THC Remediation of Hemp Extracts


The redevelopment of delta-9-tetrahydrocannabinol (d9-THC) has become a hot button problem in the US since the Drug Enforcement Agency (DEA) introduced its changes to the definitions of marijuana, marijuana extract, and tetrahydrocannabinols except for Extracts and tetrahydrocannabinols released from a cannabis plant with a dry weight of 0.3% or less d9-THC according to the Law on Controlled Substances. This is because, as a direct consequence, all extracts and tetrahydrocannabinols from a cannabis plant that contain more than 0.3% d9-THC were expressly under the purview of the DEA, including the “hemp extracts” in progress that were created as a result of the extraction process are above the d3 THC limit of 0.3% immediately after creation.

It does not address the legal implications of these changes to the definitions in the “hemp extract” market. Instead, this article focuses on the amount of d9-THC available in the plant material prior to extraction, tracing a “hemp extract” from the point it is no longer compliant to the point it is again compliant stresses the importance of the accuracy of track-n-trace logs in the processing plant. The model developed in support of this article should be academic and track the d9-THC content of a “hemp extract” over the life cycle of a typical CO2-based extract from initial extraction to THC remediation. A 2% equipment loss was used for each step.

First extraction

For this exercise, a common processing scenario of 1000 kg of plant material with 10% by weight cannabidiol (CBD) and 0.3% by weight d9-THC was modeled. This amount can be the total capacity of a facility for the day or the capacity for a single run, depending on the size of the operation. 1000 kg of plant material with 0.3% d9-THC contain 3 kg of d9-THC, which can be extracted, purified and brought onto the market. CO2 has a nominal extraction efficiency of 95%, which means that some cannabinoids remain in the plant material. The same applies to the recovery of the extract from the device. Traces of extract remain in the equipment and this small piece of material, if neglected, could potentially open an operator to legal consequences. The data for the initial extraction are shown in Figure 1.

Image 1: Summary data table for the typical CO2-based extraction of phytocannabinoids

As soon as the starting extract is produced, it does not correspond to the limit of 0.3% d9-THC to be classified as “hemp extract”, and of the 3 kg available d9-THC, the extract contains approx. 2.8 kg, as part of the d9- THC remains in the plant material and some is lost to the equipment.

Dewaxing through overwintering and solvent removal

Dewaxing a typical CO2 extract by overwintering is a common process step. A wax content of 30% by weight was used for this exercise. The wax removal process was ascribed a process efficiency of 98% and it was believed that 100% of the loss was due to the residues recovered from the equipment rather than the waxes removed. Data for overwintering and solvent recovery are shown in Figures 2 and 3.

Figure 2: Summary data table for typical winter storage of a CO2 extract
Figure 3: Summary data table for solvent removal from a CO2 extract

Two things occur during overwintering and solvent removal: non-target components are removed from the extract and there is increased loss through multiple pieces of process equipment. These steps increase the concentration of the d9-THC portion of the extract and create two streams of non-compliant waste.

Decarboxylation & degassing

Most of the cannabinoids in plant material are in acidic form. For this exercise, 90% of the cannabinoids were considered acidic forms. It is known that decarboxylation produces a mass difference of 87.7%, that is, the neutral forms are 12.3% lighter than the acid forms. Heat was modeled as the main driver and a process efficiency of 95% was used for the conversion rate during decarboxylation. To simplify the model, it is assumed that the remaining 5% acidic cannabinoids are more likely to be destroyed than broken down into other compounds, since the proportion of cannabinoids that is destroyed compared to other compounds differs from process to process.

In degassing, low molecular weight components are removed from an extract in order to stabilize it before distillation. Because the molecular constituents of cannabis resin extracts vary from variety to variety and from process to process, it was assumed that the extracts consist of 10% volatile compounds. The model combines the steps of decarboxylation and degassing to account for the full decarboxylation of the available acidic cannabinoids and ignores their weight contribution to the volatiles collected during the degassing. Destroyed cannabinoids result in a loss that can only be accounted for by a full mass balance analysis. Data for decarboxylation and degassing are shown in Figure 4.

Fig. 4: Summary data table for the decarboxylation and degassing of a CO2 extract

As the extract moves along the process line, the d9-THC concentration continues to rise. The decarboxylation further complicates traceability, as both a known mass difference is associated with the process and an unknown mass difference must be calculated and justified.


A two-pass distillation was modeled. Part of the extract was removed with each pass in order to increase the cannabinoid concentration in the material obtained. Average data for distilled “hemp extracts” were used to ensure that the model did not overestimate or underestimate the concentration of cannabinoids in the distillate. The variables used to meet these data constraints were derived experimentally to fit the model to the described scenario and do not indicate actual distillation. The data for the distillation are shown in Figure 5.

Fig. 5: Summary data table for the distillation of a decarboxylated and devolatilized extract

After distillation, it has been shown that the d9-THC concentration has increased by 874% compared to the original concentration in the plant material. About 2.2 kg of the 3 kg of d9-THC available remains in the extract, but 0.8 kg of d9-THC either ended up in a waste stream or went out the door.

Chromatography – THC Remediation Step 1

The chromatography was modeled to remove the d9-THC from the extract. Since there are several variable efficiency systems that can selectively isolate the d9-THC peak from the eluent stream, the model used a cutoff of 5% at the leading and trailing ends of the peak, i.e. 5% assumed to be 5% of the material before the d9-THC peak and 5% of the material after the d9-THC peak is collected along with the d9-THC. Chromatography data are shown in Figure 6.

Figure 6: Summary data table for the removal of d9-THC by means of chromatography

After chromatography, at least three products are made, compliant “hemp extract”, d9-THC extract and non-compliant residues that remain in the equipment. The modeled d9-THC extract contains 2.1 kg of the 3 kg available in the plant material and contains 35 wt% d9-THC, an increase of 1335% from the distillation step and 11664% from the plant material.

CBN creation – THC remediation step 2

For this exercise, the d9-THC extract was converted to cannabinol (CBN) using heat instead of being cyclized into d8-THC. However, a similar model could be used to accommodate this scenario. The rate of conversion of cannabinoids to CBN through heat degradation alone is low. Therefore, the model assumes that half of the available cannabinoids in the d9-THC extract are converted to CBN. It is believed that all of the remaining part of the cannabinoids will convert to some form of breakdown product, rather than any part being destroyed. Data for THC destruction are shown in Figure 7.

Figure 7: Summary table of data for the destruction of THC by degradation to CBN

Only after completion of the CBN cyclization step does the product that was the d9-THC extract become compliant and classifiable as “hemp extract”.

Fig. 8: Summary data table for the adjustment of the d9-THC content of the hemp extract

Loss occurs throughout the process, from the initial extraction to the final d9 THC remediation step. Of the 3 kg of d9-THC available in the plant material, only 2.1 kg were recovered and converted into CBN. 0.9 kg was either lost to the equipment, was destroyed in the process, due to the mass difference associated with the decarboxylation, or was not extracted from the plant material at all. All of these potential areas of product loss should be identified and their risk of diversion fully assessed. Not every waste stream comes at risk of diversion, but some do; When the DEA has a waste management plan, it believes a controlled substance is essential. This would be impossible without a track-n-trace program that follows the d9-THC and identifies the potential risk of diversion. The point is not to create fear, but to shed light on a very real issue that “hemp extract” producers and government regulators need to understand in order to protect themselves and their market from the DEA.



Robert Dunfee