This work investigates the role of precursor gas chemistry, SiH4/NH3/N2, on hydrogen incorporation into PECVD H:SiNx for optical applications. The largest reduction in of N-H bond density is shown to respond from SiH4 flow, indicating that all precursor gases must be optimized for low loss waveguides. A linear correlation of N-H bond density with propagation losses in SiNx waveguides is observed, allowing for a propagation loss to be estimated by N-H bond density without waveguide fabrication. With proper optimization of process parameters, we are able to obtain propagation losses as low as −1.6 dB/cm at 1550 nm without a thermal anneal.
© 2016 Optical Society of America
Low hydrogen content SiNx films are attractive for a host of applications, such as passivation, dielectric, and gate insulation layers in electronic devices [1–7]. However, there is an increasing interest in low hydrogen SiNx films for a variety of photonic applications. Photonic devices, such as waveguides, typically employ the use of low-pressure chemical vapor deposition (LPCVD) for silicon nitride and silicon oxide films. While lower hydrogen incorporation into the films is intrinsic to the LPCVD process, this deposition technique requires processing at temperatures greater than 800 °C, greatly limiting the processing schemes available . Plasma enhanced chemical vapor deposition (PECVD) is an attractive alternative approach for deposition of dielectric layers due to its significantly reduced deposition temperatures and much higher deposition rates. This lower temperature processing widens the integration approaches for photonic applications with back-end-of-line (BEOL) processing, allowing for integration of photonics with electronics .
The cost of lower temperature processing of SiNx by PECVD is higher hydrogen incorporation into the film, resulting in higher propagation loss in the near infrared (NIR). The increase in optical propagation loss is primarily attributed to a harmonic overtone of the N-H stretching mode around 3300 cm−1 [10–12], which is known to cause large wavelength dependent losses across the S and C telecommunications bands . Though the application space widens with CMOS BEOL compatible processing, high propagation losses can overburden the upside of lower processing temperatures. Standard PECVD SiNx deposition recipes employ the use of SiH4 and NH3, wherein it is suspected that a large fraction of hydrogen incorporation into the film, particularly the N-H bonds, comes from NH3. Ammonia is commonly used as the N source as it is easier to control film properties and improves deposition rates over N2. Several groups have investigated the role of hydrogen incorporation in SiNx with the use of NH3 as a process gas [1, 4, 5, 10, 14, 15]. One approach to lowering the hydrogen content has been to rely solely on N2 for the nitrogen content of the film and deposit ammonia free films in an attempt to reduce the N-H bond density, ultimately lowering the propagation loss of SiNx films for photonic applications [2, 7, 16, 17]. Complete substitution of N2 for NH3 in the plasma, however, results in much lower deposition rates and higher refractive index. This is due to the fact that it requires five times the critical power to activate N2 as it does NH3, making it very difficult to dissociate N2 into free N species .
Previous studies, such as by F. Karouta et al., have examined the effect of processing parameters on hydrogen incorporation . However, the study presented here is a more comprehensive investigation into the effect of multiple precursors on film quality and hydrogen incorporation into the film. In addition, a subset of films from each of the individual parameter sweeps were selected and used for the core of a waveguide. Waveguide propagation loss in each of these films was then measured to gain a better fundamental understanding of the role of precursors on hydrogen incorporation in SiNx with its optical response. Our comprehensive study results in the demonstration of a significant reduction in wavelength dependent loss in the S, C and L telecommunications bands to 1.6 dB/cm with BEOL compatible deposited SiNx films. Such low wavelength dependent loss will be an enabling feature for many future on-chip wavelength division multiplexing (WDM) applications.
SiNx films were deposited in an Applied Materials P5000 PECVD on 6”, low resistivity silicon wafers. The temperature of the substrates during deposition was held constant at 400 °C. While there are numerous variables that can affect film properties, we limited the deposition parameters in this study to SiH4 flow, NH3 flow, and N2 flow. Deposition rate, uniformity, stress, refractive index (measured at 633nm), and N-H and Si-H bond density were measured for all tests.
The film thickness, uniformity and refractive index were measured on a KLA-Tencor UV1280 ellipsometer at 49 points across the wafer. The radius of curvature was measured on a Tencor FLX-5200, with residual film stress calculated from the Stoney equation
2.2 Fourier transform infrared spectroscopy
The bond density of N-H and Si-H was estimated by Fourier transform infrared (FTIR) analysis. Absorption IR spectra was obtained from a Nicolet ECO-8S FTIR spectrometer, with a spectral range from 400 cm−1 to 4000 cm−1. The spectra for each sample was taken with a resolution of 4 cm−1 and averaged from 32 scans per sample. Though the FTIR chamber is N2 purged continuously, there still are CO2, H2O, and other trace molecules in the local atmosphere due to exposure to atmosphere during loading/unloading of the samples. This can lead to parasitic peaks in the spectra and can distort peaks in the sample spectra. For this reason, background spectra was taken within 1 minute of each sample spectra and subtracted from the sample spectra. Additionally, each substrate was measured prior to deposition. This allowed for the unique substrate spectra to be subtracted from that of the deposited film, thereby reducing peaks due to the Si substrate and native oxide on the substrate. This process results in FTIR spectra of only the deposited SiNx thin film. The target thickness of the SiNx film was 250nm.
While the FTIR absorbance spectra can be used to obtain a qualitative comparison of bonds, it is possible to estimate concentrations of hydrogen bonds by the Si-H and N-H stretching bond peaks in the spectra. Estimation of the bond concentration [X-H] can be determined by3]. There are numerous reports for various proportionality factors, with the most highly cited being that by Lanford and Rand, used in this study . Various groups have shown that film density, stoichiometry, and other factors can affect the proportionality factor [3, 5, 6]. The authors of this study investigated the bond density with several of these factors and found that while the absolute values were highly dependent on which proportionality factor was utilized, the overall trends reported in this paper are consistent regardless of the A(X-H) utilized.
3. Experimental results
3.1 Effect of SiH4 flow
SiH4 flow was varied from 100 sccm to 400 sccm while NH3 flow (115 sccm), N2 flow (4000 sccm), chamber pressure (4.7 Torr), and RF power (950 W) was held constant. As expected, an increase in SiH4 flow resulted in an increase in the deposition rate from about 50 Å/sec to almost 175 Å/sec. This is due to the fact that as SiH4 flow is increased, there is an increase in reactive Si species, Si-Hx where x = 1-3, available for deposition. The increase of flow from 100 sccm to 200 sccm resulted in a significant improvement of uniformity, measured at 49 sites across the wafer, from approximately 2.3% to 1.3% standard deviation. Additionally, the refractive index increased from 1.99 to 2.14. As the number of Si containing species increases in the plasma, there is more Si available for deposition resulting in a silicon-rich film. At the lowest SiH4 flow rate (100 sccm), the stress in the film was extremely compressive, about 1 GPa, and transitioned to slightly tensile at the highest flow of 400 sccm. Figure 1(a) shows the FTIR spectra for the four different SiH4 gas flows. The Si-N asymmetric stretching mode peak position in the FTIR spectra shifts with increasing SiH4 flow of 100 sccm to 400 sccm, from 856 cm−1 to 825 cm−1 respectively, Fig. 1(a). These shifts indicate a change in the local atomic strain in the film, which is a function of second neighbor N-N bond distance impacted by both Si-N bond angle and distance . As Si content increases, Si is substituted for N resulting in strain relief and thus relaxation of stress in the thin film. Furthermore as Si-H bond density increases, Si-N peak position moves to a lower wave number .
FTIR spectra indicated a decrease in the N-H peak centered at ~3350 cm−1, Fig. 1(b), and a significant increase in the Si-H bond peak centered ~2200 cm−1, Fig. 1(c), as SiH4 flow increased. Relative bond concentrations from the N-H and Si-H peak were estimated according to Eq. (3). It was observed that as the SiH4 flow increased from 100 sccm to 400 sccm, there was a sharp decrease in N-H bond density by almost an order of magnitude (Fig. 2). Conversely however, the Si-H bonds increased from less than to , Fig. 2. As SiH4 flow increases, the deposition rate increases, Fig. 3(a), due primarily to the increase in Si reactive species in the plasma and ultimately in the film, as indicated by the increase in the refractive index, Fig. 3(b). For the same reason, the increase in Si-H bond density increases in the film as well.
Sahu et al. investigated the effect of SiH4 flow on N atom density in RF plasmas using vacuum ultraviolet absorption spectroscopy (VUVAS) . Though the relative gas flows and RF power were lower than that of our study, Sahu et al. found that as the SiH4 flow increased, the N atom density increased as well. This was shown to be due to the fact that additional SiH4 increases the absorption of RF power, thereby increasing N atom density. These results are consistent with the results in our study, as well as with Martinez et al. . As the number of Si-Hx species available for deposition increases, the incorporation of these species into the film is expected. Additionally, as RF power absorption increases, dissociation of N-H is more likely to occur, reducing N-H bond density. Also in agreement with both Sahu et al.’s and Karouta et al.’s results is the total hydrogen bond density (Si-H and N-H) decreases with increasing SiH4 flow . For some applications, the overall hydrogen content is of utmost importance . However at telecommunication wavelengths, in particular at 1550nm, the N-H bond is the most deleterious [10, 22]. For optical applications, it may be possible to drive down the N-H bond density resulting in a decrease in optical loss in waveguides while still having an overall high hydrogen concentration.
3.2 Effect of nitrogen precursors
Since both NH3 and N2 are routinely used in PECVD SiNx:H deposition for the nitrogen component, we studied the effect of both precursors on film properties.
3.2.1 Effect of NH3
To investigate the role nitrogen containing reactive species, the first test varied NH3 flow from 0 sccm to 125 sccm while SiH4 flow (300 sccm), N2 flow (4000 sccm), chamber pressure (4.7 Torr), and RF power (950 W) was held constant. By increasing the NH3 flow, deposition rate increases nearly linearly (Fig. 4(a)). This indicates that at 300 sccm of SiH4, there is likely not enough N available to bind with Si species for the deposition. As expected, an increase in NH3 results in a decrease in refractive index due to the higher incorporation of N into the film (Fig. 4(b)). In addition, uniformity of the film thickness improves over the 6” wafer and the film stress becomes nearly neutral at the highest NH3 flow.
The addition of SiH4 to a N2 plasma has been shown to lower the peak energy of the electron energy probability function (EEPF) as determined by Langmuir probe, as well as decrease high energy tail . However by adding NH3 to a SiH4 and N2 plasma, the EEPF is greatly affected, with a much reduced peak energy and a very narrow high-energy tail. The EEPF peaks at around 20 eV for a N2 plasma, as opposed to about 1 eV for a N2, SiH4, NH3 plasma. This characteristically narrower probability distribution with lower peak energy indicates that the addition NH3 results in lower electron energy. Sahu et al. shows with VUVAS measurements the N atom density of approximately 2x1012 cm−3 with a SiH4 flow of 20sccm and N2 . However, with just 10 sccm of NH3 added to the plasma, the N atom density drops to 7x1011 cm−3, and then continues to climb back up with increasing NH3 flow. Due to the higher threshold energy required for ionization of NH3 (10.5 eV to 80eV) compared to SiH4, it would be expected to observe a decrease in N atom density in the plasma as the peak energy of the EEPF decreases. However, even if this drop in N atom density is occurring, the refractive index of our films (Fig. 4(b)) decreases with increasing NH3 flow, consistent with an increased incorporation of N into the SiNx film. It is likely that while the plasma N atom density decreases with small addition of NH3, N is bonding to Si at the surface as opposed to remaining in the plasma.
As shown in Fig. 5, the addition of NH3 to the SiH4 and N2 plasma resulted in a linear increase in total H bond density in the SiNx film. Interestingly, the N-H bond density stays relatively stable until above 100sccm of NH3, while the Si-H bond increases linearly with increasing NH3 flow. It was expected that a large fraction of the N-H bond density would be due to NH3. The increase in total hydrogen in the film with increasing NH3 flow confirms that NH3 is still a substantial hydrogen contributor. However, it appears from our study that the increased addition of NH3 results in additional free hydrogen in the plasma, which may be more readily bonding to active Si sites and Si dangling bonds up until a critical level of about 100sccm NH3. At this point, it is likely that the plasma is saturated with NHx species that are then incorporated into the film, resulting in a non-linear increase of hydrogen in the SiNx film.
3.2.2 Effect of N2
N2 flow was investigated additionally to determine its’ effect on N-H and Si-H bond densities. N2 flow was varied from 2000 sccm to 5000 sccm while SiH4 flow (300 sccm), NH3 flow (115 sccm), chamber pressure (4.7 Torr), and RF power (950 W) was held constant (Fig. 6). As the RF power needed to activate N2 is roughly 5X that required to activate NH3, the majority of the N component in SiNx is expected to come from the NH3 precursor . As such, it was not expected that N2 flow would particularly effect the hydrogen bond density in the film.
From 2000sccm to 4000sccm, the overall hydrogen bond density remains relatively constant. However, N-H bond density increases as the Si-H bond density decreases. This indicates that as the N atom density increases, hydrogen is more likely to create N-H bonds as opposed to Si-H bonds. It is also possible that the additional N2 results in increased dissociation of SiHx species due to a decreased mean free path in conjunction with the lower energy required for SiH4 dissociation. At 5000 sccm, total hydrogen bond density drops slightly. Increased plasma dissociation of SiH4 and NH3 could result in lower SiHx and NHx incorporation into the film. Additionally, increased surface bombardment of N2 could further dissociate hydrogen bonds, driving down total hydrogen bond density. While the main physical mechanism driving the drop in hydrogen bond density is unclear, it is of interest to note that hydrogen bond density can also be controlled with N2 gas flow.
4. Effect of precursors on waveguide absorption
In order to investigate the effect of H bond density on propagation loss, waveguides were fabricated with the four different deposition conditions described in Table 1. These four deposition conditions were chosen based on the results presented above in order to target a range of H bond density. 225 nm PECVD SiNx films were deposited on top of 3.4 µm of SiO2 on 6” silicon wafers. The SiNx waveguides were patterned and etched to a width of 1.2µm with standard UV photolithography and dry etch, landing on the SiO2 layer. 3 µm of high density plasma chemical vapor deposition SiO2 was deposited on top of the SiNx waveguide to serve as the top cladding.
The waveguide losses were measured by an effective cut-back technique. The waveguides were designed in a serpentine paperclip structure with lengths varying from 1.1 to 25 cm. We maintained a constant number of waveguide bends between the various length waveguides so that bend loss would be calibrated out of the measurement. The insertion loss was measured across the spectrum from 1.5 to 1.6 μm for the variation in waveguide lengths, and the propagation loss was calculated at each wavelength from the slope of the insertion loss in dB when plotted against propagation length. This technique effectively calibrates out all common path losses (e.g. coupling losses, bend losses) which can be determined by the y-intercept of the line fit to the insertion loss. The mask layout of four exemplar waveguide structures is shown in Fig. 7, while the results of the optical loss measurements can be seen in Fig. 8 below.
The four different deposition conditions resulted in propagation losses measured to range from −1.6 to −4.8 dB/cm at 1.55 μm (Table 1). One of the striking features of the plots in Fig. 8 is the strong resonant absorption near 1.515 μm in the SiNx films, a clear indication of N-H bonds in the material. However, it is also clear that with optimization of the film properties the resonant absorption can be significantly reduced. While several groups have attempted to decrease the N-H bond density by merely removing the precursor NH3, the studies above show that the effect of all precursors must be taken into account to develop a SiNx film with low N-H bond density [1, 2, 7, 16, 19, 23]. As such, it is interesting to note that the waveguide with the highest propagation loss and N-H bond density in our study had no NH3 added to the plasma. Even in the presence of NH3, the addition of 100 sccm of SiH4 resulted in nearly half the N-H bond density and a decrease in propagation loss of almost 60%. This further substantiates the theory that greater RF absorption into the plasma helps dissociate the N-H bonds, driving down the N-H bond density in the film. This suggests that a range of low loss SiNx films with varying refractive index and stress profiles can be tailored by appropriately tuning the deposition parameters.
In order to determine the effects of sidewall scattering losses on our optical waveguides, propagation loss was measured as a function of waveguide width for each sample. As the waveguide widens, the fundamental mode is more confined in the central portion of the waveguide, thus interacting less with the sidewalls, as can be seen in Fig. 9(a)-9(c). Figure 9(d) shows the measured material loss at 1.6 µm as a function of waveguide width. The material loss is determined by dividing the measured propagation loss by waveguide confinement factor and assumes that losses due to the oxide cladding are negligible. The plots of material loss are essentially independent of waveguide width indicating that sidewall roughness is not a major contributor to optical losses in this waveguide system.
In addition, it was also observed that there was a linear trend of N-H bond density with propagation loss (measured in dB/cm) for 3 of the 4 films investigated for waveguides (Fig. 10). The SiNx film with the lowest measured propagation loss (−1.6 dB/cm at 1.55 μm) had an N-H stretching mode at the detection limit for FTIR, and thus not included in the analysis. Over the range of propagation loss measured in our study, the ability to calculate N-H bond density on blanket films by FTIR and estimate propagation loss in a waveguide is extremely advantageous, as it allows for evaluation of deposition parameters without requiring full waveguide fabrication and optical testing. Moreover, FTIR spectra of a blanket SiNx with a N-H stretching mode at the detection limit for FTIR indicates a SiNx film capable of producing a low loss waveguide at 1.55 μm, comparable to some LPCVD films .
The effect of precursors on hydrogen incorporation into PECVD SiNx films was studied. It was found that by properly balancing the precursors, it is possible to obtain a propagation loss as low as 1.6 dB/cm at 1.55 μm, and losses below 2.5 dB/cm across the spectrum from 1.5 to 1.6 μm, with excellent film properties such as high refractive index, low film stress, high uniformity, and deposition rate without subjecting the film to any post-deposition anneals while enabling CMOS BEOL fabrication compatibility. In addition, we observed a linear trend of propagation loss (measured in dB/cm) with N-H bond density, allowing for the estimation of propagation loss of blanket SiNx films without requiring the full fabrication and optical testing of waveguides.
The authors would like to acknowledge funding from Dr. Peter Craig at the Office of Naval Research and Dr. Nicholas G. Usechak at Air Force Research Laboratory.
Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.
References and links
1. S. C. Mao, S. H. Tao, Y. L. Xu, X. W. Sun, M. B. Yu, G. Q. Lo, and D. L. Kwong, “Low propagation loss SiN optical waveguide prepared by optimal low-hydrogen module,” Opt. Express 16(25), 20809–20816 (2008). [CrossRef] [PubMed]
2. D. L. Smith, A. S. Alimonda, and F. J. von Preissig, “Mechanism of SiNxHy deposition from N2–SiH4 plasma,” Journal of Vacuum Science & Technology B 8, 551–557 (1990).
3. V. Verlaan, C. H. M. van der Werf, W. M. Arnoldbik, H. D. Goldbach, and R. E. I. Schropp, “Unambiguous determination of Fourier-transform infrared spectroscopy proportionality factors: The case of silicon nitride,” Phys. Rev. B 73(19), 195333 (2006). [CrossRef]
4. W. A. Lanford and M. J. Rand, “The hydrogen content of plasma‐deposited silicon nitride,” J. Appl. Phys. 49(4), 2473–2477 (1978). [CrossRef]
5. D. N. Wright, E. S. Marstein, A. Rognmo, and A. Holt, “Plasma-enhanced chemical vapour-deposited silicon nitride films; The effect of annealing on optical properties and etch rates,” Sol. Energy Mater. Sol. Cells 92(9), 1091–1098 (2008). [CrossRef]
6. E. Bustarret, M. Bensouda, M. C. Habrard, J. C. Bruyère, S. Poulin, and S. C. Gujrathi, “Configurational statistics in a-SixNyHz alloys: A quantitative bonding analysis,” Phys. Rev. B Condens. Matter 38(12), 8171–8184 (1988). [CrossRef] [PubMed]
7. K. Fouad, V. Kaushal, T. Jie, and J. Chennupati, “Structural, compositional and optical properties of PECVD silicon nitride layers,” J. Phys. D Appl. Phys. 45(44), 445301 (2012). [CrossRef]
8. J. Yota, J. Hander, and A. A. Saleh, “A comparative study on inductively-coupled plasma high-density plasma, plasma-enhanced, and low pressure chemical vapor deposition silicon nitride films,” J. Vac. Sci. Technol. A 18(2), 372–376 (2000). [CrossRef]
9. Y. Huang, J. Song, X. Luo, T.-Y. Liow, and G.-Q. Lo, “CMOS compatible monolithic multi-layer Si3N4− on-SOI platform for low-loss high performance silicon photonics dense integration,” Opt. Express 22(18), 21859–21865 (2014). [CrossRef] [PubMed]
10. F. Ay and A. Aydinli, “Comparative investigation of hydrogen bonding in silicon based PECVD grown dielectrics for optical waveguides,” Opt. Mater. 26(1), 33–46 (2004). [CrossRef]
11. O. P. Agnihotri, S. C. Jain, J. Poortmans, J. Szlufcik, G. Beaucarne, J. Nijs, and R. Mertens, “Advances in low temperature processing of silicon nitride based dielectrics and their applications in surface passivation and integrated optical devices,” Semicond. Sci. Technol. 15(7), R29–R40 (2000). [CrossRef]
12. F. L. Martínez, R. Ruiz-Merino, A. del Prado, E. San Andrés, I. Mártil, G. González-Díaz, C. Jeynes, N. P. Barradas, L. Wang, and H. S. Reehal, “Bonding structure and hydrogen content in silicon nitride thin films deposited by the electron cyclotron resonance plasma method,” Thin Solid Films 459(1-2), 203–207 (2004). [CrossRef]
13. D. J. Moss, R. Morandotti, A. L. Gaeta, and M. Lipson, “New CMOS-compatible platforms based on silicon nitride and Hydex for nonlinear optics,” Nat. Photonics 7(8), 597–607 (2013). [CrossRef]
14. D. L. Smith, A. S. Alimonda, C. C. Chen, S. E. Ready, and B. Wacker, “Mechanism of SiNx Hy Deposition from NH3 - SiH4 Plasma,” J. Electrochem. Soc. 137(2), 614–623 (1990). [CrossRef]
15. H. Mäckel and R. Lüdemann, “Detailed study of the composition of hydrogenated SiNx layers for high-quality silicon surface passivation,” J. Appl. Phys. 92(5), 2602–2609 (2002). [CrossRef]
16. D. H. Yoon, S. G. Yoon, and Y. T. Kim, “Refractive index and etched structure of silicon nitride waveguides fabricated by PECVD,” Thin Solid Films 515(12), 5004–5007 (2007). [CrossRef]
17. D. K. T. Ng, Q. Wang, T. Wang, S.-K. Ng, Y.-T. Toh, K.-P. Lim, Y. Yang, and D. T. H. Tan, “Exploring High Refractive Index Silicon-Rich Nitride Films by Low-Temperature Inductively Coupled Plasma Chemical Vapor Deposition and Applications for Integrated Waveguides,” ACS Appl. Mater. Interfaces 7(39), 21884–21889 (2015). [CrossRef] [PubMed]
18. S. Habermehl, “Stress relaxation in Si-rich silicon nitride thin films,” J. Appl. Phys. 83(9), 4672–4677 (1998). [CrossRef]
19. G. N. Parsons, J. H. Souk, and J. Batey, “Low hydrogen content stoichiometric silicon nitride films deposited by plasma‐enhanced chemical vapor deposition,” J. Appl. Phys. 70(3), 1553–1560 (1991). [CrossRef]
20. B. B. Sahu, K. S. Shin, S. B. Jin, J. G. Han, K. Ishikawa, and M. Hori, “Effectiveness of plasma diagnostic in ultra high frequency and radio frequency hybrid plasmas for synthesis of silicon nitride film at low temperature,” J. Appl. Phys. 116(13), 134903 (2014). [CrossRef]
21. E. Herth, B. Legrand, L. Buchaillot, N. Rolland, and T. Lasri, “Optimization of SiNX:H films deposited by PECVD for reliability of electronic, microsystems and optical applications,” Microelectron. Reliab. 50(8), 1103–1106 (2010). [CrossRef]
22. A. Gorin, A. Jaouad, E. Grondin, V. Aimez, and P. Charette, “Fabrication of silicon nitride waveguides for visible-light using PECVD: a study of the effect of plasma frequency on optical properties,” Opt. Express 16(18), 13509–13516 (2008). [CrossRef] [PubMed]
23. D. S. Kim, S. G. Yoon, G. E. Jang, S. J. Suh, H. Kim, and D. H. Yoon, “Refractive index properties of SiN thin films and fabrication of SiN optical waveguide,” J. Electroceram. 17(2-4), 315–318 (2006). [CrossRef]
24. M. Melchiorri, N. Daldosso, F. Sbrana, L. Pavesi, G. Pucker, C. Kompocholis, P. Bellutti, and A. Lui, “Propagation losses of silicon nitride waveguides in the near-infrared range,” Appl. Phys. Lett. 86(12), 121111 (2005). [CrossRef]