Laser written waveguides to the sample edge

1. Problem Statement

Femtosecond laser writing is a versatile method for creating 3D photonic circuits inside transparent materials like glass. However, a significant manufacturing bottleneck exists when interfacing these circuits with external optical fibers via butt-coupling at the sample edge.

• Edge Aberration: As the laser focus approaches the substrate edge, the focusing cone intersects both the top and side faces of the glass. This causes spatial and temporal splitting of the focus due to refraction.
• Consequences:
  • Intensity Reduction: The effective pulse energy at the primary focus drops significantly (theoretically by a factor of 4) because half the light is refracted away, and the effective Numerical Aperture (NA) is reduced, doubling the focal area.
  • Waveguide Termination: The waveguide often terminates prematurely before reaching the edge, leaving a gap.
  • Post-Processing Requirement: To achieve efficient coupling, the substrate sides must be polished after writing to remove the aberrated section. This is impractical for pre-bonded chips or complex photonic packaging where polishing is impossible.
• Limitations of Previous Solutions: While Adaptive Optics (AO) using Spatial Light Modulators (SLMs) can compensate for these aberrations, the slow refresh rates of liquid crystal SLMs make them incompatible with high-speed manufacturing (mm/s writing speeds).

2. Methodology

The authors propose a simplified, high-speed fabrication technique to write waveguides directly to the sample edge without post-polishing. The core strategy involves Amplitude Masking and Pulse Energy Compensation.

• Amplitude Masking:
  • An amplitude mask is applied to the fabrication laser beam in the pupil plane of the objective lens.
  • Specifically, half of the pupil is blocked. This prevents any light rays from intersecting the side facet of the substrate, thereby eliminating the source of the edge aberration.
  • The mask is implemented using a blazed grating on an SLM (or potentially an actuated opaque plate), which blocks the zero-order beam and allows selective control of the pupil amplitude.
• Pulse Energy Compensation:
  • Blocking half the pupil reduces the total energy reaching the focus. To compensate for the reduced effective NA and maintain the necessary intensity for waveguide formation, the laser pulse energy is increased (approximately 2.5x in experiments).
  • This ensures the waveguide can be written all the way to the edge without premature termination.
• Fabrication Process:
  1. Bulk Writing: The main waveguide is written normally using a full pupil.
  2. Transition: The laser is gated off at a specific distance (e.g., 10 µm) from the edge.
  3. Edge Writing: The amplitude mask is engaged (half-pupil), and the pulse energy is increased. The laser is re-engaged to write the final segment connecting the bulk waveguide to the physical edge of the substrate.
  4. Materials: Experiments were conducted on Fused Silica (FS) and Corning Eagle 2000 (E2K) glass.

3. Key Contributions

• Polishing-Free Edge Writing: Demonstrated a method to fabricate waveguides that terminate exactly at the sample edge, eliminating the need for post-processing polishing steps.
• High-Speed Compatibility: Unlike AO-based compensation, this amplitude masking technique is compatible with standard high-speed writing (mm/s) because it does not rely on dynamic phase correction refresh rates.
• Mode Profile Preservation: Showed that despite the change in fabrication geometry (half-pupil vs. full-pupil), the resulting guided mode profiles remain well-matched to single-mode fibers, minimizing coupling loss.
• Scalability: The technique is applicable to various depths and materials, offering a pathway for scaling up photonic chip manufacturing, particularly for pre-bonded devices.

4. Results

• Beam Profile Analysis:
  • Without Correction: As depth increases, the output beam size increases significantly, and the profile becomes distorted due to the waveguide terminating before the edge and the beam diverging within the glass.
  • With Correction: The beam size remains constant across all depths (tested from 150 µm to 750 µm). The output profile is well-matched to the input single-mode fiber mode.
• Coupling Loss:
  • Theoretical/Calculated: Without correction, coupling loss increases with depth, exceeding 3 dB at 600 µm. With the edge correction technique, coupling loss remains uniform and below 1 dB across all depths.
  • Experimental Transmission: Butt-coupling experiments at 780 nm showed significant transmission improvements. For depths >300 µm, the edge correction technique provided the most substantial gains.
• Insertion Loss (IL) and Extra Loss (EL):
  • Total insertion loss for 20 mm samples was measured at 2.6 ± 0.23 dB (FS) and 2.2 ± 0.1 dB (E2K).
  • Extra Loss (EL): When comparing the "as-written" (edge-corrected) samples to "polished-back" standard samples, the extra loss introduced by the technique was negligible for Fused Silica (<0.5 dB) and consistently around 1 dB for E2K.
  • The 1 dB loss in E2K is attributed to the difficulty in matching mode profiles in the "cumulative heating regime" (high pulse energy required for E2K), rather than the amplitude masking itself.
• Propagation Loss: Measured propagation loss was low (<0.1 dB/cm for E2K and ~0.7–0.8 dB/cm for FS) and independent of waveguide depth, confirming that amplitude masking does not degrade waveguide quality.

5. Significance

This work addresses a critical barrier in the commercialization of femtosecond laser-written photonic circuits. By removing the requirement for post-fabrication polishing, the technique enables:

• Photonic Packaging: Direct integration of laser-written circuits onto pre-bonded chips or components where polishing is mechanically impossible.
• Manufacturing Efficiency: A faster, more robust production workflow suitable for industrial scale-up.
• Versatility: The method is generalizable to other transparent materials (e.g., crystals) and internal laser processing applications, expanding the utility of 3D laser writing in diverse optical systems.