A Comparison of Analytical Systems between 3 Testing Methods for Cannabis
At their core, the 3 major testing methods GC, HPLC and HPTLC are similar – all involve running a sample through treated silica to separate the different cannabinoids from one another, and then measuring the amounts of the different cannabinoids – but the details of how this is done makes the different technologies best suited for different applications.
GC is perhaps the most common method of chemical analysis in use in the world today. In GC, the sample under study is vaporized and then pushed by a mix of gases through a long, thin, coated tube, not unlike a hollow fiber optic line up to 60 feet long. The different cannabinoids separate from each other as they travel, and are measured at the far end, usually by a detector known as an FID that burns whatever comes out of the tube and looks for the products of combustion. (One alternative detector is a mass spectrometer, much more sensitive but far more expensive and difficult to operate.) The response from the detector is compared to the response to a “reference sample” that contains a known amount of specific cannabinoids in it. Comparing the timing and size of the signals from the detector allows the analyst to calculate a number of targeted cannabinoids in the unknown sample.
GC is terrific for measuring small quantities of cannabinoids.
For cannabinoid testing, its main weakness is that, because the sample is vaporized at high temperatures when it enters the machine, it cannot distinguish THC from THC-A in a sample unless significant additional processing is done. This makes the technology impractical for testing infused products. The coated tubes cost several hundred dollars apiece and are used for hundreds or thousands of tests before replacement, leading to problems from contamination and degradation of the column.
In HPLC, the sample is pushed by liquid solvents through a short tube packed with silica particles. The separated cannabinoids are measured at the far end, usually by monitoring the output with a beam of ultraviolet light. The main drawback of this method is that the UV detector responds to many substances in addition to cannabinoids, leading to interference, and has significantly different responses to different cannabinoids, requiring calibration for each separate cannabinoid. As with GC, the columns must be re-used many times, leading to contamination and degradation problems. Finally, HPLC equipment tends to be temperamental, with significant downtime for repair and maintenance.
In HPTLC, the sample is “spotted” onto a disposable, silica-coated plate. Liquid solvents are then run across the plate, separating out the cannabinoids. The plates are treated with chemicals and scanned at a particular frequency to reveal the cannabinoids. HPTLC particularly lends itself to the analysis of complex mixtures, such as plant or food samples, as detection can be limited to specific groups of substances – in this case, cannabinoids – and the use of disposable plates means that no residues accumulate from one test to the next. The main limitation of HPTLC is that it is not as sensitive to minute levels of cannabinoids as GC or HPLC. However, even the most dilute medical marijuana products such as sodas and drinks contain enough cannabinoids to be accurately measured with HPTLC.
|GC -FID Gas Chromatography with Flame Ionization Detector|
HPLC High-Performance Liquid
HPTLC High-Performance Thin Layer Chromatography
CannLabs,- GC-Fid Gas
Rm3 Labs- HPTLC
|Limit of Quantification -THC|
<1 part per million
@1 part per million
@10 parts per million
|Risk of Cross-Contamination Between Samples|
|Sensitivity to Unwanted Materials in Sample|
|Silica Medium Degrades over Time|
|Can distinguish THC from THC-A|
Not without significant additional processing
|Relative Standard Deviation (RSD)|
<10% (Rm3: Approx. 2%)
|Resolution of different cannabinoids|
Good – Excellent
The technology used for analysis – GC, HPLC or HPTLC – is only one factor affecting the quality of analyses done by a lab. Every analysis has a number of potential sources of error; together, these potential errors add up to produce uncertainty in the results. The “final” number presented by an analysis is really the center of a likely range of results; how wide is that range depends on the quality of every step in the analysis. There are many sources of possible error, of which the analytical equipment is only one; the other sources of possible error often dwarf the range of uncertainty of the equipment. The following table lays out some of the possible sources of error or uncertainty in analyses:
Typical Error Range
Rm3 Labs using HPTLC
Food: 3-25% Plant 2-10% Hash: 0-5% Tinctures: 0%
|Sample Measurement Error|
Food: 0-10% Plant: 0-2% Hash: 0% Tinctures: 0%
Food: 0-5% Plant: 0% Hash: 0% Tinctures: 0%
|Analytical Equipment Error|
<10% (we assume 1% below)
|Reference Sample Error|
Food: 9-56% Plant: 8-37% Hash: 6-26% Tinctures: 6-21%
Food: 7-36% Plant: 6-16% Hash: 4-8% Tinctures: 4-6%
These error ranges are a percentage of the analytical results of a test.
For example, if the range of error for a plant test is plus or minus 10%, then a test result of 20.0% THC really means “the actual amount of THC in the sampled batch is likely between 18.0% and 22.0%, with the most likely number being 20.0%”.
As seen above, one of the largest sources of error is in the sample itself – either difference within a sample or differences between the sample and the batch it is taken from. This error can exceed 25%, particularly in food samples. One part of a food item can have more active ingredients than another – think of a pastry where the active ingredients are only in one layer – and such samples have to be carefully homogenized by the lab in order to get an accurate reading. Plant samples also generally show significant variation, depending on such factors as the amount of light the plant or branch received and the quality of nutrients and trimming.
Another source of error comes when samples and solvents are measured during the sample preparation process. We generally keep this error under 1%, but using less precise methods or less expensive equipment can easily bring this error to 5%.
A third major source of error arises from variations in the reference samples used to calibrate the equipment. Commercially available standards for THC, CBD and CBN are generally advertised as having variances of 5% or less, but studies have shown the actual variation can be 10% or more. As a result, we have elected to manufacture our own reference samples, reducing this variation to 2-3%.