3.5 Silicon Nitride
silicon-nitride
silicon-nitride
You can place or change feedback until 31 December 2026, 17:00
The silicon nitride photonic platform is based on a combination of (stoichiometric) silicon nitride (SiN) Si3N4 as waveguide layers, filled by and encapsulated with silica (SiO2) as cladding layers on a silicon wafer. Silicon nitride (SiN) waveguides have a number of advantages over alternative technologies, including low propagation loss, wide range of transparency from visible to the mid-infrared, low cost, good reliability for high power and industrial operating conditions, as well as good coupling to standard single-mode fiber.
The typical propagation loss of only 0.5dB/m allows to achieve filters with high finesse which are required in for example narrow line width external cavity laser but also for example to improve sensitivity of resonant micro optic gyroscopes. In sensor applications SiN has benefits in low loss over a wide range of wavelength from visible to the mid-infrared. For high-speed datacentre transceivers of 400Gb/s and beyond, loss has become even more critical as multi-bit amplitude modulation reduces the power level between the bits. So, in order to achieve the same signal to noise ratio, the power levels in the fiber and from the chip need to increase proportionally. Currently, the hypercloud datacenters experience a massive delay in the delivery of 400Gb/s transceivers due to the additional power requirements which existing suppliers cannot meet reliably. Low-loss SiN technology could meet the conflicting demands on small footprint and low loss that currently cannot be served with silicon photonics nor monolithic Indium Phosphide (InP), due to high loss for these platforms. The difficulty in making future transceivers using discrete optics lies in the challenge to fit in the small footprint, as a consequence of the larger size of the discrete optical lenses and filters, again favouring SiN as a more suitable technology platform.
For some applications it is also important to handle high power densities. This is for example the case in future 5G networks where optical mux and demux devices will be installed to connect antennas to the base station. For these applications the antennas may be at a short distance from the base station resulting in multiplexed powers of the order of Watts. These power levels can easily be handled by SiN waveguides.
All antennas in 5G networks, top of rack switches in data centers, and also many sensors, are connected by single-mode optical fibers. So for the optical chip, coupling loss to standard single-mode fiber should be better than 90%. This can be achieved by SiN technology. Using hybrid SiN technologies with joint Nitride Glass waveguides one can reach close to 99% coupling efficiency. So, for the roll-out of 5G networks, but also for future data center applications, and other products that require high power and low loss, SiN is an attractive technology. In particular, heterogeneous and hybrid integration techniques of SiN with more complex active subsystems have now been demonstrated, including stress-based modulation on the TriPleX platform of LioniX [1]. A recent comprehensive review of the latest progress on SiN technology platform and integration with active components can be found in reference [1].
SiN fabrication processes are already mature. SiN layers with uniform thicknesses, typically within 1 %, and reproducible refractive indices are deposited using both LPCVD and PECVD. For SiN waveguides both stepper lithography as well as contact lithography can be used, considering the relatively low effective index compared to Si, allowing relatively large waveguide widths while maintaining single mode characteristics. Using typical commercially available stepper lithography tools, lines/spaces of 250 nm can be fabricated. The width of 250 nm for the lines/spaces is sufficient for all applications in this platform at telecom wavelength.
Similar to silicon, the SiN platform is not efficient for lasing. LioniX International has solved this problem by hybrid integration of InP gain medium with tunable mirrors defined in their SiN platform. These hybrid lasers have high output power > 10 mW, very narrow laser linewidth and can be tuned over the whole C-band [2]. Hybrid integration has been used to integrate InP modulators and waveguide detectors with LioniX SiN plaform. With regards to switching and modulation, very reliable but rather slow modulation can be achieved through the well-known thermal phase modulation in case of SiN (limited to 1 kHz). Faster modulation up to the MHz and the GHz regimes has been achieved using PZT material [3,4]. In case of (quasi-) DC operation, these modulators furthermore reduce the required switching power to a few microWatts () only.
Multi Project Wafer (MPW) runs operating in the C-band around1550 nm are available since 2011. For application in the VIS domain, MPW shuttles are provided since 2017, by both LioniX International as well as IMEC. In these MPW services, fundamental building blocks are available, including optical waveguides, thermo-optic phase tuning elements, Y-branches, micro ring resonators (MRR’s), directional- and multi-mode interference (MMI) couplers.
In the tables below the parameters’ values of different SiN fabrication processes are given. The values of these parameters for the next 5 to 15 years are specified.
| (Epitaxial) growth/layer deposition | [unit] | 5 years | 5-10 years | 10-15 years |
| Layer thickness uniformity SiN | % | 1 | 0.5 | 0.5 |
| Layer thickness reproducibility SiN | % | 2 | 1 | 0.5 |
| Layer refractive index uniformity | RIU | 5*10-3 | 2*10-3 | 1*10-3 |
| Layer refr. index reproducibility | RIU | 5*10-3 | 2*10-3 | 1*10-3 |
| Layer thickness uniformity SiO2 | % | 2 | 1 | 0.5 |
| Layer thickness reproducibility SiO2 | % | 3 | 2 | 1 |
| Layer refr. index uniformity SiO2 | RIU | 1*10-2 | 5*10-3 | 2*10-3 |
| Layer refr. index reproducib. SiO2 | RIU | 1*10-3 | 5*10-3 | 2*10-3 |
| Dry etching | [unit] | 5 years | 5-10 years | 10-15 years |
| Side wall roughness | [nm] | 1 | 0.5 | 0.2 |
| Side wall angle | ˚ | 2 | 1 | 0.5 |
| Dry etching | [unit] | 5 years | 5-10 years | 10-15 years |
| Side wall roughness | [nm] | 1 | 0.5 | 0.2 |
| Side wall angle | ˚ | 2 | 1 | 0.5 |
| Etch depth reproducibility | [nm] | 50 | 25 | 10 |
| Etch depth uniformity | [nm] | 25 | 10 | 5 |
| Selective etch masking layers | NA | 2 | 3 | 4 |
| Minimal linewidth | [] | 0.5 | 0.2 | 0.1 |
| Minimal reproducibility | [] | 0.5 | 0.2 | 0.1 |
| Waveguide width uniformity | [] | 0.1 | 0.05 | 0.03 |
| Minimum spacing | [] | 0.5 | 0.2 | 0.1 |
| Minimum grating pitch | [] | 0.3 | 0.2 | 0.1 |
| Grating etch step uniformity | [nm] | 50 | 20 | 10 |
| Wet etching | [unit] | 5 years | 5-10 years | 10-15 years |
| Side wall roughness | [nm] | 0.5 | 0.3 | 0.2 |
| Etch depth reproducibility | [nm] | 50 | 20 | 10 |
| Etch depth uniformity | [nm] | 50 | 20 | 10 |
| Selective etch masking layers | NA | 2 | 3 | 4 |
| Minimal linewidth | [] | 0.5 | 0.2 | 0.1 |
| Minimal reproducibility | [] | 0.5 | 0.2 | 0.1 |
| Wave guide width uniformity | [] | 0.1 | 0.05 | 0.03 |
| Minimum spacing | [] | 0.5 | 0.2 | 0.1 |
| Minimum grating pitch | [] | 0.3 | 0.2 | 0.1 |
| Grating etch step uniformity | [nm] | 30 | 20 | 10 |
| Wet etch undercuts | [nm] | 500 | 750 | 1000 |
| Contact Lithography | [unit] | 5 years | 5-10 years | 10-15 years |
| Overlay accuracy | [] | 1 | 0.8 | 0.5 |
| Resolution | [] | 0.8 | 0.6 | 0.5 |
| Required Flatness requirements |
| Stepper/scanner Lithography | [unit] | 5 years | 5-10 years | 10-15 years |
| Overlay accuracy | [nm] | 500 | 250 | 100 |
| Resolution | [nm] | 300 | 200 | 100 |
| Required Flatness requirements | [nm] | 300 | 200 | 100 |
| Laser Lithography | [unit] | 5 years | 5-10 years | 10-15 years |
| Overlay accuracy | NA | |||
| Speed | NA | |||
| Resolution | NA | |||
| Required Flatness requirements | NA |
| E-Beam Lithography | [unit] | 5 years | 5-10 years | 10-15 years |
| Overlay accuracy | NA | |||
| Speed | NA | |||
| Resolution | NA | |||
| Required Flatness requirements | NA |
| Annealing | [unit] | 5 years | 5-10 years | 10-15 years |
| Temperature@time budget | [K] | 1150 | 1150 | 1150 |
| Planarization | [unit] | 5 years | 5-10 years | 10-15 years |
| Temperature@time budget | [K] | NA | NA | NA |
| Required flatness | 300 | 200 | 100 |
| Passivation | [unit] | 5 years | 5-10 years | 10-15 years |
| Temperature@time budget | [K] | 350 C | 200 C | RT |
| Dicing and cleaving | [unit] | 5 years | 5-10 years | 10-15 years |
| Position accuracy | [µm] | 2 | 1 | 0.5 |
| Metal deposition | [unit] | 5 years | 5-10 years | 10-15 years |
| Temperature budget | [C] | 250 C | 150 | 100 |
| Wafer bonding | [unit] | 5 years | 5-10 years | 10-15 years |
| Temperature budget | [C] | 500 C | 500 C | 500 C |
Routes to extend the usability of these platforms should hence address:
| Table X. Prioritized Research milestones (> 2025) | Relative priority |
| Further development of process | Critical |
| Extension of the guiding range | Regular |
| Investigation of maximum optical power | Desirable |
| Adoption of mix & match lithography | Regular |
| Table X. Prioritized Development & Implementation milestones (≤ 2025) | Relative priority |
| Integration of SiN platforms with microfluidics | Desirable |
| Increased level of integration | Critical |
| Development of transfer printing processes | Critical |
| Non-linear interactions | Desirable |
| Quantum guided applications | Desirable |
| MPW standardization for different wavelengths ranges and applications | Regular |
1. Blumenthal, D.J., Heideman, R., Geuzebroek,D., Leinse, R., and Roeloffzen, C., Silicon Nitride in Silicon Photonic. Proceedings of the IEEE, Vol. 106, No. 12, December 2018.
2. Youwen Fan, Albert van Rees, Peter J. M. van der Slot, Jesse Mak, Ruud M. Oldenbeuving, Marcel Hoekman, Dimitri Geskus, Chris G. H. Roeloffzen, and Klaus-J. Boller. OSA Publishing, Volume 28, Issue 15; July 2020; DOI: 10.1364/OE.398906
3. Briefing on PZT based stress optical tuning on TriPleX platform.
4. Koen Alexander, John P. George, Jochem Verbist, Kristiaan Neyts, Bart Kuyken, Dries van Thourhout, and Jeroen Beekman. Nature Communications 9:3444 (2018).
5. Micó, G., Bru, L. A., Pastor, D., Domenech, D., Fernández, J., Sánchez, A., ... & Muñoz, P. (2018, February). Silicon nitride photonics: from visible to mid-infrared wavelengths. In Silicon Photonics XIII (Vol. 10537, p. 105370B). International Society for Optics and Photonics.
6. Xiong, C., Pernice, W. H., & Tang, H. X. (2012). Low-loss, silicon integrated, aluminum nitride photonic circuits and their use for electro-optic signal processing. Nano letters, 12(7), 3562-3568.
7. Muellner, P.; Maese-Novo, A.; Melnik, E.; Hainberger, R.; Koppitsch, G.; Kraft, J.; Meinhardt, G. CMOS-compatible low-loss silicon nitride waveguide integration platform for interferometric sensing. In Proceedings of the European Conference on Integrated Optics, Warsaw, Poland, 18–20 May 2016. [Google Scholar]
Join the working group and help shape the IPSR chapters.
Help us improve this section for the next reader.