How lab gems are grown

Sapphire and ruby, emerald, diamond, moissaniteSources at the end

Sapphire and ruby: four ways to grow corundum

Sapphire and ruby are the same mineral, corundum, aluminium oxide (Al2O3), Mohs hardness 9. Pure corundum is colourless. A trace of chromium turns it pink to red, and that is ruby; iron and titanium together give the blue of sapphire; vanadium pushes it purple or green. Every method below grows real Al2O3 with the same hardness and the same refractive index (about 1.76 to 1.77) as a mined stone. A refractometer cannot tell them apart. The microscope can, because each process leaves its own fingerprint.

Flame fusion (Verneuil), 1902: the one in almost all cheap jewelry

Auguste Verneuil announced his process in 1902. Finely powdered alumina, with a pinch of chromium oxide for ruby or iron and titanium oxides for blue, trickles through a hydrogen and oxygen flame at over 2,000 C. The droplets land on a ceramic rod that is slowly lowered, and a single crystal builds up as a rounded cylinder called a boule, typically about 13 mm across and 25 to 50 mm long, roughly 125 carats. World output was around a tonne a year in 1907 and about 250 tonnes a year by 2000. A single factory can run close to a thousand furnaces at once, and rough flame-fusion corundum can cost a few cents per carat.

Because the crystal grows on a curved dome, the growth lines inside are curved, in concentric shells like the rings of an onion. Curved striae never occur in a natural crystal, so they are instant proof of flame-fusion origin. Round gas bubbles from excess oxygen, sometimes with a small tail, are the other tell. This is the material in scratch-resistant watch crystals and, by cost alone, in nearly every created sapphire or created ruby set in a steel ring under $30; nothing else grows corundum at that price.

Czochralski pulling, 1916: the semiconductor method

Jan Czochralski's method dips a seed crystal into a crucible of molten aluminium oxide and slowly pulls it upward while rotating, drawing out a single crystal behind it. It was developed because lasers and electronics needed cleaner crystals than the flame could make; pulled ruby can come out as clear rods inches across and feet long. Pulled stones have no curved striae. They can carry tiny metallic specks from the crucible and the odd round bubble, and they sit between flame fusion and flux in price.

Flux growth, 1958 onward: the one that looks natural

Carroll Chatham's ruby dates from 1958. Aluminium and chromium oxides are dissolved in a molten flux, typically a lead or lithium compound, in a platinum crucible held at roughly 1,000 to 1,400 C, and the corundum crystallises out over months; Chatham says about a year. Growth on flat crystal faces gives straight, angular zoning just like a mined stone, which is why flux ruby was the first synthetic that fooled jewelers. The tells are wispy, veil-like clouds of trapped flux, the "fingerprints", and triangular or hexagonal platinum platelets shed by the crucible wall. Ramaura and Kashan were the other famous flux rubies; Ramaura is no longer made.

Hydrothermal, 1992 onward: the slow, expensive one

Tairus, a Russian Academy of Sciences venture, brought hydrothermal ruby to market in 1992. Nutrient dissolves in a hot, pressurised water solution in a steel autoclave and deposits on a seed plate over weeks or months. GIA's 1997 study found growth features specific to hydrothermal crystals plus hydroxyl and carbon-oxygen groups in the infrared spectrum, which flame-fusion and flux stones do not have. It is the costliest way to grow corundum and the rarest in cheap jewelry.

How a gemologist tells lab corundum from mined

Curved striae and round bubbles mean flame fusion. Angular zoning with flux veils or platinum means flux. Roiled growth and water in the infrared means hydrothermal. Synthetic sapphire also often shows a chalky blue glow under short-wave ultraviolet, caused by titanium in the lattice at as little as one part per million, though some heat-treated natural sapphire can glow too. When the stone is clean, labs turn to trace-element chemistry: mined corundum usually carries gallium, iron and vanadium in patterns a furnace does not reproduce, though a 2026 study found hydrothermal synthetics with enough gallium that chemistry alone is no longer safe. In the US, 16 CFR 23.25 lets a seller use the word sapphire or ruby only when it is immediately preceded by "laboratory-grown", "laboratory-created", "[maker]-created" or a phrase of like meaning, and only if the stone has essentially the same optical, physical and chemical properties as the mined mineral. "Created" and "synthetic" are legally the same claim. Neither tells you the method.

Emerald: why it cannot take the flame

Emerald is beryl, Be3Al2Si6O18, coloured green by chromium or vanadium, Mohs 7.5 to 8, refractive index about 1.57 to 1.60. Natural emerald is almost always fractured and included (the inclusions are called the jardin, the garden), and well over 90 percent of the emeralds sold have been soaked in oil or resin to hide those fractures. That matters here because lab emerald is typically much cleaner than the mined stone at the same price, which is the first thing a jeweler notices.

Beryl melts incongruently: heated far enough it breaks down into other compounds before it becomes a liquid, so Verneuil's flame produces glass, not crystal. Emerald has to be crystallised from a solution, which is why it arrived decades after flame-fusion ruby and has never been cheap in the same way. The first emerald crystals of about a millimetre came from French chemists in 1888 and 1890; Richard Nacken grew flux emeralds in Germany in 1923 to 1925; IG Farben announced "Igmerald" in 1935 and produced it until 1942; Carroll Chatham grew his first crystals in 1935 and had a repeatable process by 1938, with production in the US from 1941.

Flux growth (Chatham, Gilson)

Beryllium, aluminium and chromium compounds plus silica are dissolved in a molten flux, most often a lithium molybdate or lithium vanadate, in a platinum crucible, and emerald crystallises on beryl seed plates. Chatham says a crop takes about a year; other producers ran batches of weeks. Flux emerald contains no water in its structure. Inside it shows wispy flux-filled fingerprints, small colourless phenakite crystals that grow as a by-product, triangular or hexagonal platinum platelets, and "Venetian blind" parallel growth planes. Flux-filled fingerprints are conclusive: flux cannot get into a natural fracture. Flux emerald also reads slightly low on a refractometer (about 1.560 to 1.565) with a birefringence of 0.003 to 0.004 against 0.004 to 0.010 for natural.

Hydrothermal growth (Linde, Biron, Tairus, Chinese producers)

Hydrothermal emerald reached jewelry in 1960. The Union Carbide (Linde) patent of 1964 describes an autoclave at 425 to 800 C and 7,000 to 30,000 psi, an ammonium fluoride and ammonium hydroxide solution, and growth of about a tenth of a millimetre per day on colourless beryl seed plates; commercial growth is about 0.15 mm a day, so a crystal thick enough to cut takes weeks. Russia became the major producer in the late 1980s and 1990s, and Tairus dominates now. The tells: chevron growth zoning, angular zig-zags caused by temperature swings at the seed; nail-head spicules, cone-shaped voids with a tiny crystal at the head; seed-plate remnants; and gold or platinum platelets. Chlorine from the growth solution leaves infrared features never seen in natural emerald, and the water inside is of a different type than in mined beryl, which infrared spectroscopy shows directly.

Green things that are not emerald

Green glass and flame-fusion synthetic spinel both show gas bubbles, and spinel reads 1.73 on the refractometer, far from beryl. Quartz doublets exist: GIA reported a piece made of three colourless quartz slabs glued with a 0.2 mm green cement layer, given away by quartz's refractive index of 1.54 and bubbles in the glue. Under 16 CFR 23.25, an unqualified "emerald" must be a mined stone; "laboratory-created" or "[maker]-created emerald" must be real beryl with the same properties; anything that only looks like emerald must be called imitation or simulated, and "emerald colour" or "emerald green" is a description of hue that claims nothing about what the stone is. We have not tested the green stone in the steel ring we bought, so we do not say what it is. Lab emerald costs more per carat than flame-fusion sapphire because a platinum crucible or a pressurised autoclave running for weeks is not a flame running for days.

Diamond: pressure or plasma

Diamond is carbon in the cubic crystal system, Mohs 10, refractive index 2.42, dispersion 0.044, and the best heat conductor of any gem, which is why a thermal probe identifies it. It is hard to grow because at the surface of the Earth the stable form of carbon is graphite; diamond only persists because the conversion back takes millions of years. So you either force carbon across the graphite-to-diamond boundary with pressure, or you deposit it one atom at a time in a plasma that etches away anything that is not diamond.

HPHT: high pressure, high temperature, since 1954

H. Tracy Hall at General Electric made the first reproducible synthetic diamond on 16 December 1954 in a belt press at about 1,600 C and 100,000 atmospheres. Three press designs are used today: the belt press, the cubic press with six anvils on a cube-shaped cell, and the Russian BARS split-sphere press, the most compact and economical, which holds a capsule of about two cubic centimetres inside a barrel a metre across. Graphite powder and a diamond seed sit in a molten flux of iron, nickel or cobalt at 5 to 6 gigapascals, the pressure 150 to 190 km down in the mantle, and 1,300 to 1,600 C. The seed is held slightly cooler, carbon supersaturates in the metal and crystallises onto it. A crystal takes anything from an hour to a few weeks; the largest HPHT crystal recorded is 150.42 carats (2021).

HPHT diamond grows in cube and octahedron faces at once, and its defects segregate into those sectors, so under ultraviolet it often shows a cross-shaped fluorescence pattern and a long blue-green phosphorescence that can glow for a minute after the lamp goes off. Metallic flakes of the flux metals are the other tell. More than 80 percent of colourless HPHT stones carry a trace of boron detectable in the infrared, so technically most are type IIb. Add boron on purpose and the crystal grows blue; historically, nitrogen made the early ones yellow.

CVD: chemical vapour deposition, the method behind most jewelry lab diamond

Diamond seed plates sit in a reactor at low pressure, about a quarter of an atmosphere, in a flow of hydrogen with about one percent methane; microwaves ignite a plasma and carbon lands on the seeds at roughly 800 to 1,200 C. The hydrogen is the trick: it etches away non-diamond carbon faster than diamond, so only diamond survives. Growth is measured in micrometres per hour and a jewelry stone takes three or four weeks, with the plates pulled every few days for polishing, which is why CVD stones show stop-start layering. Early CVD stones were small and brown until about 2010; the brown comes from vacancy clusters and nitrogen defects, and about 80 percent of the CVD stones GIA sees have had a post-growth HPHT anneal to bleach it. As-grown CVD crystals are flat tablets, roughly 8 mm square and a millimetre or so thick for a carat and a quarter.

CVD is now the volume method. GIA says it averages more CVD submissions per day than it once saw in a year. In 2020 the world grew 6 to 7 million carats: China about 3 million, mostly HPHT; India about 1.5 million, mostly CVD, from an estimated 4,000 to 6,000 reactors; the US about a million, CVD. A CVD reactor in India fell from around $300,000 in 2019 to under $100,000 by 2023. Under deep ultraviolet a CVD stone shows striped, layered fluorescence and growth striations; photoluminescence often shows a silicon-vacancy line at 737 nm, though that marker has faded a hundredfold over ten years as growers cleaned up their reactors.

Detection, grading and what happened to the price

A handheld thermal tester reads lab diamond as diamond, because it is diamond. It also reads moissanite as diamond. Testers with an electrical probe flag moissanite, which conducts, but a boron-bearing diamond, natural blue or HPHT, also conducts and can be misread as moissanite. Real screening starts with diamond type: essentially all colourless lab diamonds are type II and only about one percent of mined diamonds are, so a type IIa screener passes the 99 percent and refers the rest to a lab, where deep-UV imaging and spectroscopy settle it. GIA inscribes every lab diamond girdle with "Laboratory-Grown" and a number and, from 1 October 2025, stopped issuing 4Cs-style grades for lab stones in favour of "Premium" or "Standard"; IGI still issues full lab-grown reports and does most of the volume. "GRA" paper, which turns up with mass-market moissanite and melee, is a seller-facing lookup card from an entity that is not an accredited gem laboratory and has no connection to GIA or IGI.

The price fell off a cliff. Edahn Golan's wholesale index for lab diamond dropped 86.5 percent from the third quarter of 2018 to the first quarter of 2024, and a further 26 percent in 2025; a round one-carat IGI-graded stone cost retailers about $191 per carat in mid 2025, and Chinese HPHT rough reached about $10 a carat. That is why one-carat parcels of 1 mm and 2 mm CVD melee, thirty-odd stones, are sold on Amazon as a hobby item. The FTC's 2018 Jewelry Guides allow "laboratory-grown", "laboratory-created" or "[maker]-created" for a stone with essentially the same optical, physical and chemical properties as mined diamond, allow "cultured" only alongside one of those, and dropped "synthetic" from the recommended list.

Moissanite: silicon carbide by sublimation

Moissanite is the mineral name of silicon carbide, SiC. Henri Moissan found tiny crystals of it in fragments of the Canyon Diablo meteorite in Arizona in 1893 and at first took them for diamond. Natural moissanite occurs only as minute green to black grains and has never been found in a piece large enough to facet, so every gem moissanite is lab-grown; anyone selling "natural" moissanite jewelry is not. The numbers, from GIA's 1997 study: refractive indices 2.648 and 2.691 (diamond 2.417), dispersion 0.104 (diamond 0.044), Mohs 9.25, specific gravity 3.22 (diamond 3.52). It is doubly refractive, with a birefringence of 0.043, and diamond is not, which is the single most useful fact for identification.

Industrial SiC has been made in the Acheson furnace since 1893, but that gives random platelets. Lely in 1955 sublimed SiC in a sealed graphite crucible at about 2,500 C under argon and grew flake crystals in a cavity; Tairov and Tsvetkov added a seed crystal in 1978, and that seeded sublimation, called physical vapour transport, is how every bulk SiC crystal is grown today, for power electronics as much as for gems. SiC powder at the hot end, around 2,350 C, sublimes and recondenses on a cooler seed at around 2,250 C in argon at a few kilopascals, building a boule at rates in the range of a millimetre an hour. Cree Research, a semiconductor company, held the seeded-growth patents; its first described crystal was a 12 by 6 mm 6H crystal grown in six hours, and it had 50 mm boules by 1994. Gem moissanite is the 6H polytype; more than 150 polytypes of SiC exist.

Early crystals were yellow-green because atmospheric nitrogen slips into the lattice. Near-colourless growth by compensating that impurity is the 1998 Cree patent; C3 Inc., which became Charles and Colvard, sold the first gem moissanite in 1998 in colours around J to M, at 5 to 10 percent of the price of comparable diamond. The US patent expired in 2015 and the EU patent in 2016, Chinese producers (China makes over 80 percent of the world's SiC) flooded in, and Charles and Colvard's D-E-F "Forever One" launched in 2015. That is the moissanite in a cheap pair of studs today: colourless 6H silicon carbide, grown by the same process as the chips in an electric car.

Doubling, and the rest of the checklist

Stones are cut with the optic axis straight through the table, so looking straight down at the point of the stone you see no doubling. Tilt it and look through a crown facet with a 10x loupe, focusing on the far pavilion edges, and every edge becomes two lines. Diamond is cubic and never doubles. Other tells: whitish sub-parallel needles running roughly perpendicular to the table; a loose stone floats in methylene iodide (density 3.32) while diamond sinks; an electrical tester flags older stock, though post-2015 colourless material conducts so weakly it can read as diamond; and, of course, the fire. Moissanite is stable in air to about 1,700 C, survives normal jeweler's soldering and casting, and turns yellow then cherry red under a torch and back again when it cools. No gemological source we found reports it clouding or yellowing with age; the "cloudy moissanite" pages online describe dirty stones.

Why moissanite throws rainbows

Dispersion is the reason. The refractive index of any transparent material is slightly different for each wavelength, so blue light bends a little more than red on the way in and the way out, and a white ray leaving a facet has been pulled apart into a fan of colour. The number for a gem is the difference in refractive index between the red and violet ends of the spectrum: 0.044 for diamond, 0.104 for moissanite, so moissanite spreads the colours 2.36 times as far. GIA puts it as "more than twice the fire of diamond and slightly more brilliance", and notes that in stones of a carat or more the disco-ball effect is enough to give moissanite away to a trained eye.

You have probably seen the photo this diagram is drawn from: two stones on black glass under a single lamp, a rainbow starburst on the left labelled moissanite, a white one on the right labelled diamond. It is honest physics and a loaded photograph at the same time. GIA's cut research found that a point light source accentuates fire while diffuse light suppresses it, so under a spotlight the difference is at its largest, and under an office ceiling both stones look mostly white. The photo is also easy to tilt: a smaller diamond, a worse cut, a lamp a few degrees off, and the gap grows. What the diagram shows is real; how much of it you see on an ear or a hand depends on the light in the room. Whether you like the rainbow or find it too much is taste, and it is the main reason people choose one stone over the other once price is off the table.

Side by side

Mohs, refractive index and dispersion for each are on the key facts page. None of this tells you what is in a given Amazon listing; for that, read how to read a gem listing.

What this means when you buy

A created sapphire ring under $30 is flame-fusion corundum: real sapphire, grown in days, curved striae inside if you look. A created emerald at the same price deserves a question, because real lab beryl costs more to grow; ask the seller whether it is hydrothermal beryl or a simulant, and expect no answer. Lab diamond melee is CVD or HPHT and is genuinely diamond; the paper that comes with it is not a lab report unless it says GIA or IGI and the girdle is inscribed. Moissanite is silicon carbide, all of it lab-grown, and the rainbow is the physics of the material, not a coating or a trick. None of these stones holds resale value, and that is exactly why a tray of them costs what a night out does. We collect them for what they are, and this page is here so you know what that is.

Sources

• GIA: Synthetic gemstones; HPHT and CVD diamond growth processes; 2024 update on laboratory-grown diamonds; Synthetic moissanite (1997); Hydrothermal synthetic sapphires (1997); Nacken's synthetic emeralds; Emerald vibrational spectroscopy; Chinese hydrothermal emerald; Imitation gems; Grading cut quality (2004); Lab-grown services from 1 October 2025.
• Nassau, K., Synthetic moissanite: a new man-made jewel, Current Science, 2000.
• US FTC: 16 CFR 23.25 and the Jewelry Guides.
• ASME, GE belt press landmark; Union Carbide, hydrothermal emerald patent US 3,567,642; PVT growth of SiC (2024); SiC bulk growth review; Trace elements in hydrothermal synthetic corundum (2026).
• Lotus Gemology, UV fluorescence as a gemological tool; International Gem Society, Synthetic gemstone guide; Gem-A, Zerfass synthetic emeralds; JCK, Moissanite: the post-patent era and Edahn Golan on lab-grown prices; Lab-grown wholesale prices 2025; Wikipedia: Verneuil method, Synthetic diamond, Moissanite.