How Diamonds Are Made: Three Pathways from Carbon to Crystal

Natural diamonds form deep in the Earth's mantle at depths of roughly 150 to 200 kilometres, where pressures of 45 to 60 kilobar combine with temperatures of 900 to 1,300 degrees Celsius to drive carbon atoms into the diamond cubic phase rather than the layered graphite phase that is stable at the surface¹. The carbon source is typically reduced material in the upper mantle, including subducted carbonate sediments and primordial mantle carbon. Crystallisation occurs slowly. Most gem-quality natural diamonds reaching the market are between one and three billion years old.

Diamonds reach the surface only when a deep volcanic eruption transports them rapidly enough to prevent reversion to graphite. The transport mechanism is a kimberlite or, less commonly, lamproite eruption, in which magma originating below the lithosphere ascends through the crust at a rate of metres per second, freezing the diamonds into a host rock that solidifies at or near the surface as a kimberlite pipe¹. The pipe is the geological feature that is mined. Most rough diamonds extracted today come from a relatively small number of kimberlite pipes in Botswana, Russia, Canada, Angola, and a handful of other locations.

The natural pathway is therefore not just slow but contingent on a specific volcanic event. The carbon, the pressure, the temperature, and the eruption all have to coincide for a gem-quality natural diamond to exist at all. This is the structural reason natural diamond supply cannot expand at will and is the essential context for the price discussion in Chapter 7.

High-pressure high-temperature, or HPHT, is the older of the two laboratory methods. The first reproducible HPHT diamond synthesis was achieved by General Electric in 1954, and the method has been refined ever since. HPHT recreates conditions roughly comparable to the upper mantle inside a steel press².

A typical run uses operating conditions in the region of 1,500 degrees Celsius and 5 gigapascals. A small amount of carbon source, usually graphite, is placed in a sealed capsule together with a metal catalyst flux of iron, nickel, and cobalt and a tiny diamond seed crystal. The press applies pressure from multiple anvils. At temperature, the carbon dissolves into the molten metal flux, diffuses through the flux to the seed, and precipitates onto the seed in the diamond phase, atom by atom, over several days.

Three press geometries are in use. The cubic press uses six hydraulic anvils arranged on the faces of a cube. The belt press uses a pair of large opposing anvils with a circumferential die. The BARS press, originally a Russian design, uses a multi-anvil split-sphere arrangement. All three achieve broadly equivalent thermodynamic conditions; the main differences are run cost, scalability, and chamber volume.

Chemical vapour deposition, or CVD, takes a fundamentally different approach. Instead of squeezing carbon into the diamond phase, CVD deposits diamond atom by atom onto a flat seed plate from a low-pressure gas-phase reaction at sub-atmospheric pressure². The conditions are mild compared to HPHT: temperatures around 800 to 1,200 degrees Celsius and pressures well below atmospheric, typically a few tens of millibar.

The chamber contains a flat polished diamond seed mounted on a heated stage. A carrier gas, predominantly hydrogen, flows in along with a small percentage of methane (typically one to five per cent of total flow). A power source - most commonly a microwave plasma in modern reactors, sometimes a hot tungsten filament or DC arc - energises the gas above the seed, dissociating molecular hydrogen into atomic hydrogen and breaking methane into reactive carbon-bearing fragments⁴.

The atomic hydrogen does two things. It deposits chemisorbed hydrogen onto the growing surface, terminating dangling bonds and allowing the layer to remain in the diamond rather than the graphite configuration. And it preferentially etches non-diamond carbon faster than diamond, suppressing graphitic deposits. The result is layer-by-layer epitaxial growth of a single-crystal diamond on top of the seed at a rate of micrometres per hour. Growth runs of one to several weeks produce crystals of a few millimetres in thickness.

Modern microwave plasma CVD reactors are capable of producing optically clean Type IIa material at a scale that would have been industrially impossible twenty years ago. The geographic concentration of CVD production in mainland China and India reflects access to electricity, capital equipment, and downstream cutting infrastructure rather than any fundamental scientific advantage⁵.

The diamond classification scheme distinguishes Type I (containing measurable nitrogen) from Type II (containing little or none), and within each, sub-types based on the configuration of the impurity³. Most natural diamonds are Type Ia, with nitrogen present in aggregated clusters. Most CVD-grown diamonds are Type IIa, because the chamber chemistry excludes nitrogen except as a deliberate trace. HPHT-grown diamonds vary by catalyst chemistry and trapped atmosphere; they often fall into Type Ib (isolated nitrogen, sometimes giving a yellowish tint) or, with boron exposure, Type IIb.

The crystal type is not just a curiosity. It influences colour, fluorescence, and the spectroscopic signatures that gemmological laboratories use to distinguish lab-grown from natural stones. The growth-front conditions also leave morphological fingerprints: layered planar growth in CVD, octahedral or cuboctahedral facets in HPHT, and complex multi-stage growth horizons in natural stones. These patterns are visible under the cathodoluminescence and DiamondView imaging used for laboratory identification, and are covered in Chapter 13.

Operating-condition figures consistent with GIA technical descriptions of HPHT and CVD methods². Type classifications consistent with GIA Diamond Type guidance³.

All three pathways yield rough diamond. None of them yield a finished gem. From rough, every stone - natural, HPHT-grown, or CVD-grown - passes through the same downstream sequence: planning, sawing or laser-cutting, bruting (rounding), faceting, and polishing. The downstream cost per carat is broadly the same regardless of how the rough was produced, because the cutting and polishing labour and equipment do not know or care about the rough's origin. This matters for the price discussion: lab-grown saves cost in the rough but not in the cutting.

Where this fits in the reference

The next chapter, CVD vs HPHT, takes a closer look at the two laboratory methods, including apparatus geometry, growth-chemistry differences, and the morphological fingerprints that gemmological laboratories use to distinguish them. Chapter 3 explains how the resulting stones are graded, and Chapter 11 sets out the energy budget of each pathway.