Skip to content

Simulation of the ductile machining mode of silicon

MetadataDetails
Publication Date2021-05-13
JournalThe International Journal of Advanced Manufacturing Technology
AuthorsHagen Klippel, Stefan SĂŒssmaier, Matthias Röthlin, Mohamadreza Afrasiabi, Uygar Pala
InstitutionsETH Zurich, Federal Office of Meteorology and Climatology MeteoSwiss
Citations11
AnalysisFull AI Review Included
  • Core Objective: Developed a GPU-accelerated Smoothed Particle Hydrodynamics (SPH) numerical model to simulate single-grain cutting of silicon, focusing on the transition between ductile and brittle material removal modes relevant to Diamond Wire Sawing (DWS).
  • Model Limitations: The Johnson-Cook (JC) flow stress model, using parameters derived from literature, showed limited validity in accurately predicting process forces and residual depths near the ductile-to-brittle transition zone.
  • Force Prediction Discrepancy: Simulated cutting forces (Fc) were acceptably close to experimental values, but normal forces (FN) were consistently over-estimated, suggesting the model needs extensions (e.g., fracture criteria) to account for low-energy brittle fracture.
  • Phase Transformation Evidence: Simulations predicted extremely high local hydrostatic pressures (up to 60 GPa) at larger cut depths (1000 nm and 1500 nm), supporting the assumption that high-pressure phase transformations occur in the silicon workpiece.
  • Ductile Regime Extension: Experimental scratch tests using a real, non-idealized diamond grain geometry demonstrated predominantly ductile material removal at residual depths (hres) up to 1.48 ”m, significantly exceeding the typically cited critical depth (hcu,crit ~ 60 nm).
  • Future Model Improvement: Accurate prediction of the ductile-to-brittle transition requires incorporating anisotropic material behavior, phase transformation effects, and a robust fracture criterion into the constitutive model.
ParameterValueUnitContext
Workpiece MaterialMonocrystalline Silicon-(100) direction assumed isotropic in simulation.
Numerical MethodSPH (Smoothed Particle Hydrodynamics)-GPU-accelerated (NVIDIA Quadro GP100).
Cutting Speed (vc)10m/sExperimental and simulation parameter.
Simulated Cut Depths (hcu)200, 500, 1000, 1500nmRange investigated in simulations.
Lowest Experimental Residual Depth (Ductile)0.55”mScratch #1 (Ductile mode).
Highest Experimental Residual Depth (Ductile)1.48”mScratch #6 (Ductile mode).
Max Simulated Hydrostatic Pressure>60GPaObserved at hcu = 1500 nm, suggesting phase change.
Max Simulated Equivalent Stress (von Mises)10-11GPaObserved at hcu = 500 nm to 1500 nm.
Max Simulated Temperature (Tmax)1700KObserved at hcu = 1000 nm and 1500 nm (near melting point).
Silicon Young’s Modulus (E)129.09e9 - 0.01413e9 T exp(-709/T)PaTemperature dependent model used.
Silicon Thermal Conductivity (λ)-32.98 ln(T + 273K) + 251.62W/mKTemperature dependent model used.
Friction Coefficient (”)0.3-Coulomb friction assumed.
Taylor-Quinney Coefficient (ηTP)0.9-Fraction of plastic work converted to heat.
  1. Grain Geometry Measurement: A single diamond grain from a commercial wire was isolated. Its 3D geometry was captured using an Alicona optical microscope, generating a point cloud, which was converted into a 3D tetrahedron mesh (4426 elements) for use as the rigid cutting tool in simulations.
  2. Experimental Scratch Setup: Scratch tests were performed on a polished silicon wafer using a Fehlmann Picomax Versa 825 5-axis milling machine. The wafer was rotated by the spindle, while the stationary diamond grain was mounted on a Kistler 9256C three-component force dynamometer.
  3. Kinematics and Speed: The cutting speed (vc) was fixed at 10 m/s. The tool path was spiral, achieved by simultaneous vertical and horizontal feed, resulting in an increasing depth of cut over 20 spindle rotations.
  4. Data Acquisition: Process forces (Fc and FN) were measured by the dynamometer (sampling rate 211.64 kS/s). Flash temperatures were monitored using a two-color fiber optic pyrometer.
  5. Residual Depth Determination: The residual scratch depth (hres) was determined optically by analyzing the median profile height of the scratch topography relative to the unscratched surface.
  6. Numerical Modeling (SPH): Simulations were executed using the GPU-accelerated SPH code mfree_iwf. The silicon workpiece was modeled using the Johnson-Cook flow stress model, incorporating temperature-dependent material properties.
  7. Thermal and Frictional Modeling: Heat conduction was included using the Particle Strength Exchange (PSE) approximation. Coulomb friction (”=0.3) was applied, and 90% of plastic work was assumed to convert to heat (Taylor-Quinney coefficient ηTP=0.9).
  • Solar Wafer Production: Directly supports the optimization of Diamond Wire Sawing (DWS) processes for silicon ingots. Improved understanding of the ductile-to-brittle transition allows manufacturers to select optimal cutting parameters (feed rate, wire speed) to minimize subsurface damage and reduce subsequent etching costs.
  • Ultra-Precision Machining: Relevant to the micro-machining and grinding of other hard, brittle electronic and optical materials (e.g., SiC, GaAs, sapphire) where surface integrity is paramount for device performance.
  • Tool Design and Engineering: The methodology, which uses real, non-idealized grain geometries, provides a pathway for tool manufacturers to optimize the shape and distribution of diamond grains on cutting wires and grinding wheels to maximize ductile material removal.
  • Constitutive Model Development: The experimental and simulation data serve as a benchmark for developing advanced constitutive models that accurately capture complex phenomena in silicon cutting, including high-pressure phase transformations and anisotropic fracture mechanics.