Dry (mechanical) resin separation removes trichome heads from dried cannabis plant material without solvent and without water. This article covers three methods: catching kief with a screen over the trimming table, hand-rolled charas from resin that collects on the fingers, and dry ice sieving, where cold of about −78.5 °C makes the heads brittle enough to snap off cleanly. Dry ice sieving is fast and cheap, and it is the only one of the three with a serious physical hazard. Every kilogram of dry ice turns into roughly 540 litres of carbon dioxide gas as it sublimes. The grower’s manual this method usually comes from underplays that hazard, and it is far too relaxed about the quality of the input material.
Legal warning — read first. In Thailand, resin separated from cannabis flower (kief, dry sift, charas or pressed hash) will in practice contain far more than 0.2% THC. Cannabis extracts above 0.2% THC are Category 5 narcotics, and producing, possessing or supplying them without a licence is a criminal offence. Treat separated resin the same way unless the Thai FDA or ONCB tells you otherwise (see section 9). In several other Southeast Asian countries, cannabis resin offences can carry severe penalties. This article is educational and describes processing in licensed or research settings. It is not a guide to personal production.
1. Where these methods sit
| Method | Principle | Water | Temperature |
|---|---|---|---|
| Dry sift / trim-table kief | Heads pass through a screen; plant debris stays on top | None | Ambient |
| Hand-rubbed (charas) | Resin collects on the skin and is rolled into a mass | None | Ambient |
| Ice water hash / bubble | Cold water makes heads brittle; agitation separates them; screens grade them by size | Yes | Near 0 °C |
| Dry ice sieving | Extreme cold makes heads brittle; agitation through a mesh | None | Around −78.5 °C |
Dry ice is the cold extreme of the group. It makes the heads more brittle than ice water does, but like dry sift it adds no water, so there is no drying stage and no period in which wet material can support microbial growth. The trade-off is that it uses a single pass through a single screen. There is no size fractionation, so quality depends on which screen you choose, not on collecting a series of grades.
2. Trim-table kief and hand-rolled charas
Kief on the trim table. During trimming, resin heads fall onto the table together with leaf fragments. In the traditional setup, a fine screen is laid over a smooth table (glass works well). The heads pass through, the plant debris stays on top, and the resin is collected from underneath when the work is finished. The workflow does not change, and the screen catches material that would otherwise be wiped off the table and thrown away. The manual names a silk bolting-cloth grade (“T-55”) but gives no micron rating. For repeatable, documentable work, use a rated monofilament screen specified in µm, and test a few sizes on your own material. The trim must be properly dry first. Drying and the bend-and-snap test are covered in topics 1363, 1064 and 245.
Charas. Resin builds up on the fingertips during hand trimming. In the traditional technique, the leaf fragments stuck to the fingers are scraped off, the fingers are rubbed together over a small bowl until the resin balls up, and the small pieces are rolled together into one mass. Traditional charas is rubbed from the living plant. The trim-table version applies the same principle to dried material.
Hygiene. A product formed by bare-hand contact cannot meet GACP or GMP hygiene requirements, so it is not suitable for any material supplied to another person. It is described here so the technique can be understood, not recommended.
3. Dry ice sieving: equipment
- Mesh. Use nylon fabric, not pantyhose. Pantyhose tears during agitation and drops plant fragments into the resin. The manual chooses fabric by eye after testing several weaves. A facility should use food- or laboratory-grade monofilament screen in defined µm ratings, because it is repeatable, documentable and cleanable.
- Shaker vessel. A metal canister of about 2 litres. The mesh across the mouth lets gas escape. Never fit a sealed lid.
- Collection surface. A large tray lined with parchment, set on an electric heating pad on a low setting. The pad is the manual’s own refinement, and it solves a real problem: heads straight out of dry ice are hard and will not stick together, and slight warming makes them pressable. The tray should be only slightly warm.
- Dry ice, plus a freezer zip bag and a hammer to crush it.
4. Dry ice sieving: method in summary
Ratio. The manual’s figures (0.25–0.5 kg of dry ice for 30–60 g of plant material) work out to about 8 parts dry ice to 1 part input by weight. At that ratio, a 0.5 kg charge suits about 60 g of input and a 1 kg charge about 120 g.
- Pre-warm the tray and parchment on the pad for about 20 minutes.
- Prepare the input. Break it down and remove the stems as completely as possible, because stems tear mesh. Load it into the vessel.
- Crush the dry ice in the bag (eye protection on), put it on top of the plant material, and secure the mesh with an elastic band so it stays taut.
- Agitate over the parchment, shaking and inverting. A slack mesh lets debris through. The manual gives about three minutes for 30 g of input.
- Let the resin warm up. Cover it with a second tray for about 10 minutes so the very cold heads soften, then press (section 5).
Resin that will not press. Some cultivars give a powder that will not hold together. The manual’s fix is to stir it briefly in cool, clean water, strain it through a fine filter, blot it, form it, and dry it for at least two days. That puts back the water the dry method exists to avoid. At the 80–90% relative humidity common in Thailand, two days of drying in room air is a real mould risk. If you use this step at all, dry the material quickly with active airflow or in a dehumidified room, and send the batch for microbial testing.
5. Pressing and resting
- Resin can be pressed by hand in small amounts (the manual works 1–4 g at a time into the corner of a zip bag) or with a press. Keep the rest of the resin covered between batches.
- Most pressing methods use a little heat.
- Improvised presses such as hydraulic or vehicle jacks produce very large forces they were not designed for. Use a proper die and keep your hands clear.
- Laminators: some small-scale guides adapt a stationery pouch laminator. Laminator pouches are not food-contact material and should not be heated against anything meant for human use. We do not recommend this method.
- Rest. Freshly pressed material is warmer than room temperature and still losing moisture. The manual leaves it in an open bowl for about 30 minutes, then in an airtight jar in the dark for three to four days. A short rest before sealed storage is standard practice and makes sense for moisture equalisation alone. Any further benefits claimed for it are unverified.
6. Variety selection and quality markers
Trichome shape. The manual claims that trichome stalks are either “hooked” or “straight”, that the hooked ones drag leaf and stem fragments through the screen, and that the cleanest resin comes from varieties with mostly straight stalks. We could not find published morphological work that classifies cannabis trichomes this way or links stalk shape to resin purity. Treat it as the author’s observation with a plausible mechanism, not as a selection rule. The more reliable point is that varieties differ a lot in how much clean resin they give. Some suit water separation, some dry sieving, and some give most of their resin as finger resin. The only way to find out is a small test batch from each candidate plant, not looking at the flower.
Quality markers. Colour (pale gold rather than green-brown) and how the resin behaves when heated (pure resin melts and bubbles; resin with plant material in it tends to char) are rough rules of thumb used by hash makers. Neither one confirms purity or cannabinoid content. Only lab testing and a certificate of analysis (COA) do that; see topic 1318.
7. Safety
7.1 Carbon dioxide asphyxiation
Each kilogram of dry ice sublimes into roughly 540 litres of CO2 gas at room temperature. A 0.5 kg run releases about 270 litres, and a 1 kg run about 540 litres, into whatever space you are working in.
Raised CO2 levels cause headache, dizziness, confusion, shortness of breath, loss of consciousness and death. CO2 is dangerous at concentrations that have not lowered oxygen enough to set off an oxygen-deficiency alarm. It is heavier than air and collects at floor level and in low spaces.
- Ventilation is mandatory. Work in a ventilated processing area with active mechanical extraction, or outdoors. Never in a small closed room.
- Never in a cold room, refrigerator, vehicle, lift or any other confined space.
- Never in a sealed container. Gas build-up can burst containers violently. Do not store leftover dry ice in a sealed cooler, and avoid glass containers.
- An operation that runs this process regularly should have a CO2 monitor in the processing area.
7.2 Cryogenic burns
At about −78.5 °C, dry ice freezes skin and tissue quickly.
- Loose-fitting insulated cryogenic gloves that can be pulled off quickly. Never bare hands.
- Safety glasses with side shields. Wear goggles and a face shield when breaking up dry ice, which is exactly what the hammer step does.
- Long trousers and closed-toe shoes. Inspect gloves and tools before use, and never use wet or damaged equipment.
The manual’s warning about the vessel is worth repeating: the metal canister gets extremely cold, and it is the part people grab without thinking. Keep your gloves on for the whole agitation phase.
8. Input material: where we disagree with the source
The manual says spider-mite-infested and mouldy flower can be used, that the mesh filters out mites, and that the author has processed and consumed resin from heavily powdery-mildew-infected material without problems. We do not repeat that advice, and the published evidence is against it.
Gwinn et al. (2023) report that more than 100 fungal species have been found on cannabis inflorescences. Toxigenic genera of concern include Aspergillus (aflatoxins, ochratoxin A), Penicillium (ochratoxin A, citrinin, patulin), Fusarium (trichothecenes, fumonisins, zearalenone) and Mucor. The review describes case reports of pulmonary aspergillosis in cannabis users with HIV, with type 1 diabetes, and with cancer or other immunocompromised states. It also cautions that heating during smoking and vaping should kill spores but “may be insufficient to render all spores nonviable”.
- Gwinn KD, Leung MCK, Stephens AB, Punja ZK. Fungal and mycotoxin contaminants in cannabis and hemp flowers: implications for consumer health and directions for further research. Frontiers in Microbiology 2023;14:1278189. doi:10.3389/fmicb.2023.1278189
A US insurance-claims study by Benedict et al. (2020) found that cannabis users were 3.5 times more likely to have a fungal infection than non-users (adjusted odds ratio 3.5, 95% CI 2.6–4.8). The data come from claims records, and the study did not establish that cannabis use causes the infections.
- Benedict K, Thompson GR, Jackson BR. Cannabis Use and Fungal Infections in a Commercially Insured Population, United States, 2016. Emerging Infectious Diseases 2020;26(6):1308–1310. doi:10.3201/eid2606.191570
The manual’s argument also misses two points. Mycotoxins are stable chemical compounds, so killing the fungus does not remove them, and mechanical separation concentrates them along with the resin. And one person’s experience is no evidence of safety for material that could reach an immunocompromised patient.
Our position: do not process visibly mouldy material. In a GACP or GMP context, contaminated material is rejected and mycotoxin testing is a release criterion. The mesh does keep out mites and their debris, but mite-infested material is still unsuitable for anything supplied to others.
The manual is right that better input gives better output. Sugar leaf, seeded flower and lower-canopy popcorn are the economic case for these methods: low value as flower, but not as resin. Dry the material properly, store it airtight somewhere cool and dark, and break it up only when you process it.
9. Legal status and fit within a licensed operation
- The input. In Thailand, cannabis flower is a controlled herb under the Notification of the Ministry of Public Health Re: Controlled Herbs (Cannabis) B.E. 2568, issued 26 June 2025.
- The output. The Category 5 status of extracts comes from the Narcotics Code and the MoPH Notification on the specification of Category 5 narcotics B.E. 2565 (2022). That notification removed most cannabis plant parts from the list but kept extracts in Category 5 unless they contain no more than 0.2% THC by weight and come from plants grown in Thailand.
- Licensing. The Ministerial Regulation on permission to produce, import, export, sell or possess Category 5 narcotics, specifically cannabis or hemp extracts, B.E. 2569 sets the licensing rules. It is dated 26 March 2026, was published in the Government Gazette, and has been in force since 26 April 2026. Production and import/export licences are limited to Thai juristic persons that are not “foreign” under the Foreign Business Act, government agencies, and the Thai Red Cross Society. Licences issued under the earlier extract regulations remain valid until 31 December 2026. A separate regulation on controlled-herb licences (No. 2) B.E. 2569, in force since 30 April 2026, tightened licensing for the flower trade.
- The open question. It is not clear whether mechanically separated resin (dry sift, trim-table kief or dry-ice sift) legally counts as an “extract”. Confirm this with the Thai FDA or ONCB before any processing.
- Pending law. The Cannabis Control Act is still a draft. The Cabinet approved it on 22 September 2026 and sent it to the House.
Producing, supplying or selling resin requires the relevant licence. Legally grown flower does not make the resin legal.
In a licensed processing facility or a research setting, dry separation has clear attractions. It uses no solvent, so there is no solvent recovery, no residual-solvent testing and none of the fire and explosion risk that shapes hydrocarbon extraction facilities. It uses no water, so there is no drying stage. Capital cost is low. And it turns trim and popcorn into a higher-value fraction for licensed processors. What it lacks is process control. A hand-shaken canister is not a repeatable unit operation, and choosing a screen by eye is not a specification. Scaling up means rated monofilament screens, defined agitation time and temperature, and the controls in section 7 written into an SOP.
10. Checklist
Before: Licence status and the extract question confirmed. Ventilated area, with a CO2 monitor if this is a regular operation. PPE as in section 7.2. Input dry, de-stemmed and free of visible mould. Screen rated and checked for damage. Tray pre-warmed for 20 minutes.
During: Gloves on before you touch the vessel. Eye protection on before the hammer. Mesh kept taut. No sealed containers, ever.
After: Press promptly and keep material covered between batches. Rest it 30 minutes, then store airtight and dark for 3–4 days. If you used the water step, dry with active airflow, not ambient air. Ventilate until all the dry ice has sublimed, and never leave dry ice in a closed room, vehicle or cold store.
Two questions for licensed processors and researchers. Has anyone run dry ice separation and ice water separation on the same input and compared the results? The manual claims a large advantage for dry ice but gives no comparison. If you have done it, post the cultivar, screen rating in µm and yield per 100 g of input. And which regional landraces and Thai/Lao-heritage genetics give clean dry-separated resin? Nobody has evaluated them systematically, and a few real test-batch reports would be worth more than any amount of theory.
Related topics on this forum: 1315, 1313 and 323 (resin separation methods); 1363, 1064 and 245 (harvest, drying and curing); 1318 (lab testing and COAs).
This article is technical processing information for readers operating legally in their jurisdiction. In Thailand, cannabis extracts above 0.2% THC are Category 5 narcotics under the Narcotics Code and the MoPH Category 5 notification B.E. 2565; the B.E. 2569 Ministerial Regulation governs who may be licensed to handle them. Dry ice is a physical hazard that can cause serious injury and asphyxiation; the safety guidance above is a summary and is not a substitute for a proper risk assessment. Nothing here is legal or medical advice.