Diamond Wire Saw Slicing of SiC Boules for Power Electronics Substrate Production

Silicon carbide has become the material of choice for power semiconductor devices — MOSFETs, Schottky diodes, and Schottky barrier diodes for electric vehicle inverters, solar inverters, and industrial power conversion. The properties that make it attractive — wide bandgap, high breakdown voltage, thermal conductivity three times that of silicon — are properties of the crystal itself, and that crystal is expensive to grow.

A 150mm SiC boule grown by physical vapour transport takes weeks to produce and costs substantially more per unit volume than a comparable silicon ingot. The slicing operation that converts that boule into substrates is therefore not just a process step — it is a materials accounting problem. Every millimetre of kerf is crystal that was paid for and then discarded. The number of usable substrates per boule is a direct function of kerf width and slice thickness, and the economics of SiC substrate production are sensitive to both.

This sets the context for equipment and method selection in SiC slicing. It is not primarily a question of which method produces the cleanest cut — several methods can do that. It is a question of which method produces a clean enough cut at the narrowest practical kerf, with sufficient process stability to hold that performance across a full production batch.

SiC presents a specific combination of material properties that makes it more demanding to slice than silicon or sapphire. Understanding these properties is necessary to understand why wire parameter selection and wear management are the central technical challenges in SiC slicing — not just in theory, but in production.

At Mohs 9.5, SiC is among the hardest materials that wire saw cutting is used for commercially. Diamond wire cuts SiC by abrasion — the diamond particles on the wire surface remove material from the boule. But SiC is also abrasive to the wire itself. The electroplated diamond on the wire wears during cutting, and the rate of wear is substantially higher than when cutting silicon or sapphire. A wire that has worn significantly cuts differently from a fresh wire — higher cutting forces, different kerf geometry, and reduced surface quality on the substrate faces. Managing wire wear across a production batch is the central process control challenge in SiC slicing.

On a 150mm SiC boule that yields perhaps 30–50 substrates depending on thickness target, the difference between a 0.35mm and a 0.55mm kerf across the full length of the boule amounts to several additional substrates — each worth several hundred to several thousand dollars at current SiC substrate pricing. This makes kerf width not a secondary specification but a primary economic parameter. It also creates a tension with wire wear: worn wire tends to produce wider kerf. Balancing wire selection, tension, and feed rate to maintain narrow kerf across the batch while managing wear rate is the central optimisation problem.

SiC's hardness and brittleness mean that any instability in the cutting process — wire vibration, tension fluctuation, feed rate variation — produces immediate effects on the cut face geometry. On silicon, which is softer, the process has more tolerance for minor parameter variation. On SiC, minor parameter variation shows directly in TTV. Stable cutting conditions across the full slice are required, and wire wear monitoring is part of achieving that stability.

This project involved production slicing of 4H-SiC boules for power electronics substrate use. The boule diameter and target substrate thickness were in the range typical of commercial power device substrate production.

Wire selection for SiC is not the same as for silicon. Diamond particle size, electroplating density, and wire core specification are all variables that affect the balance between cutting rate, surface quality, and wire life on SiC. The wire specification used was established through qualification cuts at the start of the programme — evaluating substrate surface quality, kerf width, and wire life across a defined number of cuts before committing to the production parameters.

Feed rate was set conservatively relative to what the wire could theoretically handle at the start of a fresh wire pass — lower feed rate produces better surface quality and longer wire life at the cost of cycle time. For SiC, where material cost per substrate is high, that trade-off consistently favours surface quality and wire life over cutting speed.

Wire wear was monitored through the production run by tracking the cutting force data — a worn wire requires higher feed force to maintain the same feed rate, and the force trend across the batch gives a reliable early indicator of when wire performance is degrading before the change becomes visible in substrate quality. Wire was changed based on the force trend, not on visual inspection or fixed cut count.

Kerf measurements were taken periodically through the batch. The kerf width remained within a defined range across the production run, with no systematic widening trend that would have indicated accelerating wire wear.

The SiC slicing programme ran to completion with the following outcomes against the key production parameters:

Kerf width stayed within the defined range across the full batch. The substrate count per boule was consistent with what the kerf target predicted — the economics modelled at programme start were realised in production.

TTV across the substrate batch was within specification. The force-based wire change protocol prevented the degraded-wire TTV problems that can occur when wire changes are made on a fixed schedule rather than on a performance basis.

Subsurface damage depth was within the range expected for the wire specification and parameters used — consistent with the downstream lapping and polishing material removal budget set for this substrate type.

One observation worth making explicit: SiC slicing is not a set-and-forget process. The wire wear behaviour on SiC is different enough from other materials that production parameters developed for silicon or sapphire cannot be transferred directly. The qualification phase at programme start — establishing wire specification, feed parameters, and wire change criteria — is not a one-time overhead. For a new material grade, boule diameter, or target substrate thickness, it has to be repeated. That is the reality of SiC slicing at production scale.

Production parameters, boule source, and customer details are treated as confidential. What this article has described is the technical approach and the process control considerations specific to SiC at production scale — material properties that are publicly documented, and process logic that follows from those properties.

If you are operating a SiC substrate production programme — or evaluating wire saw cutting as an alternative to your current slicing method — the questions that matter are wire specification, kerf target, TTV requirement, and batch size. Dinosaw Machinery works with these parameters directly. Bring your production requirements and we will give you a direct technical response.

Contact us to discuss your SiC slicing scope.