OFC 2026 - Fully Automated Wafer-Level Edge and Grating Coupling Measurement System - FormFactor Inc. (FFI)
As AI and data center bandwidth demand explodes, traditional copper interconnects have reached their physical limits, and silicon photonics (SiPh) - with its high bandwidth, low power and low heat - has become key infrastructure for the 800G/1.6T era. Yet the volume-production bottleneck for photonic integrated circuits (PICs) has always been test cost. At the OFC 2026 advanced packaging and integration forum, Quan Yuan, senior optical engineer at test equipment leader FormFactor (FFI), presented a fully automated solution for "wafer-level edge and grating coupling," aiming to crack the high-cost problem of PIC volume production.
Wafer-Level Test (WLT): The Gatekeeper of PIC Profitability
Across a silicon photonics product's life cycle, post-packaging test and assembly (including fiber, laser and electrical bonding) account for 70-80% of total manufacturing cost.
Cost filter: if defects are not caught at the wafer stage, the expensive packaging steps that follow are completely wasted.
Technical pain point: optical alignment is extremely sensitive to displacement (MFD is typically only 1-10 µm), and traditional test faces tough challenges in sub-micron alignment, vibration, thermal drift and insufficient throughput.
Core Technology: The Pharos Optical Probe and 3D Trajectory Algorithm
FormFactor's Pharos optical probe series delivers key breakthroughs for the industry's two mainstream coupling schemes - edge coupling and grating coupling.
1. Edge Coupling: Breaking the Horizontal Measurement Barrier
Wafer-level edge measurement has traditionally been hard to achieve without dicing the wafer, often relying on foundries to fabricate expensive, high-loss prisms.
Periscope structure: Pharos uses micro 3D-printed lenses made with 2PP (two-photon polymerization) technology developed by Keystone Photonics, turning vertically incident light 90 degrees for horizontal coupling.
Mode field diameter (MFD) matching: the lens curvature is precisely designed to optimally match the waveguide MFD, cutting insertion loss to 0.5 dB/facet.
Working distance (WD): provides a safe working distance of 35-40 µm, greatly reducing the risk of crashing into the waveguide.
2. Grating Coupling: 10x Better Stability
Traditional measurement using a polished fiber array unit (FAU) must sit extremely close to the wafer (<15 µm) because of beam divergence, making collisions very likely.
Collimated beam: the planar wavefront produced by the Pharos lens is insensitive to Z-axis vibration, achieving measurement stability of 0.016 dBm versus 0.145 dBm for a traditional FAU - nearly 10x better.
3. Automation and 3D Incidence-Angle Extraction
An accurate incidence angle is critical to grating coupling efficiency. FormFactor's system performs XY scans at different Z heights to reconstruct the 3D beam propagation path.
Accuracy: X-Z incidence-angle repeatability of <±0.13°, and Y-Z offset-angle repeatability of <±0.12°.
Production Results: Throughput and Repeatability Data
In real production environments, FormFactor demonstrated the strong performance of its CM300xi-SiPh and TRITON platforms:
Test repeatability: optical and electrical measurement repeatability holds steady at <0.1 dB (0.05-0.07 dB on average).
Throughput optimization (skip-die strategy): by selectively skipping alignment on some dies (aligning only specific dies and applying their parameters to the rest), test time drops by 45-60%, raising throughput 1.8-2.5x.
Trade-off: the skip strategy brings a coupling power penalty of about 0.3-0.5 dB, but in high-volume production this efficiency trade is acceptable.
Simple Tech Trend View: The Last Mile of SiPh Volume Production
The Pharos probe technology FormFactor showed at this conference solves the thorniest problem in silicon photonics volume production: balancing safety and precision.
A win for 3D-printed optics: customizing lens MFD with 2PP lets the test platform adapt to different foundries' design rules - an important turning point as equipment makers move into optical component design.
Hardware-software co-optimization: beyond the physical limits of a 6-axis motion stage, FormFactor's high-frequency-domain image processing algorithm (adaptive optics) cuts focal-plane detection error from >50 µm to 2 µm - the real key to making automation work in practice.
Outlook: we expect that within the next 12-24 months, as CPO (co-packaged optics) enters pilot production, fully automated measurement systems with 3D angle extraction will become standard at Tier 1 foundries and module makers to meet increasingly stringent KGD (known good die) screening requirements.
Comments