Project

Discipline

ultra-high Q; Si3N4 integrated photonics ; high-density plasma chemical-vapor deposition; frequency combs

Lead
Les peignes de fréquences optiques ont révolutionné les domaines du chronométrage, de la métrologie, et de la spectroscopie. Leur utilisation est néanmoins restreinte aux laboratoires de recherches, due à leurs coûts, taille, et complexité. L’avancée rapide de la technologie des peignes de fréquences « intégrés » promet de changer la donne, car elle permet la fabrication en masse de dispositifs photoniques miniaturisés, basés sur les méthodes de fabrications bien établis des fonderies MEMS. A l’instar des fibres optiques, artères principales de nos réseaux de télécommunications modernes, les matériaux utilisés dans ces dispositifs intégrés doivent être aussi purs et transparents que possible pour des applications tels que les mesures ultra-rapides de distances (LIDAR), ou les systèmes de télécommunications optiques. La déposition de couches mince par ICP-CVD répond exactement à ces besoins, tout en restant à des températures compatibles avec les autres matériaux fonctionnels.
Lay summary
Encapsuler et protéger des guides d’ondes de matériaux plus sensibles, sans en affecter les performances. Par exemple le niobate de lithium, matériau très en vogue. Remplissage des interstices très étroits entre les nanostructures sans laisser de vides. Le développement de nouvelles recettes de dépositions basés sur le SiCl4 comme gaz précurseur. La chimie de réaction avec ce gaz promet un matériau sans hydrogène ; élément causant une perte de transparence pour les longueurs d’ondes pertinentes technologiquement. Le développement des circuits photoniques intégrés ouvre les portes à des nouvelles applications, autant dans le laboratoire que sur le marché industriel et technologique. La miniaturisation des sources de fréquences optiques réduira grandement la consommation en énergie des datacenters. Le développement de systèmes LIDAR miniaturisés aidera au déploiement universel de véhicules autonomes.

Direct link to Lay Summary

Last update: 30.11.2020

Number	Title	Start	Funding scheme

Inductively coupled plasma - chemical vapor deposition (ICP-CVD), commonly referred as high-density plasma CVD (HDP-CVD), is the method of choice for low temperature deposition of SiO2 and Si3N4 thin-films. Because the precursor gases are dissociated in a dense plasma generated inductively, this circumvents the need for a high thermal budget and enables the deposition of high-quality dielectric films at temperatures as low as 150°C. Compared to regular plasma-enhanced CVD (PECVD), the ICP-CVD process takes places at a lower pressure and yields denser films with less hydrogen. The biasing power is also applied independently from the plasma source; meaning that one can choose to apply low-biasing to limit ion damage, or high-biasing to encourage ion bombardment for film densification and gap filling. Overall, the strong suit of the tool resides in the incredible flexibility to control all deposition parameters independently over a wide range, to engineer the material in conditions compatible with the chosen substrate.The PI proposes to install such a unique ICP-CVD system as a platform tool at CMi at EPFL, to further expand EPFL’s research on the ultra-low loss integrated photonic platform based on Si3N4. As one of the pioneers in optical microresonators for frequency comb and soliton generation, the PI has remained at the forefront of the field, made possible by the development of the Photonic Damascene nanofabrication process [1] that is now capable of wafer-scale fabrication of ultra-low optical losses of less than 0.5 dB/m in the 1550 nm band. This has unlocked photonic integrated microcombs, whose potential have been proven in several system-level applications ranging from coherent telecommunication [2], ultrafast ranging [3], to massively parallel LiDAR [4] and astrophysical spectrometer calibration [5]. Pivotal in all these experiments has been the continued careful and artful development of the Photonic Damascene process [1, 6, 7]. This process has matured over several years and generations, producing record ultra-low loss, unity yield, and true wafer-scale devices that do not require time intensive e-beam lithography, and has reached a point where materials are the limitation. Meanwhile, the LPCVD materials employed have remained essentially the same; the windows of operation around the standard LPCVD recipes are very narrow. Most importantly, the LPCVD materials have well-known inescapable drawbacks: Namely, the high intrinsic stress in the Si3N4 film, and the concentration of undesirable hydrogen in the film. Moreover, no suitable high-quality SiO2 are available, which imposes loss limits due to the use of lower quality SiO2 such as low-temperature oxide (LTO) or TEOS.The ICP-CVD will allow to engineer the stress in thick Si3N4 layers, while remaining close to the stoichiometric composition, by varying temperature and biasing power. The high-density plasma also enables the breakdown of stable precursor molecules that would require impossibly high temperatures in LPCVD. One such particularly interesting precursor is silicon tetrachloride (SiCl4), a hydrogen-free precursor that could be used with N2 and O2 for hydrogen-free Si3N4 and SiO2 CVD processes. This compound is ubiquitous in the mass production of optical fibers, but has never been applied to integrated photonics.Beyond the material properties, this equipment will bring forth a breakthrough in the freedom to fabricate more complex structures: The possibility to deposit high-quality Si3N4 and SiO2 in the same chamber in programmed sequence will enable the fabrication of layered superstructures, either for Bragg mirrors or multilayer waveguides for advanced dispersion engineering. The ion-bombardment-assisted gap-filling capabilities will also enable the clean and void-free cladding of devices fabricated by subtractive process, even with tall elements placed very close together. An HDP-CVD platform tool of this caliber has been a long-standing need in the EPFL community, and has rapidly gained the support of other Professors working in a wide variety of research topics. For instance, Prof. Niels Quack has expressed his wish to deposit high-quality oxide sacrificial layers on pre-processes wafers that are not LPCVD compatible. Prof. Jürgen Brugger is interested to grow thick nitride films with controlled stress for nanostencil applications. Prof. Camille Brès and Prof. Andras Kis have equally expressed their interest in the equipment’s ability to deliver high-quality materials, whether it’s the study and control of induced nonlinearity in Si3N4, or strain-engineered interfaces with 2D materials.