Modern electronics — from AI accelerators to medical sensors — rely on vertical connections drilled through silicon wafers, called Through-Silicon Vias (TSVs). Today these are made with a slow, chemistry-heavy process called Deep Reactive Ion Etching (DRIE). DRIE leaves rough, scalloped walls that limit electrical performance and require multiple additional treatment steps.

STAY2ME’s proposed solution — a directly-emitting holmium-doped crystal laser at 2.1 µm in GHz-burst mode — is chosen because holmium uniquely combines high pulse energy storage, stable burst amplification, and a wavelength perfectly matched to silicon’s transparency window.

The project advances from theoretical concept to experimental proof over 48 months, building two laser demonstrator systems and establishing the scientific foundations for future industrial adoption.

If successful, STAY2ME could transform semiconductor manufacturing by enabling denser, more energy-efficient chip architectures, unlocking the hardware performance that next-generation AI, edge computing, and photonic integration demand.

EUROPEAN SOVEREIGNTY

Directly aligned with the European Chips Act and the 2025 Semiconductor Coalition, STAY2ME strengthens Europe’s strategic autonomy in chip manufacturing by developing disruptive technology in-house.

SUSTAINABILITY

Laser-based TSV fabrication consumes up to 60% less energy than DRIE and eliminates the thousands of liters of ultrapure water and toxic chemicals used per production batch.

ECONOMIC IMPACT

The global TSV market is projected to grow from €3 billion (2023) to €8 billion by 2030. 

BEYOND SILICON

The 2 µm platform opens applications in polymer microfluidics, narrow-bandgap semiconductors (InP, Ge), THz generation, and soft X-ray secondary radiation — well beyond the original TSV target.

Silicon’s hostile optical properties

Silicon’s nonlinear refractive index is 100× higher than glass. Conventional ultrafast lasers suffer self-focusing, filamentation, and energy clamping — the laser intensity saturates before it can permanently modify the material. Simply increasing power does not help.

The right wavelength has never been paired with enough power

Silicon is transparent at 2 µm — ideal for nonlinear absorption without unwanted side effects. But building a femtosecond laser that operates natively at this wavelength, in GHz-burst mode, has never been achieved. Existing approaches use complex, inefficient wavelength conversion and deliver insufficient energy.

GHz-burst mode has never been explored in silicon

In glass, GHz-burst lasers deliver trains of closely spaced sub-pulses that accumulate energy beneficially, producing smooth, cylindrical high-aspect-ratio holes. Replicating this in silicon requires a completely new interaction regime — one that must carefully balance nonlinear absorption against the clamping effect. The right parameters are unknown and must be found experimentally.

Industrial viability must be proven from scratch

The process window must be narrow enough for precision but reproducible enough for industry. The resulting vias must compete with DRIE not only in quality, but in throughput, cost, and integration speed — requirements that add technological risk to every scientific choice.