Low-confinement silicon nitride waveguides manufactured via direct glass bonding
Pith reviewed 2026-05-10 14:38 UTC · model grok-4.3
The pith
Thermal fusion bonding of glass wafers creates thick symmetric cladding for low-confinement silicon nitride waveguides.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
Low-confinement waveguides are formed by patterning nanometer-scale trenches in glass, filling them with silicon nitride to create a 50-nanometer-high core, and then using thermal fusion bonding of two glass wafers to enclose the core in thick symmetric cladding; this yields straight waveguide sections with up to 60 percent transmission at 1550 nanometers when butt-coupled to SMF-28 fiber, corresponding to 1 decibel per facet losses that align with numerical estimates.
What carries the argument
Thermal fusion bonding of two patterned glass wafers, which supplies high-quality interfaces and thick symmetric cladding around the thin nitride core to suppress radiative losses.
If this is right
- Photonic devices can be made with cladding thick enough on both sides to reduce radiative and parasitic scattering losses without complex thick-film deposition.
- The process supports simplified passive packaging for integrated chips because it avoids specialized equipment for thick dielectric layers.
- Long optical delay lines and ring resonators become more practical to fabricate since low confinement is achieved at low cost and with wafer-scale scalability.
- Straight waveguide performance matches simulations, indicating the core patterning and bonding steps preserve the intended optical properties.
Where Pith is reading between the lines
- The bonding step could be applied to patterned surfaces containing bends or splitters, provided the fusion process maintains uniform contact across non-planar features.
- Cost savings relative to deposition-based cladding would become measurable by comparing total process time and equipment needs for equivalent device lengths.
- Extending the method to curved structures would test whether bond-interface quality holds under the additional constraints of bend-induced mode shifts.
Load-bearing premise
The bonded glass interface remains free of defects, voids, or roughness that would introduce extra scattering or absorption losses beyond those predicted by standard models.
What would settle it
Fabricating longer waveguide sections and measuring propagation loss substantially above the value expected from fiber coupling alone, or imaging the bond plane and finding microscopic imperfections, would show the method adds unaccounted losses.
Figures
read the original abstract
Reducing the fabrication cost of photonic integrated circuits while maintaining low optical losses and technological simplicity is essential for their wider implementation. In conventional manufacturing methods, the dielectric cladding thickness around waveguides is usually limited to $\sim20~\mu$m, which complicates suppression of radiative losses and parasitic scattering in low-confinement geometries. In this paper, we propose and experimentally demonstrate an alternative technology for forming low-confinement waveguides in Borofloat 33 glass by thermal fusion bonding of two glass wafers. The waveguide pattern is formed by etching trenches with depths on the order of tens of nanometers into the glass, filling them with silicon nitride, removing the excess layer, and bonding the planarized glass surfaces, thereby forming a thick, symmetric dielectric cladding. As a proof of concept, we fabricated straight waveguides with a core height of 50 nm and widths from 1.3 to 3.5 $\mu$m. With butt coupling to standard SMF-28 single-mode fiber at 1550 nm, transmission of up to 60% was obtained, corresponding to input/output coupling losses of $\sim1$ dB per facet and consistent with numerical estimates. Fabry-Perot analysis of high-resolution spectra measured with AR-coated lensed fibers gave effective propagation losses down to $0.62\pm0.36$ dB/cm, depending on waveguide width and polarization. The proposed approach provides a simple and scalable route to low-confinement glass-encapsulated photonic circuits with passive butt coupling, promising for long delay lines, external-cavity laser feedback circuits, and ring-resonator sensors.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript proposes and experimentally demonstrates an alternative fabrication method for low-confinement silicon nitride waveguides using thermal fusion bonding of Borofloat 33 glass wafers. Trenches on the order of tens of nanometers are etched into one wafer, filled with SiN, excess material removed, and the structure bonded to a second wafer to form thick symmetric cladding. As a proof-of-concept, straight waveguides with 50 nm core height and widths from 1.3 to 3.5 μm are fabricated; butt-coupling to SMF-28 fiber at 1550 nm yields up to 60% transmission, stated to correspond to 1 dB per facet coupling loss and to be consistent with numerical estimates. The approach is positioned as a low-cost, scalable route for devices benefiting from reduced radiative and scattering losses.
Significance. If the bond interface indeed introduces negligible excess loss, the method could provide a practical route to thick, symmetric claddings that are difficult to achieve with conventional thin-film deposition, enabling lower-loss low-confinement waveguides for applications such as long delay lines and ring resonators. The experimental transmission data for functional devices offers initial evidence of feasibility at the proof-of-concept level.
major comments (2)
- [Abstract] Abstract: The reported maximum transmission of 60% (corresponding to 1 dB/facet coupling loss) is given for straight waveguides without any stated length, cut-back measurements, or length-dependent loss data. This prevents separation of propagation loss (potentially arising at the bond interface) from coupling loss and leaves the claim of consistency with numerical estimates unsupported by direct experimental evidence of low interface-induced loss.
- [Abstract] Abstract: No post-bonding metrology (SEM cross-sections, AFM roughness, or optical scattering maps of the bond plane) is referenced to substantiate the assumption that thermal fusion bonding produces a defect-free interface with thick symmetric cladding that adds no scattering, absorption, or radiative leakage beyond the modeled estimates.
Simulated Author's Rebuttal
We thank the referee for their constructive and detailed comments on our manuscript. We address each major comment point by point below, indicating where revisions will be made to improve clarity and strengthen the presentation of our results.
read point-by-point responses
-
Referee: [Abstract] Abstract: The reported maximum transmission of 60% (corresponding to 1 dB/facet coupling loss) is given for straight waveguides without any stated length, cut-back measurements, or length-dependent loss data. This prevents separation of propagation loss (potentially arising at the bond interface) from coupling loss and leaves the claim of consistency with numerical estimates unsupported by direct experimental evidence of low interface-induced loss.
Authors: We agree that the abstract would benefit from greater specificity on this point. In the revised manuscript we will state the length of the fabricated straight waveguides and clarify that the numerical estimates refer to the fiber-to-waveguide butt-coupling loss obtained from mode-overlap calculations that assume an ideal (lossless) structure. The measured transmission of up to 60% for these short devices is consistent with the simulated coupling efficiency, which provides indirect support that any additional propagation loss at the bond interface is small. We will add this clarification to both the abstract and the main text. Cut-back measurements were not included in this proof-of-concept demonstration but would indeed allow a more direct separation of losses; we will note this limitation explicitly. revision: yes
-
Referee: [Abstract] Abstract: No post-bonding metrology (SEM cross-sections, AFM roughness, or optical scattering maps of the bond plane) is referenced to substantiate the assumption that thermal fusion bonding produces a defect-free interface with thick symmetric cladding that adds no scattering, absorption, or radiative leakage beyond the modeled estimates.
Authors: We acknowledge that direct post-bonding metrology would provide stronger corroboration. The primary evidence presented in the manuscript is the optical transmission performance matching the ideal-structure simulations; significant interface defects would be expected to produce excess loss beyond the modeled coupling efficiency. In the revision we will expand the methods section with additional details on the thermal fusion bonding parameters (temperature, pressure, and surface preparation) and add citations to established literature on the low-defect interfaces achieved with Borofloat 33 glass bonding. We will also note that pre-bonding surface characterization was performed to ensure cleanliness and flatness. This textual expansion addresses the concern without requiring new destructive measurements on the bonded samples. revision: partial
Circularity Check
No circularity: experimental fabrication and measurement results
full rationale
The paper is an experimental demonstration of a glass-bonding fabrication process for SiN waveguides. It reports direct transmission measurements (up to 60% at 1550 nm) for fabricated devices and notes consistency with separate numerical estimates. No derivations, fitted parameters renamed as predictions, self-citations as load-bearing premises, or ansatzes appear in the provided text. The central claims rest on fabrication steps and measured data, which are independent of any internal logical reduction. This is the expected non-finding for a pure experimental methods paper.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption Thermal fusion bonding of Borofloat 33 glass wafers can produce high-quality optical interfaces without introducing significant defects or scattering losses.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.