TLDR: Cannabis oils can be remediated using heat with or without UV light, and we can take advantage of D9THC’s greater instability compared to other cannabinoids like CBD, CBN, or D8THC, to somewhat selectively remove it from a mix of them. I experimented with mostly D9 degradation from a D8 mix, but I also performed experiments to remove D9 from CBD distillate, as well as remove residual D8 from CBN.
I played around with degradation a fair bit before I stopped making D8 a few months ago, and figured I would share what I learned. I messed around with a few variables: Dish/vessel size, loading capacities, stir speeds, UV exposure, temperature, photo-catalyst concentration, input potency, cannabinoid type, oxygen bubbling, inert atmosphere, and of course, time.
CAUTION, UV light will burn your eyes and skin, please use the proper PPE when dealing with it, and under no circumstances will I be liable if you try to attempt this without doing your due diligence on UV safety. Consult a safety professional, or go without the UV light. Its that simple. It’s a good time to mention this should be performed in an area with good ventilation, I did all my work in a fume hood.
If you’re somewhat new to this, here is a brief recommended prerequisite reading list. I like beakers post for two reasons, first off, I love the nostalgia his old posts give me for the glory days of this forum. (I have had my personal account @Bearschoice710 much longer than this one, and was an adamant lurker for long before that, but I now use this business account when I am at the lab) second off, it demonstrates the extreme instability of D9THC. Feel free to add any other related sources in the comments:
To not make this post to exhaustive, here’s a quick summary of methods tried:
- Thin layer of oil in a small crystallization dish in vacuum oven on high heat (250F) for a few days
- Thin layer of oil in a small crystallization dish on a hotplate at varying temps and stir speeds, with and without exposure to ~20w of 254 nm UV light.
- Same as 2 above, but with a larger dish and a thicker layer of oil in the dish
- In an open round bottom flask with varying amounts of oil, as well as varying temperatures, and UV exposure.
- All the above methods with or without rutile TiO2, as well as varying amounts of rutile TiO2. (All the TiO2 experiments included the use of a UV light).
- All the above methods, but with bubbling a stream of air through the mix at different flow rates using a 24/40 gas inlet tube.
That’s the jist of the R&D, what I learned was that there is a balance between temperature, stir speed, and UV light exposure that needs to be balanced to achieve the best results. One of the methods that I liked the best, was heating 1kg of oil in an open four neck 5L flask to 240C on fairly high stir speed (use a rare earth stir bar) with a UV light hanging in the large open neck for direct exposure. I had custom quartz tubes made to insert the UV light into so that it was protected from splashing oil, although I didn’t notice a significant difference with or without it. If one was to scale this up, a quartz immersion well is a must, preferably jacketed. The folks over at quartz.com will make a custom jacketed immersion well for a couple hundred dollars. I was lucky to find a few lab cabinets that were made for sterilizing lab glasses that were safe for UV, and performed the UV reactions in those cabinets, it goes without saying, never open the cabinets while the UV light is on without proper PPE. I also see used Ace glass UV cabinets go up for sale now and again for a few hundred dollars, and I would have bought one of those if I didn’t find the ones that I did (the ace glass cabinets are ~$4k new).
Performing the degradation in a thin layer in a dish works well, and goes very quick (an hour or two), but at these temperatures and being open to atmosphere, you’re liable to lose a significant amount of your oil as a vapor. The degradation begins to work at an appreciable rate at temperatures above 120C, but I liked to opt for approximately 200C for quicker results. At higher temperatures your oil will darken quicker, but I found overall potency was maintained better at higher temps and quicker time, compared to lower temps and longer time.
Bubbling air through at varying speeds certainly sped the reaction up, but it hastened quinone formation, and air addition didn’t speed it up enough to warrant continuing to pursue air addition. I don’t have standards for any quinones, so this was purely speculative.
UV light certainly speeds up the reaction, and although it doesn’t affect the yield substantially, it does form some sort of polymer that is left behind in the boiling flask after the workup distillation that is much harder even than properly made WFE residue. It begins to harden at temperatures under 130C and the only thing I found to dissolve it was acetone or ethyl acetate (which worked out for me, since I already cleaned out my system with acetone after each distillation anyway). This polymer only accounted for a few percent loss, 5% max, so it was deemed worth it to continue to pursue UV light addition.
The addition of TiO2 was toyed with, and amounts varying from 0.1% up to 10% were experimented with. The TiO2 I used was of the rutile type, and was the recommended type used for photo-catalysis, I bought it for fairly cheap from a nano-materials company in Texas that I am forgetting the name of. The TiO2 didn’t make a significant enough difference to warrant using it. Even though the vast vast majority gets removed during distillation, due to its extremely small size, filtering out any remaining amounts is a major pain and in my opinion, it is not worth the potential source of contamination in the final oil.
It is important to note that for the vast majority of these experiments, I was using input oil that was heavily contaminated with delta9THC. The starting material was approximately 20% Delta9THC, and 75% Delta8. Due to this, the experiments I tried were a “worse case scenario”, and I was happy to be able to develop the workup using the worst starting material to get a really good idea for contaminants that were formed, so that I could target them for removal. At the scale I needed, doing a few hundred grams, or even a kilo, was fine in a round bottom flask. The largest I did was 3kg in a 5L RBF, and it was painfully slow. Do yourself a favor and keep the flask loading to a minimum. It goes without saying that having in house analytics is a must. Without the ability to test potency in house, you’re liable to spend a small fortune on outside testing, and you don’t want to heat your oil for longer than you absolutely have too. You will just produce more un-necessary side products that will need to get removed later.
Testing:
Depending on the experiment, samples were taken every 15 minutes.
I used a GC-FID since it was all I had ready to go at the moment. I dont have a mass spec hooked up yet with access to a large library of spectra, otherwise I would have been able to get a confirmed identity of impurities. Having a prep scale HPLC would be great for this to separate and quantify the impurities to study further before shooting em into a mass spec for identification.
Now for the workup:
Since I could not identify any impurities to confirm if they were safe or not, I figured the only option was to try to remove everything that wasn’t D8THC or CBN.
Distillation isn’t a must, but can only be omitted if you’re going straight to chromatography.
I distilled almost all of my reactions, although I went straight to orthogonal flash chromatography on the times that I omitted the initial distillation in my workup. Having the oil already hot makes for an extremely fast distillation start up time, just wipe down the ground glass joints, and than setup for normal SPD. It even helps to pre-warm the distillation head and swing down adapters in an oven so you won’t experience condenser clogs from a cold system.
For almost every experiment, the oil was distilled directly after it was confirmed to be degraded to desired levels. It was much easier to stop the experiment when the oil was confirmed to be under 0.3% than it was to take it all the way to non-detect. Starting with great input material was obviously a much cleaner and quicker reaction. When you’re only trying to degrade 1% D9, you will of course get significantly less impurities than if you’re oil is 20% D9. Either way, I cannot in good faith tell anyone that chromatography to remove potentially harmful impurities can be omitted. Sure, many folks on here will say that people have been smoking poop soup for two years now and that there can’t be any great risk from these impurities… but alas I digress. That was another reason I have been wary to make this post, as I fear many folks will simply degrade, distill, and sell. But I think starting this dialogue is important, and I am anticipating other folks pursuing this further and I am curious to know what they find. So don’t be a dick, purify your oil. Also, someone reached out to me and reminded me I had talked about making this post awhile back. Thank you for that, @Miles-Beyond
Looking ahead:
If I was to scale this reaction, I had a few ideas to try. Many companies already perform degradation on CBD oil to remove THC. To fill that need, many equipment companies have started offering large stainless steel vessels outfitted specifically for this purpose. I assume with some dialing in of the parameters, that those machines would perform this task for D8 remediation. Correct me if I am wrong, but those larger scale machines seem to be similar, albeit larger, versions of the 5L flask setup described above. I was thinking of a continuous feed machine that feeds the oil onto a large flat pan and the pan is tilted at a sufficient angle to make the residence time perfect, so that the time the oil reached the bottom it is compliant. This lead me to wonder if this could be performed in a WFE or rolled film that has the walls heated to ~200C. Simply open it up to atmosphere, maybe use a fan to push hot air up the column somehow in a counter current fashion, and dial in the feed speed and temp so that the oil that comes out of the bottom is sufficiently degraded. I of course would not volunteer my WFE for this experiment (that is, if I owned one).
I am sure I am forgetting things, as I just put this post together incredibly informally, but to keep this from turning into a novel, I am going to go ahead and end it here and start the dialogue below. I didn’t necessarily dive into an exhaustive discussion on potential impurities in this post, as there is already a post thats a good starting point linked above, but alas, it is a large part of this discussion. I found that THC degraded fairly exclusively to CBN. 20% D9 would turn into almost 20% CBN, although this isn’t likely to be a great CBN production method. When trying to remediate D8 from CBN isolate, D8 likely degraded (mostly) into D8-THCQ (HU-336), which was easy to remove from CBN by repeated re-crystallization.
