Two ways to grow it. One way to finish it.
Diamond is thermodynamically stable only under pressure, yet it can be persuaded to grow at a few kilopascals in a plasma. We use both routes, because they make different things well, and we finish the results on a single fabrication line.
High-pressure high-temperature synthesis.
- Pressure
- 5 – 6 GPa
- Temperature
- 1 300 – 1 600 °C
- Method
- temperature gradient
- Duration
- days to weeks per run
- Output
- single crystals to 15 × 15 mm plates
In a multi-anvil press a cell containing a carbon source, a metal solvent and a diamond seed is taken to the pressures and temperatures at which diamond, not graphite, is the stable form of carbon: several gigapascals and well above 1 300 °C. A deliberate temperature difference across the cell dissolves carbon at the hot end and deposits it onto the seed at the cold end, one atomic layer at a time, for days or weeks.
The crystal that emerges is a cuboctahedron with large {100} and {111} faces. Managed well, the method yields thick, low-dislocation single crystals with excellent optical homogeneity, which is why it is the route for anvils, optical windows, substrates and quantum-grade plates. Managed badly, it yields nitrogen, inclusions and strain. The difference lies in the purity of the feedstock, the stability of the gradient over hundreds of hours, and the getters that scavenge nitrogen from the cell when type IIa material is the target.
Microwave-plasma chemical vapour deposition.
- Excitation
- 2.45 GHz microwave plasma
- Gas
- H₂ / CH₄, dopants on request
- Substrate
- HPHT seed · Si · Mo
- Wafer
- up to Ø 50 mm
- Output
- SC layers · poly 30 nm → 1 mm
In a CVD reactor a hydrogen-methane mixture at a fraction of atmospheric pressure is excited into a plasma by 2.45 GHz microwaves. Atomic hydrogen keeps the growing surface clean of graphite while carbon radicals add to the lattice. Growth rates are measured in micrometres per hour, which is slow, but the area is limited only by the plasma and the substrate.
On an HPHT seed the layer continues the seed's lattice and the product is single-crystal diamond with doping controlled to parts per billion. On silicon or molybdenum the layer nucleates in many places and grows as a polycrystalline wafer up to 50 mm in diameter, from 30 nm films on a frame to free-standing plates a millimetre thick. Polycrystalline CVD is the material of large windows, heat spreaders, beam monitors and membranes.
Not rivals. Different tools.
Almost every single-crystal CVD process begins on an HPHT seed, and almost every large diamond window is CVD. The question is never which technology is better, only which one makes the part you need at the quality you need and the price you can justify. We make both, so the recommendation is not a sales argument.
| Property | HPHT single crystal | CVD single crystal | CVD polycrystalline |
|---|---|---|---|
| Maximum format | 15 × 15 mm | ≈ 10 × 10 mm | Ø 50 mm |
| Thickness | 0.25 – 2 mm | 0.05 – 1 mm | 30 nm – 1 mm |
| Dislocation density | lowest | low | grain boundaries |
| Optical homogeneity | excellent | very good | scatter at grain boundaries |
| Doping control | B to 10 ppm | ppb to ppm | ppm |
| Thermal conductivity | > 2 000 W/m·K | ≈ 2 000 W/m·K | 700 – 1 900 W/m·K by grade |
| Best for | anvils, optics, substrates | electronics, NV layers | windows, spreaders, membranes |
Five stations between the crystal and the box.
Orientation
Each crystal or wafer is mounted and oriented by X-ray Laue diffraction. The plane and miscut you specified are set here, measured again after cutting and written to the passport. For anvils and seeds this step decides whether the part works at all.
Water-jet-guided laser cutting
A laser beam is guided inside a hair-thin water jet, which cools the cut and removes debris. There is no heat-affected zone, no graphitisation and no taper, so plates, discs, anvil blanks, apertures and multi-sector outlines come off the crystal with minimal loss.
Lapping and polishing
Parts are lapped flat on both faces and then polished on cast-iron laps charged with diamond. Single-crystal faces reach Ra ≤ 5 nm routinely and ≤ 2 nm when specified; polycrystalline faces reach ≤ 10 nm. Wedges, bevels and culets are generated at this station.
Metallisation and joining
Sputtered Ti/Pt/Au, Ti/Ni/Au and Cr/Au stacks give solderable, bondable and electrically defined surfaces. Lithography patterns contacts and isolates faces. Vacuum brazing and laser welding join diamond to metal carriers and flanges, followed by nickel or gold plating where required.
Inspection and passport
Every finished part is measured for dimensions, flatness, total thickness variation, roughness and orientation, and for optical, electrical or vacuum figures where the application calls for them. The results go on the passport, the passport goes in the box.
Why the numbers are worth the price.
| Property | Value | Unit |
|---|---|---|
| Thermal conductivity, type IIa | 2 000 – 2 200 | W/m·K |
| Thermal expansion coefficient | ≈ 1.0 | 10⁻⁶/K |
| Band gap | 5.47 | eV |
| Electrical resistivity, undoped | > 10¹³ | Ω·cm |
| Dielectric constant | 5.7 | — |
| Breakdown field | > 10 | MV/cm |
| Refractive index at 633 nm | 2.42 | — |
| Transmission | 225 nm – 2.5 µm · > 6 µm | — |
| Knoop hardness | ≈ 100 | GPa |
| Young's modulus | ≈ 1 100 | GPa |
| Density | 3.52 | g/cm³ |
| Raman line | 1 332 | cm⁻¹ |
What we measure, and on what.
Inspection runs between stations as well as at the end. A crystal with inclusions never reaches the saw, and a plate with the wrong miscut never reaches the polisher.
| Method | Determines | Applied to |
|---|---|---|
| X-ray Laue diffraction | orientation and miscut | every plate, anvil and seed |
| FTIR spectroscopy | substitutional nitrogen, type | all single crystal |
| UV-Vis-NIR spectroscopy | absorption edge, transmission | optics, quantum grade |
| Cross-polarised imaging | strain, birefringence | anvils, optics, IIa plates |
| Photoluminescence | colour centres, NV content | quantum grade |
| X-ray rocking curve | crystalline perfection | substrates on request |
| White-light interferometry | flatness, TTV, parallelism, wedge | every polished part |
| Atomic force microscopy | roughness Ra, Rq | polished faces |
| Optical microscopy, DIC | inclusions, pits, chips, scratches | every part |
| Four-point probe | resistivity | doped and metallised parts |
| Helium leak test | leak rate | brazed assemblies |
| Thermal conductivity, laser flash | κ by grade | CVD wafers, per lot |
Bring us the hard specification.
The parts other suppliers describe as possible in principle are the ones we are most useful on. Tell us which constraint is binding.