]. The peak position of 490 nm is very close to that of 500 nm in the PLE spectra of porous SiC [50]. The PL excitation spectrum of the film prepared at Ud = − 5 V was measured for the fixed PL wavelength of 530 nm. There is only one broad asymmetric peak in the range of at 330–480 nm in the PLE
(n = 2.0), CNx (n = 1.4–1.7) [24] and SiC (n = 2.4–2.8) films, the refractive indexes of 1.69–1.92 can indicate the availability of the mixed bonds Si–C–N–O in the deposited Si-C-N films It is well known that the value of the refractive index depends on film density, film morphology and film chemical stoichiometry [8,20,44,45]. We have shown above that an increase in substrate bias led to: densification of the films; lowering of the oxygen content; enhancement of the Si-C network. We suppose that it is these changes in the structure of the Si-C-N films that are responsible for an increase of the reflective index with negative subtract bias (cf. Fig. 8). The optical band gap, Eg, of the films is estimated from the Tauc plot [46]: (αhν)1/2 = A(hν − Eg), where α is the absorption coefficient, A is constant, hν is the incident photon energy. The Tauc plots for the films deposited at different Ud and Eg as a function of substrate bias are shown in Fig. 9. It is clearly seen that the optical band gap decreases
Fig. 9. The dependence of (αhν)1/2 on the photon energy hν, and dependence of the Eg on the negative substrate bias (inset). 70
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
Fig. 11. PLE spectra of the Si-C-N films deposited at different negative substrate bias. Numbers − 5 and − 150 denote Ud in V. Excitation wavelengths were in the range of 275–525 nm.
spectrum. Since for the low-biased film there is the large number of Si–N bonds, we could suppose that the Si-N and Si-N-O networks are responsible for the photoemission of this film. However, here we observed only one intensive broad peak instead of three narrow ones, as in the case of the high-biased film. The high disorder of the amorphous structure and a larger number of Si–N bonds in the low-biased film can give rise to the intensive broad PLE peak shown in Fig. 11. An appearance of the additional peaks at 650 nm in the PL spectrum and at 490 nm in the PLE spectrum of the high-biased film that are absent in the spectra of the low-biased film can be assigned to the Si-C and Si-C-N networks, in agreement with the results of the FTIR and XPS measurements. So, the results presented in Figs. 10 and 11 enable us to assume that the one-peak and two-peak structures of the PL spectra of the films are caused by the N-rich and C-rich Si-C-N-O-H networks, respectively.
Fig. 12. (a) Nanohardness and (b) elastic modulus as functions of indentation depth (L) of the films deposited at various substrate biases, Ud.
3.6. Mechanical properties study The dependence of nanohardness and elastic modulus values of the films deposited at different Ud as functions of indentation depth are shown in Fig. 12. The nanohardness and elastic modulus increase from 14.5 GPa to 24 GPa and from 147 GPa to 195 GPa, respectively, as the negative subtract bias increases from − 5 V to − 250 V. The values of H and E measured at the indentation depth of 60 nm as functions of Ud are shown in Fig. 13. Negative substrate bias was found to strongly influence on the mechanical properties of the Si-C-N films: the values of H and E increase with Ud. It is known that the nanohardness of amorphous Si-C-N films depends on the concentrations of Si–N and Si–C bonds [5,6]. In our case, an increase in Ud enhances the ion bombardment that promotes the cleaning of the grooving film surface from weakly bonded C-N, Si-H and Si-O clusters, and the released silicon and carbon atoms form new strong Si–C bonds which is confirmed by both FTIR and XPS measurements. It follows that the observed increase in hardness is due to strengthening Si-C network. The densification of the high-biased films [6,30] can also lead to the observed strength enhancement.
Fig. 13. (a) Nanohardness and (b) elastic modulus as functions of substrate bias, Ud.
low surface roughness. The films were characterized by FTIR, XPS and Raman spectroscopy. It was found that Si–C, Si–N and Si–O bonds are predominant. Hydrogen Si–H, C–H and N–H bonds were also detected in the films. These bonds form the amorphous hydrogenated Si-N-C-O-H network in the films. An increase in Ud enhances the ion bombardment that promotes the cleaning of the grooving film surface from weakly bonded C–N, Si–H and Si–O clusters, and the released silicon and carbon atoms form new strong Si–C bonds. The observed increase in hardness is supposed to be due to strengthening Si-C network. The deposited films display good transparency up to 90% in visible optic region. An increase of the refractive index with Ud is a sequence of the narrowing of the band gap caused by reducing Si–O and Si–H bonds and increasing Si–C bonds. All the films demonstrate the visible photoluminescence at room temperature. The photoluminescence spectra of the films prepared at Ud = − 5 V and − 50 V, have the maximum at 530 nm caused by the N-rich Si-C-N-O-H network. A further increase in substrate bias leads to shifting this maximum towards high wavelengths and to appearing an additional PL peak at 650 nm. These changes in the
4. Conclusion Hydrogenated Si-C-N films have been deposited by using PECVD from hexamethyldisilazane at different negative substrate biases ranged from − 5 to − 250 V. The deposited films were amorphous with ultra71
Materials Science in Semiconductor Processing 88 (2018) 65–72
A.O. Kozak et al.
PL spectra are caused by strengthening Si–C bonds in Si-C-N-O-H network. The strengthening of Si–C bonds is also responsive for strength enhancement: the high-biased films demonstrate nanohardess above 24 GPa. The results of this investigation show that the Si-C-N films, prepared by the PECVD from HMDS due to their good optical properties, high hardness and low surface roughness can be recommended for their use in semiconductor devices that operate under extreme conditions.
(2011) 369–373. [21] I. Blaszczyk-Lezak, A.M. Wrobel, M.P.M. Kivitorma, I.J. Vayrynen, A. Tracz, Silicon carbonitride by remote microwave plasma CVD from organosilicon precursor: growth mechanism and structure of resulting Si:C:N films, Appl. Surf. Sci. 253 (2007) 7211–7218. [22] Y. Awad, M.A. El Khakani, M. Scarlete, C. Aktik, R. Smirani, N. Camiré, M. Lessard, J. Mouine, Structural analysis of silicon carbon nitride films prepared by vapor transport-chemical vapor deposition, J. Appl. Phys. 107 (2010) 033517. [23] A.C. Ferrari, S.E. Rodil, J. Robertson, Interpretation of infrared and Raman spectra of amorphous carbon nitrides, Phys. Rev. B 67 (2003) 155306. [24] M. Balaceanu, E. Grigore, G. Pavelescu, F. Ionescu, J.P. Roger, Optical characteristics of carbon nitride films prepared by hollow cathode discharge, J. Optoelectron. Adv. Mater. 2 (2000) 351–355. [25] S. Peter, S. Bernütz, S. Berg, F. Richter, S. Peter, FTIR analysis of a-SiCN:H films deposited by PECVD, Vacuum 98 (2013) 81–87. [26] W.F.A. Besling, A. Goossens, B. Meester, J. Schoonman, Laser-induced chemical vapor deposition of nanostructured silicon carbonitride thin films, J. Appl. Phys. 83 (1998) 544–553. [27] X.B. Yan, B.K. Tay, G. Chen, S.R. Yang, Synthesis of silicon carbide nitride nanocomposite films by a simple electrochemical method, Electrochem. Commun. 8 (2006) 737–740. [28] Z. Ma, J. Z, Z. Chen, E. Xie, Luminescence properties of terbium-doped SiCN thin films by rf magnetron reactive sputtering, Diam. Relat. Mater. 20 (2011) 475–479. [29] X. Xiao, Y. Li, L. Song, X. Peng, X. Hu, Structural analysis and microstructural observation of SiCxNy films prepared by reactive sputtering of SiC in N2 and Ar, Appl. Surf. Sci. 156 (2000) 155–160. [30] Y. Awad, M.A. El Khakani, C. Aktik, J. Mouine, N. Camiré, M. Lessard, M. Scarlete, H.A. Al-Abadleh, R. Smirani, Structural and mechanical properties of amorphous silicon carbonitride films prepared by vapor-transport chemical vapor deposition, Surf. Coat. Technol. 204 (2009) 539–545. [31] P. Gao, J. Xu, Y. Piao, W. Ding, D. Wang, X. Deng, C. Dong, Deposition of silicon carbon nitride thin films by microwave ECR plasma enhanced unbalance magnetron sputtering, Surf. Coat. Technol. 201 (2007) 5298–5301. [32] A. Majumdar, G. Das, N. Patel, P. Mishra, Debabrata Ghose, Rainer Hippler, Microstructural and chemical evolution of –CH3-incorporated (low-k) SiCO(H) films prepared by dielectric barrier discharge plasma, J. Electrochem. Soc. 155 (2008) D22–D26. [33] L. Colin, A. Cassuto, J.J. Ehrhardt, M.F. Ruiz-Lopez, D. Jamois, Adsorption and decomposition of hexamethyldisiloxane on platinum: an XPS, UPS and TDS study, Appl. Surf. Sci. 99 (1996) 245–254. [34] L. Štrbková, A. Manakhov, L. Zajíčková, A. Stoica, P. Veselý, R. Chmelík, The adhesion of normal human dermal fibroblasts to the cyclopropylamine plasma polymers studied by holographic microscopy, Surf. Coat. Technol. 295 (2016) 70–77. [35] M. Ahmed, J. Byrne, J. McLaughlin, W. Ahmed, Study of human serum albumin adsorption and conformational change on DLC and silicon doped DLC using XPS and FTIR spectroscopy, Biomater. Nanobiotechnol. 4 (2013) 194–203. [36] H.-L. Zhu, F.-D. Han, J.-Q. Bi, Y.-J. Bai, Y.-X. Qi, L.-L. Pang, C.-G. Wang, S.-J. Li, Facile synthesis of Si3N4 nanocrystals via an organic–inorganic reaction route, J. Am. Ceram. Soc. 92 (2009) 535–538. [37] V. Kulikovsky, R. Ctvrtlik, V. Vorlicek, V. Zelezny, P. Bohac, L. Jastrabik, Effect of air annealing on mechanical properties and structure of SiCxNy magnetron sputtered films, Surf. Coat. Technol. 240 (2014) 76–85. [38] A.C. Ferrari, J. Robertson, Resonant Raman spectroscopy of disordered, amorphous, and diamondlike carbon, Phys. Rev. B 64 (2000) 075414. [39] G. Ferini, G.A. Baratta, M.E. Palumbo, A Raman study of ion irradiated icy mixtures, Astron. Astrophys. 414 (2004) 757–766. [40] F. Neri, P. Tripodi, S. Trusso, G. Faggio, Optical and structural properties of silicon carbon nitride thin films deposited by reactive pulsed laser ablation, Radiat. Eff. Defects Solids: Inc. Plasma Sci. Plasma Technol. 165 (2010) 754–759. [41] V.I. Ivashchenko, V.I. Shevchenko, Effects of short-range disorder upon electronic properties of a-SiC alloys, Appl. Surf. Sci. 184 (1–4) (2001) 137–143. [42] R. Swanepoel, Determination of the thickness and optical constants of amorphous silicon, J. Phys. E: Sci. Instrum. 16 (1983) 1214–1222. [43] J. Ruiz-Perez, E. Marquez, D. Minkov, J. Reyes, J.B. Ramirez-Malo, P. Villares, R. Jimenez-Garay, Computation of the optical constants of thermally-evaporated thin films of GeSe2 chalcogenide glass from their reflection spectra, Phys. Scr. 53 (1996) 76–82. [44] E. Vassallo, A. Cremona, F. Ghezzi, F. Dellera, L. Laguardia, G. Ambrosone, U. Coscia, Structural and optical properties of amorphous hydrogenated silicon carbonitride films produced by PECVD, Appl. Surf. Sci. 252 (2006) 7993–8000. [45] P.S. Hoffmann, N.I. Fainer, O. Baake, M.L. Kosinova, Y.M. Rumyantsev, V.A. Trunova, A. Klein, B. Pollakowski, B. Beckhoff, W. Ensinger, Silicon carbonitride nanolayers—synthesis and chemical characterization, Thin Solid Films 520 (2012) 5906–5913. [46] J. Tauc, Absorption edge and internal electric fields in amorphous semiconductors, Mater. Res. Bull. 5 (1970) 721–729. [47] J. Fan, H. Li, J. Wang, M. Xiao, Fabrication and photoluminescence of SiC quantum dots stemming from 3C, 6H, and 4H polytypes of bulk SiC, Appl. Phys. Lett. 101 (2012) 131906. [48] Z. Chen, J. Zhou, X. Song, X. Xu, Y. Liu, J. Song, Z. Ma, E. Xie, The effect of annealing temperatures on morphologies and photoluminescence properties of terbium-doped SiCN films, Opt. Mater. 32 (2010) 1077–1084. [49] D. Hiller, A. Zelenina, S. Gutsch, S.A. Dyakov, L. Lopez-Conesa, J. Lopez-Vidrier, S. Estrade, F. Peiro, B. Garrido, J. Valenta, M. Korınek, F. Trojanek, P. Maly, M. Schnabel, C. Weiss, S. Janz, M. Zacharias, Absence of quantum confinement effects in the photoluminescence of Si3N4–embedded Si nanocrystals, J. Appl. Phys. 115 (2014) 204301. [50] K.-H. Lee, Y.-L. Du, T.-H. Lee, Photoluminescence and photoluminescence excitation from porous silicon carbide, Bull. Korean Chem. Soc. 21 (2000) 769–773.
Acknowledgements The authors thank the head of international laboratory “Surface Enhanced Spectroscopy” at the Institute of Physics NAS of Ukraine dr. Fesenko O.M. for the help in Raman studies; dr. Dub. S.M. (Institute for Superhard Material NASU) for help with nanoindentation measurements. References [1] I.V. Afanasyev-Charkin, M. Nastasi, Hard Si–N–C films with a tunable bandgap produced by pulsed glow discharge deposition, Surf. Coat. Technol. 199 (2005) 38–42. [2] W.-Y. Chang, C.-Y. Chang, J. Leu, Optical properties of plasma-enhanced chemical vapor deposited SiCxNy films by using silazane precursors, Thin Solid Films 636 (2017) 671–679. [3] A. Bendeddouche, R. Berjoan, E. Bêche, R. Hillel, Hardness and stiffness of amorphous SiCxNy chemical vapor deposited coatings, Surf. Coat. Technol. 111 (1999) 184–190. [4] H. Hoche, C. Pusch, R. Riedel, C. Fasel, A. Klein, Properties of SiCN coatings for high temperature applications—comparison of RF-, DC- and HPPMS-sputtering, Surf. Coat. Technol. 205 (2010) S21–S27. [5] A.O. Kozak, O.K. Porada, V.I. Ivashchenko, L.A. Ivashchenko, P.L. Scrynskyy, T.V. Tomila, V.S. Manzhara, Comparative investigation of Si-C-N Films prepared by plasma enhanced chemical vapour deposition and magnetron sputtering, Appl. Surf. Sci. 425 (2017) 646–653. [6] I. Blaszczyk-Lezak, A.M. Wrobel, D.M. Bielinski, Remote hydrogen microwave plasma chemical vapor deposition of silicon carbonitride films from a (dimethylamino)dimethylsilane precursor: compositional and structural dependencies of film properties, Diam. Relat. Mater. 15 (2006) 1650–1658. [7] C.W. Chen, C.C. Huang, Y.Y. Lin, L.C. Chen, K.H. Chen, W.F. Su, Optical properties and photoconductivity of amorphous silicon carbon nitride thin film and its application for UV detection, Diam. Relat. Mater. 14 (2005) 1010–1013. [8] S. Saloum, B. Alkhaled, Growth rate and sensing properties of plasmadeposited silicon organic thin films from hexamethyldisilazane compound, Acta Phys. Pol. A 117 (2010) 484–489. [9] T. Harada, H. Nakanishi, T. Ogata, Y. Kadotani, A. Izumi, Evaluation of corrosion resistance of SiCN-coated metals deposited on an NH3-radical-treated substrate, Thin Solid Films 519 (2011) 4487–4490. [10] H. Nakayama, M. Ito, Super H2O-barrier film using Cat-CVD (HWCVD)-grown SiCN for film-based electronics, Thin Solid Films 519 (2011) 4483–4486. [11] F. Kraus, S. Cruz, J. Müller, Plasmapolymerized silicon organic thin films from HMDSN for capacitive humidity sensors, Sens. Actuators B 88 (2003) 300–311. [12] V.I. Ivashchenko, A.O. Kozak, O.K. Porada, L.A. Ivashchenko, O.K. Sinelnichenko, O.S. Lytvyn, T.V. Tomila, V.J. Malakhov, Characterization of SiCN thin films: experimental and theoretical investigations, Thin Solid Films 569 (2014) 57–63. [13] N.-M. Park, S.H. Kim, G.Y. Sung, Band gap engineering of SiCN film grown by pulsed laser deposition, J. Appl. Phys. 94 (2003) 2725–2728. [14] P. Hoffmann, N. Fainer, M. Kosinova, O. Baake, W. Ensinger, Compilation on synthesis, characterization and properties of silicon and boron carbonitride films, Open access peer-reviewed Edited volume, Silicon Carbiden Materials, Processing and Applications in Electronic Devices, 2011 InTech, 2015, pp. 487–546, , https:// doi.org/10.5772/24330 (Chapter 21) Available from: https://www.intechopen. com/books/silicon-carbide-materials-processing-and-applications-in-electronicdevices/compilation-on-synthesis-characterization-and-properties-of-silicon-andboron-carbonitride-films. [15] O.K. Porada, V.S. Manzhara, A.O. Kozak, V.I. Ivashchenko, L.A. Ivashchenko, Photoluminescence properties of PECVD Si-C-N films, J. Nano Electron. Phys. 9 (2017) 02022. [16] W.C. Oliver, G.M. Pharr, An improved technique for determining hardness andelastic modulus using load and displacement sensing indentation experiments, J. Mater. Res. 1 (1992) 1564–1583. [17] S. Bulou, L. Brizoual, P. Miska, L. Poucques, R. Hugon, M. Belmahi, J. Bougdira, The influence of CH4 addition on composition, structure and optical characteristics of SiCN thin films deposited in a CH4/N2/Ar/hexamethyldisilazane microwave plasma, Thin Solid Films 520 (2011) 245–250. [18] Z. Khatami, C. Nowikow, J. Wojcik, P. Mascher, Annealing of silicon carbonitride nanostructured thin films: interdependency of hydrogen content, optical, and structural properties, J. Mater. Sci. 53 (2018) 1497–1513. [19] I. Ferreira, E. Fortunato, P. Vilarinho, A.S. Viana, A.R. Ramos, E. Alves, R. Martins, Hydrogenated silicon carbon nitride films obtained by HWCVDPA-HWCVD and PECVD technique, J. Non-Cryst. Solids 352 (2006) 1361–1366. [20] S. Saloum, B. Alkhaled, Structural, optical and electrical properties of plasmadeposited thin films from hexamethyldisilazane compound, Acta Phys. Pol. A119
72