Fabry-Perot-Insensitive Edge Coupling for Robust PIC Characterization
Sammy Noël Parisé, Raphaël Dubé-Demers, David Turgeon, Alireza Geravand, Sophie LaRochelle, Wei Shi · 2025
Read Fabry-Perot-Insensitive Edge Coupling for Robust PIC Characterization by Sammy Noël Parisé, Raphaël Dubé-Demers, David Turgeon, Alireza Geravand, Sophie LaRochelle, Wei Shi (2025) 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-2506-08155.txt.
Introduction
Sammy Parisé and co-authors fix a measurement wobble in photonic chip testing in a 2025 paper. When a fiber array hovers near a silicon chip, the tiny gap acts like a mirror pair that modulates light coupling. Shifts smaller than a wavelength swing readings by tenths of a decibel, wrecking repeatability.
Their remedy takes two readings at gap distances a quarter wavelength apart. The two errors carry opposite signs and cancel when averaged. Repeatability improved sharply: the spread between repeated insertion loss readings fell from 0.34 to 0.04 decibels. The trick needs no index gel, which suits wafer-scale testing lines and quantum photonics work.
Catalog record
| CREATOR | Sammy Noël Parisé, Raphaël Dubé-Demers, David Turgeon, Alireza Geravand, Sophie LaRochelle, Wei Shi |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 20 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-08155
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Fabry-Perot-Insensitive Edge Coupling for Robust PIC Characterization Sammy Noël Parisé, Raphaël Dubé-Demers, David Turgeon, Alireza Geravand, Sophie LaRochelle, Wei Shi Abstract 1 Abstract In this work, we investigate the impact of FP ( FP ) cavities formed between a fiber arrays and a silicon chips during edge-coupled testing of PIC . Our results show that subwavelength variations in the distance between the fiber array and the device under test induce coupling efficiency modulations of several tenths of a decibel—challenging current alignment precision and undermining measurement repeatability in wafer-level probing. To address these FP-induced fluctuations, we introduce a complementary dual-measurement technique that shifts the chip–fiber separation by a quarter wavelength, thereby generating phase-opposite modulation artifacts that cancel upon averaging. This approach allows a higher tolerance in probe-to-PIC positionning by substantially eliminating the FP modulation. We also demonstrate how it can be leveraged to enhance repeatability and accurate positionning during alignment, reducing the insertion loss variability between multiple measurements from 0.34 to 0.04 dB. Such precision is essential for emerging applications like quantum photonics, silicon photonic sensors, and high-bandwidth optical interconnects where measurement accuracy directly impacts device performance characterization. Ultimately, our method offers a robust and accurate solution for high-throughput PIC characterization without requiring index-matching gel, which is impractical for wafer-level testing. 2 Introduction The rapid evolution of PIC has been paralleled by an increasing demand for highly accurate and reliable characterization methods [ 6 , 4 ] . Wafer-level testing has become mandatory for high-volume manufacturing in modern optical communication systems. With the recent advancement of manufacturable platforms for photonic quantum computing [ 2 ] , the precision requirements for these testing methodologies have reached unprecedented levels. While both surface coupling and edge coupling are widely employed, edge coupling has gained significant traction in high-performance and industrial applications due to its inherently lower insertion losses and reduced sensitivity to polarization and wavelength variations. However, this approach presents significant challenges in high-throughput testing environments where measurement accuracy is critical. Inaccuracies in these measurements can lead to discrepancies between pre- and post-packaging performance, ultimately undermining the reliability of the PICs in practical applications. A critical yet often overlooked source of measurement inaccuracy in edge-coupled configurations is the formation of FP cavities between the fiber array and the silicon chip. These cavities introduce resonant interference patterns that significantly affect insertion loss measurements. What makes this particularly challenging for wafer-level testing is that even subwavelength variations in the probe-to-chip distance—on the order of nanometers—can induce measurable fluctuations in coupling efficiency (up to 0.34 dB in our measurements). Traditional calibration approaches that rely on subtracting a reference measurement assume perfectly consistent coupling conditions across all measurements. However, achieving nanometer-precision positioning control between a fiber array and multiple devices across a wafer is impractical in industrial testing environments. When testing individual dies, an index-matching gel is typically used to minimize this effect [ 7 ] . However, this approach becomes impractical when scaling to wafer or multi-die level testing, where the application and removal of gel would significantly complicate and slow the measurement process. To address this challenge, we present a CDM ( CDM ) technique that not only reveals the presence of FP -induced modulation errors in our experimental data but also proves highly effective in compensating for them through systematic phase manipulation. By performing two measurements at closely spaced, yet distinct, probe-to-PIC distances—specifically shifted by a quarter wavelength—we create datasets in which the FP modulations are phase-opposed. When combined, these complementary modulations largely cancel each other, yielding a measurement of the transmission less sensitive to positionning variation. Our results, obtained by sweeping a tunable laser from 1505 nm to 1675 nm, confirm that this CDM technique significantly mitigates insertion loss variability. We also studied the impact of the cavity on the alignment process in the x - y plane. We demonstrate how the proposed method increases repeatability in both the insertion loss measurement and the positionning of the fiber array relative to the chip. In the following sections, we detail the theoretical foundations, experimental setup, and empirical validation of our dual-measurement technique, underscoring its potential to improve the accuracy and reliability of PIC characterization in practical, high-throughput environments. 3 Theory Figure 1 : FP cavity between the die and the optical fiber As illustrated on Fig 1 , a FP cavity is formed by two partially reflective surfaces—the fiber-air interface and the oxide-aire interface—where multiple internal reflections produce resonant interference effects that depend on both wavelength and cavity length. These effects modulate transmitted light intensity, creating periodic enhancements or suppressions depending on the phase relationship between reflections [ 5 ] . In edge-coupled test configurations, the intrinsic circuit response is typically extracted by subtracting a loopback measurement from the measured response. However, both measurements are affected by FP cavities that generate interference patterns, introducing systematic, position-sensitive fluctuations in insertion loss (IL) measurements. These compounding variations create significant uncertainty in the final calculated circuit response, undermining measurement reliability. This position-dependent uncertainty—which we quantify as \sigma_{FP} — presents a fundamental challenge in wafer-level testing. While modern positioning systems excel at precise relative movements, achieving consistent absolute positioning with subwavelength precision across multiple measurements remains impractical in production environments. Our CDM technique transforms this challenging absolute positioning problem into a more tractable relative positioning solution by leveraging a controlled quarter-wavelength offset between measurements. The transmission T through a single FP cavity formed of parallel interfaces is expressed as T=\frac{T_{1}\,T_{2}}{\,1+R_{1}\,R_{2}-2\sqrt{R_{1}\,R_{2}}\cos\delta}, (1) with T_{1} and T_{2} denote the transmission coefficients and R_{1} and R_{2} the reflection coefficients at the interfaces. The phase difference \delta between successive reflections is defined as \delta=\frac{4\pi nd}{\lambda}, (2) where n is the refractive index of the medium ( \approx 1 for air) between the interfaces, d is the cavity length, and \lambda is the wavelength. To assess the position-dependent uncertainty \sigma_{FP} introduced by the FP cavity, Eq. ( 1 ) can be expanded in a Taylor series in powers of \cos\delta : T\approx\frac{T_{1}\,T_{2}}{1+R_{1}\,R_{2}}\left[1+\frac{2\sqrt{R_{1}\,R_{2}}}{1+R_{1}\,R_{2}}\cos\delta+\cdots\right]. (3) At a silica–air interface ( n_{SiO_{2}}\approx 1.44 , n_{air}\approx 1 ) as found in fiber facets and edge couplers, the normal incidence reflection coefficient is R=\left|\frac{n_{1}-n_{2}}{n_{1}+n_{2}}\right|^{2}\approx 0.032, (4) yielding T_{1}\approx T_{2}\approx 0.968 . Substituting these values into the expansion, we obtain T\approx 0.934\left(1+0.064\,\cos\delta+0.004\,\cos^{2}\delta+\cdots\right). (5) The uncertainty \sigma_{FP} is dominated by the first-order term ( \cos\delta ) corresponding to a periodic sinusoidal fluctuation of up to approximately 0.29 dB, which is significant for precision measurements. The second-order term ( \cos^{2}\delta ) contributes only about 0.018 dB, making it negligible for most practical purposes. Since the dominant contributor to \sigma_{FP} is the first-order term, our mitigation strategy focuses on eliminating this component. To mitigate it, a CDM technique is employed. By introducing a phase shift of \delta^{\prime}=\delta\pm\pi , one obtains \cos\delta^{\prime}=-\cos\delta . Since the only adjustable parameter is the distance between the interfaces, this phase shift is implemented by modifying the cavity length by translating the fiber array: d^{\prime}=d\pm\frac{\lambda_{0}}{4n}, (6) where \lambda_{0} is the design wavelength. With two measurements corresponding to cavity lengths d and d^{\prime} , the transmission coefficients are approximately \displaystyle T \displaystyle\approx 0.93\left(1+0.064\,\cos\delta+0.004\,\cos^{2}\delta+\cdots\right), (7) \displaystyle T^{\prime} \displaystyle\approx 0.93\left(1-0.064\,\cos\delta+0.004\,\cos^{2}\delta+\cdots\right). (8) Averaging these measurements, T_{\mathrm{mean}}\approx\frac{T+T^{\prime}}{2}\approx 0.93, (9) effectively cancels the first-order error term, reducing \sigma_{FP} from 0.17 dB to 0.04 dB. During