In the figures and discussion which follow, we have organized the vaping liquid examples according to the one or two most predominant cannabinoids or other distinctive features. All GC-MS chromatograms are for the “derivatized sample preparations”, in which the cannabinoids have been converted to their fully trimethylsilyated derivatives (d9THC monoTMS, CBD diTMS, THCA diTMS, etc.). However, for purposes of simplicity in the discussion, the cannabinoids will be referred to as the parent compounds. At the time of the EVALI crisis, our validated HPLC-DAD method for cannabinoids quantitation addressed a total of 13 cannabinoids which included only two of the THC isomers, d9THC and d8THC. HPLC-DAD retention times were subsequently established for the exoTHC, d6a, 10aTHC and d10THC isomers, and showed that these THC isomers did not coelute with either d9THC or d8THC. However, the d6a, 10aTHC and d10THC isomers were only sufficiently resolved from one another to allow for a qualitative assessment, and not for strict quantitation. Moreover, only qualitative not quantitative reference standards for the d6a, 10aTHC and d10THC isomers were commercially available. HPLC-DAD quantitation was also not conducted for exoTHC, as it was either not detected, or found in only minor amounts. Hence, in this work, quantitative results are provided only for d9THC, d8THC, and CBD. However, as will be demonstrated, the GC-MS peak area percentages (PAPs) for the THC isomers provide reasonable estimates of the relative amounts of all the THC isomers in the vaping liquids. We analyzed vaping liquids from both unused and used vaping product cartridges. Many labels for vaping cartridges declare fill weights or volumes of 1 g or 1 ml. In our work with emptying and weighing the vaping liquid contents from previously unused vaping cartridges, we obtained weights in the range 0.7–1 g. For the used cartridges, the remaining amount of vaping liquid recovered by our laboratory ranged from residues (less than 10 mg) to near the nominal fill weights. For each example we note the amount of vaping liquid recovered from the cartridge (see Table 1, second last column).

Figure 1 shows the structures for seven THC isomer reference standards which were used in the analysis of the vaping liquids. The structure for d9THC shows the ring letter and numbering scheme. The THC isomers vary both with respect to the position of the double bond in the terpenoid ring (C ring), and the stereoisomeric forms. ExoTHC (Δ^9,11-tetrahydrocannabinol) is the exception with the double bond being present outside the terpenoid ring. The diastereomeric form for the d9THC standard is not depicted in the figure but was the 6aR,10aR-d9THC isomer, which corresponds to the natural form found in the Cannabis plant (Hanuš et al., 2016). The d8THC and exoTHC reference standards were also the 6aR, 10aR isomers. Standards of both enantiomers of d6a, 10aTHC (9R and 9S), and two diastereomers of d10THC (6aR, 9R and 6aR,9S), were used.

Figure 2 shows the GC-MS chromatogram for a 25 μg/ml standard mix of the monoTMS derivatives of the seven THC isomers. All of the positional isomers were resolved, as was the diastereomeric pair 6aR,9R- and 6aR,9S-d10THC. The only unresolved isomers were the enantiomeric pair 9R- and 9S- d6a,10aTHC, which showed complete coelution (peak label d). These results are typical for GC-MS analysis conducted under achiral conditions. Since the 9R- and 9S- enantiomers of d6a, 10aTHC were not resolved, further discussion will only refer to “d6a,10aTHC” for this THC isomer.

Figure 3 shows the GC-MS chromatogram for a high d9THC potency vaping liquid (VL#1, Table 1) in which the predominant cannabinoid was d9THC (peak label a), and no additives were detected. The d9THC level in the vaping liquid was determined at 78% w/w (HPLC-DAD). Other cannabinoids identified include CBD, CBG, and CBN (peak labels b, c, and d respectively), and very low levels of d9THCV and CBC (peaks not visible on current scale). No d8THC or other THC isomers were identified in this vaping liquid. The figure also shows the cannabinoids retention range (10–25 min) for the current method as indicated by the double arrow (<---->). While the presence of any of the minor cannabinoids may indicate a natural Cannabis source, the presence of the varinol d9THCV in the vaping liquid is taken as stronger evidence for a plant source. The varinol series of cannabinoids are frequently found at much lower levels in Cannabis plants relative to the main cannabinoids (Hillig and Mahlberg, 2004). The presence of the varinol d9THCV, which contains a propyl side chain on the resorcinol ring (Figure 1, lowest left structure), would not be expected in any d9THC synthetic schemes, as d9THC contains a pentyl side chain in this position (Figure 1, highest center structure).

Figure 4 shows the GC-MS chromatogram for a vaping liquid (VL#3, Table 1) in which the predominant cannabinoids were d9THC and d8THC (peak labels a and b, respectively). The levels of d9THC, d8THC, and CBD (peak label d) were determined (HPLC-DAD, %w/w) as 17, 11, and 2.2%, respectively. A minor amount of exoTHC and an elevated level of olivetol (peak label c, see also Figure 1, lowest right structure) were observed. The olivetol level was considered elevated based our prior experience with THC vaping liquids in which we have not detected olivetol, or only observed trace amounts. Low levels of the varinols CBDV, d8THCV, and d9THCV (peak labels f, g, and h, respectively) were also found. Again, the presence of the varinols in this vaping liquid is evidence for a natural plant source with isomerization of d9THCV to d8THCV occurring in parallel to the isomerization of d9THC to d8THC. However, the high level of d8THC is unnatural and cannot be attributed to typical processing methods for Cannabis plants. Only minor levels of d8THC have been reported in some historical studies of processed hashish or marijuana materials, with the d8THC representing one percent or less of the combined d9THC and d8THC amounts (Mechoulam, 1970). It is unclear whether minor levels of d8THC are present in unharvested Cannabis plants or if d8THC forms during plant processing steps such as the extraction or isolation of other cannabinoids. Chemical mechanisms for the conversion of either d9THC or CBD into d8THC during isolation procedures have been proposed (Hanuš et al., 2016).

Duffy et al. (2020) also reported finding unnatural amounts of d8THC in vaping liquids associated with the EVALI outbreak. The unnatural level of d8THC shows that the Cannabis material was most likely subjected to chemical and/or thermal treatments to cause substantial conversion to d8THC. Long time Cannabis researcher Mechoulam reported an early procedure for the isomerization of d9THC to d8THC which was conducted in the presence of a strong organic acid such as p-toluenesulfonic acid (Mechoulam, 1970). Later, a group which also included Mechoulam, patented processes for conversion of CBD to either d8THC or d9THC (Webster et al., 2008). In the patented processes, conversion to d8THC was conducted using p-toluenesulfonic acid catalyst with a yield of 81% and high purity, and conversion to d9THC was conducted using boron trifluoride diethyl etherate with a yield of 57% and high purity (Webster et al., 2008).

The next three examples represent vaping liquids in which d9THC, d8THC, and several other THC isomers were found, but with different patterns for the overall profiles of the THC isomers. Figure 5 shows the GC-MS chromatogram for a vaping liquid (VL#6, Table 1) in which the predominant THC isomer was d9THC (peak label a) and d8THC (peak label b) was second most predominant. The levels of d9THC, d8THC, and CBD (peak label c) were determined (HPLC-DAD, %w/w) as 49, 5.4, and 5.2% respectively. Minor levels of three other THC isomers (not labeled in figure) were also identified as follows: d6a,10aTHC, 6aR,9R-d10THC, and 6aR,9S-d10THC. Low levels of the varinols CBDV, d8THCV, and d9THCV (expanded scale and offset, peak labels e, f, and g, respectively), were also found. Figure 6 shows the GC-MS chromatogram for a vaping liquid (VL#8, Table 1) in which the predominant cannabinoid was d8THC (peak label a), and substantial levels of both d9THC (peak label b) and d6a, 10aTHC (peak label c) were also observed. The levels of d8THC and d9THC were determined (HPLC-DAD, %w/w) as 51 and 9.8% respectively. Minor levels of other THC isomers were identified as follows: 6aR,9R-d10THC (peak label d), 6aR,9S-d10THC (not labeled in figure), and exoTHC (not labeled in figure).

Figure 7 shows the GC-MS chromatogram for a vaping liquid (VL#12, Table 1) in which the predominant THC isomers were d9THC (peak label a) and d6a, 10aTHC (peak label b). A low level of d8THC (peak label d), but marked levels of both 6aR,9S-d10THC (peak label e) and 6aR,9R-d10THC (peak label f), were also found. The level of d9THC was determined as 11% w/w, but the level of d8THC could not be determined due to a coeluting interferent. Interestingly, both varinols d9THCV and d6a, 10aTHCV were detected (expanded scale and offset, peak labels h and i, respectively). Parallel conversion of d9THCV to d6a, 10aTHCV would be expected to occur in whatever process caused the conversion of d9THC to d6a,10aTHC. No standard of d6a, 10aTHCV was available for comparison. The identification of d6a, 10aTHCV was based on comparison of its mass spectra with the standard mass spectra for d6a, 10aTHC (Figure 8, comparison of both parent compounds and monoTMS derivatives). The mass spectra for d6a, 10aTHC are unique among all of the THC isomers with fewer high mass ions and overall much less fragmentation both for the parent compound (unit mass 314) and monoTMS derivative (unit mass 386, see Figures 8A,C). The corresponding spectra for d6a, 10aTHCV show the analogous patterns, both for the parent compound (unit mass 286) and monoTMS derivative (unit mass 358, see Figures 8B,D). The elution of d6a, 10aTHCV just after d9THCV also mirrored the elution order for d6a, 10aTHC and d9THC (Figure 7).

The vaping liquids shown in Figures 3-7 represent several different patterns of THC isomer/cannabinoid compositions we encountered. These patterns are evident in Table 1, in which the peak area percentages (PAPs) of selected cannabinoids from the GC-MS chromatograms are summarized. The PAPs represent a good estimate of the relative amounts of the cannabinoids in a given vaping liquid. The selected cannabinoids include all the THC isomers, CBD, and CBN. Both CBD and CBN were included in the peak area percentage (PAP) calculations because they represent possible precursor or degradation products for the THC isomers. CBD amounts may also represent the degree of THC enrichment in the source materials. PAPs for a total of 14 vaping liquids (“VL#”) are listed in the table. Olivetol PAPs are also listed in the table (see last column); however, the olivetol peak areas were not included in the listed PAP calculations for the THC isomers, CBD, and CBN.

VL#1 and VL#2 are vaping liquids in which the only THC isomer detected was d9THC, and d9THC is also the predominant cannabinoid (PAPs near 90%). VL#3 - 5 are vaping liquids in which both d9THC and d8THC are the two predominant THC isomers, with the sum of the PAPs for d9THC and d8THC in the range 80—86%. Although the relative amounts of d9THC and d8THC vary in this group of vaping liquids, they show the same pattern with respect to the other THC isomers. Minor amounts (up to 1.0% PAP) of exoTHC were found. No d6a,10aTHC, nor either of the d10THC stereoisomers, were detected in these vaping liquids. The THC isomer profiles for VL#3 – 5 are similar to the isomer profiles disclosed in patented processes (Grondin and Ward., 2021a and 2021b) for the conversion of either CBD or d9THCA to mixtures of d9THC and d8THC. These processes use heat, and a concentrated or dilute hydrochloric acid catalyst, to effect the cannabinoid conversions on various source materials, which may include isolates, distillates, concentrates, plant extracts, or synthetic sources. The finished products show widely varying ratios of d9THC and d8THC, with minor amounts of exoTHC (Grondin and Ward, 2021b). One further note is that VL#3 - 5 all show the same trend of elevated olivetol levels, with olivetol PAPs (Table 1, third last column) substantially above those for all the other groupings of vaping liquids.

VL#6 and VL#7 are vaping liquids in which d9THC is the predominant cannabinoid (d9THC PAP 66%), with d8THC as the next most prominent (PAPs 11—16%). In these vaping liquids, minor amounts (up to 7% PAP) of the d6a, 10aTHC isomer, and both d10THC stereoisomers, were found. ExoTHC was not detected. VL#8–11 are vaping liquids in which d8THC is the predominant cannabinoid (d8THC PAPs in range 64–79%). In this group of vaping liquids, the next most prominent THC isomers are d9THC and d6a, 10aTHC (combined PAPs in range 9—21%). Minor amounts of both d10THC stereoisomers, and exoTHC were also found. Although the exoTHC isomer was present in minor amounts, this grouping of vaping liquids showed the highest levels of exoTHC. ExoTHC is known to be an impurity associated with dronabinol, a synthetic d9THC (United States Pharmacopoeia, 2008), but is also regarded as an impurity that may be found in Cannabis plant THC isolates (Cid and Van Houten, 2015). This is also the only grouping in which all six of the THC isomers listed in Table 1 were found.

VL#12–14 are vaping liquids in which d9THC and the d6a, 10aTHC are the two predominant THC isomers (combined PAPs all above 70%), with varying ratios of d9THC to d6a,10aTHC. These vaping liquids have the highest amounts of the two d10THC stereoisomers (sum of PAPs in range 10—16%) compared to the other vaping liquid groupings, and also have the lowest amount of d8THC (all PAPs less than 3%) relative to VL#3—11. ExoTHC was not detected. The cannabinoids profiles for VL#12—14 are similar to the profiles disclosed in a US patent (Siegel et al., 2021) for the conversion of d9THC to d10THC, d6a,10aTHC, and CBN. The conversion processes use intact plant materials, THC-sparse oils, or THC-rich oils as starting materials. The reaction conditions include heat and a Lewis acid catalyst such as elemental sulfur. The inventors provide a series of examples in which the finished reaction mixtures contain varying amounts of d9THC, d10THC, d6a,10aTHC, and CBN (Siegel et al., 2021). They further claim to be able to tailor the conditions to maximize or minimize residual d9THC content, and to affect the finished yield ratios of d10THC, d6a,10aTHC, and CBN. As for VL#12—14, the cited examples in the patent contain d9THC, d10THC, and d6a, 10aTHC as the three most predominant THC isomers. Interestingly, d8THC is not reported as present in the finished reaction mixtures, except for one example in which the relative level of d8THC was reported as 0.7% (Siegel et al., 2021). This is also similar to the relative levels of d8THC in VL#12—14, which ranged from 0.4—3.0% (Table 1, PAP results).

The second to last column in Table 1 shows the amount of vaping liquid recovered from each of the 14 vaping devices. Inspection of these results shows that the presence of the THC isomers, and the trends in THC isomer profiles, were not obviously correlated with the extent of vaping liquid consumed, and hence the thermal exposure of the liquid during the vaping process. For example, VL#2 represents a vaping liquid in which over 95% of the cartridge contents were consumed, yet no unnatural THC isomers were detected. The trends for both the VL#8—11, and VL#12–14 groupings, each include vaping liquids which came from unused or minimally used cartridges, as well as almost empty cartridges. This is not to conclude that the vaping process and exposure of the vaping liquid to the vaporizer heating elements cannot alter the vaping liquid composition. It is important to note that the vaping products in this work represent several different types of vaping devices. Different vaping devices may alter vaping liquid compositions to varying degrees, and this aspect is not studied here. The last column in Table 1 shows the additives identified in the vaping liquids according to the protocol described in Part 2 ¹ . Again, there is no obvious correlation between the THC isomer profiles and the additive types.

As discussed above, the THC isomer profiles for the VL#3 – 5 grouping show similarities to the reported isomer profiles for THC source materials produced by one particular commercial process (Grondin and Ward, 2021a and 2021b). In addition, the THC isomer profiles for the VL#12—14 grouping are similar to the reported isomer profiles of THC source materials produced by yet another distinct commercial process (Siegel et al., 2021). The trends in THC isomer profiles suggest that the vaping liquids were likely formulated with THC source materials which already contained unnatural THC isomers. In the course of our work with the vaping liquids, we received two different bulk THC distillates (“DST#1” and “DST#2”). The THC distillates represent examples of high THC source materials prior to incorporation into a vaping liquid composition, and thus provide data on THC isomer profiles associated with the manufacturing process. The specifics of the processes used to manufacture these distillates are not known.

GC-MS analysis for DST#1 (Figure 9A and Table 1) showed that the primary cannabinoid was d9THC, with all of the other THC isomers present except for exoTHC. The d9THC level in DST#1 was 60% w/w (HPLC-DAD), but the level of d8THC could not be determined due to a coeluting interferent. The THC isomer profile for DST#1 resembles the profiles for the grouping of vaping liquids VL#12—14, and the Siegel process (Siegel et al., 2021). GC-MS analysis for DST#2 (Figure 9B and Table 1) showed that the primary cannabinoid was d8THC, with all of the other THC isomers present. The d8THC and d9THC levels in DST#2 were 53 and 6.2% w/w, respectively (HPLC-DAD). The THC isomer profile for DST#2 resembles the profiles for the grouping of vaping liquids VL#8–11. The results for these THC distillates confirm that some manufacturing processes produce unnatural THC isomers at significant levels. The similarities between the THC distillates and the vaping liquids provides further evidence that the vaping liquids were likely formulated with Cannabis source materials which already contained unnatural THC isomers, whether or not further changes in composition occurred during the vaping process.

Additional distinct commercial scale processes have been reported for the production of THC source materials which contain unnatural THC isomers. Tegen and Cho have reported the use of p-toluenesulfonic acid catalyst to convert CBD to mixtures of d8THC and d9THC, or convert CBD to mainly d8THC or mainly d9THC (Tegen and Cho., 2019a; 2019b). The conversion processes may be conducted starting with pure CBD or hemp extracts. The inventors provided examples with crude product relative amounts of 1.1% CBD, 18.4% d8THC, 80.0% d9THC, or 96.5% d8THC, 2.5% d9THC, from pure CBD starting materials. A third example had crude product relative amounts of 2% d8THC and 98% d9THC from hemp extract starting materials.

A recent world patent application (Hartman, 2020) reports on commercial processes for the conversion of either CBD or d9THC to various combinations of d8THC, d10THC, and cannabinol. Cited starting materials include hemp isolates, marijuana isolates, CBD isolates, and hexane or butane extracts of hemp or marijuana. The isolates are obtained via CO₂ supercritical fluid extraction. Conversions are effected using heat, and may include halogen or free radical generator catalysts such as iodine. Specified reaction temperatures can range from 55—800°C. The disclosed reaction scheme shows conversion of CBD to d9THC, followed by conversion of d9THC to d8THC, d10THC, and cannabinol. Relative conversion yields of CBD to d8THC in the range 50–99% are claimed, but no conversion yields for d10THC were cited. The parallel conversions of some varinol cannabinoids such as conversion of CBDV to d9THCV, with subsequent conversion of d9THCV to d8THCV, are also disclosed.

For all of the vaping liquids and THC distillates, a similar pattern with respect to the relative isomer amounts of d6a,10aTHC, 6aR,9R-d10THC, and 6aR,9S-d10THC was observed. The amount of d6a, 10aTHC was always higher than either of the two d10THC isomers, and the amount of the 6aR,9R-d10THC isomer was always higher than the amount of the 6aR,9S-d10THC isomer. These results suggest preferential formation within this set of isomers. The formation of 6aR,9R-d10THC and 6aR,9S-d10THC from d9THC and/or d8THC under base-catalyzed conditions was reported by Srebnik et al. (1984), with the 6aR,9R-d10THC isomer being thermodynamically favored. Further isomerization of the d10THC isomers to the d6a, 10aTHC isomers was carried out under acid-catalyzed conditions (Srebnik et al., 1984), with 6aR,9R-d10THC forming the 9R-d6a, 10aTHC enantiomer and 6aR,9S-d10THC forming the 9S-d6a, 10aTHC enantiomer.

It is also of note that olivetol, which represents a structural piece (see Figure 1, lowest right structure) of the THC molecule, was only detected in vaping liquids and THC distillates in which unnatural THC isomers were also present. Olivetolic acid (the carboxylated form of olivetol), is known to be a key compound in the plant biosynthesis of the cannabinoids (Hanuš et al., 2016), and olivetol is used as a starting material in the classical synthesis of d9THC (Bloemendal et al., 2020). The significance of the olivetol finding in the vaping liquids is not clear. However, the olivetol peak area percentages appear to vary among the different vaping liquid groupings, but are rather consistent within each vaping liquid grouping (Table 1). This suggests some relation to the THC source materials, such as varying degrees of concentration of the plant constituents during purification steps, or even varying degrees of degradation of the cannabinoids among distinct chemical and/or thermal treatments. However, we could find no relevant literature to address this question.

It is not our intention to present the GC-MS peak area percentage (PAP) results for the THC isomers in the vaping liquids and distillates (Table 1) as strict quantitative values. Rather, the GC-MS PAP results are considered reasonable estimates of the relative amounts of the THC isomers, and are supported by the HPLC-DAD analysis. For reasons stated earlier, HPLC-DAD quantitative results were only obtained for the d9THC and d8THC isomers. Table 2 shows a comparison of GC-MS PAPs and HPLC-DAD quantitative analysis (%w/w) for d9THC and d8THC in the vaping liquids and distillates from Table 1. To allow for a more direct comparison of the GC-MS and HPLC-DAD results, the ratios of d9THC:d8THC were calculated for both the GC-MS PAP data, and the HPLC-DAD % w/w data (Table 2, columns 4 and 7, respectively). Inspection of the d9THC:d8THC ratio data columns shows a strong correlation between the GC-MS and HPLC-DAD results. The d9THC:d8THC PAP ratios agree with the d9THC:d8THC %w/w ratios by factors which range from 0.7—1.4 (factors calculated as the ratio of the PAP and %w/w ratios). Although quantitation was not obtained for the other THC isomers identified in the GC-MS analysis, their presence was confirmed in the HPLC-DAD analysis of the vaping liquids and distillates by retention time and UV spectral matches with the qualitative reference standards; this applies to the exoTHC, d10THC, and d6a, 10aTHC isomers.