Transforming scattering into a seed of a nanometric blade

Ultrafast lasers have steadily expanded the scope of optical manufacturing, from photonic structures in crystals to sub-20-nm surface nanowriting and the stealth dicing of hard wafers. In recent years, the lateral resolution of optical manufacturing has improved significantly through near-field, high-numerical aperture (NA) lenses and super-resolution schemes, thereby enabling advanced surface structuring, additive fabrication, and localised subsurface writing^1–13, as summarised in Table 1. However, such enhancement cannot directly translate to a small kerf when slicing crystals or transparent solids owing to the limited depth of field of a Gaussian beam. Although utilising Bessel beams may significantly extend the field depth, the resulting kerf widths measure hundreds of nanometres with rough surfaces¹⁴. This is not merely an engineering inconvenience. For high-value transparent materials such as laser crystals and semiconductor substrates, every additional micrometre of kerf translates into wasted feedstock, thicker polishing allowances, and higher costs. This bottleneck remains a longstanding problem in laser manufacturing.

Instead of pursuing conventional approaches such as optimising NAs or focusing techniques, Chai et al. transformed the physical role of scattering^4,5,8,9,13. In most laser-processing scenarios, scattering is a detrimental mechanism, which perturbs focusing, redistributes energy, and degrades precision. Chai et al. inverted this logic. They intentionally created an initial nanopore via femtosecond-laser-induced micro-explosion and then used the nanopore as a seed scatterer. When the incident beam was laterally offset from the seed within a critical distance (approximately 200 nm in fused silica; Fig. 1b–d), the scattered field interfered with the undisturbed portion of the beam, thus yielding a sufficiently strong localised intensity maximum to generate a secondary nanopore. The secondary pore became the next scatterer, and the process was repeated in a controlled cascade. In other words, the structure being written participated in writing the next structure. This interferometric-scattering mechanism achieves energy localisation independent of the original focal ellipsoid. By adjusting the pulse offset, the authors steered the trajectory and orientation of sequential nanopores. Moreover, by shifting from a point focus to a line field generated using a cylindrical lens (Fig. 1f), they translated the same principle into interferometric scattering-based optical tomoslicing (i-SOT), which is a plane-slicing strategy instead of a conventional crack-guided separation process. This strategy offers high throughput (1 m/s) and scalability, and the resulting surface simultaneously offers superior quality (Ra = 9 nm)¹³.

Fig. 1  Interferometric scattering as the basis of interferometric scattering-based optical tomoslicing (i-SOT). a–d Schematic of scatterer generation and propagation. e High-angle annular dark-field scanning transmission electron microscopy image of a horizontal nano-slit showing the 7-nm raw kerf. f Schematic of line-field generation using a cylindrical lens. g Planar wafering with high transmittance by cold splitting without etching. The wafer measures 30 × 10 mm (Images adapted from Ref. 13).

The performance metrics explain the importance of this concept. Additionally, the experiments were conducted with a modest NA of 0.6, thus avoiding the requirement for extremely high-NA optics. In fused silica, i-SOT yielded raw kerf widths down to 7 nm (Fig. 1e). Furthermore, the method accommodates geometries rarely attainable via conventional wafering, including multilayer slicing, 45° tilted planes, and wavy cross-sections. In addition to fused silica, Chai et al. applied i-SOT to crystalline yttrium aluminium garnet (YAG) and MgAl₂O₄, thus yielding ultrathin wafers that were approximately 8.6–9.5-μm thick and demonstrating that the sliced YAG plates retain notable mechanical integrity. i-SOT employs a line field instead of point-by-point raster scanning, thereby offering high throughput and precision. A 30 × 10 mm transparent slice can be created directly (Fig. 1g), and the authors predict that by adopting industrial-class femtosecond sources, the theoretical slicing efficiency may reach 400 mm²/s. Most importantly, combining ultrathin wafering with nanometric kerfs reduces mass loss from approximately 30% to below 1%, thus bringing laser wafering close to an almost-ideal, “kerf-free” manufacturing process¹³.

Perhaps the most profound impact of this study is conceptual. Over the past two decades, researchers have progressively leveraged principles from super-resolution optical imaging, nonlinear thresholding, and near-field confinement to enhance fabrication resolution. Chai et al. provided a different perspective: the medium can be actively engineered to serve as an interferometric contributor to the manufacturing process. Once scattering is seeded, directed, and cascaded, this process ceases to exert a parasitic effect and becomes a tool for energy localisation. This reversal is elegant yet practical and represents a family of laser-manufacturing strategies where sub-diffraction precision, high throughput, and low material loss are pursued simultaneously instead of through mutual compromise. For transparent solids, light scattering may have just changed from a nuisance into the blade itself ^3,5,9,13.