Winemaking guide · Cold stabilization
Crystals in wine: what they are and how to prevent them
The crystals in wine, at the bottom of the bottle or on the cork, are tartrate crystals, sometimes called wine diamonds: potassium bitartrate (KHT), a natural salt of the tartaric acid in grapes. They are not glass, and they are harmless. To keep them out of the bottle, cold stabilize the wine before bottling: one to three weeks at 0 to −3 °C (32 to 27 °F), or seed it with 4 g/L of KHT crystals and stir for at least 4 hours in the cold.
What the crystals look like
In white and rosé wine the crystals are clear or white, and look like coarse sugar or bits of glass. They sit on the bottom of the cork or at the bottom of the bottle and the glass. In red wine the crystals soak up pigment and come out a dark purple-red, sometimes almost black.
Under magnification you see small boat-shaped crystals, next to large plates that look like broken glass. That is why some buyers return bottles, sure that glass got into the wine.
A cloudy haze is a different problem, usually proteins, and belongs to heat stability.
What the crystals are made of, and why they form
They go by tartrate crystals, wine diamonds or wine stones. Chemically they are potassium bitartrate, KHT for short (potassium hydrogen tartrate). Not bicarbonate: that is baking soda, a different thing altogether.
Grapes bring tartaric acid and potassium, and in the juice they stay dissolved. Fermentation makes alcohol, and KHT dissolves less well in it. According to AWRI, about half the tartrate dissolved in the juice can no longer stay dissolved in the wine. The wine is left supersaturated: it holds more KHT than it can keep in solution over time. Enough cold, time or a first crystal, and the excess drops out.
Why it is mostly a white, rosé and sparkling problem
In white and rosé you see every crystal through the glass, and these wines are stored and served cold. In a home fridge at about 4 °C (39 °F), crystals can appear within a week. In red wine, tannins, pigments and colloids hold the crystals back, and when they do appear, they hide among the rest of the sediment.
In sparkling wine each crystal releases the CO₂ all at once, and the wine gushes when you open the bottle. According to AWRI, KHT crystals are the most common solid in these bottles. And the second fermentation adds alcohol, so a base wine that looked stable can throw crystals after it.
How a crystal forms: three stages
A crystal does not appear all at once. It goes through three stages, and every stabilization method works on one of them.
-
Supersaturation
Cold and alcohol lower solubility, and the wine holds more potassium and bitartrate ions than it can keep dissolved.
-
Nucleation
Ions cluster into a first crystal nucleus. This is the slow stage, and temperature swings slow it further. KHT seeding skips it.
-
Growth and settling
Ions attach to the nucleus, and the crystal grows until it settles. Colloids such as mannoproteins and CMC cling to the crystal faces and hold it back.
Potassium (K⁺) Bitartrate (HT⁻) Protective colloid
Cold stabilization runs all three stages in the tank, before bottling. Inhibitors such as CMC and KPA only block the third stage, and remove no KHT from the wine.
What decides whether a wine throws crystals
- Temperature colder → less stays dissolved A wine that was stable in the cellar can throw crystals in the fridge.
- Alcohol more alcohol → less stays dissolved Stronger wines are at higher risk, and sparkling needs a safety margin for the second fermentation.
- Potassium and tartaric acid more of them → more crystal material A wine rich in potassium is more likely to throw crystals. Adding tartaric acid also adds raw material.
- pH risk peaks around pH 3.7 That is where there is the most bitartrate. Below pH 3.65, crystals dropping out lower the pH a little; above it, they raise it.
- Protective colloids more colloids → crystals later Mannoproteins, tannins and polysaccharides hold crystals back. In red wine, about 1 g/L of tannin already inhibits strongly. Bentonite removes some of them.
- Calcium a different crystal: calcium tartrate Comes from concrete tanks and from deacidifying with calcium carbonate. Cold barely affects it (more below).
How much KHT dissolves, by temperature and alcohol
In a 12% alcohol solution, 2.77 g/L of KHT dissolves at 20 °C (68 °F), and only 1.11 g/L at 0 °C (32 °F). Anything above that tends to drop out, and each extra percent of alcohol lowers the line a little more.
In warm regions such as Israel: high pH and acid additions
A hot vintage gives low acidity and high pH, and many wineries correct it with tartaric acid. Add the acid before cold stabilizing, not after: acid that goes in after stabilization brings the problem back. And in a wine above pH 3.65, KHT dropping out raises the pH further, so correct first.
After stabilizing, measure pH and top up SO₂ for the new value (see how much sulfite to add to wine). And metatartaric acid, a cheap old inhibitor, breaks down within about two months at 25 °C (77 °F), so it will not hold in a wine that spends a hot summer in a warehouse or in shipping.
When: heat stability first, then cold
Cold stabilization is one of the last steps before bottling. Any change to the wine after it, such as a blend or an acid adjustment, can bring the instability back. Malolactic fermentation should be finished before it too, because it changes the acidity.
- Blend and malolactic The wine in its final makeup
- Acid adjustment Tartaric acid only now
- Heat stability Bentonite against protein haze
- Cold stability Test, stabilize, test again
- Filter and bottle No further changes
Heat stability comes first because bentonite strips out some of the colloids and changes the wine, so a cold test done before it no longer holds. And inhibitors such as CMC and KPA need a protein-stable wine, or they cause haze.
How to test whether a wine is cold stable
Chill test: three days at −4 °C (25 °F)
This is the test AWRI recommends. Filter a sample, hold it for three days at −4 °C in a brine bath or a calibrated freezer, and check for crystals. Crystals that dissolve once the sample is back at room temperature don’t count: the wine is stable. Crystals that stay mean it is not. In red wine a temporary phenolic deposit may also appear, and it usually clears on warming. For wine treated with KPA, Enartis recommends holding 6 days.
Conductivity test (mini-contact)
Electrical conductivity measures how many ions are dissolved in the wine, in microsiemens per centimeter (µS/cm). As KHT crystallizes, ions leave the solution and conductivity drops. The test takes under an hour:
- Chill about 100 mL of filtered wine to 0 °C (32 °F) for white and rosé, or 5 °C (41 °F) for red. The wine only comes out stable down to the temperature you tested at, so pick the lowest it is likely to meet.
- Measure conductivity while stirring.
- Add KHT powder and keep stirring. Many labs add 4 g/L (0.4 g per 100 mL). The AWRI method adds 1 g/L.
- Take a reading every few minutes until conductivity stops falling, usually about half an hour. Keep the temperature steady, because conductivity changes with it.
- Work out the drop as a percentage. According to AWRI, a drop of 5% or more means the wine is not stable. Some labs are stricter: in the Wineland method, for example, the limit is 4%.
An overnight freezer test is the least accurate: the ice concentrates the wine and shows crystals even in a wine that may be stable. AWRI does not recommend it as a final check.
How to cold stabilize wine
There are four ways. Two remove the excess KHT with cold, one stops it from crystallizing, and one removes the ions with no chilling at all.
Yes
No: pick a method
1. Classic chilling
Chill the wine close to its freezing point and hold it there. The usual advice: −1 to −3 °C (30 to 27 °F) for one to three weeks, or 0 to −2 °C (32 to 28 °F) until a repeat test comes out stable. Keep the temperature steady, because swings slow crystal formation.
At home, a common setup is a chest freezer with a temperature controller set to about 2 °C (36 °F), for about a month. At the end, rack or filter while the wine is still cold, and leave the crystals behind. To switch between Celsius and Fahrenheit, use the temperature converter.
How cold you can go depends on the alcohol: the more alcohol in a wine, the lower its freezing point. A common estimate for dry wine: freezing point (°C) ≈ −(alcohol − 1) ÷ 2. Move the slider:
Freezing point and chill temperature
The formula is empirical (from an ISA Lisbon review) and suits dry wine. Residual sugar lowers the freezing point further.
2. Contact process: seeding with KHT
Instead of waiting for crystals to form on their own, you give them ready-made nuclei and skip the slow nucleation stage:
- Chill to 0 °C (32 °F) for white and rosé, or 5 °C (41 °F) for red. The wine comes out stable down to that temperature.
- Add 4 g/L of fine KHT powder, about 40 microns in size.
- Stir the whole time. Without stirring the seed settles, and the process barely works.
- Most of the crystals drop out in the first hour, and in most wines it stops after about 3 hours. To be safe, hold for at least 4 hours.
- Filter at the same temperature. You can collect the crystals and reuse them, but they grow every round, and in the end you have to grind them again.
3. No chilling: crystallization inhibitors
Inhibitors don’t remove KHT from the wine. They cling to crystals early on and stop them from growing. Add them only once the wine is ready to bottle, after the bentonite and the blend, because nothing in the wine should change after them.
- KPA (potassium polyaspartate): in a study in OENO One, 100 mg/L stabilized every wine tested, white and red. It does not bind to pigments, so it suits red too. It needs a protein-stable wine (per Enartis, a difference of under 1.5 NTU, the turbidity unit, in a heat test).
- Mannoproteins: proteins from the yeast cell wall, also present in wine naturally. At 20 g/hL they inhibit well in most wines. For liquid products, dose to the maker’s instructions. Too tight a filtration can strip them out, and they do not prevent calcium tartrate.
- CMC (carboxymethyl cellulose): for white, rosé and sparkling only. Not for red: it reacts with the pigments and can cause haze and color deposits. In rosé, run a small trial first. The common dose is up to about 100 mg/L (10 g/hL), or per the maker. Since 2019 the OIV allows under 200 mg/L (before that, 100); check what your country allows. After the addition, wait 2–5 days before filtering.
4. Electrodialysis
Membranes in an electric field pull potassium and tartrate ions out of the wine. It is fast and keeps the colloids, but the equipment is expensive, and for a home winery it is not an option.
Worked example: a 1,000-liter tank of Chardonnay
- Chardonnay
- Volume 1,000 L (264 gal)
- Alcohol 13%
- pH 3.40
- Heat stable after bentonite
- Conductivity test at 0 °C: 1,480 µS/cm before seeding, 1,392 after 30 minutes. The drop: 1,480 − 1,392 = 88, so 88 ÷ 1,480 ≈ 5.9%. Over 5%: the wine is not stable.
- Freezing point: −(13 − 1) ÷ 2 = −6 °C (21 °F). A contact process at 0 °C is far enough from it.
- Seed: 4 g/L × 1,000 L = 4,000 g, that is 4 kg of KHT. In the additive dose calculator that is a dose of 400 g/hL.
- Stir continuously for 4 hours at 0 °C, and filter at the same temperature.
- Repeat test: 1,455 before, 1,431 after. The drop: 24 ÷ 1,455 ≈ 1.6%. Stable.
Seed 4 kg of KHT, and the wine comes out stable
How much KHT came out of the wine? Say the tartaric acid fell from 2.0 to 1.5 g/L. Each gram of tartaric acid binds 0.26 g of potassium and forms 1.26 g of KHT, so 0.5 × 1.26 = 0.63 g/L came out, about 630 g in the tank. The pH started below 3.65, so it fell a little: measure it and top up SO₂.
Work it out for your wine. Enter a dose (400 g/hL of KHT seed, or the dose from the inhibitor’s product sheet) and the tank volume.
Read your conductivity test
Enter the conductivity before and after seeding, at the same temperature.
Up to 4%: stable. Between 4% and 5%: borderline, confirm with three days at −4 °C. 5% or more: not stable (the AWRI limit). After an inhibitor, test as the maker instructs.
Calcium tartrate, briefly
Sometimes the crystals are calcium tartrate, not KHT. Calcium in wine ranges from 6 to more than 165 mg/L, and comes from concrete tanks, from deacidifying with calcium carbonate, and from filter aids and pads. These crystals usually appear 4–7 months after fermentation. Cold barely affects them, so neither chilling nor mannoproteins solve them. If a stabilized wine throws crystals after a few months, have a lab test it for calcium.
Common mistakes
- Stabilizing before the bentonite or the blend Any change after stabilizing cancels the test.
- Filtering after the wine has warmed up Some of the crystals dissolve back. Filter cold.
- CMC in red wine It can cause haze and color deposits. For red, choose KPA or mannoproteins.
- Sparkling base wine with no margin The second fermentation adds alcohol. Per AWRI, you can add about 1–1.5% alcohol to the sample before testing.
- Metatartaric in a wine that will sit through summer In the heat its protection is gone within months.
Frequently asked questions
I found crystals in a bottle. Can I drink the wine?
Yes. It is the same stuff as the cream of tartar used in baking: natural, harmless, and it does not change the taste. Pour gently or decant.
How do I know it is not glass?
Glass does not dissolve, and KHT crystals dissolve slowly in hot water. In red wine the color gives them away too: glass does not turn purple.
Does cold stabilization change how the wine tastes?
A little. Tartaric acid leaves with the crystals, and acidity drops. In a wine below pH 3.65 the pH falls, sometimes by up to 0.2 units, and above 3.65 it rises. Taste and measure after stabilizing.
Short glossary
- KHT
- Potassium bitartrate (potassium hydrogen tartrate). A salt of tartaric acid, not bicarbonate.
- Tartrate stability
- The state in which a wine does not throw tartrate crystals, even in the cold. Also called cold stability.
- Supersaturation
- More dissolved material than the liquid can hold over time.
- Nucleation
- The forming of the first crystal nucleus. The slow stage.
- Electrical conductivity
- How well the wine carries a current, set by the amount of dissolved ions. Measured in µS/cm.
- Protective colloids
- Large molecules, such as mannoproteins and tannins, that hold back crystal growth.
Calculators for this guide
Sources
- Potassium instability — AWRI: how the crystals look in white and red wine, the solubility table (Berg and Keefer), half the tartrate becoming unstable, pH 3.65, holding between 0 °C and −2 °C, bentonite before cold stabilization. The row and column headings of the solubility table are swapped on the page.
- Measurement of cold stability of wine — AWRI: the −4 °C test for three days, the conductivity test at 0 °C and 5 °C with 1 g/L KHT, the 5% limit, the freeze test not as a final check.
- Carboxymethylcellulose (CMC) — AWRI: not for red, a trial before rosé, protein stability and lysozyme, a 2–5 day wait before filtering.
- Cellulose gums (carboxymethylcellulose) limit, update — OIV: resolution OIV-OENO 586-2019. CMC dose under 200 mg/L, up from 100.
- Sparkling wine gushing: not a cause for celebration — AWRI: KHT crystals as a common cause, adding about 1–1.5% alcohol when testing a base wine.
- Potassium bitartrate stabilization of wines — Zoecklein and Fugelsang, Virginia Tech: contact process (4 g/L, 40 microns, 4 hours), conductivity test, pH, tannins, calcium tartrate, metatartaric acid (gone within two months at 25 °C), electrodialysis, the 1.26 ratio, pH shifts of up to 0.2.
- Recent developments in wine tartaric stabilization — ISA Lisbon: the freezing point formula, mini-contact at 0 °C with 4 g/L, CMC and mannoprotein doses.
- The cold stabilisation tests of wine — Wineland: conductivity test with up to 4 g/L KHT, 30 minutes, a 4% limit.
- Use of polyaspartates for the tartaric stabilisation of white and red wines — OENO One: KPA at 100 mg/L.
- Zenith FAQs — Enartis: KPA in red, protein stability, a 6-day test at −4 °C.
- Mannostab Liquide 200 — Laffort: dose by the level of instability, does not prevent calcium tartrate.
- Cold stabilization — Winemaker Magazine: −1 to −3 °C for one to three weeks, a home chest freezer at about 2 °C, CMC not for red or rosé (with the next article).
- How can I best rid my wine of tartrate crystals — Winemaker Magazine
- Understanding tartrate crystals — Lescombes Winery: wine diamonds, cream of tartar, crystals within a week below about 4 °C.
- Development of a fluidized-bed crystallizer for wine treatment — UC Davis: buyers who fear it is glass.
- Wine crystals — Wines of Germany: harmless, crystals on the cork.