🔬 optics

Comment on "Angle insensitive filters based on Fabry-Perot resonance structures" [J. Appl. Phys. 136, 193102 (2024)]

This Comment challenges the validity of a recent study on angle-insensitive Fabry-Perot filters by reporting failed replication attempts, the absence of a physical explanation for the claimed performance, and evidence that the original results likely stem from incorrect boundary conditions.

Original authors: Michele Cotrufo

Published 2026-05-20

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Michele Cotrufo

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a team of scientists (Cao et al.) published a paper claiming they built a special "light filter." Think of this filter like a very picky bouncer at a club who only lets in light of a specific color (wavelength) and blocks everything else. The amazing thing they claimed was that this bouncer works perfectly even if you approach the door from a steep angle, like running up to it from the side. Usually, if you change your angle, the filter stops working, but they said theirs didn't.

However, another scientist, Michele Cotrufo, looked at their work and said, "Wait a minute. I tried to build this same filter in my computer simulations, and it doesn't work the way they say it does."

Here is a breakdown of what happened, using simple analogies:

1. The "Magic" Filter vs. Reality

The original paper showed a graph where the filter's performance looked identical whether the light hit it straight on or at a sharp 70-degree angle. It was like saying a doorbell rings exactly the same way whether you press it gently from the front or slam it from the side.

Cotrufo tried to recreate this using three different, standard computer programs (like using three different calculators to check a math problem). In every single attempt, the filter failed to be angle-insensitive. As soon as the light hit the filter from an angle, the "bouncer" stopped working, and the color it let through shifted. This is normal physics; usually, things change when you look at them from a different angle.

2. The Missing Instructions and the "Ghost" Data

Cotrufo asked the original authors for their "recipe" (the exact settings they used in their computer). The authors provided some numbers and even sent the raw data (the final graph).

But when Cotrufo looked at the raw data, something felt "off."

The Perfect Overlap: In the original data, the curves for different angles were so perfectly identical that they looked like a photocopy. In the real world, even with the best equipment, there's usually a tiny bit of "noise" or variation. It was like seeing two fingerprints that matched so perfectly they looked like the same print, which is statistically impossible for two different angles.
The Wrong Blueprint: The authors claimed the filter had a specific thickness, but the pictures they showed of the "inside" of the filter (the electric field patterns) clearly showed a different, much thicker structure. It was like claiming to build a house with a 10-foot ceiling, but the blueprints they showed had a 20-foot ceiling.

3. The "Wrong Rules" Experiment

The original authors suggested that maybe Cotrufo just used the wrong "settings" in the computer, like a wrong boundary condition. In physics simulations, "boundary conditions" are like the rules of the game that tell the computer how light behaves at the edges of the screen.

To test this, Cotrufo decided to do something strange: He intentionally broke the rules.
He ran his simulations using the wrong boundary conditions (specifically, telling the computer to treat a tilted light beam as if it were coming straight on, which is physically incorrect).

The Shocking Result:
When Cotrufo used these "wrong" rules, his simulation results matched the original authors' data perfectly.

The "magic" angle-insensitivity appeared.
The strange drop in brightness at angles appeared.

The Conclusion

Cotrufo argues that the original paper's results weren't a discovery of a new physical phenomenon. Instead, they were likely the result of a simulation error. The original authors probably set up their computer model with the wrong "rules" (boundary conditions), which accidentally made the filter look like it worked at all angles.

Because the original authors refused to share their actual computer files (citing "intellectual property"), Cotrufo couldn't prove exactly what went wrong in their code. However, by showing that using incorrect rules reproduces their data, he strongly suggests that the paper's main claim is based on a mistake, not a breakthrough.

In short: The original paper claimed to find a "magic" light filter that works from any angle. Another scientist tried to build it, failed, and then realized that if you deliberately break the laws of physics in a computer simulation, you get the exact same "magic" results. This suggests the original paper was likely a mistake, not a discovery.

Technical Summary of the Comment on "Angle insensitive filters based on Fabry–Pérot resonance structures"

Problem Statement
This Comment addresses a significant discrepancy regarding the numerical results presented in a recent paper by Cao et al. (J. Appl. Phys. 136, 193102, 2024), which claimed to demonstrate angle-insensitive narrow band-pass filters using a cascaded metasurface structure. The original manuscript reported that a transmission peak with a ~5 nm linewidth remained at a constant central wavelength (approx. 1150 nm) for incident angles ranging from 0° to 70°, with only a minor decrease in peak transmission. The author of this Comment, Michele Cotrufo, asserts an inability to replicate these results despite using the exact parameters provided by the original authors, raising concerns about the validity of the reported angle-independent behavior.

Methodology
The Comment employs a rigorous comparative simulation approach to investigate the claims of the original paper:

Replication Attempts: Using the geometrical parameters disclosed by Cao et al. ( $d_g=240$ nm, $d_w=240$ nm, $d_m=310$ nm, $f=0.6$ , $\Lambda=640$ nm), the author performed simulations using three distinct electromagnetic solvers: Lumerical FDTD (matching the original software), Comsol (Finite Element Method), and Reticolo (RCWA).
Data Analysis: The author analyzed the raw data provided by Cao et al. to examine the spectral overlap and intensity dependence on angle.
Parameter and Boundary Condition Sweeps: To rule out numerical artifacts, the author conducted systematic campaigns varying mesh resolution, Fourier harmonics, and geometrical parameters.
Boundary Condition (BC) Hypothesis Testing: Specifically addressing the original authors' suggestion that the discrepancy might stem from "boundary conditions," the author intentionally simulated the structure using incorrect BCs (setting periodicity instead of the required Bloch/Floquet conditions for tilted wave injection) to test if this error could reproduce the original results.

Key Contributions and Results

Failure to Replicate Angle-Independence: Simulations using correct physical boundary conditions (Bloch/Floquet) across three different software packages consistently showed that the transmission peak shifts substantially as the incident angle exceeds 5°. By 20°, the peak had shifted away from the 1150 nm range, and additional narrow peaks appeared. This behavior is consistent with standard resonant waveguide grating physics and contradicts the flat, angle-independent response reported by Cao et al.
Analysis of Original Data: The raw data provided by Cao et al. showed an "extremely unusual" perfect overlap of transmission lineshapes at different angles, including identical weak oscillations and noise. Furthermore, the peak intensity dependence on angle in the original data fit a rescaled cosine function with an $R^2$ of 0.99996, a mathematical perfection not typical of physical resonant structures.
Identification of a Specific Error Source: The Comment demonstrates that when the author intentionally used incorrect boundary conditions (periodicity instead of Bloch/Floquet) in Lumerical, the simulation results closely matched the data presented by Cao et al. This incorrect setup effectively injects the wave at normal incidence regardless of the port angle setting, artificially lowering the amplitude (explaining the cosine-like transmission drop) while maintaining the spectral position.
Inconsistency in Reported Parameters: The Comment highlights a contradiction between the parameters provided for the transmission spectra (Fig. 7) and the field profiles (Fig. 8) in the original paper. The disclosed silica layer thickness ( $d_m=310$ nm) is thinner than the combined grating thickness, yet the field diagrams in Fig. 8 depict a silica layer significantly thicker than the metasurfaces, suggesting the figures may not correspond to the same structure.
Lack of Theoretical Explanation: The original paper offers no causal physical mechanism for the claimed angle independence, merely stating that standing wave patterns remain unchanged without explaining why such a phenomenon should occur in this specific geometry.

Significance and Conclusion
The paper concludes that the angle-independent behavior claimed by Cao et al. is likely unphysical and results from either a significant deviation between the reported and actual simulation parameters or, more specifically, the use of incorrect boundary conditions in the numerical simulations. The author notes that the original authors declined to share simulation files, citing intellectual property concerns, despite having already published the parameters and results.

The significance of this Comment lies in its demonstration that the reported "remarkable" device performance is an artifact of simulation error rather than a novel physical phenomenon. By reproducing the erroneous results through the deliberate application of incorrect boundary conditions, the author provides a plausible explanation for the discrepancy and underscores the necessity of rigorous simulation protocols and transparency in computational photonics research. The Comment was posted on arXiv following the retraction of the original article and the lack of a clear explanation from the editorial board regarding the role of this Comment in the retraction process.

1. The "Magic" Filter vs. Reality

2. The Missing Instructions and the "Ghost" Data

3. The "Wrong Rules" Experiment

The Conclusion

Technical Summary of the Comment on "Angle insensitive filters based on Fabry–Pérot resonance structures"

More like this