Analysis of Cannabidiol, Δ⁹-Tetrahydrocannabinol, and Their Acids in CBD Oil/Hemp Oil Products

In the last 2 decades, there has been an escalation in Cannabis use in the USA, with growing public popularity and pressure, together with an inconsistent and confused regulatory picture. In the USA, the use and possession of marijuana is a federal crime; however, medical marijuana legislation has been adopted in the District of Columbia and in 33 states, while recreational marijuana use has been legalized in 14 states and US territories [1]. In addition, 13 states have now passed legislation to allow certain CBD products, restricted in Δ⁹-THC content, for specific disease indications [1]. In the states in which medical marijuana has been legalized, recent studies have shown that 16–26% of medical cannabis users also consume other hemp products [2, 3]. In December 2018, the passage of the “Farm Bill” (Agriculture Improvement Act of 2018) greatly accelerated the aggressive marketing of CBD products. That legislation redefined “hemp” as Cannabis sativa containing <0.3% dry wt. of the psychoactive cannabinoid Δ⁹-THC, provided it is produced under regulations and guidelines stipulated in the statute [4]. It also removed “hemp,” so defined, as a Schedule I substance. This increasing trend of legalization has led to the production of hundreds of kinds of hemp and hemp oil products, commercialized in various forms, including oils, balms, lotions, candies, and capsules. These products contain variable concentrations of Δ⁹-THC and CBD.

Δ⁹-THC exerts its actions through interactions with the CB₁ and CB₂ receptors [5]; the CB₁ agonist activity is, however, responsible for giving the user a feeling of being “high” when consumed in moderation. The pharmacological effects of CBD are much less well known. It has been reported that CBD may act as an inverse agonist or antagonist on CB₁ and CB₂ receptors [6]; however, this varies by cell type and the agonist ligand being studied. CBD also has activity on a number of other receptors: it antagonized the G-protein-coupled receptor GPR55 and the transient receptor potential channel TRPM8 [7]. When combined with Δ⁹-THC, it may serve to counter some of the psychotropic effects of Δ⁹-THC [8, 9].

Unfortunately, the concentration levels of these 2 cannabinoids are often unknown in these products which are widely sold on the Internet, and the labels are found to omit or inaccurately list these concentrations [10, 11]. It is important to identify these values as the concentration is determinant of the dosage required for medical use, as well as for the determination of the legality of the possession of hemp products.

In this article, we report the development and validation of a GC-MS method for the identification and quantitation of the 2 most principal cannabinoids, Δ⁹-THC and CBD (Fig. 1), in CBD oil and hemp oil products. This GC-MS method is able to analyze these cannabinoids and their acid precursors down to low concentrations, with a limit of detection (LOD) and limit of quantitation (LOQ) of 0.1 μg (absolute) and 0.25 μg (absolute), respectively, in the products tested.

GC-MS analysis was performed on an Agilent Technologies 7890A gas chromatograph with an Agilent Technologies 5975C MSD and an Agilent Technologies 7693 autosampler. Separation was achieved on an Agilent Technologies 10-m × 0.18-mm DB-1 column (0.4 μ film). Helium was used as the carrier gas at a flow rate of 0.4 mL/min. The inlet was configured in splitless mode at a temperature of 250°C. The temperature program started at 180°C for 1 min and then ramped up at 15°C/min to 280°C for 5.33 min (Table 1). The total run time was ∼13 min. Retention times of all analytes are shown in Table 2. Data acquisition was performed on ChemStation G1701EA E.02.01.117. Table 3 lists the ions acquired using the SIM mode.

Hexane, chloroform, and hexane ethyl acetate (9:1) were all analytical grade. BSTFA+1% TMCS was purchased from Sigma Aldrich; 1 n HCl was prepared by diluting 10 mL of concentrated HCl to 100 mL with deionized water, and 0.2 n methanolic NaOH was prepared by combining 450 mL of MeOH with 50 mL of 2 n NaOH.

Two 1.0 mg/mL cannabinoid standard solutions of Δ⁹-THC and CBD were purchased from Cerilliant. Δ⁹-THCAA (1.0 mg/mL) was purchased from Lipomed, and CBDA (1.0 mg/mL) was prepared at ElSohly Laboratories, Inc. All four 1.0 mg/mL standard solutions, Δ⁹-THC, CBD, Δ⁹-THCAA, and CBDA, as well as a 10 µg/mL dilution of each standard were used to prepare the calibration curves.

Two 100 µg/mL internal standard solutions, d₃-Δ⁹-tetrahydrocannabinol and d₃-cannabidiol, were purchased from Cerilliant. Both internal standards were added to all samples, calibrators, and controls at a concentration of 1 µg in each test sample.

An accurately weighed 50–100 mg of material was diluted with hexane to make 10 mg/mL samples. A volume of 1 mL (straight sample, 10 mg of oil) and 0.1 mL (dilute sample, 1 mg of oil) of the hexane solutions were spiked with 10 µL of d₃-Δ⁹-tetrahydrocannabinol (100 µg/mL) and d₃-cannabidiol (100 µg/mL). The solutions were adjusted to 5 mL with hexane and vortexed. To this, 4 mL of 0.2 n sodium hydroxide was added and mixed. The solution was centrifuged, and the top layer (hexane) was discarded. A volume of 1.5 mL of 1 n HCl was added to the basic layer and mixed, with the pH checked to be between 1 and 2. A volume of 1 mL of hexane was added, and the sample was mixed and centrifuged. The top layer was transferred to a GC vial and evaporated. The sample was derivatized using N,O-bis(trimethylsilyl)-trifluoroacetamide to make a trimethylsilyl derivative, followed by analysis using GC-MS.

The validation was executed using 100 mg of a hemp oil product certified to contain Δ⁹-THC 5.5 µg/g, CBD 22.5 µg/g, Δ⁹-THCAA 7.5 µg/g, and CBDA 100 µg/g. This control was used to validate the GC-MS method with 6 replicates over a period of 6 days with 4-point calibration curves (0.25, 0.5, 1.0, and 5.0 μg absolute). The accuracy was calculated using the standard addition method. The LOD, LOQ, and upper limit of linearity (ULOL) are listed in Table 2, and the accuracy, RSD, and precision for the 2 cannabinoids are listed in Tables 4 and 5.

Linearity was calculated in 6 validation batches by using 4-point standard calibration curves (2.5, 5, 10, and 50 µg/g). The concentration-response relationship of the GC-MS method indicated a linear relationship between the concentration and response ratio with r² values of >0.99 for all the cannabinoids as follows: CBD (r² > 0.9999), Δ⁹-THC (r² > 1.0000), CBDA (r² > 0.9999), and Δ⁹-THCAA (r² > 0.9999).

The accuracy and RSD for the 4 cannabinoids were determined for within-batch and batch-to-batch (6 batches). For batch 1, the accuracy and RSD for the 5.5 μg/g control of Δ⁹-THC were calculated to be 101.21% (RSD 0.03); the accuracy and RSD for the 22.5 μg/g control of CBD were determined to be 99.26% (RSD 0.02); the accuracy and RSD for the 7.5 µg/g control of Δ⁹-THCAA were determined to be 92.67% (RSD 0.06); and the accuracy and RSD for the 100 µg/g control of CBDA were determined to be 90.78% (RSD 0.07). For batch 2, the accuracy and RSD for the 5.5 μg/g control of Δ⁹-THC were calculated to be 104.55% (RSD 0.02); the accuracy and RSD for the 22.5 μg/g control of CBD were determined to be 105.78% (RSD 0.02); the accuracy and RSD for the 7.5 µg/g control of Δ⁹-THCAA were determined to be 108.22% (RSD 0.09); and the accuracy and RSD for the 100 µg/g control of CBDA were determined to be 103.17% (RSD 0.12). For batch 3, the accuracy and RSD for the 5.5 μg/g control of Δ⁹-THC were calculated to be 95.76% (RSD 0.03); the accuracy and RSD for the 22.5 μg/g control of CBD were determined to be 100.81% (RSD 0.02); the accuracy and RSD for the 7.5 µg/g control of Δ⁹-THCAA were determined to be 112.44% (RSD 0.06); and the accuracy and RSD for the 100 µg/g control of CBDA were determined to be 90.00% (RSD 0.15). For batch 4, the accuracy and RSD for the 5.5 μg/g control of Δ⁹-THC were calculated to be 95.76% (RSD 0.08); the accuracy and RSD for the 22.5 μg/g control of CBD were determined to be 100.30% (RSD 0.07); the accuracy and RSD for the 7.5 µg/g control of Δ⁹-THCAA were determined to be 103.78% (RSD 0.06); and the accuracy and RSD for the 100 µg/g control of CBDA were determined to be 100.20% (RSD 0.10). For batch 5, the accuracy and RSD for the 5.5 μg/g control of Δ⁹-THC were calculated to be 87.01% (RSD 0.03); the accuracy and RSD for the 22.5 μg/g control of CBD were determined to be 99.63% (RSD 0.06); the accuracy and RSD for the 7.5 µg/g control of Δ⁹-THCAA were determined to be 97.11% (RSD 0.10); and the accuracy and RSD for the 100 µg/g control of CBDA were determined to be 105.00% (RSD 0.07). For batch 6, the accuracy and RSD for the 5.5 μg/g control of Δ⁹-THC were calculated to be 88.14% (RSD 0.04); the accuracy and RSD for the 22.5 μg/g control of CBD were determined to be 100% (RSD 0.04); the accuracy and RSD for the 7.5 µg/g control of Δ⁹-THCAA were determined to be 93.11% (RSD 0.05); and the accuracy and RSD for the 100 µg/g control of CBDA were determined to be 108.67% (RSD 0.04).

For the overall calculations, the accuracy for the 5.5 μg/g control of Δ⁹-THC was determined to be 97.53%; the accuracy for the 22.5 μg/g control of CBD was determined to be 100.96%; the accuracy for the 7.5 μg/g control of Δ⁹-THCAA was determined to be 101.22%; and the accuracy for the 100 μg/g control of CBDA was determined to be 99.64%. For the overall n = 36 samples, the RSD for the 5.5 μg/g control of Δ⁹-THC was calculated to be 0.06; the RSD for the 22.5 μg/g control of CBD was determined to be 0.04; the RSD for the 7.5 μg/g control of Δ⁹-THCAA was calculated to be 0.10; and the RSD for the 100 μg/g control of CBDA was determined to be 0.10. For the overall n = 6 batches, the RSD for the 5.5 μg/g control of Δ⁹-THC was calculated to be 0.02; the RSD for the 22.5 μg/g control of CBD was determined to be 0.02; the RSD for the 7.5 μg/g control of Δ⁹-THCAA was calculated to be 0.02; and the RSD for the 100 μg/g control of CBDA was determined to be 0.04.

The precision for the 4 cannabinoids was calculated for within-batch (6 batches) and overall (n = 36 samples and n = 6 batches). For batch 1, the precision for the 5.5 μg/g control of Δ⁹-THC was calculated to be 97.07%; the precision for the 22.5 μg/g control of CBD was determined to be 97.91%; the precision for the 7.5 μg/g control of Δ⁹-THCAA was calculated to be 93.78%; and the precision for the 100 μg/g control of CBDA was determined to be 92.73%. For batch 2, the precision for the 5.5 μg/g control of Δ⁹-THC was calculated to be 98.18%; the precision for the 22.5 μg/g control of CBD was determined to be 97.81%; the precision for the 7.5 μg/g control of Δ⁹-THCAA was calculated to be 90.90%; and the precision for the 100 μg/g control of CBDA was determined to be 88.36%. For batch 3, the precision for the 5.5 μg/g control of Δ⁹-THC was calculated to be 96.90%; the precision for the 22.5 μg/g control of CBD was determined to be 98.08%; the precision for the 7.5 μg/g control of Δ⁹-THCAA was calculated to be 93.74%; and the precision for the 100 μg/g control of CBDA was determined to be 85.45%. For batch 4, the precision for the 5.5 μg/g control of Δ⁹-THC was calculated to be 92.34%; the precision for the 22.5 μg/g control of CBD was determined to be 93.10%; the precision for the 7.5 μg/g control of Δ⁹-THCAA was calculated to be 93.68%; and the precision for the 100 μg/g control of CBDA was determined to be 90.06%. For batch 5, the precision for the 5.5 μg/g control of Δ⁹-THC was calculated to be 97.07%; the precision for the 22.5 μg/g control of CBD was determined to be 94.55%; the precision for the 7.5 μg/g control of Δ⁹-THCAA was calculated to be 89.59%; and the precision for the 100 μg/g control of CBDA was determined to be 92.77%. For batch 6, the precision for the 5.5 μg/g control of Δ⁹-THC was calculated to be 95.61%; the precision for the 22.5 μg/g control of CBD was determined to be 95.98%; the precision for the 7.5 μg/g control of Δ⁹-THCAA was calculated to be 94.61%; and the precision for the 100 μg/g control of CBDA was determined to be 96.48%.

For the overall 36 samples, the precision for the 5.5 μg/g control of Δ⁹-THC was calculated to be 94.33%; the precision for the 22.5 μg/g control of CBD was determined to be 95.91%; the precision for the 7.5 μg/g control of Δ⁹-THCAA was calculated to be 89.60%; and the precision for the 100 μg/g control of CBDA was determined to be 89.65%. For the overall n = 6 batches, the precision for the 5.5 μg/g control of Δ⁹-THC was calculated to be 98.02; the precision for the 22.5 μg/g control of CBD was determined to be 97.97%; the precision for the 7.5 μg/g control of Δ⁹-THCAA was calculated to be 97.90%; and the precision for the 100 μg/g control of CBDA was determined to be 96.49%.

A GC-MS method was developed and validated for the quantification of Δ⁹-THC, CBD, Δ⁹-THCAA, and CBDA in CBD oil/hemp oil products. The structures of the cannabinoids are shown in Figure 1. The GC conditions, including the temperature program (Table 1), were optimized in order to achieve the highest sensitivity of the cannabinoids’ peaks (Table 2). Chromatograms of the control (Δ⁹-THC at 5.5 μg/g, CBD at 22.5 μg/g, Δ⁹-THCAA at 7.5 µg/g, and CBDA at 100 µg/g) are shown in Figures 2-5, and a chromatogram showing these ions in a real sample (CY346) is shown in Figures 6, 7. The ions monitored for the 2 cannabinoids and internal standards are presented in Table 3.

The standard curves for Δ⁹-THC, CBD, Δ⁹-THCAA, and CBDA were linear and had correlation coefficient (r²) values of 0.9999, 1.0000, 0.999, and 0.9996, respectively. The standard calibration curves for all cannabinoids are shown in Figure 8.

The LOD for Δ⁹-THC, CBD, Δ⁹-THCAA, and CBDA was calculated to be 1.0 μg/g for each, and the LOQ for the cannabinoids was determined to be 2.5 μg/g for each. The ULOL for Δ⁹-THC, Δ⁹-THCAA, and CBDA was determined to be 250 μg and 100 µg/g for CBD. The LOD, LOQ, and ULOL of the cannabinoids are shown in Table 2.

The method was validated using 6 replicates of 100 mg hemp oil containing 5.5 μg/g Δ⁹-THC, 22.5 μg/g CBD, 7.5 µg/g Δ⁹-THCAA, and 100.0 µg/g CBDA in 6 GC-MS batches. The individual and overall accuracies, precisions, and RSD values are shown in Tables 4 and 5.

The developed and validated GC-MS method was applied for the analysis of 50 different CBD oil/hemp products, shown in Table 6. The cannabinoids were identified in each product sample based on their mass spectra and specific retention times.

It can be seen (Table 6) that other than 8 samples, all products had Δ⁹-THC and CBD ranging from 0.00002 to 0.04% in the products tested. The 8 samples were determined to have significant concentrations of Δ⁹-THC (0.006–0.797%) and CBD (0.116–17.73%) (Fig. 9; Table 6). These samples were analyzed for concentrations of both Δ⁹-THCAA and CBDA. The concentration of Δ⁹-THCAA ranged from <0.001 to 0.01%, and the CBDA concentration ranged from <0.001 to 0.44%. This indicated that most products were substantially decarboxylated.

A GC-MS method was successfully developed and validated for the analysis of Δ⁹-THC, CBD, Δ⁹-THCAA, and CBDA in CBD oil/hemp oil products. The method was reproducible for all cannabinoids and was used for the analysis of 50 commercial products. The majority of the products analyzed were oils. One of the products was butter, one was a concentrated powder capsule, and another was a hemp extract capsule. The majority of the products contained less than 0.1% CBD and less than 0.01% THC. Of the products analyzed, 3 products contained 0.1–1.0% CBD, 2 products contained 0.1–0.9% THC, and 5 products contained amounts of CBD greater than 1%. Almost all the samples appeared to be partially or totally decarboxylated, as most of the CBDA and Δ⁹-THCAA results were below 0.001%.

The authors acknowledge the efforts of Johnny Pitts, Robert Pruitt, William Harmon, and Shahbaz Gul for their assistance in the completion of this study.

No animal or human subjects were used in this study.

The authors have no conflicts of interest to declare.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. No financial support was provided for the research, authorship, and/or publication of this article.

M.A.E. was involved in study design. I.K. contributed in the acquisition of the samples. T.P.M. contributed to the analytical method development and execution of sample analyses. Both T.P.M. and W.G. contributed to reviewing the validation data. L.A.W., W.G., M.A.E., and T.P.M. organized the manuscript draft. All authors contributed to writing and finalizing the text of the manuscript.