Accepted Manuscript
Title: Commercial Cannabis Consumer Products Part 1:
GC-MS Qualitative Analysis of Cannabis Cannabinoids
Authors: Laura A. Ciolino, Tracy L. Ranieri, Allison M. Taylor
PII:
DOI:
Reference:
S0379-0738(18)30284-6
https://doi.org/10.1016/j.forsciint.2018.05.032
FSI 9327
To appear in:
FSI
Received date:
Revised date:
Accepted date:
16-11-2017
9-5-2018
19-5-2018
Please cite this article as: Laura A.Ciolino, Tracy L.Ranieri, Allison
M.Taylor, Commercial Cannabis Consumer Products Part 1: GCMS Qualitative Analysis of Cannabis Cannabinoids, Forensic Science
International https://doi.org/10.1016/j.forsciint.2018.05.032
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Commercial Cannabis Consumer Products Part 1:
Cannabis Cannabinoids
GC-MS Qualitative Analysis of
SC
RI
PT
Laura A. Ciolinoa,1,*, Tracy L. Ranieria,1, Allison M. Taylora,1
a
Forensic Chemistry Center, Food and Drug Administration, 6751 Steger Drive, Cincinnati, OH,
USA
*Corresponding author. Tel.: 1-513-679-2700.
Email addresses:
U
laura.ciolino@fda.hhs.gov,tracy.ranieri@fda.hhs.gov,allison.taylor@fda.hhs.gov
1
A
N
Tel.: 1-513-679-2700.
D
M
Forensic Chemistry Center, Food and Drug Administration, 6751 Steger Drive, Cincinnati, OH,
USA
Highlights
Cannabinoids GC-MS identification in large variety of commercial cannabis products.

Foods, candies, beverages, topicals, vapes/eliquids, oral supplements.

11 cannabis cannabinoids resolved with 35% silphenylene stationary phase.

Common interferents minimized using selective extraction and sample pretreatment.
CC
EP
TE

Substantial method validation for CBD, 9-THC, CBDA, THCA, and CBN.
A

Abstract
The recent surge in the sale of cannabis-based consumer products in the US includes foods,
candies, beverages, topicals, vapes/eliquids, oral supplements in various forms, recreational
marijuana plants, and plant extracts or preparations. The wide variety of product and sample
types has resulted in a host of new matrix interferences when conducting qualitative testing for
the cannabis cannabinoids such as cannabidiol and d9-tetrahydrocannabinol. A qualitative GCPage 1
SC
RI
PT
MS method is presented in this work, which uses a commercial 35% silphenylene phase to
provide chromatographic resolution for 11 target cannabinoids as their trimethylsilyl derivatives
(CBD, CBDA, d9THC, THCA, CBN, d8THC, CBG, CBGA, CBDV, THCV, and CBC). The
method uses variants of ethanol- and acetonitrile-based extractants to successfully minimize or
eliminate several types of interferents, and also provides protocols to address specific
interferents such as glycerin and lactose. Method validation included spike/recovery for five
cannabinoids of primary interest (spiking level 50 g/g) from a series of edible oils, foods,
beverages, candies, topicals, oral OTC pharmaceuticals, glycerin, and propylene glycol. The
minimum detectable concentration was established as 1.0 g/g. The method was applied to
about sixty diverse commercial products, as well as to recreational marijuana plants, plant
preparations, hempseed oils, and dronabinol capsules.
U
Keywords: Cannabis cannabinoids, Commercial cannabis products, Gas chromatographymass spectrometry (GC-MS), 9-tetrahydrocannabinol, cannabidiol
N
1. Introduction
EP
TE
D
M
A
The availability and use of cannabis-based consumer products in the United States has risen
steeply over the last several years. While cannabis remains a Schedule 1 drug substance
under federal law, many cannabis products are being sold in states which have legalized the
use of recreational marijuana, and also through internet web stores. There is a large variety of
products including processed cannabis plants, concentrated plant extracts, oral supplements,
foods, candies, beverages, topicals, vapes and eliquids. Many of the product labels list
cannabis plant parts or extracts as ingredients and/or make claims related to the product
content of specific cannabis cannabinoids such as cannabidiol (CBD). In addition to the surge
in cannabis-based products, the availability and use of medical marijuana is increasing, with 25
states having legalized its use [1].
A
CC
The cannabinoids are considered the active constituents in cannabis plants, and more than
100 cannabinoids have been identified in the Cannabis sativa L. plant [2]. The acidic,
carboxylated cannabinoids 9-tetrahydrocannabinolic acid-A (THCA-A, hereafter “THCA”), and
cannabidiolic acid (CBDA), are synthesized within the cannabis plant [3,4]. Both THCA and
CBDA undergo decarboxylation to their respective neutral forms 9-tetrahydrocannabinol
(hereafter “d9THC”) and cannabidiol upon heating, drying, and storage of the harvested plant
[3,4]. Recreational cannabis plants (marijuana) are characterized by relatively high THCA
and/or d9THC contents, whereas cannabis hemp varieties are higher in CBDA and/or CBD [35]. Medical varieties have traditionally had medium to high d9THC contents or similar d9THC
and CBD contents, but recently there has been great interest in high CBD varieties for medical
use (6).
Page 2
SC
RI
PT
Gas chromatography-mass spectrometry (GC-MS) has a longstanding history for the qualitative
analysis of the cannabinoids [7]. Derivatization of the cannabinoids prior to analysis is
necessary to avoid decarboxylation of the acidic cannabinoids (THCA, CBDA, etc.) in the
heated injection port [8,9], and trimethylsilyl (TMS) derivatives have been shown to be suitable
for analysis [10,11]. Most GC-MS analysis has been applied to limited sample types including
plants, and plant extracts or preparations. The emergence of cannabis-based consumer
products has introduced the need to develop alternative extraction procedures which extract the
target cannabinoids while addressing a wide variety of potential matrix interferences. Moreover,
these products are formulated across a wide range of cannabinoid concentrations, with some
products containing minute amounts and others with concentrations near 40% w/w. We have
found no published methods which meet these needs.
2. Materials and methods
D
M
A
N
U
The objective for Part 1 of this work was to develop and validate a GC-MS method for the
qualitative analysis of cannabis cannabinoids which has applicability to the current diversity of
cannabis-based samples. The scope includes all varieties of cannabis plants (recreational,
medical, and hemp), plant extracts or preparations, and consumer, medical, or illicit products.
This method will address both acidic and neutral cannabinoids, and provide chromatographic
resolution and mass spectral identification of the following eleven cannabis cannabinoids for
which standards are commercially available: CBD, CBDA, d9THC, THCA, cannabinol (CBN),
 8- tetrahydrocannabinol (d8THC), cannabigerol (CBG), cannabigerolic acid (CBGA),
cannabidivarin (CBDV), tetrahydrocannabivarin (THCV), and cannabichromene (CBC). A
further aim is to provide guidance to the practicing analyst for application of this method. An
HPLC-DAD method for quantitative analysis is addressed in Part 2 [12].
TE
2.1 Standards, chemicals, and reagents
A
CC
EP
Cannabis cannabinoids were obtained as pure substances (> 98%) or concentrated stock
solutions (10 mg/ml) in methanol or methylacetate from Cayman, Sigma, or Fluka. For the pure
substances, stock solutions were prepared in the range 0.5 – 2.0 mg/ml using methanol or
acetonitrile. Certified 1.0 mg/ml cannabinoid stock solutions from Cerilliant Corporation were
also used. For analysis, aliquots (10 – 200 l) of the 1.0 mg/ml stock standards were taken for
derivatization in the same manner as the samples (see Section 2.3). Standards of the cannabis
terpenes, fatty acids, glycerides, glycerin, propylene glycol, acetaminophen, ibuprofen,
phenibut, melatonin, and caffeine were obtained from reputable sources.
Ethanol (200 or 190 proof, USP/ACS grade) was obtained from Pharmco. Acetonitrile (HPLC
grade), and pyridine (certified ACS) were obtained from Fisher. Deionized water was obtained
from a Millipore filtration system fed by a service deionized water source. Lactase caplets were
Kroger brand Dairy Digestive Supplement (9000 FCC lactase units per caplet). BSTFA reagent
(99:1 N,O-(bistrimethylsilyl) trifluoroacetamide: trimethylchlorosilane) was obtained from
Supelco. Glycerin (certified ACS) and propylene glycol (USP/FCC grade) were obtained from
Fisher Scientific.
Page 3
2.2 Cannabis plant materials, commercial products, hempseed oils, hash oils, kief, and
dronabinol capsules
SC
RI
PT
Five THCA-rich varieties of cannabis plant materials were obtained from retail markets in
Colorado in 2016. The plant materials consisted of dried buds which were ground in a mortar
prior to extraction. Approximately 60 commercial cannabis consumer products including oral
supplements, foods, candies, beverages, vapes/eliquids, and topicals were obtained between
2015 and 2017 via purchases from internet web stores. Two different Canada-sourced
hempseed oils were obtained in 2014 or 2017. Generic dronabinol capsules (2.5 mg and 5.0
mg strengths) were obtained from a local pharmacy.
TE
2.3 Sample preparation
D
M
A
N
U
Two crude hash oils and one fine kief material were prepared in the authors’ laboratory from
one of the commercial THCA-rich bud materials. For the hash oils, 3.1 g of ground bud material
was extracted with 30 ml ethanol (hash oil 1) or isopropyl alcohol (hash oil 2). The entire extract
was filtered through a 0.45 m nylon membrane filter, after which the solvent was evaporated in
a heated water bath (45 – 50oC) under a nitrogen purge. Both hash oils were dark red amber,
with yields of 0.71 g (hash oil 1) or 0.68 g (hash oil 2). For the fine kief, 3.1 g of ground bud
material, 50 ml water, and 117 g ice were stirred in a beaker for 45 minutes. The preparation
was successively filtered through a series of five metal sieves, with mesh sizes of 10, 20, 60,
80, and 100. The materials collected on the sieves were allowed to air dry, with a yield of 0.13 g
fine kief material obtained from the 60 mesh sieve.
A
CC
EP
Sample weights and extraction volumes varied according to cannabinoid content, with sample
weights in the range 0.03 – 3.0 g, and extraction volumes in the range 0.25 – 30 ml. All
extactions were carried out using variations of ethanol (95% aqueous, or pure) or acetonitrile
(83 – 91% aqueous, or pure). Cannabis plant bud materials were extracted with pure ethanol.
Hash oils, kief, hempseed oils, dronabinol capsules, and the majority of the commercial oral
supplements, vapes/eliquids, and topicals were extracted with 95% ethanol. For highly aqueous
commercial products such as a prepared coffee beverage, and a given oral supplement, the
extractant was 100% acetonitrile. Products in the cannabinoids spiking study (see Section 2.7)
were extracted with the solvents specified in Table 1. Samples were shaken or mixed as
needed prior to sampling. A portion of sample was weighed into a glass scintillation vial or
flask, and the extraction solvent was added. The vial or flask was capped, thoroughly vortexed,
and sonicated for 15 minutes. A portion of extract was withdrawn using a polypropylene
syringe, then filtered through a 0.45 micron nylon membrane disc filter. A portion of filtrate (25 1000 l) was transferred to a GC vial for derivatization. The solvent was evaporated under a
stream of dry air on a Pierce Reactitherm block (nominal block temperature 80 oC). 200 l
Page 4
pyridine and 200 l BSTFA reagent were added to the vial, the vial was capped, mixed, and
incubated for 30 min (80oC).
U
SC
RI
PT
Variations were applied to specific matrix types as follows. For semisolid fats/oils matrices such
as butter, margarine, virgin coconut oil, chocolate bars, and various nonpolar topical ointments,
creams, or balms, the sample vessel was warmed on a hot plate after addition of the extraction
solvent to melt the matrix. The sample was then vortexed to disperse the matrix, continuing
with the general procedure. For specific high sugar and carbohydrate matrices including hard
candies, fruit preserves, and honey, the aqueous portion of the extractant was added first,
followed by warming of the sample vessel on a hot plate to cause the sugars and carbohydrates
to fully dissolve. The acetonitrile portion of the extractant was then added as pure solvent to
give a final proportion of 83 – 91% acetonitrile. The sample was then vortexed to disperse the
sample, continuing with the general procedure. See also Section 2.7.
N
2.5 CBD Conversion Experiments
EP
TE
D
M
A
Replicate 1 g portions of pure glycerin or pure propylene glycol were spiked with 65 l portions
of a 4.57 mg/ml CBD spiking solution, producing a final CBD spiking level of 300 g/g.
Separate spiked preparations were extracted with 5.0 ml 95% ethanol, pure ethanol, or pure
acetonitrile, and the extracts were filtered. For each extraction solvent, two different size filtrate
portions (200 l or 1000 l) were taken for TMS derivatization according to the general
procedure. The 200 l aliquot preparations corresponded to conditions in which there was a
molar excess of the BSTFA derivatizing agent with respect to glycerin or propylene glycol. The
1000 l aliquot preparations corresponded to conditions in which there was a molar excess of
glycerin or propylene glycol with respect to the BSTFA derivatizing agent. All sample
preparations were analyzed using the general GC-MS conditions.
2.6 Lactase pretreatment for lactose interferences
A
CC
Certain products containing lactose (yogurt, cream cheese, butter, milk containing beverage)
were subjected to reaction with lactase prior to extraction. A lactase reagent was prepared by
grinding and dissolving/dispersing 1 lactase caplet (9000 FCC lactase units) in 15 ml of pH 6.6
phosphate buffer (100 mM, prepared from potassium phosphate monobasic with final pH
adjustment). The product sample (1 g) was weighed into a glass scintillation vial, and 1 ml of
lactase reagent was added. The vial was capped and sonicated, then incubated for 90 minutes
in a water bath (nominal temperature 37 oC). After incubation, 5 ml acetonitrile was added, and
the extraction and derivation steps were carried out according to the general procedures.
2.7 Cannabinoid spike/recovery experiments from negative product matrices
Page 5
SC
RI
PT
Thirty-eight cannabinoid-negative products including edible oils, foods and beverages, topicals,
and oral OTC medicines were obtained from local stores for use in spike/recovery experiments
with commercial certified cannabinoid standard solutions. These local purchases included a
Canadian hempseed oil. In addition, a second Canadian hempseed oil (an import detention),
one pharmaceutical cream base containing no active ingredients (obtained from a local
pharmacy), pure glycerin, and pure propylene glycol were included in the study. See also Table
1.
2.8 GC-MS Analysis
TE
D
M
A
N
U
For all products, spiking experiments were conducted at a nominal level of 50 g per g product.
Two separate spiking mixes were prepared from the 1.0 mg/ml commercial cannabinoid stock
solutions. An acidic cannabinoid spiking mix (500 g/ml in each cannabinoid) was prepared by
mixing equal volumes of the 1.0 mg/ml CBDA and THCA stocks. A neutral cannabinoid spiking
mix (333 g/ml in each cannabinoid) was prepared by mixing equal volumes of the 1.0 mg/ml
CBD, d9THC, and CBN stocks. For each product, three 1 g portions (nominal) of product were
weighed into glass scintillation vials. 100 l of the acidic mix was spiked into one of the
portions, 150 l of the neutral mix was spiked into the second portion, and the third portion was
left unspiked. After spiking, the sample vial was thoroughly vortexed to mix the spike well into
the product matrix. For solid products, the solvent was allowed to completely evaporate from
the product matrix under a bench top vacuum snorkel, after which the spiked product was
extracted. For liquid and semisolid products, the extraction was performed without evaporating
the spiking solvent. The extraction solvent varied according to the product matrix (see table).
The amount of extraction solvent added was 5 ml except for the 91% acetonitrile (aq) and 83%
acetonitrile (aq) extractants for which the total extraction volumes were 5.5 or 6.0 ml,
respectively due to the additional volume from water. After addition of the solvent, extraction
and derivatization were carried out according to the general procedure.
A
CC
EP
All GC-MS analysis was carried out using an Agilent 5973N 70 eV EI GC-MS system with a
6890N MS detector. The column was a 30 meter Restek Rxi-35Sil MS (35% silphenylene) with
0.25 mm ID and 0.25 m film thickness. Injection volume was 1 l splitless with an injection
port temperature of 250oC. The carrier gas was helium with a flow rate of 1.0 ml/min (constant
flow mode). Oven program was as follows: initial temperature 60oC with 0.5 minute hold, first
ramp 25oC/min to 220oC, hold for 10 min, second ramp 10oC/min to 300oC, with a final hold
time of 15 min (run time 39.9 min). Transfer line temperature was 280oC. Solvent delay was 5
min, and MS acquisition used full scan mode with mass range 40 – 600 amu.
3. Results and discussion
Page 6
3.1 GC-MS stationary phase selection and target compounds mass spectra summary
SC
RI
PT
Figure 1 shows structures for the 11 cannabinoids included in this study. The chemical
structures of the cannabinoids generally contain aromatic, alkyl, and alcohol moieties, with the
acidic cannabinoids (CBDA, THCA, CBGA) also containing carboxylic acid groups. Given the
presence of aromatic group(s) in the cannabinoid structures, the proportion of phenyl groups in
mixed dimethylpolysiloxane-silphenylene or mixed dimethylpolysiloxane-dimethyldiphenyl
stationary phases may be expected to impact the chromatographic separation.
A
N
U
In development work, three differerent GC stationary phases were evaluated: a 35%
silphenylene phase (Restek Rxi-35Sil MS); a 5% silphenylene phase (Agilent DG5MS); and a
5% diphenyl phase (Agilent HP5MS). All three columns had 0.25 mm internal diameters and
0.25 m phase thicknesses, and the same GC-MS acquisition parameters and oven program
(see Section 2.6) were used in the comparison. While all three stationary phases provided
resolution, the 35% silphenylene phase was selected as it provided a wider retention window for
the 11 cannabinoids (10 – 23 min) vs. the other phases (13 – 23 min). Figure 2 shows a
chromatogram for a standard mix of the 11 cannabinoids with the 35% silphenylene phase.
A
CC
EP
TE
D
M
Table 2 lists retention time and mass spectral data for the 11 cannabinoid TMS derivatives.
Detailed interpretation of the mass spectra and fragmentation pathways have been reported
elsewhere for a few of the cannabinoids [13,14], including some data for the TMS derivatives.
No references which provided a concise listing of mass spectral data for the current 11
cannabinoid TMS derivatives was found, and hence a summary was provided in the table. All
of the cannabinoids are converted to the fully trimethylsilyated compounds in the derivatization
reaction. Molecular ions (M+) were observed for all 11 cannabinoid derivatives, with major
abundances (50 – 70% of base ion) for the THCV, d8THC, and d9THC compounds. The
abundances of the molecular ions ranged from 1 – 10% for the other cannabinoid compounds.
The trimethylsilyl ion (m/z 73) was the base ion for seven of the cannabinoid derivatives (see
table) and the M-15 (loss of methyl) ion was the base ion for the CBN and THCA derivatives.
For the CBGA derivative, differences in spectra acquired at different times were observed such
that the base ion varied between the trimethylsilyl (m/z 73) and M-15 ions. The base ions for the
CBC and d8THC derivatives (M-83) corresponded to larger fragment losses involving the
terpene portion of the molecules [13,14]. The last column in the table lists out additional
prominent ions (abundances > 5% of the base ion), except for THCV and d8THC for which the
listed ions include abundances > 3% of base ion. Major abundances (30 – 80% of base ion) of
the M-15 ion were observed for THCV, d9THC, CBDA, and CBGA.
3.2 Extractant selection general summary
Development work with extractants included variants of ethanol and acetonitrile, from pure
solvents to aqueous mixtures with up to 20% water. When working with a new matrix, the
Page 7
SC
RI
PT
general strategy was to use 95% ethanol as a first attempt, and then use alternative extractants
as needed to address specific matrix features or specific matrix interferences. As discussed in
some detail in Part 2 [12], ethanol-based extractants generally provided higher cannabinoid
extractability vs. acetonitrile-based extractants, and 95% ethanol provides quantitative recovery
for many product and matrix types. However, ethanol-based extractants may coextract higher
amounts of matrix components such as sugars and glycerin. While sugars and glycerin do not
cause interference for HPLC-DAD analysis [12], these compounds may interfere in the GC-MS
analysis by competing for the trimethylsilyl derivatizating agent. Because the solubility of both
sugars and glycerin are significantly lower in acetonitrile than in ethanol, these interferences
may be minimized by using acetonitrile-based extractants. Details and examples of using
acetonitrile-based extractants for specific products or matrices are given in Sections 3.4, 3.5,
and 3.6.
U
3.3 Cannabis plants, plant preparations, hempseed oils, and dronabinol capsules
TE
D
M
A
N
The method was successfully applied to several traditional types of cannabis samples (Section
2.2) including recreational marijuana plant bud materials, hash oils, fine kief, and hempseed
oils, all using 95% ethanol as the extractant. THCA, d9THC, CBGA, CBDA, and CBN were
each detected in the bud materials, hash oils, and kief. As the bud materials were fairly fresh
when analyzed, the amounts of d9THC relative to THCA were low (10% or less). Higher
conversion of THCA to d9THC was observed for the hash oils (~30%) and kief (~20%). The
predominant cannabinoids in the hempseed oils were CBDA and CBD, with minute amounts of
THCA and d9THC. No CBN was detected in the hempseed oils. The method was also applied
to generic dronabinol capsules (2.5 mg and 5.0 mg strengths, 95% ethanol extractant) with
detection of the synthetic d9THC.
EP
3.4 Commercial Cannabis consumer products
A
CC
Between 2015 and 2017, the method was applied to about 60 commercial cannabis products
including oral supplements, foods, candies, beverages, vapes/eliquids, and topicals. The oral
supplements included oral syringes containing pastes, dropper or spray bottles containing
liquids, as well as tablets and capsules. Most of the products tested positive for cannabinoids
with CBD as the predominant cannabinoid. For many of the products 95% ethanol was suitable
as an extractant, but various interferences were encountered for products with high glycerin
content, high sugar/carbohydrate content, or lactose-containing products. The nature of those
interferences and their resolution using alternative extractants or sample pretreatment are
addressed later (see Sections 3.5 and 3.6). All cannabinoid compounds were observed to be
stable throughout the analysis (extracts were analyzed up to three days after preparation when
in a sample queue).
Page 8
SC
RI
PT
Figures 3A and 3B show GC-MS chromatograms for two of the products, a whole coffee bean
(3A) and an oral liquid supplement (3B). The coffee bean and oral supplement had widely
differing CBD levels (0.81 mg/g and 145 mg/g, respectively [12]), demonstrating the flexibility of
the current method. Relative to the CBD contents, minor amounts of other cannabinoids
including d9THC, CBC, CBG, and CBN were detected in the whole bean product (up to 0.03
mg/g), and CBDV, d9THC, CBC, CBG, CBDA, and CBN in the oral supplement (up to 5 mg/g).
Detection of the minor cannabinoids in the products may provide valuable information about the
sources of the cannabis extracts/concentrates including the degree of enrichment of a particular
major cannabinoid, the degree of decarboxylation of the acidic cannabinoids (CBDA, THCA,
CBGA), and evidence as to specific cannabis varieties [3 – 5]. The large differences in
amounts of the major and minor cannabinoid(s) in a given product may require additional work
such as preparing more than one dilution level for the analysis.
A
N
U
One oral tincture product was encountered in which THCA and d9THC were the predominant
cannabinoids. Two products which contained other active constituents in addition to the
cannabinoids were also encountered. One product consisted of capsules containing both CBD
and phenibut (4-amino-3-phenylbutyric acid). Another product consisted of an oral liquid
supplement containing both CBD and melatonin. Quantitative cannabinoid data for many of the
products is provided in Part 2 [12].
M
3.5 High glycerin or propylene glycol matrices interferences and artefacts
CC
EP
TE
D
Glycerin and propylene glycol were common components found in cannabis oral supplements
and vape products. During cannabinoids analysis in a certain commercial oral supplement with
a high glycerin content, peaks corresponding to underivatized CBD, underivatized d9THC,
underivatized d8THC , CBD monoTMS, and CBD diTMS were observed (Figure 4A). A large
peak corresponding to underivatized glycerin was also apparent in the chromatogram. The GCMS sample preparation had consisted of extraction with 95% ethanol followed by the typical
TMS derivatization step. In independent analysis of the product by HPLC-DAD [12], only CBD
was found and neither d9THC or d8THC was detected. The CBD product level was determined
to be approximately 300 g/g. When the same product was extracted with pure acetonitrile and
subjected to TMS derivatization, only CBD diTMS and no underivatized glycerin was observed
in the chromatogram (Figure 4B).
A
A series of experiments (see Section 2.5) confirmed that high levels of coextracted glycerin or
propylene glycol may swamp the trimethylsilyl derivatizing reagents, creating conditions under
which the desired conversion of CBD to CBD diTMS is incomplete, and undesired side
conversions to d9THC or d8THC may occur. The problem may be avoided by taking care not
to analyze high glycerin and/or high propylene glycol products under conditions which are too
concentrated. When using the current procedure, the total glycerin and/or propylene glycol
content taken in the final aliquot for derivatization should be kept to 40 mg or less. Acetonitrile
may be used to limit the amount of coextracted glycerin in concentrated sample preparations, as
the solubility of glycerin in acetonitrile (less than 5% wt/vol) is much lower than in ethanol
Page 9
(miscible).
Ethanol-based solvents may also be used for less concentrated sample
preparations.
3.6 Validated extraction conditions for additional consumer products
M
A
N
U
SC
RI
PT
Based on the work with the commercial cannabis products, it was clear that a key method
requirement was to address the diversity of cannabis products which continue to emerge,
especially with regard to potential interferents. Another goal was to challenge the method’s
ability to address fairly low cannabinoid contents in these products. Forty cannabinoid-negative
matrices (Table 1) and two Canadian hempseed oils were spiked with five representative
cannabinoids (CBDA, CBD, d9THC, THCA, and CBN), with each cannabinoid spiked at 50 g/g
(0.050 mg/g). The acidic (CBDA and THCA) and neutral cannabinoids (CBD, d9THC, and
CBN) were tested in separate experiments, and no conversion of the acidic cannabinoids to
their neutral forms was observed for any of the matrices. As for the commercial products, all
cannabinoid compounds were observed to be stable throughout the analysis. In order to
measure the performance of the various extraction solvents, spike recoveries were measured in
the extracts using the HPLC-UV procedure [12] prior to TMS derivatization. Recovery was
calculated as (g cannabinoid recovered)/ 50 g) * 100%. The criterion for acceptable spike
recovery was a minimum of 50%. In addition, it was required that retention times and mass
spectra for the target cannabinoids from the spiked products were consistent with the
corresponding standard reference materials (see also Section 3.8).
CC
EP
TE
D
A summary of the extractants which were validated for the qualitative GC-MS analysis of the 42
products/matrices is listed in Table 1. All individual spike recoveries fell within the range 58 –
111%. The recoveries, averaged across the five cannabinoids, are listed in the table for each
product/matrix, and ranged from 67 – 99%. The recoveries for each cannabinoid averaged
across all 42 matrices were as follows: CBDA 85%, CBD 84%, CBN 77%, d9THC 79%, and
THCA 78%. The low levels of cannabinoids already present in the two hempseed oils prior to
spiking were accounted for in the spike recovery calculations. 95% ethanol was a suitable
extractant for all of the edible oils, and several of the foods, topicals, and pharmaceuticals (see
table). 100% ethanol was used for pharmaceutical products (suspension/liquid) with high water
contents. The edible oils include several (grapeseed, hempseed, liquid coconut, virgin coconut)
which have been encountered by the authors in oral cannabis supplements.
A
For products with high sugar contents, acetonitrile-based extractants were used to minimize the
coextraction of sugars. Figure 5 shows a comparison between pure ethanol (upper trace, A)
and pure acetonitrile (lower trace, B) for the extraction of a frosted brownie product. The
reduction in extraction of the sugars using acetonitrile was considerable, with the area counts
for the sucrose octaTMS peak being nearly 100 times lower with acetonitrile vs. ethanol. For a
given product matrix, coextracted sugars may or may not have caused problems in the
subsequent GC-MS analysis, and was dependent on the specific sugars and their amounts.
Trimethylsilyl derivatives of the monosaccharides (glucose, fructose, galactose, etc.) eluted well
before the retention window of the cannabinoids, causing no chromatographic interferences.
Sucrose octaTMS eluted within the retention window of the cannabinoids, but did not coelute
Page 10
with any cannabinoid. However, - and -lactose octaTMS coeluted with the TMS derivatives of
d9THC and CBN, respectively. High levels of coextracted sugars, regardless of whether they
resulted in a chromatographic interference, may swamp the trimethysilyl derivatizing agent,
limiting the derivatization of the cannabinoids.
M
A
N
U
SC
RI
PT
A matrix trapping phenomenon occurred with specific high sugar/carbohydrate products such as
hard candies, strawberry preserves, and honey. The hard candy and strawberry preserve
matrices both contain high levels of sucrose and corn syrup, and honey consists mostly of
sugars (>80% w/w) including fructose, glucose, and various disaccharides. These product
matrices did not disperse well in 95% ethanol, pure ethanol, acetonitrile, or 90% acetonitrile,
and tended to form glassy, impervious precipitates during extraction. The net result was a
physical trapping of the cannabis cannabinoids, and low cannabinoid recoveries (as low as 5%
recovery using the 90% acetonitrile extractant). The extraction procedure was modified (see
Section 2.3) to add the aqueous portion of the extractant first, followed by warming of the
extraction vessel on a hot plate to cause the sugars and carbohydrates to fully dissolve. After
matrix dissolution in water, the acetonitrile portion of the extractant was added as pure solvent
to give a final proportion of approximately 83% acetonitrile for the hard candy and strawberry
preserves, and 91% acetonitrile for the honey. More water was required for the hard candy and
strawberry preserves matrices vs. honey in order to prevent formation of the impervious
precipitates. The acetonitrile extracted the cannabinoids, while also causing formation of a finely
dispersed matrix precipitate. The modified approach resulted in acceptable recoveries (67 –
74%) of the cannabinoids for qualitative analysis.
A
CC
EP
TE
D
Lactose was a major interferent encountered in milk containing products (Greek yogurt, cream
cheese, butter, coffee beverage with milk), with two prominent peaks (- and -lactose octaTMS
derivatives) coeluting with d9THC and CBN. Attempts were made to minimize the lactose
interferences by extracting these products with pure acetonitrile, but a significant amount of
lactose was still extracted most likely due to coextraction with water from the matrices. The
lactose interference was addressed by incorporating a lactase pretreatment/lactose hydrolysis
step (see Section 2.6) prior to extraction. Lactose hydrolysis produced glucose and galactose,
which did not interfere in the analysis. The solvent composition of the finished extract was 83%
acetonitrile(aq) due to the water content of the buffer. Figure 6 shows GC-MS chromatograms
for the Greek yogurt matrix which underwent the incubation and extraction both without lactase
(Figure 6A) and with lactase (Figure 6B). The two major lactose peaks are evident in the
unhydrolyzed extract (A) but are severely diminished in the hydrolyzed extract (B), with lactose
peak area counts reduced by >99% relative to the initial amounts. Average cannabinoid
recoveries ranged from 72 – 88% for the four milk products. Note that for a milk chocolate bar
product, it was not necessary to conduct lactase pretreatment as extraction with acetonitrile was
sufficient to minimize the coextraction of sugars.
3.7 Cannabinoids identification criteria and additional validation
Page 11
SC
RI
PT
The criteria for identification of the target cannabinoids was retention time and mass spectral
correspondence to standard reference materials.
Retention time correspondence was
calculated as the %relative difference between the observed retention times (min.) for the
sample and standard as follows: %relative difference = 100% * absolute value(sample
retention time - standard retention time)/standard retention time, with an allowable variation of <
2%. Actual retention time correspondences for the cannabinoids CBD, CBDA, d9THC, THCA,
CBN, CBG, CBC, CBDV, and THCV from 50 commercial products analyzed over a two year
period ranged from 0.0 to 1.6%. Actual retention time correspondences for the cannabinoids
THCA, d9THC, CBGA, and CBN from 8 plant related materials (buds, kief, hash oils) ranged
from 0.04 to 0.90%. Average retention time correspondences for the five target cannabinoids in
the 42 products from the spiking study were as follows (ranges given in parantheses): CBD
0.09% (range 0.0 – 0.5%); d9THC 0.06% (range 0.0 – 0.4%); CBDA 0.04% (range 0.0 – 0.3%);
THCA 0.01% (range 0.0 – 0.09%), CBN 0.04% (range 0.0 – 0.2%).
M
A
N
U
Mass spectra correspondence was evaluated using expert examination and comparison to
library spectra. In most cases, it was possible to verify retention times and mass spectra for the
cannabinoids from the GC-MS total ion chromatograms.
In some cases, due to residual
interferences, it was necessary to generate an extracted ion chromatogram for a given
cannabinoid in order to obtain an accurate retention time. In those cases, the mass spectra
were evaluated after spectral subtraction of the residual interference(s).
3.8 Minimum detectable concentration
CC
EP
TE
D
Based on data for the five representative cannabinoids (CBD, CBDA, d9THC, THCA, CBN), the
minimum detectable concentration was taken as 1.0 g/g (0.001 mg/g) in all cannabis-related
samples including finished products. “This value was determined by reviewing a large amount of
both low level cannabinoid standard data (low level injections) and finished product sample GCMS data which was collected over a two year period.” For evaluation of the GC-MS sample
data, concurrent HPLC-UV assay results [12] were available and provided the actual
concentrations for the cannabinoids in the finished products. For the GC-MS low level standard
injections, the lowest detectable concentration was 0.3 g/ml. For calculation of the finished
products minimum detectable concentration, a 1 g sample weight, 5.0 ml extraction volume, and
a 0.5 ml aliquot were used. The quoted value does not take into account the possibility of
achieving lower detection limits by preconcentration of the product extracts.
A
Based on the method sensitivity, this method is suitable for the detection of cannabinoids in
hempseed oils but may not be suitable for hemp seeds. Cannabinoid levels ranging from 2 –
240 g/g have been reported [15] for various hempseed oils. In this work, we have used the
current GC-MS method to detect CBDA, CBD, THCA, and d9THC in several hempseed oils for
which our HPLC-DAD assay results ranged from 1.5 – 56 g/g. By contrast, d9THC levels less
than 1.0 g/g have been reported for d9THC [3, 16] in hemp seed kernels. While higher levels
of d9THC may be found on the surfaces of hemp seeds [3, 16], producers frequently wash the
Page 12
seeds to remove the cannabinoids. The current method was not evaluated for hemp seeds
analysis.
3.9 Retention time summary for noncannabinoid cannabis product components
N
U
SC
RI
PT
Table 3 provides a compilation of retention time data for compounds of interest which were
encountered during the analysis of the commercial products and/or the spiked consumer
products. These compounds include some of the cannabis terpenes, sugars, fats/oils
components, active constituents such as phenibut or melatonin, and other miscellaneous
compounds. The compounds are listed in order of increasing retention within each group. The
fat and oil compounds represent many of the major peaks observed during analysis of the
edible oils and butter. Some of the compounds such as the triglycerides and caffeine do not
form TMS derivatives yet are detected intact after exposure to the derivatizing agents. Most of
the compounds caused no interferences for the cannabinoids even if the elution times were
within the retention window (10 - 23 min.) for the cannabinoids. The lactose compounds did
cause interferences as was previously discussed and addressed (Section 3.4).
A
4. Conclusion
TE
D
M
There is a current need for analytical and forensic testing methods for the cannabis
cannabinoids in commercial cannabis products. This GC-MS method has been validated for
qualitative analysis of the cannabis cannabinoids in a broad range of product or sample types
including foods, candies, beverages, topicals, vapes/eliquids, oral supplements in various forms,
cannabis plants, and plant extracts or preparations. The method detects both acidic (e.g.
CBDA, THCA) and neutral cannabinoids (e.g. CBD, d9THC, CBN), and addresses multiple
interferences not encountered in traditional plant based samples. Method sensitivity is suitable
for almost any cannabis-based sample, except hemp seeds.
CC
EP
Declaration of interests: None.
A
Acknowledgements
No acknowledgements.
Page 13
References
1. A. Mead. The legal status of cannabis (marijuana) and cannabidiol (CBD) under U.S. law.
Epilepsy & Behavior 70 (2017) 288 – 291.
SC
RI
PT
2. S. A. Ahmed, S. A. Ross, D. Slade, M.M. Radwan, I. A. Khan, M.A. Elsohly. Minor
oxygenated cannabinoids from high potency Cannabis sativa L. Phytochemistry 117 (2015) 194
– 199.
3. R. Brenneisen. Chemistry and Analysis of Phytocannabinoids and Other Cannabis
Constituents. Chapter 2 in Forensic Science and Medicine Series: Marijuana and the
Cannabinoids. M.A. Elsohly, Editor. Humana Press, Totowa, New Jersey, 2007.
A
N
U
4. K.W. Hillig and P.G. Mahlberg. A chemotaxonomic analysis of cannabinoid variation in
Cannabis (Cannabacae). Am. J. Bot. 91 (2004) 966 – 975.
D
M
5. O. Aizpurua-Olaizola, U. Soydaner, E. Öztürk, D. Schibano, Y. Simsir,P. Navarro, N.
Etxebarria, A. Usobiaga. Evolution of the cannabinoid and terpene content during the growth of
Cannabis sativa plants from different chemotypes. J. Nat. Prod. 79 (2016) 324 – 331.
EP
TE
6. C. Rong, Y. Lee, N.E. Carmona, D.S. Cha, R.M. Ragguett, J.D. Rosenblat, R.B. Mansure,
R.C. Ho, R.S. McIntyre. Cannabidiol in medical marijuana: Research vistas and potential
opportunities. Pharmacol. Res. 121 (2017) 213 – 218.
CC
7. M.A. ElSohly, M.M. Radwan, W. Gul, S. Chandara, A. Galal. Phytochemistry of Cannabis
sativa L. Prog. Chem. Org. Nat. Prod. 103 (2017) 1 – 36.
A
8. T.W.M. Davis, C.G. Farmilo, W. Osadchuk. Identification and origin determinations of
Cannabis by gas and paper chromatography. Anal Chem 35 (1963) 751 – 754.
9. P.S. Fetterman, N.J. Doorenbos, E.S. Keith, M.W. Quimby. A simple gas liquid
chromatography procedure for determination of cannabidiolic acids in Cannabis sativa L.
Experientia 27 (1971) 988 – 990.
Page 14
10. J.M. Parker, B.L. Stembal. Review of gas-liquid chromatography of marihuana. Journal of
the AOAC 57(1974) 888 – 892.
SC
RI
PT
11. R. Brenneisen, M.A. ElSohly. Chromatographic and spectroscopic profiles of Cannabis of
different origins: Part 1. Journal of Forensic Sciences 33 (1988) 1385 – 1404.
12. Authors blind. Commercial cannabis consumer products Part 2: HPLC-DAD quantitative
analysis of cannabis cannabinoids. Forensic Science International (submitted).
U
13. T.B. Vree. Mass spectrometry of cannabinoids. Journal of Pharmaceutical Sciences 66
(1977) 1444 – 1450.
A
N
14. D.J. Harvey. Mass spectrometry of the cannabinoids and their metabolites. Mass
Spectrometry Reviews 6 (1987) 135 – 229.
D
M
15. M. Petrović, Ž. Debeljak, N. Kezić, P. Džidara. Relationship between cannabinoids content
and composition of fatty acids in hempseed oils. Food Chemisty 170 (2015) 218 – 225.
A
CC
EP
TE
16. S.A. Ross, Z. Mehmedic, T.P. Murphy, M.A. ElSohly. GC-MS analysis of the total d9-THC
content of both drug- and fiber-type Cannabis seeds. J. Anal. Toxicol. 24 (2000) 715 – 717.
Page 15
CBDA
CBN
A
CC
EP
TE
D
M
A
N
U
CBD
SC
RI
PT
Fig 1. Chemical structures for 11 Cannabis cannabinoids.
Page 16
d9THC
THCA
d8THC
SC
RI
PT
CBDV
THCV
A
CC
EP
TE
CBG
D
M
A
N
U
CBC
Page 17
CBGA
N
U
SC
RI
PT
Fig. 2. GC-MS total ion chromatogram (35% silphenylene phase) for standard mix of 11 cannabinoid TMS
derivatives each at nominal concentration of 25 g/ml. 1-CBDV diTMS; 2-CBD diTMS; 3-THCV
monoTMS; 4-CBC monoTMS; 5-CBG diTMS; 6-d8THC monoTMS; 7-d9THC monoTMS; 8-CBDA triTMS;
9-CBN monoTMS; 10-CBGA triTMS; 11-THCA diTMS.
A
CC
EP
TE
D
M
A
Fig.3. GC-MS total ion chromatograms for two commercial Cannabis products: a whole coffee
bean (A) and an oral liquid supplement (B). Peak assignments: 1-CBDV diTMS; 2-CBD diTMS;
3-CBC monoTMS; 4- CBG diTMS; 5-d9THC monoTMS; 6- CBDA triTMS; 7-CBN monoTMS. The lower
level cannabinoids are shown in inset chromatograms with signals multiplied 10x.
Page 18
SC
RI
PT
A
CC
EP
TE
D
M
A
N
U
Fig. 4. GC-MS chromatograms for extracts of a commercial glycerin-containing CBD oral supplement
extracted with 95% ethanol (A) or pure acetonitrile (B). CBD conversion products and underivatized
glycerin were observed in the 95% ethanol preparation (A). Peak labels: 1-CBD diTMS; 2- CBD monoTMS;
3-underivatized CBD; 4-unidentified cannabinoid; 5-underivatized d8THC; 6-underivatized d9THC. See text
for discussion.
Page 19
SC
RI
PT
CC
EP
TE
D
M
A
N
U
Fig. 5. Frosted brownie product showing much higher coextraction of sugars (peak labels
1 – 5) with pure ethanol (upper trace, A) vs. pure acetonitrile (lower trace, B). The largest sugar peak is
sucrose octaTMS (peak label 1). The abundances of the chromatograms are offset for clarity.
Fig. 6. GC-MS chromatograms for unspiked Greek yogurt matrix showing presence of major lactose
A
TMS peaks (A, -lactose peak label 1 and -lactose, peak label 2) for 83% acetonitrile extract without
lactase pretreatment, and significant reduction of the lactose interferents with lactase hydrolysis (B). The
TMS derivatives of the galactose and glucose reaction products (3, 4) do not elute in the retention window
for the target cannabinoids.
Page 20
Page 21
D
TE
EP
CC
A
SC
RI
PT
U
N
A
M
N
U
SC
RI
PT
Table 1
Validated extractant variations for products in spiking study
Extractant
Products (spiking level 50 g/g five cannabinoids average % recovery)
liquid coconut (75%), virgin coconut3(72%), grapeseed (86%), safflower (83%), sunflower
(83%), olive (77%), hempseed (two sources, 72%), sesame (79%), corn (80%)
Product Category
edible oils
foods
vegetable oil spread (74%), ground coffee (97%), creamy peanut butter (82%), loose tea (93%)
95% ethanol (aq)
hand/body lotion (84%), pharmaceutical cream base (84%), trolamine salicylate cream (75%),
menthol/methylsalicylate balm3(93%), camphor/menthol ointment3(91%)
A
topicals
M
oral OTC pharmaceuticals
100% ethanol
oral OTC pharmaceuticals
acetaminophen tablets (91%), ibuprofen tablets (85%), ibuprofen soft gels (84%)
acetaminophen suspension (84%), acetaminophen/caffeine liquid (79%)
canned green tea beverage (77%), pancake dry mix (76%), tapioca pudding dry mix (95%),
flavored gelatin dry mix (74%), powdered cocoa mix (78%), frosted brownie (74%), milk
chocolate bar3(83%), dark chocolate bar3(95%)
topicals
menthol pain gel (83%)
other
pure glycerin (99%), pure propylene glycol (98%)
91% acetonitrile (aq)1
foods
honey (67%)
83% acetonitrile (aq)1
foods
hard candy (74%), strawberry preserves (69%)
83% acetonitrile (aq) w/
lactase pretreatment2
dairy foods
Greek yogurt (86%), cream cheese spread (78%), whipped butter3(72%), bottled coffee
beverage with milk (88%)
A
CC
E
PT
100% acetonitrile
ED
foods/beverages
1Water
portion added first with matrix dissolution, see text.
in buffer portion added first with incubation, see text.
3Extraction vessel warmed after addition of extractant to melt/disperse matrix.
2Lactase
Page 22
N
U
SC
RI
PT
Unit Mass/
Molecular
Ion
Base
Ion (m/z)
cannabidivarin diTMS
10.3
430
73
362, 347, 323, 319, 309,
296, 273
cannabidiol diTMS
12.7
458
73
390, 375, 351, 337, 324,
319, 301
tetrahydrocannabidivarin
monoTMS
13.7
358
343, 315, 301, 285, 275
cannabichromene
monoTMS
16.6
M
73
386
303
371, 246
17.0
460
73
391, 377, 337, 321, 268
17.9
386
303
371, 343, 330, 318, 265
9-tetrahydrocannabinol
monoTMS
18.5
386
73
371, 343, 330, 315, 303
cannabidiolic acid triTMS
19.1
574
73
559, 491, 453
cannabinol monoTMS
20.4
382
367
310, 295, 238
cannabigerolic acid triTMS
22.2
576
73 or 561
561, 453, 417, 345
9-tetrahydrocannabinolic
acid diTMS
23.0
502
487
413, 339
Compound
cannabigerol diTMS
A
CC
E
PT
8-tetrahydrocannabinol
monoTMS
Page 23
A
Retention
Time (min.)
ED
Table 2
Retention time and mass spectral data for 11 cannabinoid TMS derivatives (35% silphenylene phase).
Additional Prominent
Ions (m/z)
N
U
SC
RI
PT
Table 3
Retention time summary for Cannabis product associated compounds.
Compound
Retention Time (min.)
underivatized
compound
TMS derivative(s)
< 3.5
---
< 3.5
---
3.7
---
4.3
< 3.5
5.1
< 3.5
6.2
---
7.7
7.5
fructose pentaTMS -furanose anomers
---
7.1, 7.6
fructose pentaTMS -pyranose anomer
---
7.2
galactose pentaTMS -pyranose anomers
---
7.4, 7.6
glucose pentaTMS -pyranose anomers
sucrose octaTMS
---
7.7, 8.1
---
17.5
lactose octaTMS -isomers
---
18.2, 20.2
---
6.9
myristic acid monoTMs
---
8.0
palmitic acid monoTMS
---
9.5
oleic acid monoTMS
---
11.9
stearic acid monoTMS
---
12.0
linoleic acid monoTMS
---
12.1
-linoleneic acid monoTMS
glycerol trihexanoate
---
12.5
18.9
---
terpenes
A
-pinene
myrcene
d-limonene
M
linalool
α-terpineol
trans-caryophyllene
ED
α-bisabolol
CC
E
PT
sugars
fats and oils
A
lauric acid monoTMS
Page 24
N
U
SC
RI
PT
glycerol monopalmitate diTMS
glycerol monooleate diTMS
glycerol trioctanoate
cholesterol monoTMS
glycerol mixed octanoate/decanoate triesters
beta-sitosterol monoTMS
---
19.8
---
22.4
25.5
---
---
27.1
27.4, 29.7
---
---
29.6
32.8
---
---
7.1, 8.7
---
7.2
9.9 (lactam)
7.9, 8.7, 9.3
11.5
---
23.8
19.8, 22.7
1,2-propylene glycol diTMS
4.3
< 3.5
glycerol triTMS
7.5
4.7
A
glycerol tridecanoate
acetaminophen diTMS, O-monoTMS
ibuprofen monoTMS
phenibut diTMS, monoTMS, triTMS
ED
caffeine
A
CC
E
PT
melatonin diTMS, monoTMS
other
Page 25
M
active constituents