{"id":2217,"date":"2026-09-04T06:37:14","date_gmt":"2026-09-04T06:37:14","guid":{"rendered":"https:\/\/ewirexon.com\/sic-wafer-metrology-ttv-bow-warp-diamond-wire-saw\/"},"modified":"2026-09-11T08:25:39","modified_gmt":"2026-09-11T08:25:39","slug":"sic-wafer-metrology-ttv-bow-warp-diamond-wire-saw","status":"publish","type":"post","link":"https:\/\/ewirexon.com\/zh\/sic-wafer-metrology-ttv-bow-warp-diamond-wire-saw\/","title":{"rendered":"SiC Wafer Metrology After Diamond Wire Saw Cutting: TTV, Bow, Warp and Damage"},"content":{"rendered":"\n\n\n\n

For a SiC wafer supplier, the cutting trial is not finished when the wire exits the boule. The useful result is a measurement record that explains whether the wafer can survive cleaning, grinding, polishing, epitaxy and device fabrication. Total thickness variation (TTV), flatness, bow, warp, roughness, edge chips and subsurface damage describe different failure modes. Treating them as one generic \u201csurface quality\u201d number hides risk.<\/p>\n\n

This matters as the industry scales 200 mm SiC and targets higher-voltage power conversion. Bosch says its Roseville site will produce and test SiC on 200 mm wafers from 2026, while Infineon has reported 200 mm product releases. At the system level, STMicroelectronics and Infineon are describing SiC in 800 VDC and high-density AI data-centre power paths. Bigger substrates and tighter reliability expectations make position-resolved metrology a manufacturing requirement.<\/p>\n\n

A diamond wire saw creates the first finished geometry, but it is only one stage in a chain. A stable measurement system lets process engineers separate incoming crystal defects from wire dynamics, fixture movement, handling damage and downstream removal. It also gives procurement teams a common basis for comparing an ewirexon endless diamond wire saw<\/em>, a production web and alternative slicing methods.<\/p>\n\n\n\n\n

\"An
Post-saw metrology must connect wafer geometry to the cutting history, rather than reduce quality to one surface number.<\/figcaption><\/figure>\n\n

Start with a clear measurement boundary<\/h2>\n\n

Write the condition of the wafer beside every result. A bow value measured on a warm wafer immediately after coolant exposure is not directly comparable with a value measured after cleaning and relaxation. Define the cleaning chemistry, drying method, temperature, waiting time, clamping state and reference plane. Use the same boundary for all suppliers and all wire conditions.<\/p>\n\n

Separate three checkpoints:<\/p>\n\n

    \n
  1. Incoming boule and blank:<\/strong> diameter, orientation, end crop, visible defects, runout and fixture datum.<\/li>\n
  2. As-sawn wafer:<\/strong> thickness, TTV, flatness, bow, warp, edge, roughness and damage before material is removed.<\/li>\n
  3. Post-finish wafer:<\/strong> the same geometry and surface metrics after the defined grinding, lapping or polishing step.<\/li>\n<\/ol>\n\n

    The delta between checkpoints is often more valuable than an isolated number. If bow falls sharply after a thin grinding pass, residual stress or a damaged layer may be driving the as-sawn shape. If a position-dependent TTV pattern remains after finishing, the root cause may be wire path, guide alignment or fixture support rather than polishing.<\/p>\n\n\n

    \"Engineering
    TTV, bow and warp describe different geometric mechanisms and should not be used interchangeably in a supplier acceptance plan.<\/figcaption><\/figure>\n\n

    Use different metrics for different mechanisms<\/h2>\n\n

    Thickness and TTV<\/h3>\n\n

    Thickness is a local dimension; TTV is the range between the thickest and thinnest measured points on the specified map. A single centre reading can miss a ring or edge pattern. Use a repeatable multi-point map that captures centre, intermediate radii and edge exclusion. For 200 mm material, record the coordinate system and keep the same clocking mark through the process.<\/p>\n\n

    When a diamond wire saw produces a periodic thickness signature, compare its phase with guide spacing, wire oscillation and feed transitions. Do not average away a pattern that could challenge downstream chucking or epitaxy. A low lot average can coexist with a high maximum TTV at a specific web position.<\/p>\n\n

    Flatness, bow and warp<\/h3>\n\n

    Flatness compares the wafer surface with a specified reference plane or best-fit plane. Bow describes the centre deviation of the median surface; warp describes the total range of that median surface. The definitions and reference conventions must be fixed before the trial. A clamped measurement can suppress the very deformation that a handler will see.<\/p>\n\n

    Record whether front and back surfaces are both included and whether edge exclusion is applied. For thin wafers, thermal gradients, residual stress, adhesive thickness and handling can change the result within minutes. Repeat measurements after a controlled relaxation period and retain the time stamp.<\/p>\n\n

    Roughness and waviness<\/h3>\n\n

    Roughness captures shorter-scale texture; waviness captures longer-scale form. Report the instrument, cutoff, scan direction, sampling length and filtering. A wire mark may look acceptable in a photograph while a directional scan reveals a process signature. Compare at least two orthogonal directions and sample more than one wafer position.<\/p>\n\n

    Edge chips and edge exclusion<\/h3>\n\n

    Edge damage can reduce the usable diameter and can initiate fracture during cleaning, grinding or robot handling. Define maximum chip size, affected arc length, corner geometry and the edge-exclusion rule. Photograph representative defects with scale and record whether they originate at entry, breakthrough, fixture contact or manual handling.<\/p>\n\n

    Make subsurface damage visible<\/h2>\n\n

    Surface appearance does not prove that the underlying SiC is sound. Brittle fracture can leave cracks and a damaged layer below the as-sawn surface. The 2026 review of ultra-fine diamond-wire slicing identifies saw marks, wafer warp, breakage and wire wear as connected industrial challenges. Earlier fixed-abrasive SiC work also explored scanning acoustic microscopy for nondestructive evaluation of subsurface damage.<\/p>\n\n

    Choose a method that matches the decision. Cross-section polishing and microscopy can measure local crack depth but are destructive. Confocal or white-light interferometry describes surface morphology, not the full subsurface. Acoustic, optical or X-ray methods may provide screening, but each requires calibration on representative SiC. Report detection limits, sampling plan and uncertainty, not just a pass\/fail label.<\/p>\n\n

    Link the measured damage layer to the finishing allowance. A supplier should explain how much material must be removed to reach the required surface and shape, and what evidence supports that allowance. If the allowance is a blanket number imported from silicon, the cost and yield model is incomplete.<\/p>\n\n\n\n\n

    \"Center,
    A spatial sampling map and preserved wire-web position make systematic process drift easier to identify.<\/figcaption><\/figure>\n\n

    Design a sampling map that finds process drift<\/h2>\n\n

    A useful map is position-aware and time-aware. For an endless cut, record entry, middle and breakthrough. For a multi-wire web, retain each wafer\u2019s wire position and order. Add samples at the centre and edge of the boule, because maximum contact length and coolant evacuation can create radial differences.<\/p>\n\n

    For each sample, retain:<\/p>\n\n