You are on page 1of 10

J.

of Supercritical Fluids 161 (2020) 104850

Contents lists available at ScienceDirect

The Journal of Supercritical Fluids


journal homepage: www.elsevier.com/locate/supflu

Extraction of cannabinoids from hemp (Cannabis sativa L.) using high


pressure solvents: An overview of different processing options
Teresa Moreno a,∗ , Fernando Montanes a , Stephen J. Tallon a , Tina Fenton a , Jerry W. King b
a
Callaghan Innovation, 69 Gracefield Road, Lower Hutt, 5010, New Zealand
b
CFS, 1965 E Spinel Link#7, Fayetteville, AR 72701, USA

h i g h l i g h t s g r a p h i c a l a b s t r a c t

• Efficient cannabinoid extraction from


raw and decarboxylated hemp with
CO2 and propane.
• High CBD extracts (up to 449 mg/g)
obtained from hemp by means of
supercritical CO2 .
• Addition of 5 % ethanol co-solvent
enhances the extraction of cannabi-
noid acid forms.
• Propane more effective than CO2 for
extraction of cannabinoid acid forms.
• Apparent solubility at 300 bar was 11
g/kg CO2 for raw hemp (dry basis).

a r t i c l e i n f o a b s t r a c t

Article history: Hemp varieties of Cannabis Sativa L. contain low levels of 9 -tetrahydrocannabinol (THC) and can be
Received 17 October 2019 used to produce therapeutic extracts rich in cannabidiol (CBD). In this work, extracts containing up
Received in revised form 25 March 2020 to 449 mg/g CBD were obtained from New Zealand industrial hemp varieties by extraction of flower
Accepted 26 March 2020
buds with supercritical CO2 . The composition of the extracts and the influence of different processing
Available online 2 April 2020
parameters (extraction pressure up to 1300 bar, use of ethanol co-solvent, decarboxylation of feed) were
determined. The apparent solubility of the extract in CO2 at different pressures was measured. Extractions
Keywords:
using near-critical propane and dimethyl-ether were also performed. Total extraction yields reached 12.0
Hemp
CBD
wt% with CO2 and 8.2 wt% with propane, whereas total cannabinoid yield ranged from 51 to 100 % with
Decarboxylation CO2 and from 74 to 99 wt% with propane. Addition of 5 % ethanol co-solvent enhances the extraction of
Propane extraction cannabinoid acid forms, as does an increase in extraction pressure.
Dimethyl-ether extraction © 2020 Elsevier B.V. All rights reserved.
CO2 extraction

1. Introduction June 2018. CBD is the dominant cannabinoid found in industrial


hemp, a term that is used for cannabis varieties that have been
The potential therapeutic uses of cannabidiol (CBD), one of the traditionally used for fibre and seed oil production and that con-
cannabinoids present in Cannabis sativa L., have recently gained tain low concentrations of the main psychoactive cannabinoid
considerable attraction, particularly after approval of the anti- (–)-trans-9 -tetrahydrocannabinol (9 -THC or THC). A number
convulsant drug Epidiolex containing 98 % CBD by the FDA in of isomers of THC such as (+)-cis-9 -THC and 8 -THC can be
found in the cannabis plant, but they are generally present in
minuscule amounts and are not considered in this work. In con-
trast to THC, CBD has no psychoactive effects but it does exhibit
∗ Corresponding author.
anti-inflammatory, analgesic, antioxidant, antimicrobial, neuro-
E-mail address: teresa.moreno@callaghaninnovation.govt.nz (T. Moreno).

https://doi.org/10.1016/j.supflu.2020.104850
0896-8446/© 2020 Elsevier B.V. All rights reserved.
2 T. Moreno, F. Montanes, S.J. Tallon et al. / J. of Supercritical Fluids 161 (2020) 104850

ever, have very low cannabinoid concentrations and are not good
candidates for medicinal cannabis extraction.
Early studies with cannabis flowers include a study of the effect
of particle size on supercritical CO2 extraction of THC [19]. Perrotin-
Brunel et al. performed the first comprehensive study on the topic,
analysing the solubility of different cannabinoids in supercritical
CO2 [20–23]. Omar et al. studied the effect of ethanol co-solvent
(0–40 %), pressure (100−250 bar), temperature (35−55 ◦ C) and flow
rate (1−2 mL/min) when extracting 100 mg of non-decarboxylated
cannabis buds using a 1 mL extraction vessel; they concluded that
only the ethanol content and its quadratic effect were significant
in the extraction of neutral cannabinoids; acid forms however
were not quantified [24]. Rovetto and Aieta analysed the effect
of pressure (170−340 bar) and ethanol co-solvent (0–10 %) in the
Fig. 1. Simplified cannabinoid synthetic pathway: decarboxylation, biosynthesis,
extraction of 500 g of different samples of non-decarboxylated, high
and oxidation reactions. THC/THCA cannabis leaves and buds using a 5 L extraction vessel
with three separator stages [25]. They obtained apparent solubili-
ties at three different pressures and studied the cannabinoid profile
protective and anticonvulsant properties [1,2]. In addition to in the different extract fractions. The use of ethanol was found to
CBD and THC, other cannabinoids commonly found in hemp are improve the overall extraction rate, but the results are hard to inter-
cannabigerol (CBG), a relatively unknown non-psychoactive phy- pret since the ethanol extracts were not evaporated nor analysed
tocannabinoid with analgesic, anti-proliferative and antibacterial for cannabinoid concentration. Brighenti et al. compared sev-
activity [3], and cannabinol (CBN), a minor constituent in fresh eral techniques, such as solvent maceration, ultrasound-assisted,
plant material that results of the oxidation of THC [2] and that is microwave-assisted and supercritical CO2 extraction (at 100 bar,
normally considered a chemical marker for poor or lengthy storage 35 ◦ C and with the addition of 20 % ethanol co-solvent) for the
conditions [4,5]. extraction of non-decarboxylated hemp buds from different cul-
These compounds are produced in the stems, leaves and tivars [26]. Ethanol extraction at room temperature was found to
flowers of growing plants in their carboxylated form, i.e. 9 - be superior to other extraction techniques in terms of giving the
tetrahydrocannabinol acid (THCA), cannabidiol acid (CBDA) and highest extraction yield of non-psychoactive cannabinoids. More
cannabigerol acid (CBGA), as indicated in the simplified synthetic recently, Ribeiro Grijó et al. compared two different strategies in
pathway shown in Fig. 1 [6] (note that this figure only shows the extraction of neutral cannabinoids: (1) decarboxylation of the
the main compounds that are relevant to this work). Cannabinoid buds followed by supercritical CO2 extraction (without co-solvent),
acids are thought to have certain interesting properties, particu- and (2) addition of 6 % ethanol co-solvent for extraction of non-
larly CBDA [5]; however, they have not been as extensively studied decarboxylated cannabis buds [27]. Although the overall extraction
as their neutral counterparts and are generally considered as non- yield was increased with the addition of ethanol, the highest con-
active cannabinoid forms that require a decarboxylation process tent of THC and CBD were obtained using the CO2 -based method
[7] to be converted into the more desirable and stable neutral ana- on decarboxylated flower. Acid forms of the cannabinoids were not
logues. These neutral forms of the cannabinoids are believed to offer quantified in this study, but higher yield levels were expected in
better bioactivity since the molecular structure of the acid forms the second strategy extracts due in part to the use of ethanol as
(i.e. the presence of the carboxylic acid group) limits access to the co-solvent.
central nervous system through the blood-brain barrier [3,8]. Neu- In this work, both decarboxylated and non-decarboxylated
tral forms are also expected to have higher solubility in supercritical hemp was extracted using supercritical CO2 at different pressures,
CO2 given their lower polarity. with and without the use of ethanol co-solvent. Liquefied propane
Decarboxylation is a slow process that occurs naturally in the and dimethyl-ether (DME) extractions were also carried out on the
plant over time, but it can be accelerated by exposure to heat, light same hemp material for comparison. Light hydrocarbon extrac-
and oxygen [7,9–11]. One of the main drawbacks of the decarboxy- tion has traditionally been carried out in the recreational cannabis
lation process is the loss of terpenes or terpenoids, i.e. volatile industry to produce the well-known butane hash oil (BHO) extract.
components of the cannabis plant having repeating units of the Light hydrocarbon extraction is traditionally performed with these
isoprene structure and that are responsible for the specific aroma liquid state solvents at very low temperatures (−20 to −30 ◦ C) and
and flavour profile associated with the different strains. These very short extraction times, which preserves the delicate terpene
compounds are not unique to cannabis; however, their synergis- profile associated with the specific hemp or cannabis cultivar in the
tic interaction with cannabinoids, often called the entourage effect, final product [28,29]. However, to the best of our knowledge there
enhances the effectiveness of the cannabinoids found in hemp and are no published works on light hydrocarbon extraction of cannabis.
is a promising new research field [12]. Although terpenes are outside the scope of this work, the extrac-
The non-polar nature of neutral cannabinoids makes them tion data obtained using propane can provide valuable information
amenable to CO2 extraction [13], hence CO2 has been used exten- in this field.
sively by the cannabis industry over the last twenty years [14,15],
and is now a well-established process with a number of dedi-
2. Experimental
cated equipment suppliers in the industry [16]. However, despite
the popularity and widespread acceptance of supercritical CO2 2.1. Materials
extraction among manufacturers and aficionados alike, scientific
publications on the subject are still scarce. Due to governmen- Hemp plant material, predominantly flower buds but containing
tal regulations, some of the published extraction studies were some additional hemp leaf and trimmings, of two different culti-
performed using either threshing residues collected after harvest- vars was supplied by three different New Zealand hemp growers.
ing and/or cleaning of industrial hemp seeds [17] or hemp waste The plant material was milled using a PM-3 pin mill from Mill Pow-
obtained from the hemp fibre industry [18]. These materials, how- der Tech Solutions (Taiwan) with a 2 mm mesh attached and then
T. Moreno, F. Montanes, S.J. Tallon et al. / J. of Supercritical Fluids 161 (2020) 104850 3

stored in a sealed bag at room temperature before its use. Since and 300 bar) and a constant temperature of 40 ◦ C was maintained
the composition of the plant material can change during storage, during the extraction. Each extraction step was run until a constant
it was characterized before each experiment. THC, THCA, deuter- extraction rate was observed and enough data points had been col-
ated THC, and CBD standards were obtained from Cerilliant (United lected to allow for calculation of apparent solubility, at which point
States). CBDA, CBG and CBN standards were obtained from Lipomed the pressure was increased. The apparent solubility, i.e. g extract
(Switzerland). CBGA standard was obtained from SPEX (United per kg of CO2 , was then calculated as the slope of the extraction
States). curve in the stationary period, i.e. yield (g extract per kg of feed) vs
CO2 (>99.8 % purity) was purchased from BOC (New Zealand). CO2 usage (kg CO2 per kg of feed). The starting feed material had a
Ethanol (>99.7 % purity) was purchased from Pacific Sphere (New moisture content of 10.3 %, measured at 120 ◦ C in a Computrac Max
Zealand). Dimethyl-ether and unodorized propane (89.9 % purity) 5000 XL moisture analyser (AMETEK Arizona Instrument, United
were purchased from Aerosol Supplies (Australia). States). Some of this moisture was co-extracted and the extracts
were dried under vacuum using a Büchi (Germany) rotary evapo-
2.2. Decarboxylation rator to determine their moisture content. The apparent solubilities
were then also calculated on a dry basis, i.e. g extract solids per kg
In some experiments, the milled plant material was decarboxy- of CO2 .
lated immediately prior to extraction by placing it on a tray in a
convection oven at 150 ◦ C for specific periods of time. 2.3.3. 10 L scale: pilot plant extractions
10 L extraction trials were carried out in a plant using a Haskel
2.3. Supercritical CO2 extraction (United States) DSF-72 air driven pump to circulate CO2 . A Rheonik
(Germany) Coriolis flow meter was used to measure CO2 flow rate.
Supercritical CO2 extractions were carried out at either 0.5, 1, 2 Two separation vessels (10 L) in series were used for fractionation
or 10 L scale. In all cases, the milled material was first loaded into of the extract. Temperature in the extraction and separation ves-
a basket with porous sintered steel end plates, and the basket was sels was set at 50 ◦ C by means of heating jackets. After passing
then loaded into the extraction vessel. The detailed procedure for through the bed, CO2 containing the dissolved extract was depres-
the individual extractions is given below. surized through a FlowServe Kämmer (United States) control valve
and passed into the first separation vessel where the first (least
2.3.1. 0.5 L scale extractions: ultra high pressure extractions volatile) extract fraction was accumulated. Following this, the CO2
The 0.5 L scale extraction plant consists of two Teledyne ISCO phase was further depressurized through a second valve into the
syringe pumps (United States) operating in tandem to provide second separator where the second (more volatile) extract frac-
a continuous supply of high pressure CO2 , and a temperature- tion was accumulated. Gas phase CO2 was then condensed and
controlled 0.505 L vessel (High Pressure Equipment Company, recirculated. Extract accumulating in the separation vessels was
United States) where the sample is placed in a basket (working manually recovered through a valve periodically during the run to
volume 0.37 L). An Emerson (United States) MicroMotion Coriolis determine the progress of the extraction. After extraction, the plant
flow meter was used to measure CO2 flow rate. The CO2 containing was depressurized, and the residual marc was allowed to degas
the dissolved extract after passing through the bed was depressur- overnight before being unloaded.
ized through a micro metering valve and passed into a separation
vessel where the extract was accumulated. Gas phase CO2 exiting 2.4. Pressurised propane and dimethyl-ether (DME) extraction
from the separator was condensed and recirculated. Extract accu-
mulating in the separation vessel was recovered through a valve The feed material to be extracted was contained in a basket
periodically during the run to determine the progress of the extrac- and placed in a 2 L extraction vessel and either propane or DME
tion. Temperature in the extraction and separation vessel was set was then passed through the vessel in upflow direction using a
at 50 ◦ C by means of heating jackets. Haskel (United States) pump. An Emerson (United States) Micro-
Motion Coriolis flow meter was used to measure solvent flow rate.
2.3.2. 1 L and 2 L scale: CO2 and CO2 +ethanol extractions and Both the extraction and separator vessels were submerged in a con-
apparent solubility measurements trolled temperature water bath held at 55 ◦ C. The extraction was
1 L extraction trials were carried out by initially filling and performed at 40 bar. After passing through the bed, the solvent
then circulating CO2 using a Newport Scientific Ltd (United States) containing the dissolved extract was depressurized through a Bad-
diaphragm pump. An Emerson (United States) MicroMotion Cori- ger Meter (United States) Research Control valve and passed into
olis flow meter was used to measure CO2 flow rate. Temperature a separation vessel where the extract was accumulated. Gas phase
in the extraction vessel was set at 50 ◦ C by means of a heating solvent from the separator was condensed and recirculated. Extract
jacket, whereas the separator was submerged in a controlled tem- accumulating in the separation vessel was recovered through a
perature water bath. After passing through the bed, CO2 containing valve periodically during the run to determine the progress of
the dissolved extract was depressurized through a Badger Meter the extraction. After extraction, the plant was depressurized and
(United States) Research Control valve and passed into a separa- the residual marc was allowed to degas overnight before being
tion vessel where the extract was accumulated and gas phase CO2 unloaded.
from the separator was condensed and recirculated. Extract accu-
mulating in the separation vessel was recovered through a valve 2.5. Cannabinoid analysis by LCMSMS
periodically during the run to determine the progress of the extrac-
tion. When used, ethanol co-solvent was introduced by a Williams Duplicate samples (either plant material or extracted resin), of
(United States) V-Series metering piston pump and its flow rate was about 0.5 g, were accurately weighed into hexane-rinsed silanised
determined by mass loss of the supply tank via gravimetry. Upon glass test tube. Five millilitres of methanol (analytical grade from
completion of the extraction, the extractor was depressurized and Fisher Scientific) were added to the samples, which were then vor-
the residual marc was allowed to degas before being unloaded. texed for 10 min and stored refrigerated overnight. The samples
Apparent solubility measurements were performed in the same were removed from refrigeration and allowed to come to room
plant described above except for use of a 2 L volume extraction ves- temperature, then vortexed for 10 min and then centrifuged for
sel. The extraction pressure was sequentially increased (100, 200 5 min. Each sample was further diluted with methanol, depending
4 T. Moreno, F. Montanes, S.J. Tallon et al. / J. of Supercritical Fluids 161 (2020) 104850

Table 1
Cannabinoid composition and moisture content of different feed materials used in this work (mg/g DB).

THC THCA CBD CBDA CBG CBGA CBN Moisture

Raw 0.74 1.62 3.62 35.9 0.21 0.78 0.01 16.1 %


Cultivar A
Decarboxylated 1.65 0.23 25.2 13.4 0.63 0.29 0.03 4.2 %
Raw 0.50 0.53 2.88 13.7 0.13 0.49 0.00 11.7 %
Cultivar B-1
Decarboxylated 1.07 0.00 15.8 0.54 0.47 0.02 0.05 2.7 %
Raw 0.20 0.00 4.40 0.56 0.14 0.02 0.01 10.3 %
Cultivar B-2
Decarboxylated 0.10 0.00 3.50 0.02 0.11 0.00 0.02 0.84 %

on type of sample (i.e. plant material or extracted resin), to ensure


all the major cannabinoids present in the extract fell within the
calibration curve.
Analysis was carried out in a Sciex (United States) 5500
Q-TrapTM quadrupole mass spectrometer operating in multiple
reaction monitoring (MRM) mode with an electrospray ion-
ization (ESI) source in positive ionization mode. A KinetexTM
Biphenyl (2.1 mm × 50 mm, 2.6 ␮m particle size) column with
SecurityGuardTM ULTRA Biphenyl guard holder-cartridge assembly
from Phenomenex (United States) was used. The column temper-
ature was maintained at 40 ◦ C. Chromatographic separation was
achieved using a multistep gradient consisting of mobile phase A
(0.1 % formic acid in water) and mobile phase B (0.1 % formic acid in
methanol). The gradient program was as follows: 0−0.5 min 30 % A,
0.5–7.5 min 30−22% A, 7.5−8 min 22−5% A, 8–9.4 min 5% A, 9.4–9.5
min 5–30 % A, 9.5−10 min 30 % A. Total analysis time was 10 min
Fig. 2. Cannabis extraction curves at different pressures:  100 bar, 䊉 200 bar, 䊏
with a flow rate of 0.5 mL/min. A calibration curve using certified
300 bar [empty symbols indicate decarboxylated feeds]. Results for each sequential
reference standards for THC, THCA, CBD, CBDA, CBG, CBGA and CBN pressure increase are calculated using an adjusted feed mass allowing for previously
was established, and deuterated THC was used as internal standard. extracted feed mass.

3. Results
and CBGA was 100, 96 and 96 % respectively, with no loss of total
cannabinoids observed after decarboxylation.
3.1. Feed material: cultivars and decarboxylation
The Cultivar B-2 sample contained a significantly lower level of
cannabinoids (5.3 mg/gDB ), comprised mostly of CBD and contain-
Three different feed materials were used throughout this work:
ing no detectable THCA. This sample was decarboxylated at 150
a sample of Cultivar A (undisclosed variety due to commercial ◦ C for 50 min, which led to a 96 and 100 % conversion of CBDA
sensitivity) and two different samples of Cultivar B (Ferimon12);
and CBGA, respectively. A loss of 30 % of total cannabinoids was
their cannabinoid composition is shown in Table 1. There is a wide
observed in the decarboxylated material.
variation in cannabinoid composition between the two cultivars,
with Cultivar A showing the highest total cannabinoid concentra-
tion. There are also clear differences between the two samples of 3.2. CO2 extraction: apparent solubility and extract composition
the same cultivar, Cultivar B. The cannabinoids in the raw (pre- at different pressures
decarboxylation) samples are predominantly in the acid form; CBD
and CBDA are the dominant cannabinoids in all cases. The apparent solubility of the cannabis extract in supercritical
The total cannabinoid concentration in Cultivar A raw material CO2 was studied at 40 ◦ C and increasing pressures using both raw
was 42.9 mg/gDB (i.e. mg per g on a dry basis); of which 8.4 % was and decarboxylated Cultivar B-2 plant material. By sequentially
CBD, and 83.7 % CBDA. After decarboxylation at 150 ◦ C for 20 min, 63 increasing the pressure, the compositional changes occurring dur-
% of the CBDA was converted to CBD, which then represented 60.8 ing the previous steps might have an impact on the extraction at
% of total cannabinoids in the decarboxylated feed. Conversion of the higher pressure end. To minimise this, the extraction in each
THCA and CBGA into their neutral forms was 86 % and 63 %, respec- step was limited to low CO2 :feed ratios that allowed collection of
tively. In both cases, the total CBD (CBD + CBDA) represents over 90 sufficient data points while keeping the remaining feed material
% of the total cannabinoids present in the hemp cultivar. The total as representative as possible. It is therefore important to note that
cannabinoid concentration remained relatively constant before and the apparent solubility reported in this work is a measure of the
after decarboxylation, indicating no major losses. The level of CBN initial extraction rate, which could be indicative of the solubility
found was very low and in most cases below the lowest calibration value under certain conditions.
point of the analysis, which suggests that the material had been Fig. 2 shows the extraction curves obtained in the constant
freshly harvested. The decarboxylation conditions used were based extraction period of each step, i.e. yield (g extract per kg of feed)
on information available online for recreational cannabis grow- vs CO2 usage (kg CO2 per kg of feed). The curves are shown on
ers as well as existing data [7]. These results suggested that the a dry basis for comparison purposes, since the raw material had
decarboxylation process was not complete and therefore the decar- a high moisture content (10.3 %) compared to the decarboxylated
boxylation time was increased for the samples of cultivars B-1 and material (0.84 %). The influence of extraction pressure and therefore
B-2. CO2 density on the apparent solubility is shown in Fig. 3. Solubility
The Cultivar B-1 sample contained a total concentration of increases linearly with CO2 density for all sets of extraction data,
cannabinoids of 18.2 mg/gDB , with CBD and CBDA representing 16 i.e. raw material on a dry and wet basis, as well as decarboxylated
and 75 % of the total, respectively. Decarboxylation for this sample material. The solubility data obtained by Rovetto and Aieta [25] is
was carried out at 150 ◦ C for 60 min. Conversion of THCA, CBDA also shown for comparison.
T. Moreno, F. Montanes, S.J. Tallon et al. / J. of Supercritical Fluids 161 (2020) 104850 5

Table 2
Apparent solubilities obtained at 40 ◦ C and different pressures.

P (bar) CO2 density (kg/m3 )a Solubility (g extract/kg CO2 ) R2

100 628.6 1.3 0.997


Decarboxylated material
200 839.8 6.8 0.999
(dry basis)
300 909.9 9.9 0.998
100 628.6 2.2 0.996
Raw material (wet basis) 200 839.8 9.2 0.997
300 909.9 12.6 0.984
100 628.6 1.7 0.996
Raw material (dry basis) 200 839.8 8.6 0.997
300 909.9 11.0 0.984
170 705.0 2.4 0.988
Rovetto and Aieta [25] b 240 801.9 4.6 0.973
340 876.0 6.7 0.988
a
NIST Chemistry Webbook.
b
Data obtained at 55◦ C.

Table 3
Composition of different fractions obtained in the solubility measurements (mg/g). THCA was not detected in any of the samples.

THC CBD CBDA CBG CBGA CBN

Marc 0.02 0.3 0.43 0.02 0.02 0.001


Extract 1 (100 bar) 1.50 84 0.65 0.95 n/d 0.044
Raw
Extract 2 (200 bar) 1.20 40 1.06 0.88 0.01 0.041
Extract 3 (300 bar) 1.20 17 1.68 0.96 0.02 0.062
Marc 0.02 0.3 0.01 0.04 0.001 0.005
Extract 1 (100 bar) 1.20 95 0.03 0.8 n/d 0.075
Decarboxylated
Extract 2 (200 bar) 1.10 75 0.01 0.76 n/d 0.073
Extract 3 (300 bar) 1.40 36 0.01 1.11 n/d 0.110

decarboxylated marc (9.2 %). This indicates that extraction of


the decarboxylated plant material was more selective towards
cannabinoids, gave a higher total cannabinoid recovery, and
extracted a smaller quantity of non-cannabinoid compounds. The
relative composition of the different fractions in Fig. 4 shows that
CBDA was the main cannabinoid left in the raw marc (54.3 % of
total cannabinoids). Acidic cannabinoids were mostly absent from
the decarboxylated feed, and the decarboxylated marc was com-
prised mostly of CBD (79.8 % of total cannabinoids) followed by CBG
(10.6 % of total cannabinoids). As expected, neither CBDA nor CBGA
were readily extracted by CO2 , particularly at the lower extraction
pressures. THC and CBG extraction was also favoured at the higher
pressure end, while CBD can be extracted at 100 bar. This agrees
with the findings of Perrotin-Brunel et al., who reported that the
solubility order of individual cannabinoids in supercritical CO2 was
Fig. 3. Apparent solubility as a function of CO2 density for different samples: 䊉 raw THC < CBG < CBD across the pressure and temperature range they
material, wet basis, 䊏 raw material, dry basis,  decarboxylated material,  Rovetto studied, and that the solubility of all components increased with
and Aieta [25] (raw material, wet basis). extraction pressure [21]. This also agrees with the lower solubility
obtained by Rovetto and Aieta [25] (Fig. 3) when they used a starting
The extraction yields and the apparent solubilities obtained plant material (of the HashBerry cultivar) containing the less solu-
with decarboxylated plant material were slightly lower than those ble THCA (12.95 %) and THC (5.28 %) as the major cannabinoids. The
obtained with raw material for all the pressures tested (Table 2). cannabinoids in the extracts obtained in the present study at 100
This is contradictory to an expected increase in solubility when bar are comprised almost entirely of CBD (97.8 % of total cannabi-
cannabinoids are decarboxylated; however, in this case the raw noids for decarboxylated flower, 96.4 % for raw material). At 300
material had a relatively low level of cannabinoids in the acid bar, the presence of non-CBD cannabinoids rises to 6.8 % in the
form even before decarboxylation, and the majority of the dis- decarboxylated extract (mainly CBG and THC) and 18.7 % in the
solved compounds in the extract are non-cannabinoids, i.e. lipid raw extract (mainly CBDA, CBG and THC).
matter, waxes, etc. The differences in solubility may be related to The extraction pressure not only affects the composition of the
the co-solvent effect of the higher water content in the raw (non- extract but also its colour and visual appearance. At low pressure
carboxylated) material improving the extraction of higher polarity (100 bar), the extract is a thin oil, rich in CBD and with a light
components from the matrix. The presence of extractable volatile golden colour. As the pressure increases, the viscosity of the extract
components such as terpenes in the non-decarboxylated material thickens, and its colour becomes brown and increasingly darker,
could also contribute to a higher extraction solubility and yield. particularly for the raw plant material extract (Fig. 5).
Nonetheless, the composition of the extracts (Table 3) shows
that the cannabinoid concentration and therefore cannabinoid 3.3. CO2 extraction: cannabinoid fractionation
recovery was significantly higher in the extracts obtained from
the decarboxylated material; and the proportion of non-extracted The differential solubility of each cannabinoid in CO2 can be
cannabinoids was higher in the raw marc (12.9 %) than in the used to attempt a partial fractionation of these compounds during
6 T. Moreno, F. Montanes, S.J. Tallon et al. / J. of Supercritical Fluids 161 (2020) 104850

Fig. 4. Relative composition of different fractions in the CO2 extraction of raw (left) and decarboxylated (right) cannabis.

Fig. 5. Visual comparison of the extracts obtained at different pressures from raw
(left) and decarboxylated (right) cannabis.

Fig. 7. Cannabis extraction curves for 䊏 raw and 䊉 decarboxylated material [empty
symbols indicate ethanol co-solvent period].

3.4. CO2 extraction: effect of ethanol co-solvent

The addition of a polar entrainer such as ethanol to supercritical


CO2 has been shown to increase extraction of cannabinoids [24,27],
particularly for the acidic cannabinoids due to their higher polarity.
To study the effect of ethanol, several experiments with Cultivar A
raw and decarboxylated material were carried out.
Fig. 6. Cannabinoid fractionation −CO2 extraction at 300 bar and two separation The first set of experiments was performed in a 1 L extraction
stages. vessel and the plant material was extracted with CO2 at 300 bar
and 50 ◦ C until a S/F = 25 (solvent to feed, wet basis) ratio had
been reached. At this point, 5% ethanol was added to the CO2 (1:1
ethanol:feed) and the extraction continued with the extract col-
extraction. One way to achieve this is by using a two-step extraction lected separately to the extract obtained with CO2 only. Following
process where the pressure is initially low in order to favour extrac- this stage, CO2 without co-solvent was circulated in order to flush
tion of CBD, and is subsequently increased to obtain the less soluble residual ethanol and extract from the system. The extraction curves
components [21]. An alternative approach is a staged depressuri- for the raw and decarboxylated material (Fig. 7) show an identical
sation process, where two or more separators are used in series at initial extraction rate followed by a slowdown after approximately
decreasing pressures (i.e. decreasing solvent density) [25] to sepa- S/F = 5 (dry basis); this decrease in extraction rate was more acute
rate the different components. This was studied in the present work for the decarboxylated material, leading to a lower extract yield
using decarboxylated Cultivar B-1 plant material and two separa- in the initial CO2 period. The raw material extract obtained in
tion stages, with varying pressure of the first stage, (90-130 bar) to this period contained 17.6 % free water by weight. As discussed in
study its effect on cannabinoid fractionation (Fig. 6), while the sec- Section 3.2, it is possible that the presence of water and other com-
ond stage remained constant at 50 bar. Less soluble cannabinoids ponents (e.g. terpenes) in the raw material could lead to a higher
such as THC and CBG along with heavier compounds such as waxes extraction yield.
and lipids are preferentially collected in the first separator (S1), The yield obtained in the ethanol co-solvent extraction stage
whereas CBD and other more volatile compounds are collected in was also slightly higher for the raw material (2.6 % vs 2.2 %, dry
the second separator (S2). basis), which is likely caused by the higher concentration of CBDA
When a lower pressure was used in S1 (90 bar), the extract from present in the feed. The cannabinoid composition of the extracts
the first separator contained 159 mg/g CBD (89.5 % of the total (mg/g) in the CO2 period (extract A) and the CO2 + ethanol period
cannabinoid concentration) and 18.7 mg/g non-CBD cannabinoids (extract B) is displayed in Table 5. Fig. 8 shows this data expressed as
(THC, CBG and CBN; 10.5 % of the total cannabinoid concentration). g per 100 g of feed (dry basis) to allow better comparison between
When the first stage separation pressure was increased to 130 bar, the fractions as well as to illustrate the recovery of each cannabi-
the ratio of non-CBD cannabinoids in S1 increased to 16.9 %. Under noid. The concentration of neutral cannabinoids (THC, CBD, CBG and
these conditions, the fractionation of THC and CBG between S1 and CBN) in both extracts obtained from the raw material was almost
S2 was slightly more pronounced (Table 4). identical, whereas the addition of ethanol co-solvent led to con-
T. Moreno, F. Montanes, S.J. Tallon et al. / J. of Supercritical Fluids 161 (2020) 104850 7

Table 4
Composition of S1 and S2 under different pressure conditions (CO2 extraction at 300 bar).

Concentration in S1 (mg/g) Concentration in S2 (mg/g)


First separator pressure
CBD THC CBG CBN CBD THC CBG CBN

90 bar 159.1 12.1 5.9 0.7 340.7 16.2 6.7 0.8


110 bar 187.1 15.4 7.8 0.6 309.2 17.4 8.3 0.6
130 bar 94.2 13.4 5.1 0.7 296.1 21.3 8.3 1.1

Fig. 8. CO2 (extract A) followed by CO2 + ethanol (extract B) extraction of cannabis: composition (left) and relative cannabinoid content (right) of the different fractions for
raw (top) and decarboxylated (bottom) plant material.

Table 5
Composition of different fractions obtained in the CO2 + ethanol extraction of cannabis (mg/g DB).

THC THCA CBD CBDA CBG CBGA CBN

Marc 0.01 0.11 0.05 2.37 0.00 0.14 0.00


Raw Extract A 5.49 11.4 24.7 278 1.85 3.01 0.04
Extract B 5.64 24.0 25.3 442 1.79 15.3 0.04
Marc 0.01 0.02 0.13 0.71 0.01 0.05 0.00
Decarboxylated Extract A 23.0 1.78 342 144 8.91 1.10 0.30
Extract B 8.84 5.72 100 260 4.36 8.42 0.16

centrations of THCA, CBDA and CBGA in extract B that were 2.1, times and/or the extraction of compounds that otherwise have lim-
1.6 and 5.1 times higher than those in extract A, respectively. In ited solubility in CO2 [30,31]. Several experiments were performed
the decarboxylated material, the available concentration of neutral on Cultivar A raw and decarboxylated plant material with extrac-
cannabinoids in the feed was much higher, and these were concen- tion pressures up to 1300 bar. It is important to note that these
trated in extract A, while the acidic forms were again concentrated extractions were not performed to completion but rather for a fixed
in extract B. The positive effect of ethanol addition on extraction extraction duration of S/F = 20. Separate experiments were carried
yield for acid cannabinoids was most pronounced for CBGA, fol- out, using fresh feed material, for each pressure condition studied.
lowed by THCA and CBDA. Based on the non-extracted material Cannabinoid concentration in the CO2 extracts for both raw
remaining in the marc, cannabinoid recovery was slightly higher and decarboxylated material is shown in Fig. 9 (left). In the raw
for the decarboxylated plant material (98 % vs 94 %). The highest extracts, all key cannabinoids follow a generally increasing trend,
CBD concentration was obtained in the decarboxylated extract A with higher pressures favouring the extraction. In the decarboxy-
(over 34 % by weight), representing 65.7 % of the total cannabinoid lated extracts, CBD extraction was favoured at low to mid pressures
concentration in this fraction (Table 5). (200−500 bar) whereas the extraction of the other three key
cannabinoids (CBDA, THCA and THC) benefitted from higher extrac-
3.5. CO2 extraction: effect of extraction pressure tion pressures (1000 bar). This agrees with previous results by
Perrotin-Brunel et al. [21]
Extractions performed at pressures above about 700 bar may The extraction yields obtained when using the raw material
be referred to as ultra-high pressure supercritical fluid extraction consistently increase with pressure, reaching 9.8 % at 1300 bar.
(UHPSFE) [30]. The main advantage of the higher pressure is the However, cannabinoid extraction is not complete in these runs, and
increased solubility of certain lipids, leading to shorter extraction ranges between 51 (at 200 bar) and 79 % (at 1300 bar), caused by
8 T. Moreno, F. Montanes, S.J. Tallon et al. / J. of Supercritical Fluids 161 (2020) 104850

Fig. 9. Ultrahigh pressure CO2 extraction of cannabis: composition (left) and relative cannabinoid content (right) of the different fractions for raw (top) and decarboxylated
(bottom) plant material. Lines in the extraction yield (left) added to guide the eye.

poor extraction of the predominant acidic forms (i.e. CBDA) in CO2 .


These results are consistent with what was obtained in the CO2
extraction period as described in Section 3.4. With the decarboxy-
lated material, where water had been mostly removed during the
decarboxylation step, the overall extraction yield peaked at 8.6 % at
500 bar, and then decreased with pressure. A large proportion of the
acidic forms had been transformed into the more easily extractable
neutral forms, and thus the overall cannabinoid yield in these runs
was improved, ranging from 78 (at 1300 bar) to 97 % (at 500 bar),
but there was still a small amount of non-extracted acidic forms.
In terms of CBD concentration in the extract, the highest concen-
tration was achieved with decarboxylated material at the lowest
tested pressure (200 bar), with 45 g CBD per 100 g of extract
(Table 6); however, the overall extraction yield under these con-
ditions was low (6.2 %), so the CBD extract concentration in terms
of CBD g per 100 g of dry feed (as seen in Fig. 9 bottom left) was actu- Fig. 10. Effect of ethanol co-solvent addition in extract composition at different
ally lower at 200 bar than at 500 bar. Some of the extracts in Fig. 9 pressures.
show higher amounts of CBD than what was originally available in
the feed material (expressed in g per 100 g of feed DB), leading to
CO2 while offering excellent solubility for lipophillic compounds
>100 % recoveries for this compound. This indicates that the CBDA
[29,32]. Increased solubilities normally lead to shortened extrac-
present in the feed is likely being partially decarboxylated during
tion times. Fig. 11 shows a comparison of extractions performed
the extraction, generating additional CBD in the process.
using different solvents and materials. Plant material from Culti-
Two additional experiments were performed in this plant
var A was extracted with propane in raw and decarboxylated form.
with addition of 5 % ethanol co-solvent. A comparison between
Unlike what was observed with CO2 , the corresponding extraction
extractions carried out on decarboxylated material in equivalent
curves (dry basis) and yields were identical for raw and decar-
conditions with and without ethanol addition can be seen in Fig. 10,
boxylated material. The solubility of water in propane is low and
showing the composition of the extracts in terms of the four major
very little water is co-extracted in the propane extraction of raw
cannabinoids (CBD, CBDA, THC, THCA). Addition of ethanol signif-
material, supporting the co-solvent effect of water observed in CO2
icantly increased the extraction of CBDA and THCA. Regardless of
extractions that led to higher overall extraction yields.
the presence of co-solvent, increasing the pressure from 200 to 500
Raw plant material from a different cultivar (Cultivar B-1) was
bar led to improved extraction of all major cannabinoids.
also extracted using liquefied propane. This cultivar had a signif-
icantly lower cannabinoid concentration (18.2 mg/g feedDB ) than
3.6. Other pressurised solvents: propane and dimethyl-ether Cultivar A (40.3 mg/g feedDB ) and resulted in a much lower total
extraction extraction yield (4.8 % vs 8.2 %). For comparison, the extraction
curve obtained with CO2 at 300 bar in the 10 L plant using decar-
Near-critical propane or dimethyl-ether (DME) are alternative boxylated Cultivar B-1 material as described in Section 3.3 is also
extraction solvents that require lower operating pressures than shown in Fig. 11 (extraction yield is a combination of extracts S1 and
T. Moreno, F. Montanes, S.J. Tallon et al. / J. of Supercritical Fluids 161 (2020) 104850 9

Table 6
Composition of different extracts obtained from Cultivar A with different CO2 extraction pressures (mg/g DB).

Extraction pressure THC THCA CBD CBDA CBG CBGA CBN

200 bar 6.36 51.7 234 2.09 1.03 0.12 0.00


500 bar 10.1 51.6 263 1.69 2.09 0.11 0.00
Raw 700 bar 12.3 52.5 269 1.66 1.97 0.13 0.00
1000 bar 12.1 51.5 254 1.96 2.23 0.12 0.00
1300 bar 10.4 55.3 265 2.04 2.29 0.10 0.00
200 bar 24.3 0.80 450 67.4 7.97 0.50 0.33
500 bar 20.6 0.67 367 69.9 9.11 0.45 0.36
Decarboxylated
1000 bar 24.0 1.09 338 108 10.7 0.68 0.33
1300 bar 19.6 0.72 353 74.3 7.20 0.57 0.37

Table 7
Composition of different cannabis extracts obtained with near-critical propane and DME (mg/g DB).

THC THCA CBD CBDA CBG CBGA CBN

Raw 6.58 15.5 37.9 419 2.45 6.41 0.06


Cultivar A Propane
Decarboxylated 18.1 1.18 315 86.6 5.76 0.71 0.36
Propane Raw 11.3 12.1 46.1 204 2.27 3.39 0.04
Cultivar B-1
DME Decarboxylated 18.0 0.03 229 5.03 6.94 0.17 0.82

Supercritical CO2 and liquefied propane allow efficient and flexi-


ble extraction of cannabinoids from raw and decarboxylated plant
material. Total gross extract yields ranged from 5.8 to 12.0 wt% with
CO2 and from 4.8 to 8.2 wt% with propane, whereas total cannabi-
noid yield (i.e. the proportion of total cannabinoids available in the
feed material that are extracted) ranged from 51 to 100 % with CO2
and from 74 to 99 wt% with propane. CBD concentration reached
449 mg/g in a CO2 extract obtained at 200 bar. Decarboxylating
the plant material prior to extraction led to a slightly lower total
extraction yield, but it had a positive effect on the cannabinoid yield,
since the decarboxylated form is more soluble in non-polar sol-
vents than the acid form, and on the concentration of cannabinoids
in the extract. Addition of 5 % ethanol co-solvent to supercritical
CO2 enhances the extraction of acid forms, as does an increase in
Fig. 11. Cannabis extraction curves for different cultivars and solvents (× propane, extraction pressure. Propane is also more effective than CO2 for the
Cultivar A, decarboxylated;  propane, Cultivar A, raw; ♦ propane, Cultivar B-1, raw; extraction of the acid forms, and it offers a similar extract compo-
䊐 DME, Cultivar B-1, decarboxylated;  CO2 , Cultivar B-1, decarboxylated).
sition (in terms of cannabinoids extracted) to the extracts obtained
with CO2 + ethanol for non-decarboxylated material. For decar-
S2). The CO2 extraction reached a similar final yield to the propane boxylated material, the propane extract was richer in CBD, with
extraction of the same material, albeit taking considerably longer. CBD representing 73.6 % of total cannabinoids present in the extract
Finally, this material (decarboxylated Cultivar B-1) was also vs 57.2 % in the CO2 + ethanol extract. A partial fractionation of the
extracted using liquefied DME. The yield obtained in this case extracted cannabinoids can be achieved by using two separation
is increased with respect to CO2 and propane due to the higher stages.
polarity of DME, which leads to extraction of a broader range of
compounds. DME is very efficient at extracting cannabinoids, but it Declaration of Competing Interest
is also able to extract larger amounts of waxes and pigments, which
produces heavily coloured, thick extracts that will require further The authors declare no conflict of interest for this work.
purification to obtain a higher concentration of cannabinoids.
When the propane extraction of raw Cultivar A plant material
Acknowledgements
was compared to the equivalent CO2 extraction of the same mate-
rial (as described in Section 3.4), the extraction yield obtained with
The authors wish to acknowledge the New Zealand Institute of
propane (8.2 %) was slightly lower than with CO2 (9.3 %), and the
Environmental Science and Research for their analytical services,
propane extract was more concentrated in the key cannabinoids,
and Rua Bioscience and Kanapu Holdings for supplying some of the
particularly CBD and CBDA (Table 7). Propane appears to be more
raw material.
efficient than CO2 at extracting the acidic cannabinoids. For decar-
boxylated Cultivar A plant material the extraction yield obtained
with propane (8.2 %) was higher than the equivalent yield obtained References
with CO2 (5.9 %), and the individual cannabinoid concentrations in [1] D.C. Hao, X.-J. Gu, P.G. Xiao, 11 - phytochemical and biological research of
the extract were slightly lower. Cannabis pharmaceutical resources, in: D.C. Hao, X.-J. Gu, P.G. Xiao (Eds.),
Medicinal Plants, Woodhead Publishing, 2015, pp. 431–464, ISBN
978-0-08-100085-4.
4. Conclusions [2] A.A. Izzo, F. Borrelli, R. Capasso, V. Di Marzo, R. Mechoulam, Non-psychotropic
plant cannabinoids: new therapeutic opportunities from an ancient herb,
Cannabinoids, and in particular CBD, offer an exciting new Trends Pharmacol. Sci. 30 (2009) 515–527, http://dx.doi.org/10.1016/j.tips.
2009.07.006.
opportunity in the supercritical extraction business in light of [3] E.B. Russo, J. Marcu, Chapter Three - Cannabis pharmacology: the usual
the ongoing regulatory changes and market trends worldwide. suspects and a few promising leads, in: D. Kendall, S.P.H. Alexander (Eds.),
10 T. Moreno, F. Montanes, S.J. Tallon et al. / J. of Supercritical Fluids 161 (2020) 104850

Advances in Pharmacology, Academic Press, 2017, pp. 67–134, ISBN [19] L. Eöry, B. Dános, T. Veress, Supercritical fluid extraction of
1054-3589. tetrahydrocannabinol from marihuana - study of the effect of particle size,
[4] J.M. McPartland, E.B. Russo, Cannabis and Cannabis extracts: greater than the Problems of Forensic Sciences 47 (2001) 322–327.
sum of their parts? J. Cannabis Ther. 1 (2001) 103–132, http://dx.doi.org/10. [20] H. Perrotin-Brunel, Sustainable production of cannabinoids with supercritical
1300/J175v01n03 08. carbon dioxide technologies, in: Mechanical, Maritime and Materials
[5] J.A. Hartsel, J. Eades, B. Hickory, A. Makriyannis, Chapter 53 - Cannabis sativa Engineering, TU Delft, 2011.
and hemp, in: R.C. Gupta (Ed.), Nutraceuticals, Academic Press, Boston, 2016, [21] H. Perrotin-Brunel, M.C. Kroon, M.J.E. Van Roosmalen, J. Van Spronsen, C.J.
pp. 735–754, ISBN 978-0-12-802147-7. Peters, G.J. Witkamp, Solubility of non-psychoactive cannabinoids in
[6] A. Hazekamp, J.T. Fischedick, M.L. Díez, A. Lubbe, R.L. Ruhaak, 3.24 - chemistry supercritical carbon dioxide and comparison with psychoactive cannabinoids,
of Cannabis, in: H.-W. Liu, L. Mander (Eds.), Comprehensive Natural Products J. Supercrit. Fluids 55 (2010) 603–608, http://dx.doi.org/10.1016/j.supflu.
II, Elsevier, Oxford, 2010, pp. 1033–1084, ISBN 978-0-08-045382-8. 2010.09.011.
[7] M. Wang, Y.-H. Wang, B. Avula, M.M. Radwan, A.S. Wanas, J. van Antwerp, J.F. [22] H. Perrotin-Brunel, P.C. Perez, M.J.E. van Roosmalen, J. van Spronsen, G.J.
Parcher, M.A. ElSohly, I.A. Khan, Decarboxylation study of acidic Witkamp, C.J. Peters, Solubility of 9-tetrahydrocannabinol in supercritical
cannabinoids: a novel approach using ultra-high-performance supercritical carbon dioxide: experiments and modeling, J. Supercrit. Fluids 52 (2010)
fluid chromatography/photodiode array-mass spectrometry, Cannabis 6–10, http://dx.doi.org/10.1016/j.supflu.2009.12.001.
Cannabinoid Res. 1 (2016) 262–271, http://dx.doi.org/10.1089/can.2016.0020. [23] H. Perrotin-Brunel, M.J.E. Van Roosmalen, M.C. Kroon, J. Van Spronsen, G.J.
[8] E. Rosenthal, G. Zeman, Terpenes and cannabinoids, in: Beyond Buds, Next Witkamp, C.J. Peters, Solubility of cannabinol in supercritical carbon dioxide, J.
Generation: Marijuana Concentrates and Cannabis Infusions, Quick American Chem. Eng. Data 55 (2010) 3704–3707, http://dx.doi.org/10.1021/je100245n.
Archives, 2018, pp. 14–41, ISBN 9781936807383. [24] J. Omar, M. Olivares, M. Alzaga, N. Etxebarria, Optimisation and
[9] C. Citti, B. Pacchetti, M.A. Vandelli, F. Forni, G. Cannazza, Analysis of characterisation of marihuana extracts obtained by supercritical fluid
cannabinoids in commercial hemp seed oil and decarboxylation kinetics extraction and focused ultrasound extraction and retention time locking
studies of cannabidiolic acid (CBDA), J. Pharm. Biomed. Anal. 149 (2018) GC-MS, J. Sep. Sci. 36 (2013) 1397–1404, http://dx.doi.org/10.1002/jssc.
532–540, http://dx.doi.org/10.1016/j.jpba.2017.11.044. 201201103.
[10] H. Perrotin-Brunel, W. Buijs, J.V. Spronsen, M.J.E.V. Roosmalen, C.J. Peters, R. [25] L.J. Rovetto, N.V. Aieta, Supercritical carbon dioxide extraction of
Verpoorte, G.J. Witkamp, Decarboxylation of 9-tetrahydrocannabinol: cannabinoids from Cannabis sativa L, J. Supercrit. Fluids 129 (2017) 16–27,
kinetics and molecular modeling, J. Mol. Struct. 987 (2011) 67–73, http://dx. http://dx.doi.org/10.1016/j.supflu.2017.03.014.
doi.org/10.1016/j.molstruc.2010.11.061. [26] V. Brighenti, F. Pellati, M. Steinbach, D. Maran, S. Benvenuti, Development of a
[11] T. Veress, J.I. Szanto, L. Leisztner, Determination of cannabinoid acids by new extraction technique and HPLC method for the analysis of
high-performance liquid chromatography of their neutral derivatives formed non-psychoactive cannabinoids in fibre-type Cannabis sativa L. (hemp), J.
by thermal decarboxylation: I. Study of the decarboxylation process in open Pharm. Biomed. Anal. 143 (2017) 228–236, http://dx.doi.org/10.1016/j.jpba.
reactors, J. Chromatogr. A 520 (1990) 339–347, http://dx.doi.org/10.1016/ 2017.05.049.
0021-9673(90)85118-F. [27] D. Ribeiro Grijó, I.A. Vieitez Osorio, L. Cardozo-Filho, Supercritical extraction
[12] E.B. Russo, Taming THC: potential cannabis synergy and strategies using CO2 and ethanol to obtain cannabinoid compounds from
phytocannabinoid-terpenoid entourage effects, Br. J. Pharmacol. 163 (2011) Cannabis hybrid flowers, J. Co2 Util. 30 (2019) 241–248, http://dx.doi.org/10.
1344–1364, http://dx.doi.org/10.1111/j.1476-5381.2011.01238.x. 1016/j.jcou.2018.12.014.
[13] E. Rosenthal, G. Zeman, CO2 extracts, in: Beyond Buds, Next Generation: [28] E. Rosenthal, G. Zeman, BHO: Butane Hash Oil - Shatter, Wax, Sauce and
Marijuana Concentrates and Cannabis Infusions, Quick American Archives, Beyond, in: Beyond Buds, Next Generation: Marijuana Concentrates and
2018, pp. 149–168, ISBN 9781936807383. Cannabis infusions, Quick American Archives, 2018, pp. 87–116, ISBN
[14] I.R. Flockhart, G.W. Wheatley, S. Dring, L. Archer, Method of preparing 9781936807383.
cannabidiol from plant material, 2004, WIPO Patent 2004026802A1. [29] J.W. King, The relationship between cannabis/hemp use in foods and
[15] B. Whittle, C.A. Hill, I.R. Flockhart, D.V. Downs, P. Gibson, G.W. Wheatley, processing methodology, Curr. Opin. Food Sci. (2019), http://dx.doi.org/10.
Extraction of pharmaceutically active components from plant materials, 2008, 1016/j.cofs.2019.04.007.
US Patent 7344736B2. [30] J.W. King, Supercritical fluid extraction at high pressures (& 700 bar):
[16] K. Hawthorne, Leading the Way: Eden Labs Brings Precision and Safety to the theoretical considerations and practical applications, in: III Iberoamerican
New Cannabis Era, Cannabis Brightline Focus Media Group, 2019, pp. 14–17. Conference on Supercritical Fluids, Cartagena de Indias (Colombia), 2013.
[17] V. Kitrytė, D. Bagdonaitė, P. Rimantas Venskutonis, Biorefining of industrial [31] F. Montañés, O.J. Catchpole, S. Tallon, K.A. Mitchell, D. Scott, R.F. Webby,
hemp (Cannabis sativa L.) threshing residues into cannabinoid and Extraction of apple seed oil by supercritical carbon dioxide at pressures up to
antioxidant fractions by supercritical carbon dioxide, pressurized liquid and 1300 bar, J. Supercrit. Fluids 141 (2018) 128–136, http://dx.doi.org/10.1016/j.
enzyme-assisted extractions, Food Chem. (2017), http://dx.doi.org/10.1016/j. supflu.2018.02.002.
foodchem.2017.09.080. [32] O. Catchpole, T. Moreno, F. Montañes, S. Tallon, Perspectives on processing of
[18] T.M. Attard, C. Bainier, M. Reinaud, A. Lanot, S.J. McQueen-Mason, A.J. Hunt, high value lipids using supercritical fluids, J. Supercrit. Fluids 134 (2018)
Utilisation of supercritical fluids for the effective extraction of waxes and 260–268, http://dx.doi.org/10.1016/j.supflu.2017.12.001.
Cannabidiol (CBD) from hemp wastes, Ind. Crops Prod. 112 (2018) 38–46,
http://dx.doi.org/10.1016/j.indcrop.2017.10.045.

You might also like