How a diamond is grown.

There are two industrial ways to make gem diamond, and they could hardly be less alike. One dissolves carbon in molten metal and squeezes it inside a press the size of a room until diamond is the form carbon prefers. The other holds a wafer-thin seed at a fraction of atmospheric pressure and rains carbon onto it, atom by atom, out of a glowing ball of plasma. The press method is called HPHT — high pressure, high temperature. The plasma method is called CVD — chemical vapour deposition.

Both produce diamond: the same carbon lattice, the same hardness, the same optics as mined stone. Both are graded for the same properties and both are disclosed as laboratory-grown; the report formats differ by laboratory, and grading conventions for grown material have been changing. What differs is the machine, the chemistry, the growth habit of the crystal — and the faint fingerprints each process leaves, which is how laboratories tell them apart.

Growing from a gas — the CVD route

A CVD reactor is a floor-standing cabinet roughly the size of a large wardrobe — about 1.85 m tall — but the chamber inside it is surprisingly small: a stainless drum only 140–200 mm across, pumped down to a fraction of atmospheric pressure — growth runs at 10–200 Torr, between about one and a half percent and one quarter of an atmosphere. Inside that drum sits a tray of diamond seed plates. Microwaves ignite the gas above them into plasma, and over the following weeks each plate slowly thickens into a block of single-crystal diamond. Explore the machine:

The feed gas is mostly hydrogen with a few percent methane. In the plasma, hydrogen molecules crack into atomic hydrogen — and that is the ingredient that makes the whole process work. Atomic hydrogen etches away graphite-like carbon far faster than diamond-bonded carbon, and it caps the growing surface, opening and closing attachment sites. Methane fragments — chiefly methyl radicals — land on those sites and bond. Carbon that lands in the wrong, graphitic arrangement is preferentially etched back off by the hydrogen before it can be buried — the selectivity is strong, but not perfect. What survives, layer by stepped layer at a few microns an hour, is diamond.

Growth runs at roughly 700–1200 °C at the seed surface, for two to six weeks per cycle — about a month in the reactor yields four to six carats of cuttable rough. Small amounts of nitrogen in the gas speed growth but tint the stone; many as-grown CVD stones carry a brownish cast traced to vacancy clusters and related point defects, which growers reduce or remove with a post-growth anneal — HPHT annealing most often, low-pressure high-temperature (LPHT) treatment in some plants. Either way the result is a treated stone, and is disclosed as such. The cleanest material is grown slowly, needs no treatment, and commands a premium for it.

Growing under pressure — the HPHT route

At ordinary conditions graphite, not diamond, is carbon's stable form — every diamond at the surface of the Earth is metastable, kept from reverting only by the enormous activation barrier between the two structures. An HPHT press recreates the pressures of the cratonic mantle, where that preference reverses: five to six gigapascals — fifty to sixty thousand times atmospheric pressure — at around 1350–1600 °C. The two are coupled, not independent: the hotter the cell runs, the more pressure it takes to stay on the diamond side of the boundary, from roughly 5.1 GPa at 1350 °C to about 5.7 GPa at 1600 °C. The dominant machine is the six-anvil cubic press:

Pressure alone is not enough — solid carbon rearranges far too slowly. The trick is a molten metal flux: an iron-, nickel- or cobalt-based alloy — Fe–Ni and Ni–Mn–Co are the common families — sits between a graphite source and a small diamond seed. Carbon dissolves into the metal at the hot end of the cell and precipitates onto the seed at the cool end, driven by a temperature difference of a few tens of degrees. A gem cycle runs from about ten days for a one-carat stone to six weeks for premium material, and longer still — up to ten weeks — for the largest crystals. The stone grows into the cuboctahedral habit — cube and octahedron faces together — that marks HPHT growth, quite unlike the octahedra nature favours or the flat tabular blocks CVD produces.

Chemistry controls colour. Nitrogen dissolved from the cell materials enters the lattice as isolated atoms and turns the stone canary yellow; growers add nitrogen “getters” — titanium or aluminium — to trap it and grow colourless material. Add boron instead and the stone grows blue. The 10.02 ct colourless stone GIA examined in 2015 — then the largest colourless HPHT-grown diamond on record — came from Russia's New Diamond Technology, grown in modified cubic presses rather than in a split-sphere BARS.

The machines

Four machine families grow effectively all of the world's laboratory diamond. Gem growers in Surat and the United States run MPCVD reactors by the thousand; China's Henan cluster runs halls of six-anvil presses; BARS spheres grow one premium crystal at a time; belt presses — the 1954 original — still serve industrial abrasives at enormous scale.

The four diamond types

Gemmology sorts all diamond — mined or grown — into four types by its trace chemistry. The classification matters commercially because it is the first screening cue: almost all natural diamonds are Type Ia, while nearly all colourless grown stones are Type IIa.

Telling them apart

Each growth environment leaves fingerprints. HPHT stones grow in cube-and-octahedron sectors that take up impurities differently, so under deep-ultraviolet imaging they show geometric, cross-shaped fluorescence patterns; they can carry tiny trapped beads of the metal flux — occasionally enough to attract a magnet — and colourless HPHT material often phosphoresces after the lamp goes off. CVD stones grow layer by layer, and that layering shows as fine striations under the same instruments. Screening devices in the trade exploit the type divide: they clear the Type Ia majority as natural and refer everything else — Type II and Type Ib stones alike — to a laboratory. A referral is not a verdict; most referred stones prove to be natural Type IIa. It is the laboratory, with growth-structure imaging and photoluminescence spectroscopy, that settles origin. Many CVD stones also carry a silicon-vacancy signature — a doublet near 737 nm — in photoluminescence, picked up from silicon in the reactor's quartz components. It is occasionally seen in HPHT-grown and, very rarely, in natural diamond, but not at the intensities routine in CVD material.

Optically, a grown diamond is diamond: refractive index 2.417, dispersion 0.044, the same critical angle and the same Fresnel reflectance — the quantities that drive everything on the optics page — because those properties belong to the carbon lattice, not to its origin. This is why a physically-based render computed from cut geometry and material constants represents a grown stone exactly as faithfully as a mined one: the physics cannot tell the difference, and neither can the ray tracer.

Nothing here bears on value or disclosure policy. Laboratory-grown diamond must be disclosed as laboratory-grown wherever it is sold — in the United States under the FTC Guides for the Jewelry, Precious Metals and Pewter Industries, and elsewhere under national consumer-protection law together with the CIBJO nomenclature most trade bodies apply. The wording required, and who must give it, differ by market; check the rules that apply where you sell rather than relying on this page, which is not legal advice. This page takes no position on markets, pricing, or how consistently that requirement is met. The detection section describes what laboratories look for; it is not a field guide. None of those cues can be read reliably outside a laboratory, origin determination needs instruments and reference spectra a trade counter does not have, and no stone should be called grown or natural on the strength of this page. Machine figures are class-typical ranges from manufacturer specifications, not guarantees about any particular producer's equipment. Individual stones vary — inclusions, strain and treatment history are properties of a specific stone that no process description, and no render from specification, can claim to show. Figures and grading practice current as of July 2026.

Machine specifications compiled from manufacturer data and trade documentation: Seki Diamond Systems (MPCVD) · Henan press makers (Hanfa, Qiming, Huanghe Whirlwind) · GIA Gems & Gemology on BARS growth and the New Diamond Technology stones · the BARS apparatus literature (Novosibirsk Institute of Geology & Geophysics) · General Electric belt-press lineage (Tracy Hall, 1954). Full citations in the FacetCAM machines reference.

Related · Why gems sparkle · Faceting glossary