I need help with an analytic problem

One of the idea’s im working on the preparative high pressure chromatography and to do the method i plan on it requires an in situ monitoring method capable of telling me when compounds elute. Luckily ever since i heard that summit was coming out with a “fraction finder” for short path distillation ive been researching in situ analytic methods so I have some idea’s but i thought id ask the braintrust we have here lol

All i need is the ability to tell when cannabinoids r in solution not anything specifics about compounds at all and i dont need any kind of quantity either just purely differential measurements will work for what i need if i can get my setups noise to signal ratio low enough. so I was leaning towards going with either single wavelength UV absorbance or full on UV spectroscopy with a high pressure flow cell. I dont know if UV absorbance would be able to pick up small concentrations of cannabinoids in solution or if the difference in the signal will be large enough for me to pick it out. Im not gonna buy a new high pressure spectroscope, I have found many possibilities in used equipment from ebay and and some new ones from alibaba, so if i go that route im worried that it might now be sensitive enough but thats just cause i dont have any experience this kind of spectrophotometer and i couldnt guarantee the equipment. i know @beaker has beat me the the punch in this endeavor and he got his ebay uv spectroscopy working but he’s using it on purged extract and i would be using it on a possibly dilute solution. And now im seeing more and more fourier transformed near infrared and roman talked about and dont know if they could be as good or better option if i can find a good one. Any thoughts would be greatly apprecieted
Just found some crazy stuff on ebay today including these,


https://www.ebay.com/itm/Wedgewood-AF42-05-TC-1MM-3-8-ID-Flow-Through-UV-Optical-Sensor-Analyzer-Assy/152971937025?rt=nc&_trkparms=aid%3D222007%26algo%3DSIM.MBE%26ao%3D2%26asc%3D44040%26meid%3D30a3f3a46601489a95d7ed04f4c85fc5%26pid%3D100005%26rk%3D5%26rkt%3D12%26sd%3D152951888248%26itm%3D152971937025&_trksid=p2047675.c100005.m1851

ru saying buying these is DIY? cause i can get these from ebay including the controller for a couple hundred but a new one is alot more than that and these would require basically zero modification. Plus im not planning on using these forever but i need to know if the idea is gonna work before i spends thousands on them

I would get the one you like the best, and try it out. Worst comes to worst you resell it in ebay.
Either way your going to learn what works and what doesn’t .

You will easily be able to tell when a fraction has THC in it since the UV absorbance is going to be strong and obvious. However, that will not tell you how pure your fraction is. It might have lots of THC, but it could still have many other contaminants. It depends on what level of purity you are going for whether or not this matters.

The uv absorbance is going to be that strong even if its very dilute? Purity is not a issue in this application as i just need to know when there are cannabinoids present. If it wasnt at high pressures and i was doing it in glass then i prob wouldnt need this but this is just a cheap as a high pressure view cell and should work much better

Yea thats kind of my mindset too plus ive never met a sensor i couldnt interface with so unless the diode array is faulty or it doesnt hold pressure im pretty sure i can get something out of it. Plus calling this spectroscopy is kind of giving it to much credit in my opinion cause u just shine a light thru it at a sensor theres no diffraction grating or optics like the fancy ones. Ive found some crazy things online in the last two days
this website says this one is ~80$
https://www.labx.com/item/rainin-dynamax-absorbance-detector-model-uv-c-wavelength/LV38562521
and this one is ~185$
https://www.labx.com/item/waters-associates-450-variable-wavelength-detector/3663986

that doesnt include shipping but still. at that price buying one just to take apart and scavenge parts from is feasible. I guarantee that waters unit contains valves u couldnt by individually for that price

I need it to know how to proceed so if i did tlc id have to stop the pause the whole process to depressurise and prepare a sample for tlc then wait for it to develop. I know my use case isnt typical but tlc isnt an option

UV is going to be a much stronger signal than IR.

What pressure do you need this rated for?

the extinction coefficient for THC should be available, so you should be able to maths your way out of your sensitivity question.

ie what conc THC will this detector see as 0.100 absorbance? at 280mn if that’s the wavelength you go with.

usually the addition of co2 messes with easy calculations like that cause it might have pressure dependent effects. maybe this wouldnt i dont know why it would but theres also the matter finding the solubility and saturation point of cannabinoids in co2-ethanol mixtures at different pressures and temps. The other reason is im not very familiar with those calculations and while i could muddle thru it even if i did i wouldnt have confidence in my result. Plus i dont know how i should rate or derate used equipment

I agree, there is a gold mine out there in used equipment.

I’ve been finding the craziest stuff the last few days on ebay and other reseller sites, ill find the list i made and post some

Sounds fun

looks like you can get the optek sensors rated for up to 100bar.

temp range might be a bigger problem with -30C as the low end, and only if you get the low temp rated sensor.

https://www.tandfonline.com/doi/abs/10.1080/10826070500187558?src=recsys&journalCode=ljlc20 which might already be in the data dump, should have the extinction coefficients you need to figure out sensitivity.

reading the specs on the optek you posted above suggests that a usable signal might be at 0.05 rather than 0.10 (but don’t quote me on that).

the paper referenced above (which is going in the data dump next) is using data generated at 0.01mg/ml cannabinoids to look at absorbence spectra and extinction coefficients. suggesting to me that your lower detection limit should be at least a hair lower than that.

EDIT: corrected math!
If you extract 10lbs of cannabis at 20% THC in 20gal of EtOH you’re looking at 2lbs THC in 20gal or about 900g in about 75liters. == 0.012mg/ml

Suggesting to me you’re in the right ballpark. just… :slight_smile:

edit: I can’t figure out how to add that paper to the data dump. not sure that uploading it here is the same trick. hazekamp2005.pdf (723.1 KB)

1 Like

thank you for the input, and im not planning on going cold during a chromatography that would only be for extraction so this sensor shouldnt ever be exposed to large temp swings. I was also thinking of the a fiber optic uv spec but that optek is just so convenient if it works and i just inquired about a new process crossview cell and was told between 3 and 4 thousand without any sensor or anything so thats fucking yikes. I would probably end up buying the optek or another one similar used and scavenge the crossview cell anyway so might as well see how well it works first lol
https://www.google.com/url?hl=en&q=https://oceanoptics.com/product/process-cross-flow-cells/&source=gmail&ust=1524072810274000&usg=AFQjCNEouNe6APFAZbw0mw5LsGMl8ayAhQ

When I first read this I think I pondered for hours about the whole idea. I dab, walk slow to coffee, and think… it occurred to me that all the needed info is available in ways already measured to find fractions automatically via computer control. Adding optical gear might work great but I realized that a solution could be implemented in a more reliable and less complex manner.

If a small computer like an Arduino were fed the inputs of thermocouples placed at just four locations I have in mind and the input from the Pirani based system sensor (my unit outputs simple ASCII numbers from a serial port that the Arduino is native to) then I am certain a highly reliable fraction finder could be implemented in software.

However it would certainly require the mantle control to be totally computer operated. The computer would likely need to control things for it to work. This is just a thought based on how I do it manually now. Unplanned adjustments to the mantle would throw off the computations I have in mind.

There is a direct correlation between pressure, system wide temps, and time at a set and known energy input level that can identify the gas(es) being evolved. A fingerprint of sorts combining time, temp, and pressure can be stored with each run and within a few runs the operator could then set “peaks” in that fingerprint into the computer identifying set points. When a setpoint is reached the computer then acts to notify the operator, auto change the receiver, adjust temps, or shut down.

It would be pretty simple program to sketch up and all the libraries for PID and such are written so no hard core programming. What do you think?

1 Like

what i think? what i think is omg yes lol
I just randomly saw that video where he talks about it and I literally did the same thing, and i sure what ur talking about wouldnt be hard to achieve at all and probably could be retrofitted onto almost any short path setup. (except the part about automating the swapping of recovery flasks, that might be tricking to do with a deep vacuum system without alot of doing. i’m thinking a specialized component like a high vacuum rotary seal for a cow adapter or a valve to divert the condensate in different directions would be the best bet)
i always thought paying for a “chiller” that u only use to heat water was ridiculous so i was building my own and i thought about a system for controlling condenser coolant that takes data from thermocouple at the inlet and the outlet of condenser and used the difference to to control both the flow rate and/or the heating. i bring this up cause that could delta temperature could also be a good data input for this system and uses the same idea of using thermocouples and control theory.
One thing is ud need all the components together before programming basically any code. and if ur doing it with prebuilt modules that have libraries thats doubly true because there is another layer and almost anything u write is likely need to be adjusted.
this is a great idea and super fascinating to think about but im working on high pressure systems right now in large part to avoid such low pressure systems that are required for distillation. i got alot of the idea after watching ur chromatography videos on youtube lol

I am fairly certain the Arometrix Fraction Finder used for SPD would also work for a chromatography elution system. The nice thing about that unit (I believe) is it is a scattering type detector, meaning it emits light and detects absorbance from the same side, so the thickness of the sample flowing past it does not change the reading. A flow cell with emission on 1 side and detector on the opposite side would have to remain totally full at all times to read accurately.

4 Likes