Engineer trying to play chemist with cannabis testing

Basic questions from students; resources for projects and reports.

4 posts Page 1 of 1
Hi folks. Mechanical engineer working on a startup, playing chemist. Making progress, but as expected in need of help. I am doing cannabis research. FYI I am following a method from a paper.

I have a HP5890 series 2 FID that has shown me some decent result with gasoline. In setup, I have a hydrocarbon gas trap on the helium, and am using UHP air and H2. I have used an old school He leak detector to leak check and am reporting leak free. I am using .25mm id columns and configured my gas flows by the manual. I'm having to hit some pretty high pressures to achieve 2ml/min flow rates. I can't recall where I came up with 2ml/min, but thats where i am at.

Initially, I configured with a Zebron semi-volatile column and ran injections of gasoline. I was able to achieve decent peaks. I changed over to restock 5MS-SiL 30m .25mm id column to start my cannabis attempts. These are old stock columns I bought off of eBay (original rubber cap on the column).

I am reading ~9.5 on my output (I can't recall he units, pA?). My signal shows some occasional spikes with values less than +- 0.5 units. I am operating in a residential building with typical dirty power. With my cannabis runs, my peaks are registering very low (~20pA) and I have some drift in my baseline (+- 1 unit). My linear velocity is somewhere around 20 cm/s and I am getting a good solvent peak, but no significant cannabis peaks (thc&CBD/terpenes). I should certainly be getting a significant THC peak w/ cannabis @23+% thc.

My expectation is that if I am loading the column properly, my peaks should be orders of magnitude higher than my baseline. My first thought to getting better peaks would be to operate my split valve in split less mode for a second or so after injection, but the valve is non functional. How would you all recommend going about modifying a method to improve my column loading to achieve better peaks?
The spikes in your baseline could be due to a dirty flame jet. You can take it out and blow some solvent through it (I use methanol) to clean it. I usually soak them in a beaker of methanol and sonicate them for 30 minutes. Then, I heat them in a vacuum oven at 60 °C for 30 minutes or so to evaporate most of the methanol. This will save you some time getting it to stabilize once you reassemble it.

This is a handy tool for calculating your flows/pressures: ... alculators

At 40 °C, you need a head pressure of 14.6 psig to get a 31 cm/s linear velcocity. That's in the middle of the "happy place" for helium carrier gas.

What is the flow out of the split vent? Your lack of sensitivity could be due to the fact that you're blowing it all out the split vent. The flow calculator can help you with that too.
Thanks for the feedback.

The flow rates were way out of spec. I got it tuned back in and sensitivity returned. Everything seems to be functioning well.

I ran a long calibration curve on serial dilution with cleaning blanks between calibration standards. The noise stabilized after about 4 hours of running time. I've been told older machines tend to need time to warm up and that the noise will drop off. Confirmed.

I'm ordering my standards tonight for my target analytes. Should be off and running in a week or so.

The next question I have run into is how many injections to run between Quality Control runs. I had a suggestion of a very inefficient pattern that brought machine efficiency down to nearly 38%. I'd like to achieve 70%.

This is my current patern to achieve 67%. I am running my blanks at a much higher ramp to achieve this. I'm hitting a three point standard curve on alternating sides of the run (picks me up 4% efficiency). I will pick a deviation limit for determining that the curve needs to be run-run completely.

Run Count
Startup Blanks x 10
Initial Blank Count x 3
QC Standard x1
Standard Curve 1 x1
Standard Curve 2 x0
Standard Curve 3 x1
Blanks x3
Samples x20
Blanks x3
QC Standard x1
Standard Curve 1 x1
Standard Curve 2 x1
Standard Curve 3 x0

I'm wondering how I go about validating the column performance through this system? Any suggestions as to a better sampling/QC pattern and how to do the validation of the methodology would be appreciated?

I don't know how many column runs I'm going to get.
Sorry. Can't help you with most of that. Are there regulatory guidelines you must follow for this analysis? If yes, then that will dictate what you need to do. If not, I guess you're just going to have to come up with your own.

Validation of column performance is usually done on a critical pair of analytes. You need to verify that you get adequate resolution between them as well as good peak shapes for your analytes. I don't know what that might be for your samples. Again, it would help if you had some regulatory guidelines to follow here.

Good luck. I'm glad your sensitivity is back.
4 posts Page 1 of 1

Who is online

In total there are 6 users online :: 1 registered, 0 hidden and 5 guests (based on users active over the past 5 minutes)
Most users ever online was 200 on Fri Nov 10, 2017 4:15 pm

Users browsing this forum: Baidu [Spider] and 5 guests