CO2 vs Ethanol Extraction

Two solvents, two very different results

Somewhere around 80% of the commercially available cannabis extracts you'll read about in Australian regulatory filings , whether accessed via the Special Access Scheme or prescribed through an Authorised Prescriber , were produced using one of two dominant extraction methods: supercritical CO2 or ethanol. The difference between them isn't just a manufacturing footnote. It shapes which compounds survive, which are lost, and what the final chemical profile actually looks like.

I've spent a fair bit of time in the primary literature on this, partly because colleagues keep asking me which method is "better." The honest answer is: it depends entirely on what you're trying to preserve.

Solvent behaviour: the basic chemistry

Every extraction method works on the same underlying principle , a solvent is introduced to plant material, it dissolves target compounds based on polarity and solubility, then it's removed, leaving a concentrated extract behind. Where CO2 and ethanol diverge is in how selectively they do that job.

Ethanol is a polar solvent, though it sits somewhere in the middle of the polarity scale. That middle-ground position means it's quite promiscuous chemically speaking: it pulls cannabinoids efficiently, but it also co-extracts chlorophylls, waxes, and water-soluble plant pigments. Refiners typically do a post-extraction winterisation step; chilling the crude extract to precipitate out the waxes, but some heat-sensitive compounds can degrade during that process. Terpenes, in particular, are volatile and prone to evaporative loss during ethanol recovery.

CO2 extraction, in its supercritical form, works differently. At temperatures above 31.1°C and pressures above 73.8 bar, CO2 enters a supercritical state where it has properties of both a gas and a liquid. In this state it becomes a remarkably tuneable solvent: by adjusting temperature and pressure, operators can selectively target different compound classes. Lower-polarity cannabinoids, cannabidiol, tetrahydrocannabinol, cannabigerol; dissolve readily. At different parameter settings, you can preferentially capture terpene fractions.

When the pressure is released, the CO2 simply reverts to a gas and dissipates. No solvent residue. That's the key practical advantage for pharmaceutical-grade production.

What survives, and what doesn't

The compound profile of the finished extract is where this distinction matters most for anyone reading formulation data.

Ethanol, because it extracts broadly, tends to produce a high-yield crude that's rich in total cannabinoids. The cannabinoid acid forms, THCA and CBDA; survive reasonably well in cold ethanol extraction (sometimes called "cold crash" extraction), since lower temperatures slow decarboxylation. The tradeoff is terpene loss, which is often significant. The heat involved in solvent evaporation and recovery, even under vacuum, degrades compounds like myrcene, linalool, and beta-caryophyllene, all volatile monoterpenes and sesquiterpenes with relatively low boiling points.

Supercritical CO2 can be gentler on terpenes when run at lower temperatures, and because terpene fractions can be captured in separate collection vessels (using subcritical parameters at the start of a run), some operators return them to the final extract. Whether that produces a chemically accurate reconstruction of the original plant profile is debated. I'd argue it's a reasonable approximation rather than a true preservation, the terpene ratios in a recombined extract are partly an engineering decision, not purely a botanical one.

Minor cannabinoids; things like THCV, CBG, CBN, and CBC, appear in both extract types, though their relative concentrations differ. The selectivity of CO2 extraction means operators targeting specific minor cannabinoids can adjust parameters accordingly. Ethanol doesn't offer that kind of tunability; you get everything, then you remove what you don't want.

Full-spectrum, broad-spectrum, isolate: how extraction method connects to product type

This is where the extraction chemistry connects to the label taxonomy most people encounter. A full-spectrum extract is defined by retaining the full range of cannabinoids, terpenes, and minor phytochemicals present in the source material. A broad-spectrum extract typically has THC removed post-extraction while retaining other compounds. An isolate is a single purified compound, usually CBD or THC; with everything else removed.

Ethanol extraction, because of its high total-yield profile, is often used as a starting point for isolate production. The crude is refined through chromatography, crystallisation, or distillation to isolate a single compound. CO2 extraction, with its tuneability, is more commonly used when the goal is to preserve a fuller phytochemical profile, though neither method is exclusively tied to either outcome.

The relevance to pharmacology is real. The entourage effect hypothesis, that the combined presence of cannabinoids, terpenes, and other phytochemicals produces different pharmacological outcomes than isolated compounds alone; is an active area of research. Whether extraction method meaningfully affects clinical outcomes in a standardised product is a separate and harder question, and the evidence base is not yet settled. But the chemistry underpinning the hypothesis is at least coherent: different compounds, different interactions with the endocannabinoid system.

GMP, residual solvents, and the TGA framework

Under Australian Good Manufacturing Practice requirements, any cannabis extract included in a product on the ARTG, or supplied via the Special Access Scheme, must meet strict specifications for residual solvent levels. The International Council for Harmonisation Q3C guideline, which the TGA adopts, classifies solvents into three categories by toxicity risk. Ethanol sits in Class 3 (low toxicity, preferred), but residual ethanol levels in a finished product must still be controlled and documented.

CO2 leaves no residual solvent by definition, which simplifies that part of the dossier. But CO2 extraction equipment is capital-intensive, and the technical expertise required to run it consistently is not trivial. In practice, both methods can produce pharmaceutical-grade extracts when validation is done properly; the regulatory obligation is to demonstrate that, regardless of which solvent you use.

I'll add one practical observation from the literature: standardised extracts; where the cannabinoid profile is verified by third-party testing and held consistent batch to batch, are the relevant benchmark for any serious pharmacological analysis. The extraction method is upstream of that standard, not a substitute for it.

Bioavailability and formulation downstream

The extraction method also has indirect effects on bioavailability in the final formulation. Cannabinoids are highly lipophilic, and the presence or absence of natural waxes and phospholipids in a crude extract affects how well downstream formulation chemists can achieve stable emulsification. Ethanol extracts, post-winterisation, are relatively wax-free, which some formulators prefer for nanoemulsion work. CO2 extracts, depending on extraction parameters, may retain more of the native lipid fraction.

How the body handles cannabinoids after ingestion, including the well-documented first-pass metabolism effect that substantially reduces oral bioavailability; is a separate pharmacokinetic question that applies regardless of extraction method. But the formulation vehicle the extract ends up in (oil-based, emulsion, sublingual, inhaled) is partly determined by the starting extract chemistry. It's a chain, not a series of isolated decisions.

Between birdwatching trips I've been working back through some of the older chromatographic analyses of cannabis extracts from the early 2000s, there's a decent run of papers in the Journal of Chromatography A that documented terpene degradation under different solvent recovery conditions, and honestly, some of the data there holds up better than more recent marketing literature I've seen. Primary sources age better than claims.

Sources

• Guidelines for Medicinal Cannabis, Manufacturing Principles; Therapeutic Goods Administration (TGA)
• Bautista JL et al., "Extraction and Analysis of Cannabis", National Center for Biotechnology Information (NCBI/PubMed)
• Rovetto LJ & Aieta NV, "Supercritical carbon dioxide extraction of cannabinoids from Cannabis sativa L.", NCBI/PubMed
• Guide to Medicinal Cannabis Product Manufacturing; Therapeutic Goods Administration (TGA)

, Tom Whitfield, Health librarian, references & fact-checking