- Xanadu and ASML collaborating on advanced lithography for photonic quantum chips
- Partnership targets line edge roughness to reduce optical losses
- Xanadu achieved 0.085 dB/facet edge-coupling loss in June 2024
- Optical loss remains primary barrier to fault-tolerant photonic quantum computing
(cite index=”1-7,19-8″>Line edge roughness from lithography causes optical loss in photonic quantum processors, and (cite index=”1-5″>Xanadu, founded in 2016, just enlisted the world’s dominant lithography supplier to help fix it. (cite index=”23-1″>Xanadu and ASML announced a collaboration to develop advanced lithography processes for Xanadu’s photonic quantum hardware, aiming to reduce the scattering losses that destroy quantum information before it completes a computation.
The partnership comes seven months after (cite index=”2-1″>Xanadu and Tower Semiconductor expanded their collaboration to develop advanced silicon photonics on Tower’s high-volume platform. ASML brings a different capability: precision patterning control through deep ultraviolet (DUV) and extreme ultraviolet (EUV) exposure systems that can minimize nanometer-scale sidewall imperfections.
Optical loss sets the quantum error threshold
(cite index=”10-1,10-2″>A lost photon cannot be reset or recovered, making loss reduction a central focus of photonic quantum hardware development. (cite index=”22-1,22-2″>Line edge roughness induces Rayleigh scattering, causing photons to scatter out of integrated optical waveguides. Every scattered photon is lost quantum information—and unlike superconducting or trapped-ion qubits, there’s no way to reinitialize it mid-computation.
(cite index=”3-8,3-9″>Xanadu’s Aurora photonic quantum computer identified optical loss as the next major hurdle, with the biggest impact found in optimizing chip design and fabrication. (cite index=”8-2,8-5″>Xanadu demonstrated 0.085 dB/facet edge-coupling loss in June 2024. But (cite index=”17-6,17-7″>current optical path transmission rates of 78-82% must exceed 99.5% to meet fault-tolerance requirements.
The difference between 82% and 99.5% transmission sounds marginal. It’s not. A photonic quantum chip routes photons through dozens of components—beam splitters, waveguides, phase shifters—and loss compounds at each step. If a chip requires 50 operations and each has 95% transmission, overall photon survival drops to 7.7%. At 99.5% per operation, survival is 77.8%. That gap determines whether error correction can keep pace with physical errors, or whether the system drowns in noise before completing a single useful calculation.
ASML targets patterning precision at nanometer scale
(cite index=”1-1,1-2″>The work will examine how lithography and process conditions affect line edge roughness and optical losses, aiming to identify manufacturing approaches that improve patterning control. (cite index=”22-3″>The partnership aims to establish optimized patterning conditions by leveraging ASML’s advanced exposure systems and computational lithography.
ASML’s involvement signals that photonic quantum computing has matured enough to warrant attention from the semiconductor industry’s infrastructure providers. The company dominates advanced lithography—it’s the sole supplier of EUV systems—and photonics applications represent a growth vector beyond traditional logic and memory chips. (cite index=”18-2,18-3″>Stan Baron, Senior Vice President at ASML, said the collaboration is an opportunity to explore advanced lithography processes and better understand how ASML’s technology can support next-generation photonics applications.
Tower and Corning complete the manufacturing stack
Xanadu isn’t relying on lithography alone. (cite index=”2-3,2-4″>Xanadu and Tower co-engineered a unique production flow for Xanadu’s custom material stack, designed to sustain scalability and performance as systems grow in complexity. (cite index=”8-7″>Xanadu also has a joint development agreement with Corning to develop customized fiber and fiber-array solutions engineered for low-loss networking of photonic quantum computing chips.
The manufacturing strategy resembles traditional semiconductor fabless models: Xanadu designs the chips, Tower fabricates them on silicon photonics platforms, ASML optimizes the lithography step, and Corning handles chip-to-fiber interconnects. (cite index=”3-1″>Xanadu’s Aurora quantum computer consists of 35 photonic chips and 13 km of fiber optics operating at room temperature, so packaging and interconnect losses matter as much as on-chip waveguide quality.
Lithography partnerships won’t solve photonic quantum computing alone, but line edge roughness is one of the few loss mechanisms that advanced manufacturing can address directly. Xanadu has already demonstrated edge-coupling losses below 0.1 dB/facet and built functional 12-qubit systems. The ASML collaboration targets waveguide sidewall quality, which compounds across hundreds of components in a fault-tolerant architecture. If the partnership can push transmission fidelity above 99% per component through better patterning control, photonic quantum systems move closer to error-correction thresholds that make million-qubit machines plausible. Until then, optical loss remains the binding constraint.
Why is line edge roughness the primary lithography challenge for photonic quantum chips?
(cite index=”22-1,22-2″>Line edge roughness is a nanometer-scale sidewall imperfection introduced during photolithography and etch processing that induces Rayleigh scattering, causing photons to scatter out of integrated optical waveguides. Unlike electrical signals that can be amplified or regenerated, scattered photons are permanently lost, destroying the quantum information they carry. Smoother waveguide edges reduce scattering and improve photon survival rates across multi-component circuits.
How does Xanadu’s approach differ from competing photonic quantum architectures?
(cite index=”17-10,17-11,17-12″>PsiQuantum prioritizes operational redundancy with fusion measurements and hardware fault-tolerance thresholds above 10% photon loss, while Xanadu prioritizes encoding efficiency using GKP logical qubits that resist Gaussian noise but demand extreme optical path transmission fidelity below 0.5 dB loss. Xanadu’s architecture requires better component-level performance but offers deterministic gate operations. PsiQuantum tolerates more loss per component but requires higher photon generation rates and more complex fusion networks.
Article Source: Xanadu, ASML to Advance Lithography for Photonic Quantum Hardware
