Many processes have been adopted for extraction and refinement of cannabis. Much like traditional botanical extractions, the aim of these techniques is to provide the desirable qualities of cannabis in a more readily usable form for the delivery method of choice. The wide range of options available has left processors questioning the best method for their extractions. A goal-oriented approach, focused on finding the best fit between process and physicochemical product attributes is critical. Pairing this goal-oriented process development with analytically guided optimization will allow for the development of an ideal process to deliver the best end product.
The heart of any cannabis processing plant is the extraction process. This unit operation will have a major impact on every physicochemical attribute of the cannabis extract being processed and will likely represent a significant portion of the capital outlay associated with a processing plant.
Light hydrocarbons, specifically butane and propane, have been used for cannabis extraction for some time (1). This solvent system provides fast extractions and is fairly selective for cannabinoids and terpenes, meaning it is possible to avoid coextracting other potentially unwanted compounds from the cannabis raw material. Light hydrocarbon extraction systems tend to be middle of the road in terms of capital cost, but depending on local safety requirements it may demand a serious infrastructure investment to create a safe working space.
From the lens of goal-oriented process development, light hydrocarbon extractions have two limitations. The first involves tunability. Compared to other extraction processes like supercritical fluid extraction, there are a smaller number of variables that can be adjusted when producing a light hydrocarbon extract, which means the processor has less impact on the properties of the extract produced. The upshot of this lack of tunability is an ease of use and shallower learning curve for operators. Secondly, while butane and propane are highly volatile compounds, a post-processing step, usually referred to as purging, is required to remove the solvent from the extract. This purge step is often carried out in vacuum ovens and involves heating the extract under deep vacuum conditions to drive off the remaining solvent. Through regulations, each state will dictate the degree to which these solvents must be removed. Colorado, for example, has just reduced its limit from <5000 ppm butane to <1000 ppm (2).
Again, analysis becomes a critical step for driving the development of an optimized process. If simple efficiency and throughput is the aim of our process optimization, it is critical to determine which vacuum pressure and temperature combination results in the fastest removal of butane from our extract to achieve a product that meets the regulatory specifications. To determine this, a residual solvent analysis (RSA) must be performed on samples from the purge process to determine how much butane remains at various times under a given set of purging conditions. If, for example, the RSA result on a sample comes back above the regulatory specification, the process developer knows they must either increase the purge time, increase the temperature of the purge, or pull a deeper vacuum.
It is quite unlikely, though, that the only goal of this process is efficiency. Delivering a high level of terpenes is critical both to the bioactive and organoleptic properties of the extract. This challenge adds a new layer of complexity because the methods for improving purge efficiency (increased time and temperature) will also result in greater loss of terpenes. In this case, it is critical to analyze both the residual solvent content and the terpene content throughout the purge process. An example of the type of data that could be generated for this process development is shown in Figure 1. (See upper right for Figure 1, click to enlarge.) During a purge process at a given temperature and pressure, samples can be pulled and analyzed for terpene and residual solvent content. This step not only gives the process developer a sense of how long it will take to produce material within specification, but also the expected loss of terpenes. Similar charts would then be produced at different temperatures and pressures until the developer finds a set of conditions that meets their goal of minimizing purge time and terpene loss.
Ethanol extraction has become increasingly popular because of its fast extraction, low capital costs, and clear designation as a food-grade solvent. Ethanol is a less selective solvent than butane and propane, but depending on the goals of the extraction this characteristic could be either a positive or negative attribute. If the goal is to strip a wide range of compounds from the cannabis raw material beyond just cannabinoids and terpenes, ethanol will certainly accomplish that. Because of this, it is not uncommon for ethanol extraction to be coupled with post-extraction purification processes to remove some unwanted coextractives. Just like with light hydrocarbon extraction, solvent removal becomes the most time-consuming step in the process. For ethanol extracts, it is more common for rotary or falling film evaporation to be used because ethanol requires significantly more energy to evaporate compared to butane. This means that the vast majority of volatile aroma compounds will be stripped from the extract during solvent removal. If those compounds are not important to the end product, then this potential problem is a non-issue, but if your formulation is meant to leverage the entourage effect of cannabinoids and terpenes acting synergistically to drive a condition specific effect, then this approach may not be the ideal process.
Supercritical fluid carbon dioxide (CO2) extraction has risen to a place of prominence in the cannabis industry. Its steep learning curve and high capital spend mean it is not the ideal choice for some, but the depth of tunability afforded by its solvent system gives the processor a level of control on the extract produced that no other process can provide. This tunability is due to a unique property of CO2 in the supercritical phase. By altering the pressure and temperature conditions of the extraction, the operator can alter the solvent characteristics of the CO2 (4). What this means is that the process developer can design an extraction process to select only the compounds that meet the goals of their end product.
Additionally, CO2 requires no further work for solvent removal because it simply evaporates from the extract as it is collected from the extractor. This means that after a method is developed for the extraction of cannabis with supercritical fluid CO2, there is no additional solvent removal process where volatile compounds can be lost. This gives the operator the ability to capture truly nature identical aroma profiles from their raw material. Figure 2 illustrates this point by displaying two chromatograms back to back. (See upper right for Figure 2, click to enlarge.) The top chromatogram represents the terpene profile of a cannabis raw material before extraction, while the bottom chromatogram (intentionally flipped for ease of viewing) represents an 18-fold dilution of the terpene extract produced by supercritical fluid CO2 extraction. Although only a few of the notable compounds are labeled specifically, the comprehensive mirroring of the two spectra clearly shows how powerful this extraction technology is at producing truly nature identical extracts.
- J.C. Raber et al., The Journal of Toxicological Sciences 40(6), 797-803 (2015), doi:10.2131/jts.40.797.
- Colorado Department of Revenue, Marijuana Enforcement Division, “Recently Adopted Retail Marijuana Permanent Rules - Effective January 1, 2018.” https://www.colorado.gov/pacific/enforcement/med-rules, p. 123.
- C. Sweeney, “Cannabis Extraction and Refinement in Colorado: A 5,280 Ft. View” presented at the Cannabis Science Conference, Portland, Oregon, 2017.
- M. Mukhopadhyay, Natural Extracts Using Supercritical Carbon Dioxide (CRC Press, Boca Raton, Florida, 2015).
How to Cite This Article
C. Sweeney, Cannabis Science and Technology 1(1), 54-57 (2018).