Wavelength-agnostic 3D-Nanoprinted coupler

Chunhui Yao; Huiyu Huang; Qixiang Cheng; Richard Penty; Zhitian Shi

REVIEW 5 major objections 6 minor 16 references

Wavelength-agnostic 3D-Nanoprinted coupler

T0 review · 5 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A 3D-nanoprinted coupler using two ellipsoidal reflectors transfers light from fibre to chip with a 1 dB bandwidth exceeding 800 nm, the widest reported for any photonic coupler.

desk verdict Genuinely broad bandwidth nanoprinted coupler, but the 'wavelength-agnostic' mechanism is asserted, not derived; empirical result outweighs the theory. read the letter →

arxiv 2506.17825 v1 pith:U7SCKE6G submitted 2025-06-21 physics.optics

Huiyu Huang , Zhitian Shi , Chunhui Yao , Richard Penty , Qixiang Cheng This is my paper · ORCID

classification physics.optics

keywords wavelength-agnosticcouplingfibre-to-chipcouplerdual-ellipsoidalreflectortotalinternalreflection3Dnanoprintingtwo-photonpolymerizationmatrixopticsbroadbandphotonics

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that a 3D-nanoprinted coupler with two ellipsoidal reflectors can transfer light from an optical fibre to a photonic chip with a 1 dB insertion-loss bandwidth exceeding 800 nm, the widest ever reported for any fibre-to-chip coupler, at a minimum loss of 1.3 dB. The design abandons guided-wave coupling altogether: light expanded into the polymer volume propagates as a quasi-free-space beam, and two total internal reflections on ellipsoidal facets focus it onto the target waveguide. A sympathetic reading is that this makes the coupling fundamentally wavelength-independent, breaking the bandwidth ceiling of edge and grating couplers. The paper also contributes a fast matrix-optics pre-design step that cuts the parameter search before full FDTD simulation, and demonstrates the device experimentally in fibre-to-fibre and fibre-to-chip configurations with thermal cycling.

What carries the argument

The central mechanism is a pair of confocal ellipsoidal reflectors that redirect light through two total internal reflections at the polymer–air interface. The design condition given by Eq. (5), $R_1 + R_2 = 2d_2$, with $R_1$ and $R_2$ as the radii of curvature of the two reflection surfaces and $d_2$ as the propagation distance between them, encodes an aberration-free imaging condition that forces rays from one focal point to reconverge with minimal divergence. The matrix-optics ray-transfer model of Eq. (3) treats the optical path as seven segments—two refractions, three free-space propagations, and two reflections—and is used to quickly explore the (x-focus, y-focus) parameter space before full-vectorial FDTD simulation, with a spherical collimating lens of radius 15 μm added at the input to approximate a quasi-parallel beam.

What would settle it

Calculate the ray-transfer output at the two band-edge wavelengths using the paper's Eq. (1): the divergence half-angle at 1670 nm is roughly double that at 800 nm for the same fibre, so if the matrix model predicts a wavelength-dependent focus position larger than the tolerance implied by Eq. (4), the achromaticity premise is falsified; alternatively, measure the fibre-to-chip insertion loss at 700 nm and 1750 nm and a 1 dB bandwidth narrower than 800 nm would directly disprove the record claim.

Watch

Extended reading notes

Core claim

The central claim is that coupling efficiency of this reflector-based interface is essentially flat across an 800 nm-plus window because the light stops being a confined mode. After the guided beam exits the fibre and enters the printed polymer, it expands and reflects twice at the air–polymer interface, and the ellipsoidal geometry images the waveguide facet onto the fibre facet with minimal aberration. The paper states this as quasi-free-space propagation, inherently eliminating wavelength dependencies, and supports it with FDTD simulations over 800 nm to 1670 nm, with a simulated minimum loss of about 0.5 dB for the optimized baseline. Experimentally, the fibre-to-chip coupler shows a minimum insertion loss of about 1 dB and a 1 dB bandwidth spanning more than 800 nm, which the authors call a record for any reported photonic couplers. The paper further claims that the same geometry can be retuned by adjusting the focal positions to match different fibre and waveguide mode sizes, and that the 3D-printed envelope provides passive alignment and thermal robustness.

Load-bearing premise

The claim stands on the assumption that once the light leaves the waveguide and expands inside the printed polymer, it behaves like a free-space beam, so the ellipsoidal two-reflection geometry focuses all wavelengths onto the output identically; if that quasi-free-space assumption fails near the band edges, the 800 nm 1 dB bandwidth claim would not hold.

Editorial extensions

If this is right

A coupler with more than 800 nm of 1 dB bandwidth would let one photonic chip interface with sources and sensors across the O-band through the U-band, covering much of the telecom spectrum and the near-infrared sensing window.
Because the coupler is printed with two-photon polymerization directly onto silicon-nitride or silicon-on-insulator chips, the same fabrication process could be extended to other photonic platforms with only geometric adjustments.
The matrix-optics pre-design step reduces the FDTD parameter space, making rapid design iteration practical for industrial packaging workflows.
The demonstrated thermal robustness—surviving cycling to 150 °C with negligible loss change—shows the device can operate outside a laboratory setting.
Since multiple couplers can be spatially overlapped and the reflector occupies only a small part of the structure, the design could support high-density, multi-port fibre-chip interconnects.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Inference: if the quasi-free-space assumption holds, the same dual-ellipsoid principle might be scaled to other spectral regions such as the visible or mid-infrared, limited mainly by the transparency and printable geometry of the polymer rather than by the coupling mechanism.
Inference: the paper reports performance only for TE polarization; a natural testable extension is whether a symmetric version or a polarization-splitting variant can maintain the same bandwidth for TM modes.
Inference: the residual loss at the band edges is attributed to chromatic aberration, so a direct quantitative check would be to measure the output beam profile at 800 nm and 1670 nm and compare the wavefront curvature, isolating the achromaticity assumption independently of the loss measurement.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

You should know: this paper's real contribution is an empirical one — a nanoprinted dual-ellipsoid coupler with a measured 1 dB bandwidth beyond 800 nm and around 1.3 dB fiber-to-chip loss. The 'wavelength-agnostic' framing is not supported by the theory as written; the bandwidth is a property of the specific optimized geometry, not of quasi-free-space propagation per se.

What's genuinely new: the dual-ellipsoidal TIR geometry for fiber-chip coupling, the matrix-optics pre-design step that narrows the FDTD parameter space, and the integrated plug-in funnel for passive fiber alignment. The experimental work is solid in its breadth: SLD measurements from 840 to 1670 nm, fiber-to-fiber and fiber-to-chip variants, and thermal cycling data. That's a serious demonstration.

The soft spots are in the interpretation. The paper claims the design 'inherently eliminate[s] wavelength dependencies' because light becomes quasi-free-space. But Eq. (1) makes the divergence angle wavelength-dependent, and Eq. (5) R1+R2=2d2 is only the collimation condition C=0 for the ABCD matrix; it does not make the mode overlap achromatic. The authors themselves admit in Section 3.4 that the band-edge loss increase is due to 'residual chromatic aberrations' in the ellipsoidal geometry. So the 'wavelength-agnostic' label is overstated. Also, the geometry was tuned in FDTD and then measured; no independent prediction, no error bars, no quantitative comparison against prior edge or grating couplers. The 'record' claim needs at least a table of loss/bandwidth for existing couplers.

These are fixable. Add a wavelength-dependent mode-overlap derivation or a careful discussion of why the optimized geometry approximates achromaticity over the measured range. Report measurement uncertainty. Bench against edge couplers and photonic wire bonds. Then the paper would be much stronger.

This is a device paper for the integrated-photonics packaging community; anyone working on broadband sensing or frequency combs should know about it. Worth a serious referee and, with revisions, publication. I'd bring it to a reading group to discuss the gap between device demonstration and mechanism claims.

Referee Report

5 major / 6 minor

Summary. The paper reports a fiber-to-chip coupler made by two-photon polymerization, consisting of a dual-ellipsoidal reflector system that collimates and refocuses light via two total internal reflections. The authors claim that because the guided mode expands into the polymer volume and propagates in a quasi-free-space fashion, the coupler is intrinsically wavelength-agnostic. They present a matrix-optics pre-design model to narrow the parameter space before FDTD optimization, and they demonstrate fibre-to-fibre and fibre-to-chip coupling with a 1 dB bandwidth exceeding 800 nm and insertion loss as low as 1.3 dB, which they state is a record for any reported photonic coupler. Experimental tests also include thermal cycling and broadband measurements from 840 nm to 1670 nm.

Significance. If the demonstrated performance is robust, the device is a valuable contribution to broadband photonic packaging, particularly for sensing and full-band communication applications. The use of 3D nanoprinting to fabricate freeform reflectors and integrated alignment funnels is practical and promising. The paper also provides a useful example of combining analytic modeling with FDTD optimization. However, the central theoretical claim that the design is intrinsically wavelength-agnostic is not rigorously established; the paper itself qualifies the mechanism in Section 3.4. The record claim and the absolute loss values need quantitative benchmarking and uncertainty analysis before the result can be fully assessed.

major comments (5)

[Eq. (3)] Equation (3) writes the reflection matrices as [1 0; 2/R_i 1], but the standard ray-transfer matrix for reflection from a spherical surface of radius R is [1 0; -2/R 1]. The subsequent collimation condition R1+R2=2d2 in Eq. (5) is only obtained with the standard negative sign. Please correct the sign convention in Eq. (3) or explicitly justify the alternative convention, as this is a central equation of the proposed pre-design model.
[Section 2, Eq. (5)] The claim that quasi-free-space propagation makes the coupler intrinsically wavelength-agnostic is not supported by the presented model. Equation (1) explicitly gives a wavelength-dependent divergence half-angle, and even with C=0 in the ABCD matrix, a Gaussian beam transforms with q'=(Aq+B)/(Cq+D); the output spot size and curvature still depend on wavelength through the input q parameter and the wavelength-dependent mode-field diameter. Section 3.4 itself attributes band-edge loss to residual chromatic aberrations. The authors should either provide a full wavelength-dependent mode-overlap derivation for the dual-ellipsoidal geometry or temper the claim to 'reduced wavelength sensitivity' rather than 'inherently eliminating wavelength dependencies.'
[Abstract and Conclusion] The statement that the device achieves 'to the best of our knowledge, a record for any reported photonic couplers' is not substantiated. No quantitative comparison with existing couplers (edge couplers, grating couplers, photonic wire bonds, or the OFFCHIP couplers cited as Ref. [15]) is provided. A table listing bandwidth, insertion loss, and coupling configuration of competing devices should be included to support the record claim.
[Section 3.1 and 3.2] The experimental loss measurements lack error bars and uncertainty estimates. The broadband loss curves in Figs. 4(g) and 4(h) exhibit fluctuations of roughly 1 dB, so the reported '1 dB bandwidth exceeding 800 nm' is highly sensitive to how the curve is smoothed or thresholded. Please specify the measurement procedure, the reference used for normalization, and the repeatability of the measurements. Also reconcile the abstract value of 1.3 dB with the '~1 dB' minimal insertion loss reported for the fibre-to-chip configuration in Section 3.2.
[Section 2 (hybrid modelling workflow)] The matrix-optics model is used only to narrow the parameter space before FDTD optimization, and the final geometry is then optimized in FDTD and tested experimentally. This makes the measured performance an optimized demonstration rather than an independent prediction of the design principle. To strengthen the paper, please include a comparison between the matrix-model predictions, FDTD results, and experimental data (e.g., predicted optimal foci positions vs. simulated/measured optima), which would quantify the model's predictive value.

minor comments (6)

[Abstract] There are several typographical errors: 'employs' should be 'employ', 'losses its form' should be 'loses its form', and 'wide range of' needs the article 'a'. Please proofread throughout.
[Section 2.1] The notation is inconsistent: 'θ' appears both as a variable and as part of the vector [x θ]; please clarify that θ is the ray angle in the vector representation. Also, the text refers to 'Figure . 2' with a stray period.
[Section 2.1 and 2.2] The refractive index of the resin is given only at 1310 nm. For a device claimed to operate over 800–1670 nm, the material dispersion of the polymer should be reported, as it directly affects the chromatic aberration discussion.
[Section 2.2] The FDTD simulation details are not provided: mesh size, boundary conditions, source injection method, normalization, and whether the simulated loss includes the collimator lens. These are needed for reproducibility.
[Section 3.2] The description of how the total coupling loss is obtained from the power difference is incomplete. Please specify how the input power to the device was measured and how on-chip waveguide losses were subtracted.
[Section 3.3] The thermal robustness test only covers two thermal cycles. The claim of 'excellent stability' would be strengthened with more cycles and quantification of transmitted power variation.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the 800 nm bandwidth is an empirical, FDTD-optimized and experimentally measured result, not an output forced by the matrix-optics model or by self-citation.

full rationale

The claimed 1 dB bandwidth exceeding 800 nm is not derived from an input that already contains it. The matrix-optics pre-design (Sec. 2.1) uses Eq. (1) for the wavelength-dependent divergence angle and Eq. (5), R1+R2=2d2, as a collimation/aberration condition; neither equation encodes a wavelength-flat coupling efficiency or a specific 1 dB bandwidth. The actual bandwidth is produced by full-vectorial FDTD optimization (Sec. 2.2) and then measured on fabricated devices (Secs. 3.1-3.2), so the experiment validates an optimized geometry rather than recovering a fitted parameter. The self-citations (refs 7 and 8; also ref 2) are used only to motivate broadband sensing and do not support the coupler's mechanism or performance. I therefore find no step where a prediction reduces by construction to its input. Separately, the paper asserts without derivation that quasi-free-space propagation 'inherently eliminat[es] wavelength dependencies' (Sec. 2), and Sec. 3.4 later concedes 'residual chromatic aberrations present in the current ellipsoidal geometry'; this is an evidentiary/overclaim weakness, not circularity, because the conclusion is not a restatement of the premise. Score 0.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard paraxial optics plus an asserted achromatic quasi-free-space premise. Several geometry parameters are tuned in FDTD before experiments, so the measured record is not an out-of-sample prediction. No raw data, code, or geometry files are shipped.

free parameters (4)

Ellipsoid x-axis focus position = 60 um
Optimized in FDTD baseline design (Section 2.2) to minimize fiber-to-fiber insertion loss; this is a fitted geometric parameter, not derived.
Ellipsoid y-axis focus position = 48 um
Second tuned focal length in the FDTD optimization; together with the x-focus it sets the reflector shape.
Spherical collimator radius = 15 um
Simulations found optimal collimation and coupling efficiency at this radius (Section 2.2); it is an input to the design, not independently predicted.
Output reflector geometry for chip mode = not stated (adjusted for roughly 4 um mode)
The second ellipsoid was modified by fine-tuning the x-intercept and y-focal parameters to match the SiN waveguide mode (Section 2.2); exact values are not given.

assumptions (5)

standard math Ray-transfer matrix model with thin spherical mirror and refraction matrices describes the dual-TIR ellipsoidal system sufficiently for pre-design (Eq. 3).
Paraxial ABCD matrices are a standard tool; their validity for this strongly curved, high-NA geometry is assumed, not proven.
domain assumption The two TIR events at the air-polymer interface behave as ideal spherical mirrors with radii R1 and R2, and TIR is lossless over the full wavelength range.
Section 2 treats reflections as mirror matrices with no wavelength-dependent loss; surface roughness and dispersion are ignored.
ad hoc to paper Once the guided mode expands, light propagates quasi-free-space, making coupling intrinsically wavelength-agnostic.
This is the paper's central enabling premise (Section 2, first paragraph) and is stated without derivation; it is precisely the premise the experiments are meant to support.
domain assumption A Gaussian beam divergence model (ISO 11146-1, Eq. 1) applies to the fiber and waveguide emission, and the collimated input approximates a point source at the first ellipsoid focus.
Used to set the angular spread in the matrix model; real waveguide modes are not exactly Gaussian and the collimating lens introduces its own aberrations.
domain assumption The optimized FDTD geometry transfers to the fabricated IP-n162 printed structure without significant shrinkage or index mismatch.
The paper acknowledges resin shrinkage only for the funnel diameter (Section 3) and does not quantify how fabrication error shifts optical performance.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Wavelength-agnostic 3D-Nanoprinted coupler." pith.science (2026). https://pith.science/paper/U7SCKE6G

@misc{pith2026250617825,
  author       = {Pith},
  title        = {Pith review of: Wavelength-agnostic 3D-Nanoprinted coupler},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/U7SCKE6G}},
  note         = {Machine review of arXiv:2506.17825}
}

read the original abstract

We present a photonic coupler that exhibits effectively wavelength-agnostic performance for ultra-broadband optical interfacing. By incorporating a dual-ellipsoidal geometry, the design facilitates quasi-free-space optical propagation. We further propose a hybrid modelling workflow employs a matrix optics-based approach as an efficient pre-design tool, capturing critical geometry-to-mode mapping characteristics, significantly narrowing the parameter space required for subsequent full-vectorial finite-difference time-domain (FDTD) simulations. Our design achieves a 1 dB bandwidth exceeding 800 nm coupling from fibre to chip, with an insertion loss as low as 1.3 dB,to the best of our knowledge, a record for any reported photonic couplers. The additive manufacturing approach via 3D nano-printing enables flexible geometry customization and sub-micron integrated alignment features, facilitating seamless integration with photonic chips and optical fibers. Experimental validation demonstrates excellent stability and thermal robustness across diverse operational conditions, highlighting the design's suitability for integration into wide range of broadband photonic systems.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

[15]

Yu, S., Zuo, H., Sun, X., Liu, J., Gu, T., & Hu, J. (2020). Optical Free-Form Couplers for High-density Integrated Photonics (OFFCHIP): A Universal Optical Interface. Journal of Lightwave Technology, 38(13), 3358–3365

work page 2020
[1]

K., Duan, G

Dong, P., Chen, Y. K., Duan, G. H., & Neilson, D. T. (2014). Silicon photonic devices and integrated circuits. In Nanophotonics (Vol. 3, Issues 4–5, pp. 215–228). Walter de Gruyter GmbH

work page 2014
[2]

(2022) Advances in cost-effective integrated spectrometers

Li A, Yao C, Xia J, Wang H, Cheng Q, Penty R, Fainman Y, Pan S. (2022) Advances in cost-effective integrated spectrometers. Light: Science & Applications.11(1):174. Fig. 6 Possible methods to further broaden the coupler's operating bandwidth. By quantizing the ellipsoidal reflector geometry, the output beam profile can be more precisely engineered, paving...

work page 2022
[3]

Alimi, I., Patel, R., Silva, N., Sun, C., Ji, H., Shieh, W., Pinto, A., & Muga, N. (2021). A review of self- coherent optical transceivers: Fundamental issues, recent advances, and research directions. Applied Sciences (Switzerland), 11(16)

work page 2021
[4]

Huang, Z., Wang, P., Liu, J., Xiong, W., He, Y., Xiao, J., Ye, H., Li, Y., Chen, S., & Fan, D. (2021). All- Optical Signal Processing of Vortex Beams with Diffractive Deep Neural Networks. Physical Review Applied, 15(1)

work page 2021
[5]

Yang, Y., Luo, H., Zhang, R., Yang, F., Wu, B., Qiu, K., & Wen, F. (2023). Semiconductor Optical Amplifier (SOA)-Driven Reservoir Computing for Dense Wavelength-Division Multiplexing (DWDM) Signal Compensation. Sensors, 23(12)

work page 2023
[6]

M., & Lončar, M

Hu, Y., Yu, M., Buscaino, B., Sinclair, N., Zhu, D., Cheng, R., Shams-Ansari, A., Shao, L., Zhang, M., Kahn, J. M., & Lončar, M. (2022). High-efficiency and broadband on-chip electro-optic frequency comb generators. Nature Photonics, 16(10), 679–685

work page 2022
[7]

(2023) Integrated reconstructive spectrometer with programmable photonic circuits

Yao C, Xu K, Zhang W, Chen M, Cheng Q, Penty R. (2023) Integrated reconstructive spectrometer with programmable photonic circuits. Nature Communications. 14(1):6376

work page 2023

Show all 16 references

[8]

Yao, C., Zhang, W., Bao, P., Ma, J., Zhuo, W., Chen, M., Shi, Z., Zhou, J., Ye, Y., Ming, L., Yan, T., Penty, R., & Cheng, Q. (2024). Chip-scale sensor for spectroscopic metrology. 15(1):10305

2024
[9]

Mu, X., Wu, S., Cheng, L., & Fu, H. Y. (2020). Edge couplers in silicon photonic integrated circuits: A review. Applied Sciences (Switzerland), 10(4)

2020
[10]

Cheng, L., Mao, S., Li, Z., Han, Y., & Fu, H. Y. (2020). Grating couplers on silicon photonics: Design principles, emerging trends and practical issues. In Micromachines (Vol. 11, Issue 7). MDPI AG

2020
[11]

S., Lee, K

Bhandari, B., Im, C. S., Lee, K. P., Kim, S. M., Oh, M. C., & Lee, S. S. (2020). Compact and broadband edge coupler based on multi-stage silicon nitride tapers. IEEE Photonics Journal, 12(6)

2020
[12]

J., Reed, G

Marchetti, R., Lacava, C., Khokhar, A., Chen, X., Cristiani, I., Richardson, D. J., Reed, G. T., Petropoulos, P., & Minzioni, P. (2017). High-efficiency grating-couplers: Demonstration of a new design strategy. Scientific Reports, 7(1)

2017
[13]

R., Blaicher, M., Hoose, T., Dietrich, P.-I., Marin-Palomo, P., Lindenmann, N., Nesic, A., Hofmann, A., Troppenz, U., Moehrle, M., Randel, S., Freude, W., & Koos, C

Billah, M. R., Blaicher, M., Hoose, T., Dietrich, P.-I., Marin-Palomo, P., Lindenmann, N., Nesic, A., Hofmann, A., Troppenz, U., Moehrle, M., Randel, S., Freude, W., & Koos, C. (2018). Hybrid integration of silicon photonics circuits and InP lasers by photonic wire bonding. Op...

2018
[14]

E., Muñ oz, M

Elsherif, M., Salih, A. E., Muñ oz, M. G., Alam, F., AlQattan, B., Antonysamy, D. S., Zaki, M. F., Yetisen, A. K., Park, S., Wilkinson, T. D., & Butt, H. (2022). Optical Fiber Sensors: Working Principle, Applications, and Limitations. Advanced Photonics Research, 3(11)

2022
[16]

Napiorkowski, M., Kasztelanic, R., & Buczynski, R. (2024). Optimization of spatial mode separation in few- mode nanostructured fibers with generative inverse design networks. Engineering Applications of Artificial Intelligence, 133

2024

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [15]

Yu, S., Zuo, H., Sun, X., Liu, J., Gu, T., & Hu, J. (2020). Optical Free-Form Couplers for High-density Integrated Photonics (OFFCHIP): A Universal Optical Interface. Journal of Lightwave Technology, 38(13), 3358–3365

work page 2020

[2] [1]

K., Duan, G

Dong, P., Chen, Y. K., Duan, G. H., & Neilson, D. T. (2014). Silicon photonic devices and integrated circuits. In Nanophotonics (Vol. 3, Issues 4–5, pp. 215–228). Walter de Gruyter GmbH

work page 2014

[3] [2]

(2022) Advances in cost-effective integrated spectrometers

Li A, Yao C, Xia J, Wang H, Cheng Q, Penty R, Fainman Y, Pan S. (2022) Advances in cost-effective integrated spectrometers. Light: Science & Applications.11(1):174. Fig. 6 Possible methods to further broaden the coupler's operating bandwidth. By quantizing the ellipsoidal reflector geometry, the output beam profile can be more precisely engineered, paving...

work page 2022

[4] [3]

Alimi, I., Patel, R., Silva, N., Sun, C., Ji, H., Shieh, W., Pinto, A., & Muga, N. (2021). A review of self- coherent optical transceivers: Fundamental issues, recent advances, and research directions. Applied Sciences (Switzerland), 11(16)

work page 2021

[5] [4]

Huang, Z., Wang, P., Liu, J., Xiong, W., He, Y., Xiao, J., Ye, H., Li, Y., Chen, S., & Fan, D. (2021). All- Optical Signal Processing of Vortex Beams with Diffractive Deep Neural Networks. Physical Review Applied, 15(1)

work page 2021

[6] [5]

Yang, Y., Luo, H., Zhang, R., Yang, F., Wu, B., Qiu, K., & Wen, F. (2023). Semiconductor Optical Amplifier (SOA)-Driven Reservoir Computing for Dense Wavelength-Division Multiplexing (DWDM) Signal Compensation. Sensors, 23(12)

work page 2023

[7] [6]

M., & Lončar, M

Hu, Y., Yu, M., Buscaino, B., Sinclair, N., Zhu, D., Cheng, R., Shams-Ansari, A., Shao, L., Zhang, M., Kahn, J. M., & Lončar, M. (2022). High-efficiency and broadband on-chip electro-optic frequency comb generators. Nature Photonics, 16(10), 679–685

work page 2022

[8] [7]

(2023) Integrated reconstructive spectrometer with programmable photonic circuits

Yao C, Xu K, Zhang W, Chen M, Cheng Q, Penty R. (2023) Integrated reconstructive spectrometer with programmable photonic circuits. Nature Communications. 14(1):6376

work page 2023