DIY CO2 Extraction System Based on Carbon Dioxide (CO2)

I’ve been interested in using supercritical compressed carbon dioxide (CO2 SFE) for cannabis extraction for quite some time. I was eager to try oils produced with this extraction technique, but unfortunately, the price of professional SFE equipment is far beyond the reach of the average grower. That’s why I decided to try building my own CO2 SFE system from scratch using readily available parts.

What Is Supercritical Carbon Dioxide?

According to Wikipedia, supercritical carbon dioxide is a liquid state of CO2 that occurs at its critical temperature and pressure. Under standard atmospheric conditions, CO2 is a gas. When frozen, it becomes dry ice. When both temperature and pressure are increased to the critical point, CO2 enters a state with properties of both a gas and a liquid. Specifically, after passing the critical temperature (31°C) and critical pressure (72.9 atm), CO2 expands to fill its container like a gas but has the density of a liquid.

Currently, supercritical CO2 is mainly used for chemical extraction because it is a powerful solvent with low toxicity to humans and the environment. Its relatively low temperature and stability allow CO2 SFE to extract many compounds without damaging or denaturing them. Additionally, the solubility of substances in CO2 SFE depends on pressure, enabling selective extraction by adjusting system pressure.

In simple terms, SFE is just CO2 heated and pressurized. In its supercritical state, CO2 becomes a “super solvent” with high penetration ability, which can be tuned to extract specific compounds by changing temperature, pressure, or mixing with other solvents like ethanol or hexane.

Controlling Pressure in CO2 SFE

Pressure in CO2 SFE can be controlled in several ways. In industry, a booster pump is usually used to achieve high pressure. Alternatively, a cryo-pump can be used, combining CO2 with dry ice and heating them in a closed space.

The first concept of my homemade SFE system, which is relatively affordable for a regular user, uses standard CO2 heated and pressurized to 1500 PSI. By using a dip tube with attachments for both gas and liquid, I was able to eliminate the need for pumps to create pressure.

The system was designed to use stainless steel pipes, 7 inches in diameter, that would fit my X36 lathe. Initially, I planned to use a hydraulic cylinder, but I found that at the extremely low temperatures involved, high-carbon steel quickly degrades. Pressure in the extraction tube would be created by heating, using standard band heaters. Pressure control would be managed by a standard, though expensive, back-pressure regulator valve. A detailed diagram of this CO2 extraction system, created about five years ago, is provided below.

Refining the Design

After reading new patents on SFE CO2 extraction systems, I began to doubt the effectiveness of CO2 as a solvent for cannabis. I also concluded that such systems would consume a lot of solvent and be relatively slow for oil extraction. Later, after hearing rumors that CO2 extraction worsens the taste and effect of the material, I decided to put my plans for a working SFE system on hold.

However, I heard about interesting experiments with dry ice extraction by DIY enthusiasts. Their design could maintain high enough pressure in the tube to prevent rapid heating and melting of the dry ice, successfully extracting pure limonene from orange peels.

This led me to reconsider the use of supercritical pressure for cannabis oil extraction. After much research and consulting with experts, I concluded that the best results come from using subcritical pressure and mixing CO2 with other solvents.

With these new insights, I completely redesigned my setup, making it simpler. The new system consists of an extraction column and a separate atmospheric chamber. The column is a tube with a pressurized section containing dry ice and a section above it loaded with cannabis. A tight screw-on lid is placed on top. The extractant rises from the ice container by regulating heat and pressure between tube sections with a band heater.

By rotating the compact extraction column, you can quickly fill the cannabis section with CO2. Rotating it back returns the material to its end of the tube. This soaking process can be repeated several times to extract more cannabinoids. The extract is then drained into the atmospheric chamber, where pressure manipulation removes residual CO2.

Building the Prototype

I decided to build a prototype, as I unexpectedly gained access to resources and helpers. An old acquaintance offered his lab, previously used for atmospheric testing of aviation equipment. Another friend provided his cryogenic workshop and expertise in SFE extraction. After agreeing to help them with other engineering projects, I got to work.

First SFE CO2 System Diagram

The new, simplified model consists of two sections: the lower section holds dry ice (the CO2 source), and the upper section contains the extraction material in a metal mesh. Band heaters turn the ice into liquid, which is transferred to the adjacent section by simply flipping the column. After thorough soaking, the column is mounted, and the lower part (with the cannabis oil and CO2 mixture) is connected to the atmospheric chamber via a valve connector. Open the valve, release the pressure, and pour the liquid into the chamber for further purification. If the valve fails, the lower section can be removed entirely.

After discussions with a colleague, I calculated that my prototype would operate at 5000 PSI or less. Instead of a stainless steel tube, I used a piece of stainless bar stock, machined so the homemade column was no more than 1 inch in diameter.

Expert Feedback and Commercial Systems

I recently spoke with people working with commercial SFE systems about potential operational issues. Here’s an excerpt from an expert’s letter:

Hi Graywolf! Here are photos of the system I’ve used for over 8 years. It’s designed for optimal extraction results, rated for 10,000 PSI, and made of stainless steel. We place a mesh with material in the column. I’m considering using a fabric pouch to catch more material during extraction. From our experience, SFE extraction works best at high pressure (up to 10,000 PSI) with high column heat (up to 100°C), or at lower pressure (up to 5000 PSI) with heat up to 60°C. High pressure and temperature affect the product’s effect and flavor, but high-pressure extraction yields more oil with less CO2. When purifying oil from CO2 using alcohol, freezing, and filtration, the product loses all flavor. I recommend removing CO2 from the oil without solvents. Also, why use other solvents like hexane or alcohol? Pure CO2 extracts all active cannabis compounds and doesn’t dissolve chlorophyll or terpenes, resulting in a higher concentration of actives. Maybe you want to extract a specific cannabinoid? Next week, I’ll test additional solvents to see if they speed up extraction or separation, but I doubt I’ll find anything new. Other engineers may have suggested it, but I don’t see the logic. Additives only lower oil quality. Also, using slightly moist material (up to 10% moisture) doesn’t affect product quality. Water just evaporates or collects on the oil’s surface and can be removed. No other effects were observed.

Additional Observations from Another Expert

• Subcritical pressure (CO2 as a liquid): 10–12°C, 65–70 bar, process took about 6 hours.
• Supercritical pressure (CO2 as liquid and gas): 40–45°C, 230 bar, process took 4 hours.
• Fats and sugars were removed by freezing, chromatography, and high-speed centrifuge.
• Product purification from CO2 can be done with a chromatography column or by evaporation (simpler and faster).
• Both methods yielded product with 95–99% pure THC (delta-6, -8, and -9 molecules). Delta-9 purity ranged from 78–97% depending on initial parameters.
• No solvents other than CO2 were used, except for small amounts in chromatography analysis.

Why Use SFE CO2 Extraction?

In 2013, after two years of research, I finally moved to active testing of my system, inspired by similar Terpenator projects. I spent 2012 refining information and solving tough problems, and in hindsight, I’m glad I didn’t rush. While working on a cheap, accessible SFE CO2 extraction system, I wondered why anyone would need this method when BHO extraction is already popular. The answer is simple: SFE is most effective for selective extraction. With flexible control over heat and pressure, and by adding co-solvents, SFE can extract specific compounds, leaving unwanted ones behind.

For medical users, it’s important that cannabis oils retain the full spectrum of plant terpenes, especially diterpene cannabinoids with healing properties. SFE extraction can destroy these compounds, but with proper knowledge and experience, medical extracts can be produced much faster than with BHO, without sacrificing quality.

One major issue with the system is that using a single column for both material and CO2 means the extraction process strictly follows all temperature and pressure changes in a single vessel, eliminating flexible control. This can be solved by separating the dry ice/CO2 chamber from the material section, allowing you to heat and pressurize the CO2 before introducing it to the material. However, this complicates the system’s design.

After years of working with SFE, I’ve learned that even the most unexpected solutions can lead to successful extraction. For example, one inventor simply unscrewed the valve from an empty CO2 tank, filled it with dry ice and ground buds, replaced the valve, brought it to critical pressure, then released the gas and poured out the extract. He achieved the best results using additional solvents and subcritical pressure around 5500 PSI. Unfortunately, I couldn’t discuss his results further as he was banned from the forum for spamming.

Owners of commercial SFE systems report successful extraction at low pressures (350–850 PSI), though I haven’t tasted their products to judge quality. Fortunately, two more experts with their own systems joined our work on a home-budget SFE extractor prototype.

Even Simpler DIY CO2 Extraction System

Here’s a practical example of a homemade system provided by our friend “jyndustries.” This version uses minimal resources and, like the previous example, consists of two standard CO2 tanks. The empty tank is filled with material, then connected to a full CO2 tank, transferring the liquid CO2 to the cannabis. Finally, the tank with the material is heated and brought to supercritical pressure. More details can be found on the author’s blog, along with photos of the real prototype.

Testing and Results

In August 2013, we finally tested the CO2 extraction device and produced a small batch of vacuum-purified product. The batch was a solid resin with no aromatic properties but retained the taste of the raw material and had a noticeable effect even in small doses. Since the batch was small, I couldn’t freeze it for further filtration, so I placed it in an atmospheric chamber, evacuated the air, and heated it to 46°C. Streams of evaporating CO2 began to escape from the resin. At 46°C, the material underwent minimal decarboxylation. This cleaning method also works for BHO extracts. However, the final product was a semi-resinous goo, so I plan to try other purification methods. Here are photos of the resin and the resulting goo:

Visiting Eden Labs and Commercial Prototypes

Last week, my colleagues and I toured Eden Labs in Seattle. Patrick, one of their sales consultants and a former student of our BHO courses, invited us to see Eden’s CO2 extractor developments. We met Fritz, a local SFE system developer, who had a professional working prototype and promising improvements. The prototype, made of stainless steel, had been successfully tested in both subcritical and supercritical modes and could also work with gases like butane and propane. It featured a vacuum pump for solvent transfer and several pressure intensifiers. The machine had a simple capsule closure system and a water jacket for temperature control using commercial heaters or antifreeze. The base price for this prototype was $60,000, not including solvent costs—clearly too expensive and professional for home enthusiasts. If you’re serious about extraction as a business, consider buying or building such a system, and don’t forget about a closed-loop CO2 recycling system with vacuum pumps.

Fritz notes that the best quality and aroma extracts are obtained at low chamber pressures—stick to 800–3500 PSI. While some have achieved aromatic products at 9500 PSI, lower pressure is generally better.

Recent Developments

In October 2013, we were lucky again—JYNdustries agreed to provide his extractor model for testing and will demonstrate its use. I continue to stay in touch with Eden Labs developers and work on my own designs. Ultimately, the machines will be compared not only by product quality but also by build and operating costs. We were also invited by Sound CO2 to see their new Apeks extractor. After a quiet period, it’s been an eventful year!