Non-Linear FID response to C18:0 FAMEs

[II5tickmanII]

Hello All,

I am currently calibrating a YoungLin 6100GC for FAME analysis of C4:0 through to C24:0 including common isomers such as C18:1c9 C18:2c9,12 and C18:3c9,12,15. Nothing out of the ordinary for FAME analysis. I am experiencing a problem with the C16:0 to C20:0 calibration however as the responses are not linear, not even close.

Brief overview of the set up is
Split injector, 1ul Inj, 50:1 split, 250C,
Column DB-23 30M x 0.32mm x 0.25um
Oven Program start 100 slow ramp to 240
FID Detector 250C
Hydrogen Carrier Gas from Generators

I am using fresh, pure standards from a reputable company and preparing 0.25mg/ml, 0.50mg/ml, 1.00mg/ml and 2.00mg/ml standard solutions in Iso-Octane. For C4:0 to C15:0 and C22:0 to C24:0 linear responses are seen though the C22:0 and C24:0 responses are very much reduced. The C16:0 to C20:0 responses exhibit a plateu effect to 2.0mg/ml then see a somewhat linear rise in response to 5.0mg/ml (examples below, first column after name is the RT, then 0.25mg/ml response through to 2.0mg/ml response mV.s)

FAME C12:0 4.005 76.323 144.532 271.643 507.382
FAME C16:0 6.130 82.430 150.352 268.325 348.683
FAME C17:0 6.750 82.370 143.618 180.634 243.405
FAME C18:0 7.450 75.485 101.217 122.042 177.280
FAME C18:1 7.630 81.805 130.967 153.704 207.213
FAME C18:2 8.010 84.652 137.970 166.071 222.016
FAME C18:3 8.510 86.241 133.995 164.447 211.727
FAME C20:0 9.040 40.857 45.593 55.789 84.901
FAME C22:0 10.935 16.441 24.124 36.026 64.605
FAME C24:0 13.780 8.582 13.919 29.163 56.145

Is there an obivous cause for this as i have not encountered it before? I believe the system set up to be fine and doubt it is due to analytical error. Any help would be appreciated. Thanks

[Alexandre]

Is it splitless or 50:1 split?

Looks like you have discrimination towards the last eluting compounds.
T of injector/detector can be higher. Concentration is too high 2,000 ppm, you need more split; you may experience non linearity due to saturation.

[Peter Apps]

If the results that you show really are mV, then you are measuring peak height - can you confirm this ?

If you are measuring peak height (as opposed to area) then you may be seeing something related to peak broadening for whatever reason - the later peaks are getting short and wide (this is in addition to an expected discrimination against the high boilers). One common reason for late eluting peaks to be wider is that they are eluting on the plateau at the end of a temperature programme.

So to summarize this might be a combinbation of measuring peak height, inlet discrimination, the higher levels being out of the linear range of the detector and the ramp rate on the programme being too fast.

Peter

[II5tickmanII]

Thankyou for the replies. Apologies, it is a 50:1 split system NOT splitless, also it is the peak areas that are listed (quoted as mV.s within the clarity software used).

The oven program starts at 100C
ramps at 20C/min to 160C
then ramps at 10C/min to 180C
then ramps at 5C/min to 200C
4.5min hold at 200C
final ramp at 20C/min to 220C with 2.5min hold (burnoff)

I thought the inj/detector temperatures would be sufficient as the final oven temp is 220C?

The samples ran in routine analysis typically give C16:0 to C18:3 peak areas around 190mV.s to 250mV.s, as such i based my calibration solutions on these responses.

I would expect to see some saturation at the higher concentrations but find it odd that there is a typical initial response then as concentration doubles from 0.25 to 0.50 then 1.00 then 2.00mg/ml only a small increase in response is seen (for the problem C16:0 to C20:0 chains) yet this is not replicated in the other FAMEs (low C-chains and high C-chains). Why would this effect only be evident in some and not others when a linear response can be achieved at response values up to approx 500mV.s (for C4:0)?

Will be doing more work to get to the root of the problem this week.

[Peter Apps]

quite a puzzle !

How do you make up the different levels of the calibration ?, particularly at stage do you mix the solutions of the individual acids ?

Peter

[II5tickmanII]

quite a puzzle !

Indeed. To make the calibration solutions i have a neat >99% pure standard from a reputable supplier for each individual C-chain. These were sealed in glass ampules and freshly opened on the day i made the solutions so highly unlikely they are compromised. In total each analysis in the finished method can identify and quantify 24 components (if all are present).

The standard solutions were made by weighing out x mg of standard then dissolving in x ml of Iso-Octane to make a 2.0mg/ml solution (for instance 10.0mg C18:0 in 5.0ml Iso-Octane). A series dilution was then performed to make the corresponding lesser concentration solutions again using Iso-Octane as the diluent. I would have preferred to make an individual solution for each concentration however due to the number of (24 x 4 = 96) this was impractical. As can be seen from the responses though i feel the series dilution has not greatly impacted on accuracy. I know standard solutions can be purchased which contain numerous C-chains however thought this a better approach as i had the opportunity to do it.

[Peter Apps]

I am still not clear how you get from the individual dilutions of each pure acid to a mixed standard with alll of the acids in it at a known concentration.

If you take 10mg of C18:0 dissolved in 5 ml of octane, and the equivalent solutions for all the other 23 acids and mix them together you end up with, 2 divided by 24 mg/ml of each acid - in round figures about 0.1 mg/ml.

Or are the data from separate series of injections for each acid ? If they are, what happens if you make a mixed stock solution, and then dilute that down ?

Peter

[II5tickmanII]

Or are the data from separate series of injections for each acid ? If they are, what happens if you make a mixed stock solution, and then dilute that down ?

Yes, the data is from a seperate series of injections for each individual acid. For instance the 2.0mg/ml C18:0 only has that acid at that concentration in the solution and so on.

I have not made a mixed stock solution yet (trying to do routine analysis alongside this which is keeping me very busy) but intend to try a mixed injection at different concentrations this week as i still have all the stocks in the fridge, they should still be ok. But surely injecting single solutions rather than mixed is not the cause of the issue i am experienceing?

[Peter Apps]

Or are the data from separate series of injections for each acid ? If they are, what happens if you make a mixed stock solution, and then dilute that down ?

Yes, the data is from a seperate series of injections for each individual acid. For instance the 2.0mg/ml C18:0 only has that acid at that concentration in the solution and so on.

I have not made a mixed stock solution yet (trying to do routine analysis alongside this which is keeping me very busy) but intend to try a mixed injection at different concentrations this week as i still have all the stocks in the fridge, they should still be ok. But surely injecting single solutions rather than mixed is not the cause of the issue i am experienceing?

As soon as all the instrumental and method usual suspects have been eliminated (which they have been in this case) human error is the next most likely cause. I wanted to look at the possibility that one or two dilution steps for the aberrant acids had gone wrong. If you inject a mixed standard then every peak must experience exactly the same instrument conditions - this should be true for separate injections also, but who knows ? - which elimates some possible causes.

Two things to consider - I presume that you ran each acids calibration set sequentially - i.e. you ran each level of C12 than each level of C14 etc etc. Could this have led to the problem acids being run at a time when there might be something different - room temperature for example ?.

The shorter chain acids are liquids at room temp, the longer ones are solids (from C14 upwards iirc) - could this affect anything that you did - volumetrics vs weighing etc ?

Peter

[II5tickmanII]

The lab is temperature controlled and the rcords show nothing out of the ordinary. For the majority of the c-chains they are liquid at room temperature only a few are solid. I will reinject some C18:0 standards using an older GC program and compare responses then take it from there. This should highlight an issue with either the gc settings or the standards. Thanks for the suggestions Peter

[II5tickmanII]

Success!!

After a bit of trial and error and eliminating as much as possible could step by step (much tedium) i have now got linear responses for all compounds by simply dropping the injector temperature from 250C to 200C. When checking % areas against standards the results also match up as would be expected.

Now although this works i am at a loss as to why. Many of the example chromatograms supplied by column manufacturers have the injetor temp above 200C some up at 280C (all split injectors). If there was an issue with the injector temperature and the c-chains surely you would see this effect across all of the c-chains also not just C16:0 to C20:0?

Any ideas? Thanks as always

[Peter Apps]

Your success is even more puzzling than the original problem ! I would have expected exactly the opposite result.

Are you still gettting discrimination against the heavier molecules ?

What are you injector settings - sample volume, pre-inject delay, injection speed, post injection delay etc etc ? I could be that you were getting evaporatio in the syringe needle before injection.

I am not familiar with the YoungLin instruments - is there anything unusual about the inlet ?

Peter

[Bigbear]

On first read I said overloading liner, but for iso octane @250 and assuming 10 psi the vapor volume would be about 155ul. This volume would be critical if your liner has a capacity of 200ul or so.
A couple of other things you could try are:
1 do a pressure pulsed injection
2 try a liner with wool so that the needle injects into the wool.

[krickos]

mmm

got a vague memory of a similar analysis with the same column, as I recall the RSD% was poor and consequently linearity was not excellent with our deafault prepacked liners, switching to a "pestigrade" manual packed liner however significatly improved RSD% (<<1,0%) and linearity.

[Alexandre]

Obviously chromatography is not right. There is 10 fold discrimination of C24 compare to c12.
Dropping inj T down is contradictory with intention to improve skewness.
You need to transfer more heavy stuff into the column. Is any big hump/baseline rise with T? If yes then these are your C24 and Co. You need to sweep injector well after each injection with art not to transfer sweep into the column. Can you dilute, can you dilute more and do splitless?