GradeZ
Zero-MPD
For large-area power devices and anything where a micropipe kills the die. The grade you need for 1200 V+ MOSFET production.
- MPD
- ≤0.2 cm⁻²
- TTV / Bow
- ≤2.5 / ≤25 µm
- Warp
- ≤35 µm
- Price index
- highest
03Semiconductor substrates
Wafers bought against
a number, not a
grade label.
Silicon, 4H-SiC and sapphire from polishing houses we have walked through. Flatness, resistivity and particle count arrive as a measured certificate for the lot — and the grade definition goes into the purchase order, not into a footnote.
01Silicon
Prime silicon, by the numbers
The most counterfeited category on this page — not because silicon is hard to make, but because "prime" is a word, not a measurement. Below is the envelope we buy to, and the values we put on the certificate.
Table 1 · Prime silicon wafer envelope
| Parameter | 100 mm | 150 mm | 200 mm | 300 mm | Measured by |
|---|---|---|---|---|---|
| Thickness µm | 525 ±20 | 675 ±20 | 725 ±20 | 775 ±20 | Capacitive / optical gauge |
| TTV µm, max | ≤2 | ≤2 | ≤2 | ≤3 | Full-surface scan |
| Bow µm, max | ≤20 | ≤20 | ≤20 | ≤30 | Unclamped, 3-point |
| Warp µm, max | ≤25 | ≤25 | ≤25 | ≤40 | Unclamped, full surface |
| Orientation | 〈100〉 〈111〉 〈110〉, ±0.5° · off-cut to drawing | XRD | |||
| Resistivity Ω·cm | 0.001–10,000 depending on dopant · P (B) or N (P, As, Sb) | 4-point probe, centre + edge | |||
| Front roughness | Ra ≤0.5 nm · AFM RMS <0.2 nm on 5×5 µm | AFM, site map | |||
| Particles count/wafer | ≤10 at ≥0.2 µm (200 mm) · ≤20 at ≥0.2 µm (300 mm) | Laser scatter, both sides | |||
| Edge | Flat ≤150 mm · notch ≥200 mm · bevel T or R | Profile projector | |||
| Backside | Polished, etched or as-lapped · laser mark per SEMI M12/M13 | Visual + OCD | |||
Scroll the table sideways for all columns →
Growth matters more than diameter. CZ covers most of the market; MCZ is what you ask for when oxygen precipitation is eating your yield; FZ is the only way to reach the high resistivity needed for RF and detector work. We quote the growth method explicitly, because it is the first line most traders drop.
02Silicon carbide
4H-SiC, and what "grade" buys you
Micropipe density is the price. A zero-MPD substrate can cost several times a production-grade one of the same diameter, so the grade has to be chosen against the device — not against the budget alone.
GradeP
Production
The default for Schottky diodes and smaller devices. MPD is present but controlled and counted per wafer.
- MPD
- ≤2 cm⁻²
- TTV / Bow
- ≤5 / ≤40 µm
- Warp
- ≤60 µm
- Price index
- mid
GradeD
Dummy / test
For epitaxy trials, furnace seasoning and process development. Never for devices you intend to sell.
- MPD
- ≤15 cm⁻²
- TTV / Bow
- ≤15 / ≤40 µm
- Warp
- ≤60 µm
- Price index
- lowest
Table 2 · 150 mm 4H-SiC substrate, conductive vs semi-insulating
| Parameter | 4H-N conductive, N-doped | 4H-SI semi-insulating |
|---|---|---|
| Diameter | 149.5–150.0 mm | 149.5–150.0 mm |
| Thickness | 350 ±15 µm | 500 ±15 µm |
| Orientation | 4.0° off-axis ±0.5° | on-axis ±0.5° |
| Resistivity | 0.015–0.028 Ω·cm | ≥1×10¹⁰ Ω·cm |
| Micropipe density | ≤0.2–2 cm⁻² by grade | ≤1–15 cm⁻² by grade |
| TTV / Bow / Warp | ≤2.5 / ≤25 / ≤35 µm | ≤2.5 / ≤15 / ≤30 µm |
| Surface, CMP | Ra ≤0.2 nm | Ra ≤0.2 nm |
| Dislocations | TSD ≤500 cm⁻² · BPD ≤1,000 cm⁻² KOH-etched | reported per lot |
| Edge exclusion | 3 mm | 3 mm |
Scroll the table sideways for all columns →
Dislocation data is destructive. TSD and BPD counts come from KOH-etched witness wafers, not from the wafer you receive. If a supplier quotes a dislocation figure and cannot say which witness wafer it came from, treat it as marketing.
03Sapphire
Sapphire, C-plane and off-cut
Sold by the micron of bow as much as by the inch. For LED and micro-LED work the number that decides your lithography yield is TTV and warp after the epi step, so we buy the flatness, not just the polish.
Table 3 · C-plane sapphire substrate, epi-ready
| Parameter | 2 in | 4 in | 6 in |
|---|---|---|---|
| Diameter | 50.8 ±0.1 mm | 100 ±0.1 mm | 150 ±0.2 mm |
| Thickness | 430 ±15 µm | 650 ±25 µm | 1,000–1,300 ±25 µm |
| Orientation | c-plane (0001), 0.2° off toward M ±0.1° · A-plane, R-plane, M-plane on request | ||
| Primary flat | 16 ±1 mm | 30 ±1 mm | 47.5 ±1 mm or notch |
| Front finish | Epi-polished, Ra ≤0.2 nm · DSP available | ||
| Back finish | Lapped or polished, Ra 0.6–1.2 µm | ||
| TTV | ≤5 µm | ≤8 µm | ≤10 µm |
| Bow | −8 to 0 µm | −10 to 0 µm | −15 to 0 µm |
| Warp | ≤10 µm | ≤15 µm | ≤20 µm |
| Purity | Al₂O₃ ≥99.99% · Kyropoulos or HEM growth | ||
Scroll the table sideways for all columns →
Negative bow is not a defect. A slight concave bow (0 to −10 µm) is normal for SSP sapphire and helps contact during epi. A positive bow of the same magnitude is a different conversation — specify the sign, not just the magnitude.
Also quoted on request
- GaAs
- Semi-insulating and N-type, 2–6 in, 〈100〉 off-cut 2° or 15°
- InP
- Fe-doped SI and S-doped N, 2–4 in, epi-ready
- Ge
- 〈100〉 and 〈111〉, optical and detector grade
- GaN
- GaN-on-Si and GaN-on-SiC epi wafers, 100–200 mm
- LiNbO₃ / LiTaO₃
- SAW and optical grade wafers, 3–6 in, black-LN on request
- Reclaim
- Prime-geometry reclaim with a stated thermal history and re-polish spec
04Qualification
What we check before a lot ships
Wafers are the easiest category to fake on paper and the hardest to argue about afterwards. So the checks are about method as much as result — the same number measured two ways is two different promises.
01
Grade is a document
A certificate of conformance with the measured values for that lot, signed, in English, with the lot number matching the box. No CoC, no order.
02
Flatness, measured where
TTV quoted over 5×5 mm sites and over 10×10 mm sites are not the same number. We insist the site size and edge exclusion are printed next to the value.
03
Particle size threshold
"≤10 particles" means nothing without the threshold. At ≥0.2 µm and at ≥0.3 µm the counts differ by an order of magnitude on the same wafer.
04
Resistivity across the radius
A single centre-point reading hides the edge. We ask for a five-point map, and for the dopant species to be named rather than "N-type".
05
Epi-ready vs polish-only
Epi-ready implies a specific final clean and a measured oxide. If the price is 20% below everyone else, this is usually the line that was quietly removed.
06
Packaging for the class
Single-wafer container for 150 mm and below, FOUP or double-stack box for 200 and 300 mm, vacuum bag, class 100. We open one box at PSI and photograph the seal.
05Where it goes wrong
Three failures we see most often
Trap 01
Test grade sold as prime
Same diameter, same polish, different flatness and a lower particle spec. The only way to catch it is to compare the certificate against the grade definition — which is exactly the page most buyers never ask for.
Trap 02
Roughness from one 5×5 µm scan
"Ra <0.2 nm" measured at the centre of one wafer tells you almost nothing about the lot. We ask for a five-site map and the scan area before we accept the claim.
Trap 03
Reclaim with an unknown history
Reclaimed wafers are legitimate and useful — but thermal history and the number of re-polish cycles decide whether your process is repeatable. Undisclosed reclaim is the most expensive saving on this page.
Other categories
What travels with the wafer
05 · Handling
Wafer handling consumables
Cassettes, FOUP-compatible boxes and baskets that protect the surface you just paid for.
View category 02 · CrystalsPiezoelectric & scintillation crystals
LiNbO₃ and LiTaO₃ wafers, plus scintillators with measured light yield.
View category All fiveOptics, ceramics & the rest
Precision optics and quartz glass, plus machined technical ceramics.
Back to overviewSend the spec you were quoted against.
We will tell you which grade it actually describes, where the wording is ambiguous, and what the same lot costs from a polishing house we have audited.