How Lab-Grown Gems Are Grown
Grown, not found · Field guide
ApparatusAlmost every grown gem material on the market comes out of one of nine processes or a close variant. The method sets the price, the size, the flaws, and the fingerprints a laboratory will look for. Here they are in cross-section.
The processes sort into five families, by what the material is dissolved in — or whether it's dissolved at all.
Family one: melt growth
Melt the compound and freeze it back in an orderly way, so the atoms land on a single growing crystal instead of a million random ones. Fast and cheap, but you need the compound to melt without decomposing, and the crystal has to survive cooling through a thousand-plus degrees.
Flame fusion (Verneuil)
- 1Alumina powder hopper, tapped so grain falls at a metered rate
- 2Oxy-hydrogen burner, concentric tubes
- 3Flame, roughly 2000–2200 °C
- 4Boule. Builds up from a sintered cone as melt freezes on top; splits lengthwise when cooled
- 5Ceramic pedestal, lowered slowly as the boule lengthens
The oldest commercial method, and still the cheapest by a wide margin. Fine oxide powder is dusted through an oxy-hydrogen flame onto the top of a growing boule sitting on a pedestal that creeps downward. A boule takes a couple of hours. No crucible is involved at all, so nothing contaminates the melt, and in 1902 that also solved the problem of containing corundum at 2050 °C.
Because the melt film is a curved cap on top of the boule, the growth layers come out curved. That's one of the classic identification features in gemology: curved banding means a melt-grown synthetic, nearly always flame fusion, though pulled crystals can show curved striae too. Verneuil boules also carry round gas bubbles and split lengthwise on cooling from internal strain — which is why the rough often arrives as half-carrots.
Makes: corundum, spinel, rutile, strontium titanate.
Crystal pulling (Czochralski)
- 1Pull rod. Lifts a few mm per hour while rotating
- 2Seed crystal, then a thin neck that strains out dislocations
- 3Shoulder, flared out to the target diameter
- 4Boule body, grown as a single crystal
- 5Iridium crucible holding the melt, counter-rotated, RF heated
Jan Czochralski found this in 1916 while measuring how fast molten metal crystallizes. It became the process that makes most of the world's silicon wafers, and it was adapted to oxides in the 1960s. A seed crystal is touched to the surface of a melt in an iridium crucible, then lifted and rotated while the melt freezes onto it. The thin "neck" just below the seed is deliberate — it strains out crystal defects before the boule widens.
Slower and far more expensive than flame fusion, but it gives large, low-strain crystals of high optical quality, and it accepts dopants cleanly. Almost every engineered gem material — YAG, LuAG, GGG, GAGG, YAP, lithium niobate, YVO₄, Czochralski alexandrite — comes off a puller.
The other melt methods
The first three below exist mainly to make sapphire in industrial quantity, and colorless gem rough is often the offcut. The last two serve other materials:
- Kyropoulos. The seed is barely withdrawn; the melt is cooled in place and the crystal grows down into it. Produces the huge low-strain sapphire boules used for watch crystals and smartphone camera windows.
- Heat exchanger method (HEM). A helium-cooled finger under the crucible pulls heat out from one point, and the crystal freezes outward from it. Nothing moves. Makes very large sapphire boules for armor and windows, though the size record now sits with Kyropoulos, at 350 kg by 2017.
- Edge-defined film-fed growth (EFG). Melt wicks up a shaped die and freezes in that profile. Makes sapphire ribbon, rod and tube directly, no coring required.
- Bridgman–Stockbarger. A sealed crucible is lowered slowly through a temperature gradient. Used for fluoride crystals, some scintillators, and some LuAG.
- Floating zone. No crucible at all — a molten band is held between two rods by surface tension and swept along. Used when crucible contamination cannot be tolerated.
Skull melting
- 1Water-cooled copper fingers. The crucible stays cold
- 2A shell of unmelted zirconia freezes against the wall and becomes the container
- 3Molten pool at roughly 2750 °C, heated directly by radio-frequency coupling
- 4Columnar crystals grow upward as the assembly is withdrawn from the coil
A solution to an impossible problem: zirconia melts at about 2750 °C, hotter than any practical crucible can stand. So you make the crucible out of the zirconia itself. A ring of water-cooled copper tubes keeps the outside frozen while a radio-frequency field heats the interior directly. A shell — the "skull" — of unmelted powder forms against the cold wall and contains the melt. Lower the whole assembly out of the coil and columnar crystals grow upward through it.
Makes: cubic zirconia, and essentially nothing else in the gem trade.
Family two: solution growth
Dissolve the compound in something at a temperature well below its melting point, then let it come back out slowly onto seeds. Much slower — weeks to a year — and much more expensive, but it produces crystals that grew the way natural ones do, with flat faces and natural-looking inclusions. This is the method used when the goal is to fool people, and also the only method that works for compounds that decompose before they melt.
Flux growth
- 1Platinum crucible inside a furnace held to a fraction of a degree
- 2Molten flux (lithium molybdate, vanadate, lead oxide) with gem oxides dissolved in it
- 3Crystals nucleate and grow over weeks to months as the melt cools very slowly
The solvent is a molten salt or oxide — lithium molybdate, vanadate, lead oxide, borate — held around 1000–1300 °C in a platinum crucible. Gem oxides dissolve into it, and as the furnace cools very slowly over weeks to months, they crystallize out. Yields are low, crucibles are expensive, and the run takes a season.
The signature: flux inclusions, wispy veils and fingerprints of trapped solvent that look strikingly like the healed fractures in natural stones — plus occasional flakes of platinum from the crucible, which nothing natural contains.
Makes: emerald (Chatham, Gilson, Lennix), ruby and sapphire (Kashan, Ramaura, Douros), alexandrite, spinel, KTP.
Hydrothermal growth
- 1Thick-walled steel autoclave. About 350–400 °C for quartz, 500–620 °C for beryl, at 700 atmospheres or more
- 2Seed plates hanging in the cooler upper zone, where growth happens
- 3Perforated baffle. Sets the temperature split between the two zones
- 4Nutrient feedstock in the hotter lower zone, dissolving into the fluid
The solvent is hot water under enormous pressure. Quartz grows at about 350–400 °C and a thousand or more atmospheres in an alkaline solution of sodium hydroxide or carbonate. Emerald needs about 500–620 °C at 700–1,500 atmospheres. This is essentially a laboratory copy of how quartz and beryl grow in nature. Nutrient dissolves in the hot lower zone, convects up through a perforated baffle into the cooler upper zone, and deposits on hanging seed plates.
The signature: a visible seed plate running through the middle of the crystal, and chevron or "nailhead" growth patterns on the faces. Hydrothermal stones grow on a flat plate, so the rough is tabular rather than blocky.
Makes: quartz in enormous quantity, beryl (emerald, aquamarine, morganite, red beryl), some corundum.
Family three: vapor growth
Skip the liquid. Get the material into the gas phase and let it condense onto a seed.
Physical vapor transport
- 1Seed wafer mounted on the cooler lid
- 2Crystal thickens downward, one polytype at a time
- 3Source powder at about 2200–2500 °C sublimes and re-deposits on the seed
Silicon carbide doesn't melt at any usable pressure — it sublimes. So SiC powder is heated to 2200–2500 °C at the bottom of a graphite crucible, the vapor drifts up a temperature gradient, and it re-condenses on a seed wafer mounted on the cooler lid. Growth is measured in fractions of a millimeter per hour and a boule takes days.
Makes: moissanite.
Chemical vapor deposition
- 1Hydrogen and a trace of methane at low pressure
- 2Microwave-driven plasma cracks the gas into reactive carbon and atomic hydrogen
- 3Diamond seed plates. The crystal builds up in layers, roughly 10–30 µm per hour
A low-pressure atmosphere of hydrogen with a percent or so of methane, energized by microwaves into a plasma. The plasma cracks the gas into reactive carbon and atomic hydrogen; carbon lands on diamond seed plates and the atomic hydrogen etches away anything that isn't diamond. The crystal builds up layer by layer at maybe 10–30 microns an hour.
Makes: diamond. CVD stones grow as flat tablets, so the rough looks nothing like a natural octahedron.
Family four: extreme pressure
High pressure, high temperature
- 1Carbon source at the hot end of the capsule
- 2Molten metal solvent (iron, nickel, cobalt) that carbon dissolves into
- 3Diamond seed at the cooler end. Carbon crosses the gradient and crystallizes on it
The original synthetic diamond route, from General Electric in 1955. A capsule containing a carbon source, a molten metal solvent and a diamond seed is squeezed to 5–6 GPa — roughly the pressure 150 km down in the mantle — at 1300–1600 °C. Carbon dissolves in the metal at the hot end, crosses the temperature gradient, and crystallizes on the seed at the cool end.
HPHT crystals grow as cuboctahedra with a distinctive mix of cube and octahedron faces that natural diamond rarely shows, and they usually carry traces of the metal solvent — which makes them faintly magnetic, a tell no natural diamond has.
Family five: controlled devitrification
Glass-ceramic (ceramming)
- 1A silicate melt with added nucleating agents, poured and cooled to a plain transparent glass
- 2Reheating grows crystals only to tens of nanometers — far below the wavelength of light, so the block stays clear
The odd one out: it doesn't grow a crystal, it grows crystals inside a piece of glass. A silicate melt loaded with nucleating agents is poured and cooled to an ordinary transparent glass, then reheated on a precise schedule — a hold to seed nucleation sites, then a hotter hold to let them grow. Stop the growth while the crystals are still tens of nanometers across, far below the wavelength of visible light, and the block stays perfectly clear while gaining hardness.
Makes: nanosital and its cousins. Also, in the industrial world, ceramic cooktops, telescope mirror blanks and missile radomes.
Which method makes what
| Method | Timescale | Typical materials | Cost signal |
|---|---|---|---|
| Flame fusion (Verneuil) | Hours | Corundum, spinel, rutile, strontium titanate | Cents per carat |
| Czochralski | Days | YAG, LuAG, GGG, GAGG, YAP, LiNbO₃, YVO₄, alexandrite, corundum | Dollars per carat |
| Kyropoulos / HEM / EFG | Days–weeks | Industrial sapphire; gem rough as offcut | Low, driven by tech demand |
| Bridgman–Stockbarger | Days | CaF₂, scintillators, LuAG | Moderate |
| Skull melting | Hours–days | Cubic zirconia | Cents per carat |
| Flux | Weeks–months | Emerald, ruby, sapphire, alexandrite, spinel | Tens to hundreds per carat |
| Hydrothermal | Weeks–months | Quartz, beryl, some corundum | Low for quartz, high for beryl |
| Physical vapor transport | Days | Moissanite | Tens per carat |
| Microwave plasma CVD | Weeks | Diamond | High, falling fast |
| HPHT press | Hours–days | Diamond | High, falling fast |
| Melt + ceram | Days | Nanosital | Low |
Fingerprints the method leaves behind
| If you see… | It was probably grown by… |
|---|---|
| Curved color bands or curved striae | Flame fusion, occasionally Czochralski. Natural corundum zones are straight and angular |
| Round or teardrop gas bubbles | Flame fusion, or glass |
| Wispy veils, fingerprints, tiny flux droplets | Flux |
| Bright metallic platelets (platinum) | Flux, from the crucible |
| A flat seed plate through the middle of the crystal | Hydrothermal |
| Chevron or nailhead growth patterns | Hydrothermal |
| A perfectly flat tabular crystal with parallel striations | CVD |
| Cuboctahedral crystal, faint magnetism, hourglass color zoning | HPHT |
| An aggregate reaction in a polariscope on a clear, isotropic-looking stone | Glass-ceramic (nanosital) |
| Cross-hatched 'tabby' strain extinction in a cubic stone | Verneuil spinel |
Grown, not found: the field guide
Start here
Nature barely does
Nature never does
Reference
A reference to lab-grown gem materials, written for cutters. Values are typical published figures; ranges vary with composition, dopant and measurement wavelength. Nothing here is a substitute for a laboratory report on a stone whose identity matters commercially.