- Email: [email protected]
Characterization of bias magnetron-sputtered silicon nitride films
Thin Solid Films 478 (2005) 256 – 260 www.elsevier.com/locate/tsf
Characterization of bias magnetron-sputtered silicon nitride films S. Guruvenketa, Jay Ghatakb, P.V. Satyamb, G. Mohan Raoa,* a
Department of Instrumentation, Indian Institute of Science, Malleswaram, Bangalore 560 012, India b Institute of Physics, Bhubaneswar 751 005, India Received 14 February 2004; received in revised form 7 October 2004; accepted 25 October 2004 Available online 26 November 2004
Abstract Influence of the deposition parameters and the substrate bias voltage on the optical, compositional and the surface properties of DC magnetron-sputtered silicon nitride thin films are studied. Silicon nitride thin films are deposited on silicon (100) and quartz substrates at different partial pressures of nitrogen and discharge currents. The variation in the refractive index and the optical band gap of these films is studied. Compositional variation has been studied using Rutherford backscattering spectroscopy (RBS). Silicon nitride thin films deposited at 3102 Pa partial pressure of nitrogen with 2.5 mA/cm2 cathode current density showed an optical band gap of 4.3 eV and refractive index of 2.04 (at 650 nm). Nitrogen to silicon ratio in the film is 1.31, and the roughness of the films is 2.3 nm. Substrate bias during deposition helped in changing the optical properties of the films. Substrate bias of 60 V resulted in films having near stoichiometry with N/Si ratio 1.32, and the optical band gap, refractive index, and the roughness are 4.8 eV, 1.92 and 0.78 nm, respectively. D 2004 Elsevier B.V. All rights reserved. Keywords: Magnetron sputtering; Silicon nitride; Ion-assisted deposition; Substrate bias; Rutheford backscattering
1. Introduction Silicon nitride is an important material in the fabrication of the microelectronic devices [1]. Its properties, like high thermal stability, chemical inertness, hardness, and good dielectric behavior, make it more attractive and most commonly used material [2]. Transparency of silicon nitride over a wide spectral range from near-ultraviolet (UV) to infrared (IR) region along with the passivating property makes it suitable for many optical applications [3]. Conventionally, silicon nitride thin films are deposited at high substrate temperatures (700–900 8C) by chemical vapor deposition (CVD) [4]. In most of the abovementioned applications, a lowtemperature deposition is preferred. However, films deposited with the low-temperature CVD technique show hydrogen entrapment, which deteriorates the properties of
* Corresponding author. Tel.: +91 80 2932349; fax: +91 80 3600135. E-mail address: [email protected] (G.M. Rao). 0040-6090/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2004.10.031
the silicon nitride thin films [5]. Several low-temperature techniques, like plasma-enhanced CVD and reactive sputtering, etc., have been adopted to deposit silicon nitride thin films. Reactive sputtering at low temperature has been found to be the best alternative, which can result in films with minimum hydrogen content [6]. In sputter deposition, additional ion assistance during the film growth helps in improving the properties of the growing film. Ion assistance can be accomplished either by using a separate ion source (ion-beam-assisted deposition) or by biasing the substrate during deposition (bias sputtering). In the present study, we investigated the effect of the substrate bias and the influence of the partial pressure of nitrogen gas, together with cathode current density on the optical, surface, and compositional properties of silicon nitride thin films prepared by DC-reactive magnetron sputtering. We correlate the composition and microstructure of the deposited films to the observed changes in the properties of the films, which are due to substrate bias effects.
S. Guruvenket et al. / Thin Solid Films 478 (2005) 256–260
257
2. Experiment section We used a homemade magnetron-sputtering system for this study. A planar silicon disk (99.99% pure and 100 mm diameter) is mechanically clamped to the magnetron sputtering cathode assembly. The ultimate pressure obtained in the sputtering chamber is 103 Pa using a combination of diffusion and rotary pump. Sputtering was done in Ar+N2 mixed gas, and the total pressure during the sputtering was maintained at 101 Pa. The partial pressures of the two gases are controlled individually. The distance between the substrate and the target is maintained at 100 mm. Polished Si (100) (p-type, 10 Vcm resistivity) was used as the substrate, which was maintained at the ambient temperature without any deliberate heating. For the optical studies, films were deposited on quartz substrates. The deposition was carried out at different partial pressures (2102, 3102, and 4102 Pa,) of nitrogen while maintaining the total pressure at 101 Pa and at different cathode current density of 1.9, 2.5, and 3.10 mA/cm2. A bias potential was applied to the substrate and varied in steps of 20 V from 0 to 120 V for the films deposited at 3102 Pa partial pressure of nitrogen and with the cathode current density of 2.5 mA/ cm2. Optical characterization of the silicon nitride films was carried out using Fourier transform infrared (FTIR) spectroscopy (Perkin-Elmer) with 2 cm1 resolution and UV–VIS spectroscopy (Hitachi 330A). Compositional analysis of the films was carried out using Rutherford backscattering spectroscopy (RBS), using 2 MeV, He+ ion beam. Spectra were analyzed using GISA simulation code [7]. Surface roughness of the films was measured using atomic force microscopy (AFM). Thickness of the silicon nitride films used in this studies varied from 270 to 300 nm (measured using the stylus profiler).
Fig. 1. FTIR spectra of the silicon nitride film deposited at different nitrogen partial pressures (a) 2102, (b) 3102, and (c) 4102 Pa.
3.1. Spectral studies
substrate bias voltages. In the case of ion-assisted deposition, it has been generally observed that the packing density will be high, and hence the water vapor absorption would be less. It can be observed from Fig. 2 that, with the increase in the bias voltage, there is no absorption band at 3340 cm1. It can also be observed that only the films deposited at higher negative bias voltage (above 40 V ) show absorption band at 2190 cm1, which is due to Si– N2 stretching band [9]. All the films showed an absorption band in the range 840–870 cm1, which corresponds to the stretching vibration mode of the Si–N bond. In Fig. 2, it can be noticed that, with increase in the substrate bias voltage, the absorption band of Si–N shifts towards the higher energy value. This shift could be due to the increased nitrogen incorporation in the film. Optical band gap of the silicon nitride films has been calculated from the optical spectra using the Tauc’s relation [10]. ðahmÞ1=2 ¼ A hm Eg ;
FTIR spectra have been used as the basic tool for identifying silicon nitride formation in the films deposited under different conditions. All the spectra recorded have been corrected for the absorption due to substrate. Fig. 1 shows the FTIR spectra of the silicon nitride films deposited at a cathode current density of 2.5 mA/cm2 with nitrogen partial pressure of 2102, 3102, and 4102 Pa. These spectra show a small absorption band around 3340 cm1, which corresponds to hydrogen bonding groups. As no hydrogen-based gas has been used in the process, this band could be due to the water vapor absorption in the silicon nitride films after the deposition [8]. Fig. 2 shows the FTIR spectra of silicon nitride thin film deposited at 3102 Pa partial pressure of nitrogen and cathode current density of 2.5 mA/cm2 at different
where a is the absorption coefficient, A is a constant, hr is the photon energy, and E g is the band gap. Optical band gap value for the films deposited at various conditions is shown in Table 1. It can be seen from Table 1 that, at lower cathode current density, lower optical band gap value of 2.5 eV is obtained. With increase in the partial pressure of nitrogen and cathode current density, optical band gap value increased up to 4.5 eV. Variation in the band gap as the function of the substrate bias voltage for the films deposited at the cathode current density of 2.5 mA/cm2 and 3102 Pa partial pressure of nitrogen is shown in Fig. 3. When the substrate bias voltage is increased, it is observed that the optical band gap value increases gradually from 4.3 eV and attains a constant value around 4.8 eV at the bias voltage of 60 V, which is much closer to the optical band gap of the
3. Results and discussion
258
S. Guruvenket et al. / Thin Solid Films 478 (2005) 256–260
Fig. 3. Variation in the refractive index and the optical band gap as a function of substrate bias voltage (cathode current density of 2.5 mA/cm2 and at a nitrogen partial pressure of 3102 Pa). Fig. 2. FTIR spectra of the silicon nitride film deposited at nitrogen partial pressure of nitrogen 3102 Pa and cathode current density of 2.5 mA/cm2 at (a) 40, (b) 60, and (c) 80 V bias voltage.
bulk silicon nitride (5.0 eV). Higher substrate bias voltage of 100 V showed decrease in the energy band gap to 4.6 eV, which may be due to the increases in the structural disorder in the films (caused by energetic ion bombardment) and also due to reduced nitrogen content as discussed later. Kim and Chung [11] reported similar observations. Refractive index of the silicon nitride films was calculated using envelope technique of Swanepoel [12]. The variation in the refractive index (at 650 nm) with the deposition conditions is shown in Table 2. Higher refractive index values (2.24 and 2.15) at the higher cathode current density may be due to the excess silicon present in the film (refractive index of bulk silicon nitride film is 2.02). Fig. 3 shows the change in the refractive index of the silicon nitride thin film as a function of substrate bias voltage (deposited at 3102 Pa partial pressure of nitrogen, 2.5 mA/cm2 cathode current density). With increase in the substrate bias voltage to 40 V and above, it can be noticed that the refractive index value decreases from 2.04 and attains saturation value of about 1.92. The observed refractive index values are in good agreement with the reported data. Deenamma Vargheese and Rao [8] have reported a refractive index between 2.2 to 1.8 for the films deposited by electron-cyclotron-resonance (ECR)-assisted sputtering. Kim and Chung have reported a refractive index
between 1.9 and 2 for the films deposited by reactive bias sputtering [11]. 3.2. Compositional analysis The composition of the films was determined using RBS analysis. Table 3 shows the composition (N/Si ratio) of the films deposited at a nitrogen partial pressure of 3102 Pa and different cathode current densities. It can be clearly seen that the composition is sensitive to the cathode current, and optimum ratio of 1.3 is achieved for a current density of 2.5 mA/cm2. Fig. 4 shows the RBS spectra of the silicon nitride thin films deposited at 3102 Pa partial pressure of nitrogen at cathode current density of 2.5 mA/cm2. As discussed earlier, a near stoichiometric film is obtained for the films deposited under these conditions. This condition is chosen for further studies, and the substrate bias voltage is changed from 20 to 120 V in steps of 20 V, and the change in the properties is studied. The effect of substrate bias on the composition has been seen to be insignificant, where the N/Si ratio was about 1.3. Thus, it is obvious from the above discussion that stoichiometric silicon nitride films could be deposited at 3102 Pa of nitrogen partial pressure, cathode current density of 2.5 mA/cm2, and a substrate bias of 60 V. The RBS spectra of the films deposited under these optimized conditions is shown in
Table 1 Change in the optical band gap as a function of nitrogen partial pressure and cathode current density
Table 2 Change in the refractive index as a function of nitrogen partial pressure and cathode current density
Cathode current density (mA/cm2)
Partial pressure of nitrogen (Pa) 3102
4102
Cathode current density (mA/cm2)
Partial pressure of nitrogen (Pa)
2102
2102
3102
4102
1.90 2.50 3.10
3.1 2.8 2.8
3.3 4.3 4.4
3.7 4.4 4.5
1.90 2.50 3.10
1.84 1.98 2.24
1.90 2.04 2.15
1.95 2.12 2.08
S. Guruvenket et al. / Thin Solid Films 478 (2005) 256–260
259
Table 3 Ratio of N/Si in silicon nitride thin film deposited at 3102 Pa nitrogen partial pressure as a function of cathode current density Cathode current density (mA/cm2)
N/Si
1.9 2.5 3.1
1.2 1.3 1.4
Fig. 5. The change in the composition of the films is reflected in the properties of the film as discussed earlier. This kind of compositional variation is observed earlier by Deenamma Vargheese and Rao, with change in the ECR power instead of the applied bias voltage [8]. Vila et al. has observed the change in the composition of silicon and silicon nitride target sputtered with and without nitrogen in argon ambient [13]. They observed a small amount of trapped argon in the silicon nitride thin films, but, in our case, we did not observe the presence of Argon, but all the films showed presence of a small amount of oxygen (about 0.05). This could be due to the residual oxygen present in the sputtering chamber as the base pressure before deposition was only 103 Pa. 3.3. Surface morphology Surface roughness of the silicon nitride thin films deposited at the different substrate bias voltages (deposition condition 3102 Pa with 2.5 mA/cm2) were measured using an AFM. Sampling area over the surface of the film was 33 Am, and the images were checked at different scan regions on the substrate. With increase in the bias voltage, the roughness value (mean square root) of the film decreased. Films deposited without bias showed a roughness value of 2.3 nm, and, with increase in bias voltage to 60 V, the film roughness decreased to 1.4 nm, and further increase in the bias to 80 V bias resulted in a roughness of 0.8 nm. Beyond 80 V bias voltage, the
Fig. 4. RBS spectrum of the silicon nitride film deposited at a cathode current density of 2.5 mA/cm2 and at a nitrogen partial pressure of 3102 Pa.
Fig. 5. RBS spectrum of the Silicon nitride film deposited at. cathode current density of 2.5 mA/cm2 and at a nitrogen partial pressure of 3102 Pa with 60 V substrate bias voltage.
roughness increased to 1.0 nm, as shown in Fig. 6. The surface features of the film deposited without any bias exhibits a rough surface, which corresponds to the coarse grains. This leads to a columnar microstructural growth with voids because of the shadowing effect and low mobility of the ad atom. With increase in the substrate bias voltage to 60 V, it can be clearly observed that the surface roughness decreases, which is due to the increase in the ad atom mobility. With further increase in the bias voltage to 80 V, the surface becomes smoother with minimum voids. This clearly indicates the densification of the film due to the ion bombardment. It is well known that, with the increase in the substrate bias, the ions present in the glow will be accelerated towards the substrate and helps in improving the ad atom mobility, which in turn decreases the shadowing effect, thus leading to densification of the film. Similar changes have been
Fig. 6. Variation in the surface roughness with the applied bias voltage (at a cathode current density of 2.5 mA/cm2 and at a nitrogen partial pressure of 3102 Pa).
260
S. Guruvenket et al. / Thin Solid Films 478 (2005) 256–260
observed earlier in case of silicon nitride films deposited by ECR-plasma-assisted sputtering [8]. Furthermore, the changes in densification reported here are in agreement with the simulated data of Muller [14]; Kim and Chung also observed a similar transition in the surface morphology in bias sputtered silicon nitride films [11].
4. Conclusions Thin films of silicon nitride were deposited by reactive DC magnetron sputtering at different partial pressures of nitrogen and at different cathode current density. It is observed that the films deposited at 3102 Pa partial pressure of nitrogen with 2.5 mA/cm2 cathode current density had a near stoichiometry ratio of N/Si, optical band gap and refractive index values nearer to the bulk values. The substrate bias voltage was applied during the deposition of the film at the abovementioned condition, and its effect over the composition, optical, and surface property is studied. The substrate bias was found to have minimum effect on the composition, except at higher bias where the nitrogen content in the film reduced due to resputtering effects. However, the surface roughness was dependent on the substrate bias. A minimum surface roughness value of 0.78 nm is observed at 80 V bias. It is found that the films deposited at 60 V bias showed films with better optical properties (refractive index and optical band gap values nearer to bulk values). It is also observed that N/Si ratio is about 1.3 near to the stoichiometry of silicon nitride thin film. This effect of improvement in the properties due to substrate bias is
contributed to the ad atom mobility induced as the effect of bias voltage.
Acknowledgment The authors are thankful to Prof. K.V. Rao, KTH, Stockholm, and I. Nithiya Priya for their help in the AFM studies.
References [1] T.P. Ma, IEEE Trans. Electron Devices 45 (1998) 680. [2] E. Poloura, K. Nauka, J. Lagowsky, H.C. Gatos, Appl. Phys. Lett. 49 (1986) 97. [3] M. Vetter, Thin Solid Films 337 (1999) 118. [4] C.S. McCormick, C.E. Weber, J.R. Abelson, G.A. Davis, R.E. Weiss, V. Aebi, J. Vac. Sci. Technol., A, Vac. Surf. Films 15 (1997) 2770. [5] M.F. Lambrinos, R. Valizadeh, J.S. Colliyon, Appl. Opt. 35 (1996) 3620. [6] S.A. Awan, R.D. Gould, S. Gravano, Thin Solid Films 355–356 (1999) 456. [7] J. Saarilahti, E. Rauhala, Nucl. Instrum. Methods Phys. Res., B Beam Interact. Mater. Atoms 64 (1992) 734. [8] K. Deenamma Vargheese, G. Mohan Rao, J. Vac. Sci. Technol., A, Vac. Surf. Films 19 (2001) 1336. [9] Lin Huang, K.M. Hipps, J.T. Dickinson, Ursula Mazur, X.D. Wang, Thin Solid Films 299 (1997) 104. [10] J. Bauer, Phys. Status Solidi, A Appl. Res. 39 (1977) 411. [11] J.K. Kim, K.W. Chung, J. Appl. Phys. 83 (1998) 5831. [12] R. Swanepoel, J. Phys. E 16 (1983) 1214. [13] M. Vila, C. Prieto, J. Garica-Lopez, M.A. Respuldiza, Nucl. Instrum. Methods Phys. Res., B Beam Interact. Mater. Atoms 21 (2003) 199. [14] K.H. Muller, J. Appl. Phys. 62 (1987) 1796.
Characterization of bias magnetron-sputtered silicon nitride films S. Guruvenketa, Jay Ghatakb, P.V. Satyamb, G. Mohan Raoa,* a
Department of Instrumentation, Indian Institute of Science, Malleswaram, Bangalore 560 012, India b Institute of Physics, Bhubaneswar 751 005, India Received 14 February 2004; received in revised form 7 October 2004; accepted 25 October 2004 Available online 26 November 2004
Abstract Influence of the deposition parameters and the substrate bias voltage on the optical, compositional and the surface properties of DC magnetron-sputtered silicon nitride thin films are studied. Silicon nitride thin films are deposited on silicon (100) and quartz substrates at different partial pressures of nitrogen and discharge currents. The variation in the refractive index and the optical band gap of these films is studied. Compositional variation has been studied using Rutherford backscattering spectroscopy (RBS). Silicon nitride thin films deposited at 3102 Pa partial pressure of nitrogen with 2.5 mA/cm2 cathode current density showed an optical band gap of 4.3 eV and refractive index of 2.04 (at 650 nm). Nitrogen to silicon ratio in the film is 1.31, and the roughness of the films is 2.3 nm. Substrate bias during deposition helped in changing the optical properties of the films. Substrate bias of 60 V resulted in films having near stoichiometry with N/Si ratio 1.32, and the optical band gap, refractive index, and the roughness are 4.8 eV, 1.92 and 0.78 nm, respectively. D 2004 Elsevier B.V. All rights reserved. Keywords: Magnetron sputtering; Silicon nitride; Ion-assisted deposition; Substrate bias; Rutheford backscattering
1. Introduction Silicon nitride is an important material in the fabrication of the microelectronic devices [1]. Its properties, like high thermal stability, chemical inertness, hardness, and good dielectric behavior, make it more attractive and most commonly used material [2]. Transparency of silicon nitride over a wide spectral range from near-ultraviolet (UV) to infrared (IR) region along with the passivating property makes it suitable for many optical applications [3]. Conventionally, silicon nitride thin films are deposited at high substrate temperatures (700–900 8C) by chemical vapor deposition (CVD) [4]. In most of the abovementioned applications, a lowtemperature deposition is preferred. However, films deposited with the low-temperature CVD technique show hydrogen entrapment, which deteriorates the properties of
* Corresponding author. Tel.: +91 80 2932349; fax: +91 80 3600135. E-mail address: [email protected] (G.M. Rao). 0040-6090/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2004.10.031
the silicon nitride thin films [5]. Several low-temperature techniques, like plasma-enhanced CVD and reactive sputtering, etc., have been adopted to deposit silicon nitride thin films. Reactive sputtering at low temperature has been found to be the best alternative, which can result in films with minimum hydrogen content [6]. In sputter deposition, additional ion assistance during the film growth helps in improving the properties of the growing film. Ion assistance can be accomplished either by using a separate ion source (ion-beam-assisted deposition) or by biasing the substrate during deposition (bias sputtering). In the present study, we investigated the effect of the substrate bias and the influence of the partial pressure of nitrogen gas, together with cathode current density on the optical, surface, and compositional properties of silicon nitride thin films prepared by DC-reactive magnetron sputtering. We correlate the composition and microstructure of the deposited films to the observed changes in the properties of the films, which are due to substrate bias effects.
S. Guruvenket et al. / Thin Solid Films 478 (2005) 256–260
257
2. Experiment section We used a homemade magnetron-sputtering system for this study. A planar silicon disk (99.99% pure and 100 mm diameter) is mechanically clamped to the magnetron sputtering cathode assembly. The ultimate pressure obtained in the sputtering chamber is 103 Pa using a combination of diffusion and rotary pump. Sputtering was done in Ar+N2 mixed gas, and the total pressure during the sputtering was maintained at 101 Pa. The partial pressures of the two gases are controlled individually. The distance between the substrate and the target is maintained at 100 mm. Polished Si (100) (p-type, 10 Vcm resistivity) was used as the substrate, which was maintained at the ambient temperature without any deliberate heating. For the optical studies, films were deposited on quartz substrates. The deposition was carried out at different partial pressures (2102, 3102, and 4102 Pa,) of nitrogen while maintaining the total pressure at 101 Pa and at different cathode current density of 1.9, 2.5, and 3.10 mA/cm2. A bias potential was applied to the substrate and varied in steps of 20 V from 0 to 120 V for the films deposited at 3102 Pa partial pressure of nitrogen and with the cathode current density of 2.5 mA/ cm2. Optical characterization of the silicon nitride films was carried out using Fourier transform infrared (FTIR) spectroscopy (Perkin-Elmer) with 2 cm1 resolution and UV–VIS spectroscopy (Hitachi 330A). Compositional analysis of the films was carried out using Rutherford backscattering spectroscopy (RBS), using 2 MeV, He+ ion beam. Spectra were analyzed using GISA simulation code [7]. Surface roughness of the films was measured using atomic force microscopy (AFM). Thickness of the silicon nitride films used in this studies varied from 270 to 300 nm (measured using the stylus profiler).
Fig. 1. FTIR spectra of the silicon nitride film deposited at different nitrogen partial pressures (a) 2102, (b) 3102, and (c) 4102 Pa.
3.1. Spectral studies
substrate bias voltages. In the case of ion-assisted deposition, it has been generally observed that the packing density will be high, and hence the water vapor absorption would be less. It can be observed from Fig. 2 that, with the increase in the bias voltage, there is no absorption band at 3340 cm1. It can also be observed that only the films deposited at higher negative bias voltage (above 40 V ) show absorption band at 2190 cm1, which is due to Si– N2 stretching band [9]. All the films showed an absorption band in the range 840–870 cm1, which corresponds to the stretching vibration mode of the Si–N bond. In Fig. 2, it can be noticed that, with increase in the substrate bias voltage, the absorption band of Si–N shifts towards the higher energy value. This shift could be due to the increased nitrogen incorporation in the film. Optical band gap of the silicon nitride films has been calculated from the optical spectra using the Tauc’s relation [10]. ðahmÞ1=2 ¼ A hm Eg ;
FTIR spectra have been used as the basic tool for identifying silicon nitride formation in the films deposited under different conditions. All the spectra recorded have been corrected for the absorption due to substrate. Fig. 1 shows the FTIR spectra of the silicon nitride films deposited at a cathode current density of 2.5 mA/cm2 with nitrogen partial pressure of 2102, 3102, and 4102 Pa. These spectra show a small absorption band around 3340 cm1, which corresponds to hydrogen bonding groups. As no hydrogen-based gas has been used in the process, this band could be due to the water vapor absorption in the silicon nitride films after the deposition [8]. Fig. 2 shows the FTIR spectra of silicon nitride thin film deposited at 3102 Pa partial pressure of nitrogen and cathode current density of 2.5 mA/cm2 at different
where a is the absorption coefficient, A is a constant, hr is the photon energy, and E g is the band gap. Optical band gap value for the films deposited at various conditions is shown in Table 1. It can be seen from Table 1 that, at lower cathode current density, lower optical band gap value of 2.5 eV is obtained. With increase in the partial pressure of nitrogen and cathode current density, optical band gap value increased up to 4.5 eV. Variation in the band gap as the function of the substrate bias voltage for the films deposited at the cathode current density of 2.5 mA/cm2 and 3102 Pa partial pressure of nitrogen is shown in Fig. 3. When the substrate bias voltage is increased, it is observed that the optical band gap value increases gradually from 4.3 eV and attains a constant value around 4.8 eV at the bias voltage of 60 V, which is much closer to the optical band gap of the
3. Results and discussion
258
S. Guruvenket et al. / Thin Solid Films 478 (2005) 256–260
Fig. 3. Variation in the refractive index and the optical band gap as a function of substrate bias voltage (cathode current density of 2.5 mA/cm2 and at a nitrogen partial pressure of 3102 Pa). Fig. 2. FTIR spectra of the silicon nitride film deposited at nitrogen partial pressure of nitrogen 3102 Pa and cathode current density of 2.5 mA/cm2 at (a) 40, (b) 60, and (c) 80 V bias voltage.
bulk silicon nitride (5.0 eV). Higher substrate bias voltage of 100 V showed decrease in the energy band gap to 4.6 eV, which may be due to the increases in the structural disorder in the films (caused by energetic ion bombardment) and also due to reduced nitrogen content as discussed later. Kim and Chung [11] reported similar observations. Refractive index of the silicon nitride films was calculated using envelope technique of Swanepoel [12]. The variation in the refractive index (at 650 nm) with the deposition conditions is shown in Table 2. Higher refractive index values (2.24 and 2.15) at the higher cathode current density may be due to the excess silicon present in the film (refractive index of bulk silicon nitride film is 2.02). Fig. 3 shows the change in the refractive index of the silicon nitride thin film as a function of substrate bias voltage (deposited at 3102 Pa partial pressure of nitrogen, 2.5 mA/cm2 cathode current density). With increase in the substrate bias voltage to 40 V and above, it can be noticed that the refractive index value decreases from 2.04 and attains saturation value of about 1.92. The observed refractive index values are in good agreement with the reported data. Deenamma Vargheese and Rao [8] have reported a refractive index between 2.2 to 1.8 for the films deposited by electron-cyclotron-resonance (ECR)-assisted sputtering. Kim and Chung have reported a refractive index
between 1.9 and 2 for the films deposited by reactive bias sputtering [11]. 3.2. Compositional analysis The composition of the films was determined using RBS analysis. Table 3 shows the composition (N/Si ratio) of the films deposited at a nitrogen partial pressure of 3102 Pa and different cathode current densities. It can be clearly seen that the composition is sensitive to the cathode current, and optimum ratio of 1.3 is achieved for a current density of 2.5 mA/cm2. Fig. 4 shows the RBS spectra of the silicon nitride thin films deposited at 3102 Pa partial pressure of nitrogen at cathode current density of 2.5 mA/cm2. As discussed earlier, a near stoichiometric film is obtained for the films deposited under these conditions. This condition is chosen for further studies, and the substrate bias voltage is changed from 20 to 120 V in steps of 20 V, and the change in the properties is studied. The effect of substrate bias on the composition has been seen to be insignificant, where the N/Si ratio was about 1.3. Thus, it is obvious from the above discussion that stoichiometric silicon nitride films could be deposited at 3102 Pa of nitrogen partial pressure, cathode current density of 2.5 mA/cm2, and a substrate bias of 60 V. The RBS spectra of the films deposited under these optimized conditions is shown in
Table 1 Change in the optical band gap as a function of nitrogen partial pressure and cathode current density
Table 2 Change in the refractive index as a function of nitrogen partial pressure and cathode current density
Cathode current density (mA/cm2)
Partial pressure of nitrogen (Pa) 3102
4102
Cathode current density (mA/cm2)
Partial pressure of nitrogen (Pa)
2102
2102
3102
4102
1.90 2.50 3.10
3.1 2.8 2.8
3.3 4.3 4.4
3.7 4.4 4.5
1.90 2.50 3.10
1.84 1.98 2.24
1.90 2.04 2.15
1.95 2.12 2.08
S. Guruvenket et al. / Thin Solid Films 478 (2005) 256–260
259
Table 3 Ratio of N/Si in silicon nitride thin film deposited at 3102 Pa nitrogen partial pressure as a function of cathode current density Cathode current density (mA/cm2)
N/Si
1.9 2.5 3.1
1.2 1.3 1.4
Fig. 5. The change in the composition of the films is reflected in the properties of the film as discussed earlier. This kind of compositional variation is observed earlier by Deenamma Vargheese and Rao, with change in the ECR power instead of the applied bias voltage [8]. Vila et al. has observed the change in the composition of silicon and silicon nitride target sputtered with and without nitrogen in argon ambient [13]. They observed a small amount of trapped argon in the silicon nitride thin films, but, in our case, we did not observe the presence of Argon, but all the films showed presence of a small amount of oxygen (about 0.05). This could be due to the residual oxygen present in the sputtering chamber as the base pressure before deposition was only 103 Pa. 3.3. Surface morphology Surface roughness of the silicon nitride thin films deposited at the different substrate bias voltages (deposition condition 3102 Pa with 2.5 mA/cm2) were measured using an AFM. Sampling area over the surface of the film was 33 Am, and the images were checked at different scan regions on the substrate. With increase in the bias voltage, the roughness value (mean square root) of the film decreased. Films deposited without bias showed a roughness value of 2.3 nm, and, with increase in bias voltage to 60 V, the film roughness decreased to 1.4 nm, and further increase in the bias to 80 V bias resulted in a roughness of 0.8 nm. Beyond 80 V bias voltage, the
Fig. 4. RBS spectrum of the silicon nitride film deposited at a cathode current density of 2.5 mA/cm2 and at a nitrogen partial pressure of 3102 Pa.
Fig. 5. RBS spectrum of the Silicon nitride film deposited at. cathode current density of 2.5 mA/cm2 and at a nitrogen partial pressure of 3102 Pa with 60 V substrate bias voltage.
roughness increased to 1.0 nm, as shown in Fig. 6. The surface features of the film deposited without any bias exhibits a rough surface, which corresponds to the coarse grains. This leads to a columnar microstructural growth with voids because of the shadowing effect and low mobility of the ad atom. With increase in the substrate bias voltage to 60 V, it can be clearly observed that the surface roughness decreases, which is due to the increase in the ad atom mobility. With further increase in the bias voltage to 80 V, the surface becomes smoother with minimum voids. This clearly indicates the densification of the film due to the ion bombardment. It is well known that, with the increase in the substrate bias, the ions present in the glow will be accelerated towards the substrate and helps in improving the ad atom mobility, which in turn decreases the shadowing effect, thus leading to densification of the film. Similar changes have been
Fig. 6. Variation in the surface roughness with the applied bias voltage (at a cathode current density of 2.5 mA/cm2 and at a nitrogen partial pressure of 3102 Pa).
260
S. Guruvenket et al. / Thin Solid Films 478 (2005) 256–260
observed earlier in case of silicon nitride films deposited by ECR-plasma-assisted sputtering [8]. Furthermore, the changes in densification reported here are in agreement with the simulated data of Muller [14]; Kim and Chung also observed a similar transition in the surface morphology in bias sputtered silicon nitride films [11].
4. Conclusions Thin films of silicon nitride were deposited by reactive DC magnetron sputtering at different partial pressures of nitrogen and at different cathode current density. It is observed that the films deposited at 3102 Pa partial pressure of nitrogen with 2.5 mA/cm2 cathode current density had a near stoichiometry ratio of N/Si, optical band gap and refractive index values nearer to the bulk values. The substrate bias voltage was applied during the deposition of the film at the abovementioned condition, and its effect over the composition, optical, and surface property is studied. The substrate bias was found to have minimum effect on the composition, except at higher bias where the nitrogen content in the film reduced due to resputtering effects. However, the surface roughness was dependent on the substrate bias. A minimum surface roughness value of 0.78 nm is observed at 80 V bias. It is found that the films deposited at 60 V bias showed films with better optical properties (refractive index and optical band gap values nearer to bulk values). It is also observed that N/Si ratio is about 1.3 near to the stoichiometry of silicon nitride thin film. This effect of improvement in the properties due to substrate bias is
contributed to the ad atom mobility induced as the effect of bias voltage.
Acknowledgment The authors are thankful to Prof. K.V. Rao, KTH, Stockholm, and I. Nithiya Priya for their help in the AFM studies.
References [1] T.P. Ma, IEEE Trans. Electron Devices 45 (1998) 680. [2] E. Poloura, K. Nauka, J. Lagowsky, H.C. Gatos, Appl. Phys. Lett. 49 (1986) 97. [3] M. Vetter, Thin Solid Films 337 (1999) 118. [4] C.S. McCormick, C.E. Weber, J.R. Abelson, G.A. Davis, R.E. Weiss, V. Aebi, J. Vac. Sci. Technol., A, Vac. Surf. Films 15 (1997) 2770. [5] M.F. Lambrinos, R. Valizadeh, J.S. Colliyon, Appl. Opt. 35 (1996) 3620. [6] S.A. Awan, R.D. Gould, S. Gravano, Thin Solid Films 355–356 (1999) 456. [7] J. Saarilahti, E. Rauhala, Nucl. Instrum. Methods Phys. Res., B Beam Interact. Mater. Atoms 64 (1992) 734. [8] K. Deenamma Vargheese, G. Mohan Rao, J. Vac. Sci. Technol., A, Vac. Surf. Films 19 (2001) 1336. [9] Lin Huang, K.M. Hipps, J.T. Dickinson, Ursula Mazur, X.D. Wang, Thin Solid Films 299 (1997) 104. [10] J. Bauer, Phys. Status Solidi, A Appl. Res. 39 (1977) 411. [11] J.K. Kim, K.W. Chung, J. Appl. Phys. 83 (1998) 5831. [12] R. Swanepoel, J. Phys. E 16 (1983) 1214. [13] M. Vila, C. Prieto, J. Garica-Lopez, M.A. Respuldiza, Nucl. Instrum. Methods Phys. Res., B Beam Interact. Mater. Atoms 21 (2003) 199. [14] K.H. Muller, J. Appl. Phys. 62 (1987) 1796.