i got a few questions about the linear range of detectors, especially the FID. May be someone could help me, plz.
I've read, that the linear dynamic range of the FID is about 10^7. I guess it is the response of my compounds. But I've found nothing about the lower limit. Is it only my detection limit and the detector is up to 10^7 linear?
Ok, if my upper limit is 10^7, what range do you prefer. Are you tring to reach the limit or something between?
It confuses me a bit, if I get a non-linear curve. Is this non-linearity only caused by the detector and it's limit or maybe by the compounds (VOCs) itself?
I got a few standards and every standard has 10 compounds. The lowest concentration has about 0,1% of each compound, the highest was 5%.
With 0,1% i got response of 50.000, but with the 5% I'm over 10.000.000 and the curve fit isn't really ok now.
From 0,1% to 3,0% i got a linearity factor of 0.999x and including the 5% STD my corr.factor decreases to 0,9x.
I have to check the linearity up to 20-25%, but how to do that? Is deluting the STD/Samples the only option i have - cause i need less response?
But what happens then to very low concentrations, that i also want to detect.
Is it a good idea to through out the lower concentrations and force linear curve fit through zero (0,0)?
Im doing that with a "HS/GC-FID/MS-Full Evaporation Technique" - method, by the way.
By Ron on Friday, June 18, 2004 - 07:36 am:
0.1% is 1000 ppm, and I have done HS/GC-FID down to 5 ppb concentration. You have used up most of the linear range, so the best solution I can offer at this point is to do the headspace injection with a higher split ratio. Headspace is a way to increase sensitivity, and you are at a point where you should have plenty of sensitivity already. If you can do the sample by direct injection I would do that, if not use less sample and increase the split.
By Anonymous on Saturday, June 19, 2004 - 07:47 am:
I think the FID should be linear over these ranges. Could it be a solubility problem. In other words, at the lower level you get less of a response because a fraction of your analytes is dissolved. Whereas at the higher levels - where the solubility is exceeded - so most of the material is in the headspace.
If your calibration curve appears as a smooth curve, one option might be to use a polynomial curve fit.
By Anonymous on Monday, June 21, 2004 - 02:54 am:
"I think the FID should be linear over these ranges."
I dont think so. Maybe quantification works with other regressions but you can not work with an const. response factor.
Why can a curve be non-linear?
-Limit in detection (upper/lower)
By Ron on Monday, June 21, 2004 - 06:36 am:
OK, let's do a little math. If you can see 5 ppb on an FID using full evaporation HS, which is typical, and your curve is linear up to 3% (30,000,000) or slightly above, simple division shows the FID is linear over a range of at least 6x10^, which I would consider to be approximately 10^7, especially since all we know is that the non-linearity occurs between 3 and 5%. At a minimum you are exceeding the linear range of the detector, and probably overloading the column as well.
Basically, I don't think the method as described above will ever work, as you are fighting physics if you try to inject that much material on column.
By Anonymous on Friday, June 25, 2004 - 02:57 am:
increasing the split ratio didn't help.
I tried to dilute the sample 1:1000, now linearity is ok over the necessary range.
Sample Amount is 1ml in a 10ml vial. Pressurizing time is 0.3 minutes at a vial temp of 90°C with 45 minutes equilibration time.
But I'm not sure, if it is a good idea to use water for diluting, because I got another problem.
Now, I can't determine n-methylpyrrolidon.
By Anonymous on Friday, June 25, 2004 - 03:22 am:
n-methylpyrrolidon is anyway a bad client for HS.
By jeffo on Friday, June 25, 2004 - 05:25 am:
You can acheive less column overload by decreasing the injection volume by 50%. You are using split mode right? Splitless mode could never handle the high concentrations. Try increaing the split ratio. Is your carrier gas helium? Helium has a maximum linear velocity of 65cm/sec.
How many cal points are required? you can always throw out any outlier cal points, this will improve your R2 value thus improving linearity. Good luck!
By Anonymous on Tuesday, June 29, 2004 - 03:03 am:
my split ratio is 50:1, I don't want to increase it and waste a lot of gas.
Why is NMP a bad HS compound? Is this due to its low vapor pressure and the slow rate of evaporation.
Could increasing pressurizing and equilibration time help?
By Ron on Wednesday, June 30, 2004 - 07:04 am:
You stated that you were using a full evaporation technique, and later said the sample size was 1 mL into a 10 mL vial. For full evaporation you should use no more than 10 uL, and preferably no more than 5 uL for the full evaporation technique. I would expect the method to work much better with a few microliters of sample.
By Anonymous on Thursday, July 1, 2004 - 02:02 am:
first, i did FET with sample amount of 2,5µl.
Now I use 1ml of a dilution in H20. The concentration levels for the calibration: 1ppm 10ppm 25ppm 50ppm and 100ppm (R² ~ 0,999x).
With FET I didn't get good linearity, therefor I tried that. Linearity was ok, till a certain level of concentration in the STD.
I don't know why, I don't know if I overload my column or detector. But increasing split didn't help.
Pressurizing time is 0,2min, injection time is 0,1min.
Vial temp 100°C, loop 110°C, tr.line 120°C. Equilibration time = 15 min
Split 50:1 with EPC
- STD contains 10 diffent alcohols/vocs
- 10%STD has 100% VOC summed
- 0-10% is the necessary range i need to cover
By Anonymous on Thursday, July 1, 2004 - 12:32 pm:
"Why is NMP a bad HS compound? Is this due to its low vapor pressure and the slow rate of evaporation."
Yes and to his high boiling point,regarding the other 10 diffent alcohols/vocs ...
"Could increasing pressurizing and equilibration time help? "
not really then the other alcohols/vocs will also be influenced...
By Ron on Thursday, July 1, 2004 - 01:23 pm:
Please give the instrument conditions of the analysis. The column appears to be overloaded, as decreasing the amount of analyte going on the column by diluting gives the desired linearity. If you have a GC with a gas saver option try running a very high split for 1 minute during the injection, then decrease the split during the rest of the run.
To get the desired linearity I would increase the split ratio to 200 or 250:1, use a 1 uL sample size with full evaporation, and use the smallest sample loop available for your headspace sampler. If you are using a 1 mL loop go down to 250 uL or smaller if available. For the high concentrations you are working with the smaller the loop the better.
In regards to not using excessive carrier gas by running high split, which is more costly to your company, extra carrier gas or bad analytical data?
By Anonymous on Friday, July 2, 2004 - 05:17 am:
yes, i know, that diluting works fine.
But, again, I cant find NMP. I tried H20, THF, and H20/PEG-1:1-mix as solvent. Maybe someone has an idea why NMP can't detected when it's diluted. Ethylhexanol works well.
EPC-Split-Mode, 250°C, Pressure:12psi, Total flow: 63 ml/min, Split ratio 50:1, Gas saver:20ml/min
HP5-MS, He-flow: 1 ml/min, v=25cm/sec.
60°C inital, 5°C/min till 16min, 8°C/min till 25 min, 15°C/min till end
Vial temp: 100°C, loop: 110°C, tr.line:120°C
Vial equilibration time: 5min (2.5µl - 45min at 1ml), Pressurizing time: 0,2min, Loop fill time: 0,1min, loop equilibration=0,02min, injection time=0,1min
1.How do you select your vial pressurizing time? 2.Is it possible to dilute the sample by increasing the pressurizing time, when I work with FET?
By Anonymous on Friday, July 2, 2004 - 06:41 am:
...and it's a 1ml loop, i forgot
By Anonymous on Sunday, July 4, 2004 - 11:57 pm:
sorry, it was a software problem. the method already works!