Ultra-broadband, Low-loss Wavelength Combiners and Filters: Novel Designs and Experiments in Thin-film Lithium Niobate Journal: opticajournal Publication type: Research Article
Robert Kwolek 1, Parash Thapalia 2, Aditya Tripathi 1, Pooja Kulkarni 2, Jaber Balalhabashi 4, Farzaneh Arab Juneghani 2, Michael Bullock 1,5 Oanh Hoang Vo 1,6, Sasan Fathpour 2,7, and Rajveer Nehra 1 , 3 , 4 , 5 ∗ {}^{1,3,4,5^{\ast}} · 2026
Read Ultra-broadband, Low-loss Wavelength Combiners and Filters: Novel Designs and Experiments in Thin-film Lithium Niobate Journal: opticajournal Publication type: Research Article by Robert Kwolek 1, Parash Thapalia 2, Aditya Tripathi 1, Pooja Kulkarni 2, Jaber Balalhabashi 4, Farzaneh Arab Juneghani 2, Michael Bullock 1,5 Oanh Hoang Vo 1,6, Sasan Fathpour 2,7, and Rajveer Nehra 1 , 3 , 4 , 5 ∗ {}^{1,3,4,5^{\ast}} (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-27034.txt.
Introduction
Kwolek and co-authors design wavelength combiners and filters on thin-film lithium niobate with very low loss. Scaling photonic circuits needs clean routing of broad pump and signal bands. Their closed-form designs use fast quasi-adiabatic driving to merge or split wavelengths. At the 1550 nanometer band, loss stays below 0.06 dB over 90 nanometers. Extinction ratios pass 25 dB there, and 19 dB at 775 nanometers. Fabricated devices added under 0.1 dB of loss across the band. The parts suit high-fidelity quantum photonic circuits.
Catalog record
| CREATOR | Robert Kwolek 1, Parash Thapalia 2, Aditya Tripathi 1, Pooja Kulkarni 2, Jaber Balalhabashi 4, Farzaneh Arab Juneghani 2, Michael Bullock 1,5 Oanh Hoang Vo 1,6, Sasan Fathpour 2,7, and Rajveer Nehra 1 , 3 , 4 , 5 ∗ {}^{1,3,4,5^{\ast}} |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34338 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.27034
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Ultra-broadband, Low-loss Wavelength Combiners and Filters: Novel Designs and Experiments in Thin-film Lithium Niobate Journal: opticajournal Publication type: Research Article Robert Kwolek 1, Parash Thapalia 2, Aditya Tripathi 1, Pooja Kulkarni 2, Jaber Balalhabashi 4, Farzaneh Arab Juneghani 2, Michael Bullock 1,5 Oanh Hoang Vo 1,6, Sasan Fathpour 2,7, and Rajveer Nehra 1 , 3 , 4 , 5 ∗ {}^{1,3,4,5^{\ast}} Abstract Thin-film lithium niobate (TFLN) has emerged as a leading platform for large-scale programmable photonic circuits for quantum and classical applications. As circuits scale in complexity, low-loss routing of broadband pump and signal fields becomes essential. Here, we present closed-form analytical models and experimentally demonstrate compact, fast-quasi-adiabatic driving-optimized wavelength combiners and filters operating at the fundamental harmonic (FH, 1550 nm) and second-harmonic (SH, 775 nm) wavelengths. Our designs achieve ultra-low loss below 0.06 dB 0.06~\mathrm{dB} across a 90 nm 90~\mathrm{nm} bandwidth at FH, while maintaining extinction ratios exceeding 25 dB 25~\mathrm{dB} . At SH, the loss remains below 0.12 dB 0.12~\mathrm{dB} over a 45 nm 45~\mathrm{nm} bandwidth with extinction ratios greater than 19 dB 19~\mathrm{dB} . Devices fabricated on a 300 300 -nm TFLN platform exhibit added loss below 0.1 dB 0.1~\mathrm{dB} across 1550 1550 – 1600 nm 1600~\mathrm{nm} , with minimum values of 0.04 dB 0.04~\mathrm{dB} around 1580 nm 1580~\mathrm{nm} and 0.021 dB 0.021~\mathrm{dB} at 775 nm 775~\mathrm{nm} . Combined with recent advances in on-chip quantum state generation, low-loss interferometers, and detection, these results enable high-fidelity quantum photonic circuits on the TFLN platform. 1 Introduction Nanophotonics offers a practical route to build robust large-scale information processing and sensing systems by integrating light sources, programmable phase-stabilized interferometers, filters, and detectors on a compact chip [ 3 ] . Silicon nitride and silica are mature nanophotonic platforms with ultra-low propagation loss and excellent fabrication processes, making them highly effective for many linear and nonlinear operations. However, their relatively weak third-order ( \chi^{(3)} ) nonlinearity typically relies on resonant enhancement, which introduces a trade-off between interaction strength and bandwidth, potentially limiting scalability in broadband frequency- and time-multiplexed photonic systems. Thin-film lithium niobate (TFLN) has recently emerged as a promising nanophotonic platform, combining strong second-order nonlinearity [ 12 , 17 ] , large electro-optic effect [ 18 ] , quasi-phase-matching via periodic poling, and dispersion engineering to achieve single-pass traveling-wave nonlinear devices with orders of magnitude higher efficiencies and bandwidths compared to bulk LN devices [ 1 , 16 , 21 , 10 , 7 ] . As integrated photonic systems grow from function-specific devices to configurable large-scale circuits, efficient routing and filtering of optical fields across multiple wavelengths becomes increasingly important. In particular, nonlinear quantum photonic circuits often require simultaneous handling of pump and signal fields spanning octave-scale spectral separation, such as the telecom-band fundamental and its second harmonic. Achieving broadband wavelength multiplexing with low insertion loss and high extinction ratio is therefore critical for preserving quantum coherence and minimizing loss-induced degradation in nonlinear and quantum processes. However, conventional directional couplers and wavelength-selective elements often face trade-offs between footprint, bandwidth, and loss. Conventional approaches typically employ adiabatic directional couplers due to their large fabrication tolerance and broadband operation [ 4 , 19 , 10 ] . In these devices, the waveguide supermodes evolve adiabatically at the fundamental harmonic (FH) while remaining weakly coupled at the second harmonic (SH). Achieving this dual condition generally requires long interaction lengths, which increases propagation loss. To reduce the device footprint, fast quasi-adiabatic driving (FAQUAD) methods constrain the maximum adiabatic coefficient during the evolution of the waveguide parameters [ 14 , 9 ] . Previous implementations have largely focused on optimizing the dominant constant-gap interaction region [ 14 , 6 ] (Region I, Fig. 1 a). However, the coupler transition regions can strongly influence overall device performance [ 11 , 15 ] , and imperfections in these sections can degrade extinction ratios, increasing losses that are particularly detrimental for quantum photonic applications. In this work, we extend conventional constant-gap FAQUAD approaches to the full device evolution and present closed-form analytical models for wavelength combiners and filters operating simultaneously at FH (1550 nm) and the SH (775 nm). We propose and experimentally demonstrate an ultra-broadband, low-loss quasi-adiabatic coupler with an arbitrary splitting ratio. The optimized designs exhibit total losses below 0.06 dB over 90 nm at the fundamental harmonic (FH) and 0.12 dB over 45 nm at the second harmonic (SH), representing an order-of-magnitude improvement in extinction-ratio-limited loss and bandwidth compared to conventional adiabatic designs. The devices are implemented on a 300-nm-thick thin-film lithium niobate platform with a waveguide width of \sim 1.2~\text{\textmu m} , a sidewall angle of 65^{\circ} , and an etch depth of \sim 100 nm. Experimentally, we measure a total added loss <0.1 dB across the 1550 – 1600 nm wavelength range, with optimal performance of 0.04\pm 0.02 dB at the FH and 0.021\pm 0.003 dB at the SH. 2 Proposed Design and Analysis Framework We follow the analysis described in Ref. [ 11 ] . Two parallel waveguides with weak interaction can be formalized with coupled mode equations, with interacting mode amplitudes F_{1} and F_{2} \frac{d}{dz}\begin{bmatrix}F_{1}\\ F_{2}\end{bmatrix}=-i\begin{bmatrix}-\beta_{1}&\kappa\\ \kappa&-\beta_{2}\end{bmatrix}\begin{bmatrix}F_{1}\\ F_{2}\end{bmatrix}, (1) where \beta_{1} and \beta_{2} are isolated propagation constants, and \kappa is the coupling coefficient. This can be made relative to the reference frame of the supermodes through G_{1,2}=F_{1,2}e^{-i\frac{(\beta_{1}+\beta_{2})}{2}z} , leading to \frac{d}{dz}\begin{bmatrix}G_{1}\\ G_{2}\end{bmatrix}=-i\begin{bmatrix}-\Delta\beta&\kappa\\ \kappa&\Delta\beta\end{bmatrix}\begin{bmatrix}G_{1}\\ G_{2}\end{bmatrix}, (2) where \Delta\beta=\frac{\beta_{1}-\beta_{2}}{2} . Using polar coordinates, one can define \Gamma and \chi as the coupling magnitude and angle. As a result, we have \frac{d}{dz}\begin{bmatrix}G_{1}\\ G_{2}\end{bmatrix}=-i\Gamma\begin{bmatrix}-\cos(\chi)&\sin(\chi)\\ \sin(\chi)&\cos(\chi)\end{bmatrix}\begin{bmatrix}G_{1}\\ G_{2}\end{bmatrix}=-i\hat{H}\ket{G}. (3) The eigenvalues of \hat{H} are \pm\Gamma , from which the eigenvectors corresponding to the instantaneous supermodes are \ket{+}=\begin{bmatrix}\sin\left(\frac{\chi}{2}\right)\\ \cos\left(\frac{\chi}{2}\right)\end{bmatrix},\ket{-}=\begin{bmatrix}\cos\left(\frac{\chi}{2}\right)\\ -\sin\left(\frac{\chi}{2}\right)\end{bmatrix}. (4) Thus, adiabatic transfer of the optical field from the bottom to the top port requires that the coupling angle \chi be varied continuously from 0 to \pi while suppressing excitation to the other supermode. Under this condition, optical power remains entirely in the \ket{+} eigenstate throughout device evolution. Expressing the system in the instantaneous supermode basis via the transformation \ket{G}=w_{+}\ket{+}+w_{-}\ket{-} , the coupled-mode dynamics can be written as \displaystyle\frac{dw_{+}}{dz} \displaystyle=-i\Gamma{}w_{+}+\frac{1}{2}\frac{d\chi}{dz}w_{-} (5) \displaystyle\frac{dw_{-}}{dz} \displaystyle=-\frac{1}{2}\frac{d\chi}{dz}w_{+}+i\Gamma{}w_{-}, (6) from which it is clear that the interaction strength between modes is governed by the ratio \left|\tfrac{1}{2}\chi^{\prime}(z)\right|/\Gamma . This motivates us to define the adiabaticity parameter [ 8 ] \eta(z)=\frac{1}{2\Gamma}\frac{d\chi}{dz}. (7) The FAQUAD approach is to let \eta(z) be a constant \eta\ll 1 [ 9 ] . To establish an analytical solution, we use a model to relate the coupling dynamics to waveguide parameters, namely the difference in top width (TW) between waveguides \Delta{} TW and the top gap between waveguides g(z) . Since \Delta{}\beta is solely a function of isolated modes, it does not contain a gap dependence and is solely a function of \Delta{} TW. For the geometries considered here, this function is linear for small values of \Delta{} TW, though this assumption is not essential and the design readily generalizes to other models. For weakly coupled identical waveguides, \kappa(z)=\kappa_{0}e^{-\frac{g(z)}{g_{0}}} . We can approximate \kappa to be constant with respect to \Delta{} TW for small variations in top width and extend the 1D expression. We characterize the accuracy of our reduced model in Section A of our Supplemental Document,