Application Spotlight: Stealth Dicing of Sapphire with Femtosecond Bessel Beam – Zero Kerf, Zero Taper, 200 nm Surface Finish

Image source: Comparison of femtosecond Bessel beam sapphire cutting quality at different scanning speeds. (a) 1 mm/s shows cracks, (b) 3 mm/s crack-free with improved quality, (c,d) 5-7 mm/s demonstrate optimal pulse separation with uniform surface finish. Adapted from Li et al., Optics & Laser Technology 2021, 135, 106713.

Industrial

Application Spotlight: Stealth Dicing of Sapphire with Femtosecond Bessel Beam – Zero Kerf, Zero Taper, 200 nm Surface Finish

February 13, 2026
Stealth Dicing
Sapphire
Bessel Beam
LED Manufacturing

Research Article

Source: Optics & Laser Technology

View original publication

Research published in Optics & Laser Technology demonstrates a breakthrough sapphire cutting method using femtosecond Bessel beam stealth dicing, achieving 200 nm Ra surface roughness, nearly an order of magnitude improvement over previous laser Bessel beam approaches, with zero kerf width, zero taper, and crack-free cutting suitable for LED substrate manufacturing.

The Sapphire Cutting Challenge

Sapphire (α-Al₂O₃) combines extraordinary properties with extreme processing difficulty:

Material Properties:

  • Hardness: Mohs 9.0 (second only to diamond at 10.0)
  • Optical transmissivity: 0.15-5.5 μm wavelength range
  • Density: 4 g/cm³
  • Melting temperature: 2050°C

LED Industry Requirements:

  • Zero debris/chipping to reduce light losses
  • Zero taper for high manufacturing yield
  • Low surface roughness (sub-micron) for 300-550 nm GVR LED efficiency
  • High sidewall uniformity

Existing Methods & Limitations

Diamond Saw Cutting:

  • Surface roughness: ~500 nm Ra
  • Kerf width: 100-200 μm (high material waste)
  • Generates debris, delamination, chipping
  • Low efficiency, large tool wear

Laser Fusion Cutting (ns fiber/DPSS):

  • Surface roughness: ~1000 nm Ra
  • Kerf width: 20-80 μm
  • Kerf taper: ~1-2°
  • Large heat-affected zones → micro-cracks up to 10 μm
  • Striations, burr, splatters unavoidable

Laser Full Ablation Cutting (ps/fs):

  • Surface roughness: ~400 nm Ra
  • Kerf width: 20-50 μm
  • ~50% slower than fusion cutting
  • Generates dust, chipping occurs
  • Tapering difficult to eliminate

Laser Scribing & Cleaving (ns/fs):

  • Surface roughness: <50 nm Ra (non-uniform: only cleaved area)
  • Kerf: ~45 μm on top surface
  • Generates debris and dust
  • Crack formation tendency
  • Cutting path deviation, breakage plane not perpendicular

Traditional Stealth Dicing (ns/ps/fs focused beam):

  • Surface roughness: ~50 nm Ra (highly non-uniform)
  • Kerf width: ~0 μm
  • No debris, no chipping, no taper
  • Limitation: Illuminated areas damaged, cleaved areas smooth, poor uniformity

Previous Bessel Beam Stealth Dicing:

  • Surface roughness: ~2000 nm Ra (non-uniform, dash-line technique used)
  • Kerf width: ~0 μm
  • Rough and non-uniform sidewalls despite crack orientation control

Breakthrough: Optimized Femtosecond Bessel Beam Approach

This work achieves nearly order-of-magnitude improvement in surface quality through highly uniform non-diffractive Bessel beam generation and optimized pulse separation strategy.

Laser System (Ti:Sapphire):

  • Wavelength: 800 nm
  • Pulse duration: 100 fs
  • Repetition rate: 1 kHz
  • Maximum average power: 1 W
  • M² beam quality: <1.5
  • Pulse energy: 450 μJ (optimized)
  • Polarization: Circular (via quarter waveplate for omnidirectional processing)

Bessel Beam Generation Strategy:

  • Beam reducer: 0.25× magnification (5 mm → 1.25 mm diameter)
  • Axicon lens: Fused silica, 20° physical angle
  • Rationale: Direct beam reduction (not demagnification system) maximizes efficiency and purity without aberration
  • Central core diameter (D): 1.8 μm (calculated)
  • Central core length (L): 3.6 mm (calculated)

Why This Configuration Works:

  • Lower physical angles → longer central core but low pulse energy efficiency
  • Higher physical angles → shorter central core, insufficient depth of focus, poor uniformity, highly localized fluence generates cracks along crystal orientations
  • 20° angle → optimal balance for 0.38-1.5 mm sapphire thickness range

Critical Finding: Pulse Separation Controls Quality

Scanning speed directly controls pulse overlap/separation, which fundamentally determines cut quality.

Low Speed (0.1 mm/s): High Pulse Overlap

  • Large ablation area on top surface inhibits Bessel beam propagation into material
  • Cracks observed within and around processing trajectory
  • Heat accumulation + laser fluence → crack generation
  • Sidewall structures: particles (~10 μm diameter) attached to sidewall, generally rough surface
  • Re-solidification of ablated sapphire during expansion/evacuation

Medium Speed (1 mm/s): Transitional Phase

  • Bessel beam pulses gradually separate; laser-irradiated areas interrupt pulse connection
  • Sidewall structures: particles enlarge to ~30 μm diameter

Medium-High Speed (3 mm/s): Partial Separation

  • Cracks disappear as pulses separate further
  • Sidewall structures: crescent formations 40×110 μm appear
  • Long axis of crescents perpendicular to Bessel beam scanning direction

High Speed (5-9 mm/s): Optimal Separation

  • No unwanted structures observed
  • Energy from previous pulse dissipates before next pulse arrives
  • Heat accumulation reduced
  • Smooth sidewall with visible trace of each Bessel beam pulse interaction
  • Surface roughness: 200 nm Ra achieved

Laser Spot Characteristics:

  • Measured diameter: ~1.9 μm (close to theoretical 1.8 μm central core)
  • At 5-7 mm/s: slight high-order ring visible around central core (Bessel feature) but no impact on dicing quality

Material Modification Mechanism

Six-Photon Absorption Process:

  • Multi-photon absorption occurs inside sapphire (high transmissivity enables internal focusing)
  • Coulomb explosion and optical damage breaks electron bonds and crystalline structure
  • Local pressure: ~10 TPa generated within tenths of picoseconds (sapphire → plasma transformation)
  • Nano-Joule pulse energy sufficient for interior modification without damage outside focusing area

Cleavage Process:

  • Three-point bending applied post-laser processing
  • Support span: 35 mm between first and third points
  • Motor speed: 0.01 mm/s
  • Highly uniform Bessel beam + uniform material modification → zero taper cleavage
  • Unirradiated areas between pulses cleave along laser scanning contour → low sidewall roughness

Thickness-Dependent Flexural Strength Results

Mechanical separation force requirements show non-intuitive thickness dependence.

Sapphire Thickness 0.38 mm:

  • Flexural strength required: up to 229 MPa (highest)
  • Observation: Obvious elastic bending/deformation before separation
  • Explanation: Thin sheet absorbs more energy as elastic deformation before breakage
  • Optimal surface roughness: achieved at 3 mm/s scanning speed

Sapphire Thickness 1 mm:

  • Flexural strength required: up to 42 MPa
  • Optimal surface roughness: achieved at 3 mm/s scanning speed

Sapphire Thickness 1.5 mm:

  • Flexural strength required: up to 184 MPa
  • Optimal surface roughness: 200 nm Ra achieved at 7-9 mm/s scanning speed
  • Unexpected result: Easier to achieve low roughness for thick samples requiring higher breaking force

General Trend: Flexural strength gradually increases with scanning speed (non-linear) for all thicknesses.

Surface Quality Characterization

Measurement Methods:

  • White light interferometry (Contour GT-K; Veeco): 860 μm × 650 μm measurement area
  • Scanning electron microscopy (Quanta 200; FEI): 10 nm Au/Pd coating for conductivity

Crystal Orientation Independence:

  • C-M 0.2° crystal orientation sapphire tested
  • Two adjacent random cutting directions: 680 nm Ra and 610 nm Ra measured
  • Confirmed: Crystal orientation influence on surface roughness is limited

Demonstrated Applications

Production Parameters (1.5 mm sapphire):

  • Pulse energy: 450 μJ
  • Scanning speed: 4 mm/s (optimized for low roughness + low bending force)

Straight Line Cutting:

  • Zero taper achieved
  • Debris-free
  • Crack-free
  • Uniform surface finish

Curved Line Cutting:

  • Radius: 4 mm arcs demonstrated
  • Strategy: Extra contours designed connecting arc points (0°, 45°, 90°) to reduce continuous arc length
  • Result: 7 out of 10 arc points successful with straight-line prototype bending fixture
  • Improvement path: Shorter-curve-shape bending fixture or additional connecting contours

Complex Geometries:

  • Combination of straight lines and curves demonstrated
  • Square shapes cut successfully

Key Success Factors

Achieving 200 nm Ra surface roughness requires optimization of multiple interdependent parameters:

  • Bessel beam uniformity: High-purity generation via direct beam reduction through 20° axicon
  • Pulse separation: Appropriate scanning speed (5-9 mm/s for 1.5 mm thickness) to eliminate heat accumulation while maintaining modification continuity
  • Material thickness: Thicker samples (1.5 mm) paradoxically easier to achieve low roughness despite higher cleavage force
  • Cleavage of mixed zones: Both laser-irradiated and unirradiated areas cleave together along contour
  • Circular polarization: Eliminates directional distortion for omnidirectional processing

LED Manufacturing Impact

Sapphire substrate cutting quality directly impacts LED performance:

  • Light extraction efficiency: Sub-micron surface roughness critical for 300-550 nm wavelength GVR LEDs
  • Manufacturing yield: Zero taper ensures dimensional consistency
  • Reliability: Crack-free, debris-free edges eliminate failure initiation sites
  • Cost reduction: Zero kerf width maximizes substrate utilization

Comparison to Alternative Methods

This Work vs. Previous Bessel Beam Stealth Dicing:

  • Surface roughness: 200 nm vs. 2000 nm (10× improvement)
  • Uniformity: High vs. poor (dash-line technique eliminated)
  • Crack control: Speed-based prevention vs. polarization/direction control

This Work vs. Traditional Stealth Dicing:

  • Uniformity: High (entire sidewall) vs. poor (illuminated/cleaved zone mismatch)
  • Surface roughness: 200 nm uniform vs. 50 nm non-uniform

This Work vs. Ablation Methods:

  • Kerf width: 0 μm vs. 20-80 μm
  • Debris: None vs. significant
  • Taper: Zero vs. present
  • Surface quality: 200 nm vs. 400-1000 nm

Technical Challenges & Considerations

Very Thin Sapphire (<0.4 mm):

  • High flexural strength requirement (229 MPa) due to elastic deformation
  • Alternative approaches may be more suitable: Bessel beam fracture plane generation or improved bending jig design

Curved Cutting Limitations:

  • Standard straight-line bending fixtures achieve 70% success rate (7/10 arc points)
  • Solution: Custom curve-matched bending fixtures or increased connecting contour density

Process Scalability:

  • 1 kHz repetition rate suitable for precision but limits throughput
  • Higher repetition rates would require heat accumulation management

Future Optimization Opportunities

  • Higher repetition rate systems with thermal management for increased throughput
  • Adaptive bending fixtures for complex curve geometries
  • Multi-beam systems for parallel processing
  • In-line quality monitoring via interferometry or acoustic emission

This breakthrough demonstrates that careful optimization of Bessel beam generation and pulse separation strategy can achieve surface quality approaching mechanical polishing (~200 nm Ra) while maintaining the advantages of laser stealth dicing: zero kerf, zero taper, no debris, and no cracks.

Reference: Li, Z., et al. "Stealth dicing of sapphire sheets with low surface roughness, zero kerf width, debris/crack-free and zero taper using a femtosecond Bessel beam." Optics & Laser Technology 2021, 135, 106713. https://doi.org/10.1016/j.optlastec.2020.106713

Interested in This Application?

Contact us to learn how Clear Photon workstations can enable your research or production process.

Get in Touch

Related Applications