Paper no.

133
Chemeca 2018
30 September – 3 October 2018, Queenstown, New Zealand

Extraction and fractionation of cannabinoids from Cannabis Sativa

S. Tallon1* T.Moreno1 F.Montanes1 and M.Watson2
1 Callaghan Innovation, Integrated Bioactive Technologies, Industrial Research Limited, 69 Gracefield Road, 5040

Lower Hutt, New Zealand

2 University of Canterbury, Department of Chemical and Process Engineering, Private Bag 4800, Christchurch
8140, New Zealand
*Corresponding author. Email: [email protected]

Abstract: There is a rapidly growing market for therapeutic and medicinal products containing cannabinoids from
Cannabis, driven in part by a growing acceptance of their therapeutic benefits, and by global regulatory change
that allows more widespread use of both recreational and prescribed products. There are a number of known
and established methods for preparation and concentration of extracts containing cannabinoids and other
natural compounds from the plant material; however, in many cases these are still poorly developed and
understood from a large scale industrial manufacturing perspective. Product composition preferences and
preferred formulation options are also still very much in development.

This paper gives a review of the current state of the art for extraction and fractionation of Cannabis Sativa. One
of the emerging methods of choice is extraction using supercritical fluid CO2 as it offers flexibility and selectivity
in terms of the extract composition and is a low temperature solvent-free process that produces a high quality
product. Here we present experimental data on the extraction of two New Zealand grown industrial hemp
cultivars using supercritical CO2 and subcritical (liquid) propane. The effect of decarboxylation of the plant
material on the yields and extract composition is also presented.
Keywords: Cannabis, Cannabidiol, Hemp, Extraction, Supercritical CO2, Propane.

1 Introduction
Cannabis is widely grown in New Zealand and globally to make legal as well as illicit products. Strains of cannabis
that have less than 0.3% levels of the psychoactive cannabinoid THC are commonly referred to as hemp and are
grown legally in New Zealand, under licence, for industrial fibre and edible seed products. Hemp seeds do not
contain cannabinoids and the seed oil, made by cold pressing the seeds or solvent extraction, can be used as a
culinary oil and the de-oiled seed meal has applications as a high protein flour. Hemp seed oil is currently
manufactured and sold in New Zealand, and hemp seed meal products will soon be available for human
consumption.
Products containing cannabinoids are a new growth area in New Zealand. Cannabinoids are known for the
psychoactive high associated with recreational use of the plant, particularly strains high in THC content, but the
various cannabinoids that are present in the plant also have a range of clinical and therapeutic applications. This
includes prescription for pain relief, control of seizures or fits, and treatment of nausea following chemotherapy
treatment [McPartland and Russo].
There are a wide range of identified cannabinoids in cannabis. The ones that are present at the highest
concentration, and the most studied, include tetrahydrocannabinol (THC), cannabidiol (CBD), and to a lesser
extent cannabigerol (CBG) [McPartland and Russo]. These compounds are present in the leaf and flower, and are
in highest concentration in flowering parts of the plant. Total cannabinoid content can exceed 10% of the dry
weight of flowering tops, and can vary in composition between high THC and low THC content.
Cannabinoids in their natural state exist in a carboxylated (acid) form. This form decomposes rapidly at high
temperatures and cannabinoid products are often sold in de-carboxylated form. De-carboxylation can be carried
out in a controlled way during industrial preparation of the product by holding the plant material, or extracts
from the plant, at temperatures typically in the range 120 to 150C for up to an hour [Veress et al, Iffland et al].

1
Decarboxylation will occur at temperatures as low as 80-90C over much longer periods. Temperatures higher
than 150 are generally avoided to minimise further decomposition of the cannabinoids. De-carboxylation will
also occur during smoking or vaping of cannabinoid products.
Products containing only CBD (defined in New Zealand as CBD constituting >98% of total cannabinoids) can be
imported and prescribed in New Zealand by registered practitioners and do not require Ministry of Health
approval (https://www.health.govt.nz/our-work/regulation-health-and-disability-system/medicines-
control/medicinal-cannabis/prescribing-cannabis-based-products). Other cannabinoid products may only be
prescribed on a case by case basis if approved by the Ministry. One exception is the imported product Sativex
(http://www.medsafe.govt.nz/profs/RIss/Sativex.asp ) which contains CBD and THC in approximately 1:1 ratio
and can be used without ministerial approval for treatment of Spasticity. Cannabinoids are classified as class B
drugs in New Zealand and fall within the control of the Misuse of Drugs Act. Manufacture and export of
cannabinoid products cannot currently be carried out in New Zealand.
Globally there are a variety of approaches for regulating cannabis use. Medicinal use is legal in an increasing
number of countries, and personal use in many of these is either decriminalised or laws prohibiting its use are
not enforced.

2 Extraction of cannabinoids
Extraction of a plant material is generally a balance between extraction of as much of the desired component as
possible while minimising co-extraction of other compounds. Compounds in hemp that may extract along with
the cannabinoids include terpenes, which are responsible for much of the aroma often associated with cannabis
extracts, as well as a complex mix of other phenolics, waxes, chlorophylls, and other low polarity compounds.
The degree of extraction of each of these depends on the solvent(s) and extraction conditions used. The solvent
selected should also be non-toxic, and inexpensive. In practice, solvents that are practical to use include ethanol,
carbon dioxide, light hydrocarbons (propane, butane). Hexane and other low polarity organic solvents like
chloroform, acetone, and ethyl acetate will extract cannabinoids but are less desirable due to solvent toxicity
concerns. Cannabinoids can also be extracted into other oils to make an infused oil product but these methods
generally produce a product that contains only low concentrations of cannabinoids and a low yield of the
cannabinoids present in the plant.
Medicinal use of cannabinoids typically requires preparation of a controlled and reproducible product with
known composition. In practice this requires extraction and concentration of the cannabinoids and minimisation
of other plant components. Increasing scale of production and pharmaceutical manufacturing regulations also
create a need for dedicated industrial manufacturing facilities and good manufacturing practice to be employed.
There is a reasonable body of published work on analysis of cannabinoids, including methods for quantitative
extraction of samples for analysis (see for example Citti et al), however there is limited published literature on
industrial extraction of cannabinoids. Existing industrial knowledge is largely ad-hoc and small scale.
Eöry et al describes the effect of sieving milled plant material into different size fractions on extraction of THC
using supercritical CO2. Different size fractions were shown to contain different concentrations of extractable
THC with the highest concentration in the 63 to 125 micron size range. Extraction was slower from larger particle
size fractions, consistent with expectation.
Kitryte et al describe supercritical carbon dioxide, pressurized liquid, and enzyme-assisted extractions of cannabis
plant material containing a mix of leaf, flower fragments and immature seed. CO 2 extraction conditions ranging
from 35 to 70 °C and 100 to 500 bar were investigated. The highest total extraction yield of 8.3% was achieved
at the highest pressure and temperature conditions tested, and using a solvent to feed ratio of approximately
50:1. CBD and (predominantly) CBDa constituted about 28% of the extract, and 93% of the available CBD and
CBDa was extracted under these conditions. The concentration of CBD and CBDa was similar under other
conditions however total yield was lower. Subsequent extraction was carried out by Kitryte et al using
ethanol/water mixtures but was not targeted at extraction of cannabinoids.
The solubility of THC, and other cannabinoids, in supercritical CO2 has been reported by Perrotin-Brunel et al
[Perrotin-Brunel et al, 2010a, Perrotin-Brunel et al, 2010b]. Solubility was reported for temperatures ranging
from 42 to 72 °C, and pressures from 113 to 251 bar. Molar solubilities of the order of 10 -4 were reported, with
solubilities for different cannabinoids generally increasing in the order THC, CBG, CBD, CBN.

2
Rovetto and Aeita describe extraction of four different high THC strains of cannabis sativa using supercritical CO2.
Conditions ranging from 170 to 340 bar at 55 °C were evaluated, and extraction using an ethanol co-solvent with
CO2 was also investigated. Total extract yields of up to 18% were reported, with higher yields at higher pressures.
Extraction efficiency up to 92% of total THC was reported.
Extraction was also more rapid at higher pressure. Packed bed extractions were carried out and showed a
constant extraction rate period in early stages of extraction from which an apparent solubility was calculated.
Values ranging from 0.2% at 170 bar to 1.4% at 340 bar were reported, although the measured solubility values
also varied substantially with plant variety.
Use of ethanol as a co-solvent gave higher total yields, mainly additional extraction of non-cannabinoid
compounds, and more rapid extraction.
The use of ultrasound to assist solvent extraction is described by Agarwal et al. Extractions were carried out with
different solvent ratios of ethanol in water, and for different time periods and ultrasound power levels. Only total
phenolic and total flavonoid yields were reported. Ultrasound was demonstrated to give an improved yield,
including qualitatively for major cannabinoids, under the conditions used.
There are a number of commercial suppliers of extraction equipment that are targeted towards extraction of
cannabis, although many are still relatively small scale. These include ethanol extraction units sold under the
trademarks COLDFINGER (Eden Labs, https://www.edenlabs.com/) and Ethos 4 (Capna Fabrication,
http://capnafabrication.com/the-ethos-4). Cold ethanol extraction is often used to supress co-extraction of
waxes and terpenes. The extraction unit is typically coupled with a vacuum evaporation step to remove ethanol.
Supercritical fluid CO2 extraction is a well established industry, and established equipment manufacturers such
as Natex (https://www.natex.at/), Supercritical Fluid Technologies (http://www.supercriticalfluids.com/), and
Eden Labs ( https://www.edenlabs.com/) market their equipment as suitable for extraction of cannabinoids from
cannabis.
Concentration of cannabinoids, including separation of THC, can be achieved by a number of methods. These
include crude separation methods such as winterisation to remove waxes and adsorption of minor compounds
like chlorophylls, or the use of a selection of solvent systems to selectively separate terpenes before or after
extraction of cannabinoids. More refined methods for concentrating cannabinoids include molecular distillation,
chromatography, and selective adsorption on specialised resins (for example Molecularly Imprinted Polymers).
Concentration methods are not discussed further here.

3 Extraction of Cannabinoids from NZ hemp varieties

We present here experimental data using CO2 and propane for extraction of leaf and flower material from two
hemp varieties that are currently grown in New Zealand. The varieties represented here are primarily grown for
fibre and seed products and do not reflect newer varieties that are being established specifically for high
concentrations of CBD. For commercial sensitivity reasons the varieties are referred to here only as Variety A and
Variety B. The effects of decarboxylation are also explored.
3.1 Methods
Plant materials were received as air dried sample that were already separated into leaf and flower fractions. The
flower buds were generally free of seed.
Supplied plant material was milled as received using a knife mill with a 2mm mesh screen. Milled material was
then either extracted as-is or was first decarboxylated by placing the milled material in a convection oven at
150 C for 20 minutes. These conditions were sufficient to decarboxylate 60-80% of the THC and CBD acid forms
(THCa and CBDa). Longer time or higher temperatures would have enabled higher conversion. A reduction in
water content also occurred during decarboxylation.
The extractions described here were carried out using two different solvents: Supercritical CO2 and subcritical
propane. The method was similar for each solvent, but the scale of equipment varied. In each case, milled plant
material was first loaded into a cylinder with porous sintered steel end plates, and the cylinder was then loaded
into a pressure vessel. Packing density was approximately 230 kg/m3.
Pressurised solvent was pumped through the vessel containing the plant material. After passing through the
plant material, any extracted material dissolved in the solvent was recovered by depressurising and vapourising
the solvent into a separator vessel in which dissolved solids precipitated and were recovered. In the case of CO2

3
extraction the pressure was reduced in two stages and recovered from two separate separators. Results for the
combined extracts are reported here.
After the separators the gas phase solvent was cooled to condense it and recirculated. Circulation of the solvent
continued until no further extracted material was being observed in the separator.
Extraction using propane was carried out in a 2L extraction vessel at a pressure of 40 bar and 50 °C. Extractions
using CO2 were carried out at either 10L scale or 375mL scale, and results reported here were carried out at a
pressure of 300 bar except where noted, and at a temperature of 50 °C.
Extract samples were analysed by HPLC-MS to determine the concentration of the main cannabinoids present,
THC, CBD, CBG, CBN, and their acid forms. CBG, and in particular CBN, were only present at very low levels and
are not discussed further here.
3.2 Extraction results
Figure 1 shows the progress of the extraction with solvent throughput for extraction of type B milled leaf using
CO2 at 300 bar and 50 °C. Extraction of decarboxylated leaf begins to level off after a yield of approximately
15g/kg and use of a 70:1 CO2 to feed mass ratio. The extraction curve for as-supplied leaf continues to increase
above 40 g/kg yield however the majority of the additional extract is water from the leaf which continues to
extract at a rate limited by the solubility of water in CO2. Free water in the as-received extract was observed and
was able to be decanted. The resulting yield, along with additional extract recovered from within the extraction
plant after the run, was 3.9wt%. The decarboxylated leaf contained lower moisture content and no free water
was observed in the extract. Final yield was 2.9wt%. Yields are reported here on a dry feed weight basis.
Figure 2 shows a comparison of extraction curves for extraction of flower heads using CO 2 and propane. Propane
extraction occurs more rapidly than with CO2, with the majority of the extraction occurring using less than 20:1
solvent to feed ratio. The propane extraction for type A flower shows a much higher yield than for type B, which
is a reflection of the higher content of total extractable material in the type A flowers. CO2 extraction was slower,
with extract yield still increasing after use of 120:1 solvent to feed ratio. There is a lag in the early stages of the
CO2 extraction before extract is observed, which is due to the higher solvent residence time in the extraction
plant used for CO2 extraction compared to the propane extraction plant. The rate of extraction after this lag
period however is a function of the solvent properties of CO2.

Figure 1: Extraction curve for CO2 extraction of type B cannabis sativa leaf. Extract mass includes any co-
extracted water.

4
 Figure 2: Extraction curve for solvent extraction of cannabis sativa flowers

3.3 Analysis results

Yield and analysis results are shown in Table 1. Total extract yields ranged from 2.7 to 3.9 wt% for leaf, through
to 5.8 to 8.6 wt% for flower. Total extract yield and cannabinoid yields are higher from flower, due largely to the
higher content present in the flowers compared to the leaf. Cannabinoid yields were higher for the type A variety,
in both the leaf and flower.
Decarboxylation had a mixed effect on the total extracted yields observed here. The profile of cannabinoids in
the extract however shifted towards the decarboxylated form, and the proportion of the total cannabinoids
available in the feed material that was extracted increased. This is consistent with expectation as the
decarboxylated form is more soluble in non-polar solvents than the acid form.
Extraction efficiency of up to 93% was observed for propane and 92% for CO 2.
Table 1: Total extract yield and yield of key cannabinoids
Total Total
extract CBD THC CBDa THCa cannab.
Wt% of feed
Wt% of dry feed weight
cannab.
Type A leaf as received, Propane 2.7 0.22 0.02 0.16 0.00 83
Type A flower as received, Propane 8.2 0.31 0.05 3.43 0.13 93
Type A flower decarboxylated, Propane 8.2 2.59 0.15 0.71 0.01 80
Type A flower as received, CO2 6.3 0.32 0.06 1.65 0.06 49
Type A flower decarboxylated, CO2 8.6 3.17 0.18 0.60 0.01 92
Type B leaf as received, CO2 3.9 0.03 0.01 0.27 0.01 62
Type B leaf decarboxylated, CO2 2.9 0.17 0.02 0.21 0.01 75
Type B flower as received, CO2* 7.6 0.37 0.06 0.59 0.02 58
Type B flower decarboxylated, CO2* 5.8 1.15 0.10 0.20 0.00 88
* Extractions carried out at 500 bar.

5
4 Conclusions
Products containing cannabidiol and other cannabinoids from Cannabis Sativa are a growing global and New
Zealand market opportunity, particularly for medicinal cannabis products. To meet this market new high yielding
strains of cannabis are being developed along with improvement in scale and knowledge of industrial processing
methods.
Results presented here demonstrate the ability of CO2 and/or propane to produce a crude extract from milled
plant material with a high overall recovery of available cannabinoids. Propane has a higher capacity to solubilise
cannabinoids and extraction is more rapid than for CO2. CO2 however has the advantages of being non-flammable
and provides some additional flexibility to adjust the composition of the extract.

Acknowledgement
The authors would like to acknowledge Hikurangi Enterprises for the preparation and supply of some of the raw
materials used in this work.

References
Agarwal C., Máthé K., Hofmann T., Csóka L., 2018. Ultrasound-assisted extraction of cannabinoids from cannabis
sativa l. Optimized by response surface methodology. J. Food. Sci., 83, 700-710.
Citti C., Braghiroli D. Vandelli M.A., Cannazza G., 2018. Pharmaceutical and biomedical analysis of cannabinoids: A
critical review. J. Pharm. Biomed. Anal., 147, 565-579.
Eöry L., Dános B., Veress, T., 2001. Supercritical fluid extraction of tetrahydrocannabinol from marihuana study of
the effect of particle size. Zagadnien Nauk Sadowych 47, 322-327.
Iffland K., Carus M., Grotenhermen F., 2016, European Industrial Hemp Association (EIHA) paper on:
Decarboxylation of Tetrahydrocannabinolic acid (THCA) to active THC, nova-Institut GmbH, Hürth (Germany).
Kitryte V., Bagdonaite D., Rimantas Venskutonis P., 2017. Biorefining of industrial hemp (cannabis sativa l.)
threshing residues into cannabinoid and antioxidant fractions by supercritical carbon dioxide, pressurized
liquid and enzyme-assisted extractions. Food Chem.
McPartland J.M., Russo E.B., 2001. Cannabis and Cannabis Extracts, Journal of Cannabis Therapeutics, 1:3-4, 103-
132.
Perrotin-Brunel H., Kroon M.C., Van Roosmalen M.J.E., Van Spronsen J., Peters, C.J., Witkamp, G.J., 2010a. Solubility
of non-psychoactive cannabinoids in supercritical carbon dioxide and comparison with psychoactive
cannabinoids. J. Supercritical Fluids, 55, 603-608.
Perrotin-Brunel H., Perez P.C., van Roosmalen M.J.E., van Spronsen J., Witkamp G.J., Peters C.J., 2010b. Solubility
of δ9-tetrahydrocannabinol in supercritical carbon dioxide: Experiments and modeling. J. Supercritical Fluids,
52, 6-10.
Rovetto L.J., Aieta N.V., 2017. Supercritical carbon dioxide extraction of cannabinoids from cannabis sativa l. J.
Supercritical Fluids, 129, 16-27.
Veress T., Szanto J.I., Leisztner L., 1990. Determination of cannabinoid acids by high-performance liquid
chromatography of their neutral derivatives formed by thermal decarboxylation. I. Study of the
decarboxylation process in open reactors, J. Chromatogr. A. 520, 339-347.