Choosing an Extraction Platform

The first decision sets every later one industry practice, not published data

Extraction platform selection is not one decision among many, it is the decision that constrains all the others. A supercritical CO2 run on dried flower gives a crude that is heavy in waxes and lipids, which means winterization is not optional and a rotary evaporator or falling-film unit must be sized for the ethanol that winterization consumes. A cold-ethanol run gives a crude that is cleaner in wax but has already lost most of its monoterpenes to the recovery step, which means a terpene product has to be captured separately or bought in. A cold hydrocarbon run on fresh-frozen material gives the best terpene retention available from a solvent process and gives a crude that often needs little or no winterization, and it costs you a classified electrical room, continuous gas monitoring, a closed-loop vessel with documented pressure ratings, and a licence that in many jurisdictions is harder to obtain than the cultivation licence. A solventless process has no residual-solvent panel to fail and the lightest regulatory footprint of the four, and it caps your yield well below anything a solvent can reach. There is no dominant platform. There is only a platform that matches your input material, your product mix, your capital, and the regulatory environment you can actually operate in.

• Input form matters as much as platform: fresh-frozen, dried-and-cured, and dried trim behave differently in every one of the four.
• Product mix matters more than yield: an isolate house and a live-resin house make opposite platform choices.
• The platform you can be permitted for is the only platform you have.

Sources: Composite: extraction-equipment vendor documentation 2026* · Angelicalist (Van Kush Family) 2026*

CO2: subcritical and supercritical contested

Carbon dioxide has a critical point at 31.1 °C and 73.8 bar. Above both of those it is a supercritical fluid: gas-like diffusivity and viscosity with liquid-like density, and a solvent power that can be tuned continuously by changing pressure and temperature because density is the variable that sets solvating strength. Below the critical point but above the vapour pressure it is a liquid, and liquid CO2 — subcritical operation, typically run in the region of 5 to 25 °C and 55 to 70 bar — is a weaker, more selective solvent that favours the volatile and lower-molecular-weight fraction, which is why subcritical runs are used to pull a terpene-rich first fraction before a supercritical stage takes the cannabinoids. Supercritical cannabinoid extraction is commonly run somewhere in the 45 to 80 °C and 200 to 350 bar band, with higher density pulling more but pulling less selectively: push the density up and the waxes, lipids, chlorophyll degradation products and long-chain plant material come with the cannabinoids. Published academic envelopes and vendor recommendations differ substantially, and a processor tunes their own by running a pressure and temperature grid against assay rather than adopting a number from a paper.

• Density is the control handle: solvating power tracks CO2 density, not pressure or temperature alone.
• Co-solvent (entrainer) use, most often 1 to 10 percent ethanol, increases polarity and raises cannabinoid recovery, at the cost of pulling more polar plant material and reintroducing a residual-solvent question.
• Fractional collection into separate separator vessels at stepped pressures is a real advantage of the platform: you can bank a terpene fraction and a cannabinoid fraction from one run.
• Almost every CO2 crude requires winterization before distillation.
• High capital cost, and the pressure vessels bring ASME-style code, relief-device and periodic-inspection obligations.
• Low solvent-safety burden by comparison with hydrocarbon: CO2 is non-flammable. It is not harmless — it is an asphyxiant heavier than air, so low-level CO2 monitoring and room ventilation are required, and a high-pressure release is a mechanical hazard in its own right.

Contested — caveat. The critical point (31.1 °C, 73.8 bar) is a fixed physical constant and is not in dispute. The operating envelopes are: published supercritical cannabinoid extraction conditions range from roughly 40 to 90 °C and 100 to 500 bar across the literature and vendor documentation, with yields and selectivity reported against different input materials, particle sizes, moisture contents and flow rates. Treat the bands given here as orientation and establish your own envelope against assay.

Sources: Span R 1996* · Rovetto LJ 2017* · Qamar S 2021* · Composite: extraction-equipment vendor documentation 2026*

Ethanol: cold versus warm industry practice, not published data

Ethanol is a polar protic solvent that dissolves cannabinoids readily and also dissolves chlorophyll, waxes, sugars and water-soluble plant material. Temperature is the selectivity handle. Warm or ambient ethanol — anything from room temperature upward — extracts fast and extracts nearly everything, giving a high-yield, dark, chlorophyll-heavy crude that will need both winterization and a hard polish before it distils to a pale product. Cold ethanol, typically chilled to −20 °C to −40 °C and sometimes to −60 °C or lower, is a poorer solvent for chlorophyll and for the plant waxes, so a short cold contact time co-extracts much less of both. The polarity consequence does not disappear: even cold ethanol pulls more polar material than hydrocarbon or subcritical CO2 will, so an ethanol crude is characteristically greener and more viscous than a hydrocarbon crude from the same biomass. The platform wins on throughput and on simplicity: ethanol extraction scales to tonnes per day with centrifuges and jacketed tanks, the equipment is ordinary process equipment, and the solvent is recoverable in high yield. The economics live and die on that recovery percentage, because ethanol is bought by volume and consumed by the litre per kilogram of biomass, so a falling-film evaporator that returns 95 percent or better of the solvent is the difference between a viable and an unviable operation. Ethanol is a Class 3 solvent under ICH Q3C, the lowest-concern class, with a correspondingly generous permitted daily exposure; that is a real advantage when a residual-solvent panel is the gate on saleability.

• Cold ethanol favours cannabinoids over chlorophyll and wax; warm ethanol maximises yield and maximises cleanup.
• Contact time matters as much as temperature — a long cold soak starts behaving like a warm one.
• Ethanol is flammable (flash point about 13 °C for anhydrous ethanol) and its vapour is denser than air, so the room still needs ventilation, ignition-source control and appropriate electrical classification, even though the burden is far below hydrocarbon.
• Water content in the biomass ends up in the ethanol and changes its polarity; wet input silently shifts your selectivity.
• Class 3 under ICH Q3C: low toxic potential, high permitted daily exposure, which is why an ethanol platform has an easier residual-solvent conversation than a hydrocarbon one.

Sources: International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use 2021* · Composite: extraction-equipment vendor documentation 2026* · Armarego WLF 2017*

Hydrocarbon: butane, propane and blends — and the hazard that must be stated plainly

Light hydrocarbons are non-polar and are excellent, selective solvents for cannabinoids and terpenes and poor solvents for chlorophyll and for much of the polar plant material. n-butane is the workhorse; propane, with a much higher vapour pressure, pulls a lighter, more terpene-weighted fraction and is often blended with butane at 20 to 30 percent to lift terpene recovery and drop viscosity. Run cold on fresh-frozen material, a hydrocarbon process gives the best terpene retention of any solvent platform, which is precisely why live resin and live rosin sit in that part of the market. The crude is often light enough in wax that winterization can be reduced or skipped, which shortens the whole downstream chain. Recovery is fast because the solvent boils near or below room temperature. Against that: the safety and licensing burden is the heaviest in the industry, and unlicensed hydrocarbon extraction is the single most common cause of serious injury in this field. That is not a disclaimer, it is the operating fact. Butane vapour is denser than air, so it pools at floor level and travels; its lower explosive limit is in the region of 1.8 percent by volume in air, meaning a small release in an enclosed space reaches an ignitable mixture quickly and invisibly; and the ignition energy required is trivially small — a light switch, a refrigerator compressor, a phone, a static discharge from a synthetic sleeve. The clinical literature on the resulting burn admissions is unambiguous, and the emergency-department series from jurisdictions that liberalised cannabis before they regulated extraction are the reason the modern rules exist. The operator corpus records the same history from inside it: people setting themselves on fire and destroying houses doing indoor butane extractions before the laws were written.

• Closed-loop only. An open blast tube venting hydrocarbon into a room is the configuration that produces the injuries.
• Classified electrical area (commonly specified as Class I Division 1 for the extraction room, Division 2 for adjacent space) with all equipment, wiring and lighting rated for it.
• Continuous lower-explosive-limit gas detection, interlocked to alarm and to shut down ignition-capable equipment, with alarm set points well below the LEL.
• Mechanical ventilation designed for a vapour heavier than air — extraction at floor level, not at ceiling level.
• Peer-reviewed engineering documentation of the vessel and relief devices, pressure-rated components, and documented hydrostatic testing; relief-valve discharge routed outside.
• Bonding and grounding of every vessel and transfer line; static is a sufficient ignition source.
• No solvent-wet material in an oven, no solvent recovery into an open vessel, no warm water bath heated by anything with a flame or an unrated element.
• Jurisdiction-specific licensing on top of all of the above; in many states the hydrocarbon endorsement is a separate and harder permission than the extraction licence itself.

Sources: Bell C 2015* · National Fire Protection Association 2024* · Occupational Safety 2024* · Angelicalist (Van Kush Family) 2026* · Composite: extraction-equipment vendor documentation 2026*

Solventless: ice-water hash and rosin industry practice, not published data

Solventless processing does not dissolve anything. Ice-water extraction agitates cold plant material in ice water so that the trichome heads, which are denser and more brittle when cold, break at the stalk and sink, and then classifies the collected material by size through a stack of mesh bags — the familiar 220, 160, 120, 90, 73, 45 and 25 micron series, where the 45 to 90 micron fractions are usually the most desirable because that is the size band of intact mature trichome heads. The wet hash is then freeze-dried, because air-drying a wet hash at ambient temperature invites microbial growth and oxidative loss. Rosin is the second half of the discipline: heat and pressure applied to flower, to kief or to dried ice-water hash forces the resin out of the trichome without a solvent. Flower rosin is typically pressed in the region of 80 to 100 °C, hash rosin lower, commonly 70 to 90 °C, with hold times measured in tens of seconds to a couple of minutes and pressure applied progressively rather than all at once; the trade rule of thumb is that lower temperature and longer hold gives better terpene retention and a more stable product, and higher temperature gives a higher, faster yield and a runnier, darker result. The operator corpus records the lineage of this method from dry-ice and bubble-bag kief work through the ice-water technique associated with Matt Rize, which used a small camping washing machine for controlled agitation. The regulatory consequence is the real prize: there is no solvent, therefore there is no residual-solvent panel to fail, therefore the capital, the room classification, the gas detection and much of the licensing burden simply do not apply. The cost is a hard yield ceiling — solventless recovers only what the trichome will give up mechanically, so it cannot approach a solvent platform on grams per kilogram, and it demands input material of a quality that solvent platforms do not, because you cannot refine your way out of poor starting material.

• Cold is the whole mechanism on the wash side: warm water softens the trichome and it deforms instead of breaking.
• Freeze-drying the wet hash is not optional at scale — it is the step that protects the terpenes and the microbial result.
• Rosin is a physical squeeze, so everything in the input that is not resin stays behind in the bag rather than being dissolved out with the target.
• Yield ceiling and input-quality dependence are the trade for the clean regulatory position.
• Low but non-zero hazard: hydraulic press pinch points and high surface temperatures, and heavy cold-water handling.

Sources: Angelicalist (Van Kush Family) 2026* · Van Kush Family Research Institute 2026* · Composite: extraction-equipment vendor documentation 2026*

Platform comparison contested industry practice, not published data

The table compresses the decision. Read it as relative ranking within this industry rather than as absolute figures: the yield column in particular is dominated by input material, moisture, particle size and operator skill, and any specific gram-per-kilogram number quoted for a platform without stating the input is meaningless. What is stable across operations is the ordering and the shape of the trade-offs.

Contested — caveat. Yield and terpene-retention rankings vary with input form (fresh-frozen versus dried), cultivar, and the exact operating parameters chosen. Namdar and colleagues showed that extraction method alone changes the recovered cannabinoid and terpenoid profile from the same biomass. Use this table to frame the decision, not to predict your numbers.

Sources: Composite: extraction-equipment vendor documentation 2026* · Angelicalist (Van Kush Family) 2026* · Namdar D 2018*

What this shelf does not cover: conversion chemistry historical / ethnographic

One category of chemistry is deliberately absent from this shelf and from every page on it. Acid-catalysed isomerisation of CBD to Δ8-THC or Δ9-THC, and side-chain homologation to produce THCP, THCJD and the other extended-chain analogues, are preparative conversions that create an intoxicating molecule the plant did not supply in that quantity. They are documented in the chemical literature from Adams and colleagues in 1940 onward and, on the Δ9 side, from the Gaoni and Mechoulam partial synthesis of 1964; the isomerisation in particular is chemically facile and the procedures exist in those and later citations. No reagent, catalyst, solvent, equivalent, concentration, temperature, reaction time, work-up or yield for any of them appears anywhere on this shelf. The molecules themselves are described structurally and pharmacologically on cannabinoids/isomers and cannabinoids/side-chain-series, their legal position on regulatory/thcp-thcjd, and what an unspecified converted product actually contains — the reaction by-products, the unidentified peaks, the reason a COA on a converted product needs reading differently — on safety/converted-cannabinoid-products. Everything on this shelf is separation and purification of what is already there. That distinction is the line, it is drawn on purpose, and it is the only thing this shelf withholds.

Sources: Adams R 1940* · Gaoni Y 1964*