🔬 optics

Goos-Hanchen-Shift Photonic Sensor for Nanometer-Scale Delayering and Tamper Detection in Semiconductor Packages

This paper proposes a co-packaged photonic sensor that utilizes the Goos-Hanchen shift of a reflected optical beam to rapidly and sensitively detect nanometer-scale delayering and localized drilling in semiconductor packages, offering significantly higher sensitivity and spatial encoding compared to conventional total internal reflection methods.

Original authors: Mia Mohammad Shoaib Hasan, Mohamed Elkabbash

Published 2026-08-13

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Mia Mohammad Shoaib Hasan, Mohamed Elkabbash

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 you are trying to protect a secret treasure chest, but the lock is so complex that a thief could pick it without you ever hearing a click. In the world of computer chips, this "treasure" is the sensitive data inside a semiconductor package. To steal it, a hacker might try to slowly scrape away the layers of the chip's protective shell, a process called "delayering," to get a peek at the circuits inside. Traditional security systems are like a metal mesh wrapped around the chest; if the mesh is cut, an alarm sounds. But clever thieves have learned to cut the mesh and weave it back together so perfectly that the alarm never goes off.

To solve this, scientists are looking at light itself. Two key ideas make this new approach possible. First, there is "Total Internal Reflection," which happens when light hits a surface at a steep angle and bounces back completely, like a stone skipping across a pond. Second, there is a quirky phenomenon called the "Goos–Hänchen shift." When light bounces off a surface, it doesn't just bounce straight back; it actually slides a tiny bit sideways along the surface before returning. This slide is usually invisible to the naked eye, but if you can measure it with extreme precision, it becomes a super-sensitive ruler. If the surface the light is bouncing off changes even a tiny bit—like if a layer of material is scraped away—the size of that sideways slide changes in a predictable way. This paper explores how we can use this tiny, invisible slide to catch thieves in the act of peeling back a chip's skin, layer by microscopic layer.

The researchers, MIA MOHAMMAD SHOAIB HASAN and MOHAMED ELKABBASH, propose a new kind of security sensor that acts like a high-tech "slip detector" for computer chips. Instead of using a metal mesh that can be tricked, they use a laser beam and the Goos–Hänchen shift to watch the thickness of the chip's packaging. They designed a system where a laser beam is shot at the chip's surface at a sharp angle. Normally, the beam would bounce right back. However, they added a special high-index layer that allows the light to "tunnel" slightly into the material before reflecting. This creates a specific amount of sideways slide, or shift, for the reflected beam.

The team used computer simulations to test how this sensor reacts when an attacker tries to drill or scrape away the material. They found that as the material gets thinner, the sideways slide of the light beam grows in a very straight, predictable line. In their simulations, for every single nanometer of material removed, the beam shifted by 5.3 nanometers. To put that in perspective, this is nearly three times more sensitive than older methods that just use a simple bounce. This means the sensor can detect a thief removing a layer of material that is incredibly thin—down to about 28 nanometers—long before the thief can actually reach the sensitive circuits inside.

The paper also tackles a major worry: what if the environment changes? If the chip gets hot, or if the laser's color shifts slightly, could the sensor get confused and scream "Thief!" when no one is there? The authors simulated these conditions, testing what happens if the temperature changes drastically or if the laser's wavelength drifts. They found that their sensor is very tough. Even with big changes in temperature or light color, the relationship between the material thickness and the beam's slide remains stable. This is because they are using a specific "off-resonant" range where the physics is simple and reliable, avoiding the messy, jumpy behavior that happens when things are perfectly tuned to a specific frequency.

Finally, the researchers checked if this sensor could catch a thief who tries to be sneaky by only drilling a tiny hole in one spot, rather than scraping the whole surface. The simulations showed that the sensor still works. Even a narrow drill creates a change in the light's slide, and the wider the drill, the bigger the change. This suggests the sensor can spot both slow, uniform scraping and sudden, localized drilling attacks.

In short, this paper suggests a way to turn a tiny, weird quirk of light physics into a powerful shield for computer chips. By measuring exactly how much a laser beam slides sideways, this sensor can detect the removal of nanometer-thin layers of material with high speed and high sensitivity. It offers a way to catch tampering that is much harder to fake than current electronic alarms, potentially stopping attackers before they can ever see the secrets hidden inside the chip.

Technical Summary: Goos–Hänchen-Shift Photonic Sensor for Nanometer-Scale Delayering and Tamper Detection

Problem Statement
Physical tampering of semiconductor packages poses a critical threat to system integrity, enabling attackers to expose internal circuitry, probe signals, or modify device operation. Current state-of-the-art anti-tamper solutions, primarily electronic active meshes, face limitations: they are vulnerable to sophisticated FIB-assisted removal or bypass strategies, offer only die-level protection, and fail to address the expanded attack surface created by advanced packaging (e.g., chiplets, interposers). Furthermore, indirect thermal or electrical signature sensors can be spoofed by reproducing environmental conditions. There is a distinct lack of protection methods capable of rapid, nanometer-scale sensing of thin-film delayering that generates an optical signature difficult for an attacker to emulate.

Methodology
The authors propose a co-packaged photonic sensor based on the Goos–Hänchen (GH) shift—the lateral displacement of a finite optical beam upon total internal reflection (TIR). The sensor architecture integrates a laser or VCSEL and a detector array directly onto the package.

Mechanism: A Gaussian beam is launched at an oblique angle into a high-index sensing layer. The sensor utilizes Frustrated Total Internal Reflection (FTIR) to couple the beam into this high-index layer.
Physics: The beam consists of transverse-wavevector components ( $k_{\parallel}$ ) that acquire a thickness-dependent propagation phase. Upon recombination, this phase gradient results in a lateral shift ( $D_{GH} \approx -\partial\phi/\partial k_{\parallel}$ ).
Detection Logic: Progressive delayering (removal of material) or localized drilling alters the multilayer complex reflection coefficient. This modifies the GH shift, changing the beam's centroid and transverse intensity distribution. These changes are measured by a CMOS photodetector array to trigger zeroization or shutdown.
Simulation: The study employs Finite Element Method (FEM) analysis (COMSOL Multiphysics) to model the interaction between a Gaussian beam (incident at 65°) and the sensing layers, varying thickness from 20 nm to 820 nm.

Key Contributions and Results

Enhanced Sensitivity via FTIR: The paper demonstrates that using an FTIR-based structure with a high-index sensing layer significantly outperforms conventional low-index TIR structures.
- Low-index TIR: Exhibits a delayering sensitivity of 1.8 nm of beam shift per nanometer of material removed.
- High-index FTIR: Achieves a sensitivity of 5.3 nm of beam shift per nanometer of material removed (nearly three times higher).
- Detection Limits: Based on a benchmark beam-position accuracy of 150 nm, the FTIR sensor reduces the minimum detectable delayering depth from ~83 nm (low-index) to **28 nm**.
Linear Operating Regime: The sensor exhibits an approximately linear response for sensing-layer thicknesses below 250 nm. In this regime, the relationship between beam displacement and remaining layer thickness is directly invertible, allowing for straightforward calibration without complex nonlinear corrections.
Robustness to Environmental Variations: The authors evaluate the sensor's stability against variations that typically cause false positives in other systems:
- Refractive Index: The study simulated a total refractive-index variation of $\Delta n = 0.01$ for both dielectric layers. For the high-index layer ( $Al_2O_3$ ), this index change corresponds to a hypothetical temperature excursion of approximately 364°C based on its thermo-optic coefficient. For the low-index layer ( $SiO_2$ ), the same $\Delta n = 0.01$ corresponds to a hypothetical temperature change of approximately 900°C. In both cases, the change in the GH-shift response was negligible.
- Wavelength: Variations of 1540–1560 nm (representing a 25–85°C VCSEL drift) show nearly overlapping GH-shift curves in the linear regime.
- Incidence Angle: Variations of ±1° around the nominal 65° angle produce no significant change in calibration, provided the system operates off-resonance (away from the Fabry–Pérot peak near 600 nm).
Localized Attack Detection: The sensor is not limited to uniform delayering. Numerical results indicate that localized drilling produces a monotonic GH-shift change that increases with both drill depth and width. Wider drills are detected at shallower depths due to the larger fraction of the beam perturbed.

Significance and Claims
The paper posits that this GH-shift readout offers a rapid, spatially encoded, and difficult-to-emulate approach for semiconductor-package tamper detection.

Emulation Difficulty: Unlike scalar electronic sensors, an attacker cannot easily spoof the signal. To emulate an untampered response after delayering, an adversary would need to precisely reconstruct the entire spatial intensity profile of the reflected beam (attenuating excess light and injecting specific missing light distributions), a task significantly more complex than bypassing a single electrical trace.
Security Architecture: The approach provides continuous security from the package boundary to the protected dies. The authors suggest that combining this distributed package-level sensor with localized die-level optical sensors could extend detection to FIB-scale attacks with feature sizes of a few tens of nanometers.
Performance: The system offers a detection latency potentially on the order of 10 ns (limited by readout electronics), significantly shortening the time-to-detection compared to current methods.

The work establishes that FTIR-based GH shifts can enhance delayering sensitivity without sacrificing robustness against environmental and mechanical perturbations, addressing a critical gap in physical security for advanced semiconductor packaging.

Technical Summary: Goos–Hänchen-Shift Photonic Sensor for Nanometer-Scale Delayering and Tamper Detection

More like this