Lost of Linearity_P&T-GC-MS

[Celiagg]

Hello Everybody!!

I'm in problems : I'm developing a method and validating it and I've found that I 've lost linear range.....Can someone tell me if it may be a fault in the GC column performance?.Peak shapes are OK. I tried changeing the trap and it does not improve..

Thank you

Celia

cegrunder@yahoo.es

_________________

[JI2002]

More information please.

What are the target compounds?
What is the calibration range?
What are the GC/MS and P &T parameters?

[Celiagg]

Hello!
First at all, thanks for replying me.

I'm analyzing VOC, with a polyethylene glycol ( 60m*0.25mm*0.5um column) with (40-260ºC) limits of operation temp;

The oven temperature program goes from 40º C to 220ºC, an aux intrephase with MSD Temp of 230 ºC.
The injection system is a Purge&Trap Tekmar 3100 with a Sample Injection Volumen of 5 mL that goes to 1 uL in the CG inlet. The liner is clean at eye sight. The interphase between CG and P&T is at 200º C and the purge temp is 260º C , because the trap usedis a Vocarb 3000, and those are the conditions recomended by the manufacturer.

The los of linearity is found in every compound, not only in THM.

I can't unserstand...

Celia

[Celiagg]

I forgot to tell you that calibration range is from 1 ppb to 15-20 ppb.

Thanks

Celia

[Peter Apps]

Is it the high or the low end of the calibration that deviates from linearity ? If you can, post a calibration graph (instructions are at the top of the LC page)

Peter

[Celiagg]

Hello again,

It's the top of the range: at the beguining, it was lineal to 20 ppb; after, it stopped to be lineal at 15 ppb. But now it does not reach 15 ppb either.

The linearity is good with regression coefficient of "0.999997" to some of the compounds to 10 ppb, with : 1, 2, 5, 10 ppb as points of calibration. I revised the method of integration, and everthing is ok ...

The curve is linear but starts to be a sigmoideal at 15 ppb , as if it was a saturated the detector capacity. But the source it's been cleaned recently and autotunes are made each week and results in concetration are reproducibles between days in this range. With a deviation of 20-30 % .

I work with and internal standard method to avoid instrunmental drift.

Peak shape is correct with no apparent asymmetry. Baseline does not goes over 2000 conts.

I'm lost. It has never happend to me before.

Regards,

Celia

[AICMM]

Celiagg,

Are you connecting the 3100 transfer line to the split/splitless inlet on the GC or directly to the column? I assume split/splitless inlet. If so, what is the column flow rate, what is the split ratio, and are you running split the whole time? Also, what is the EM running at and what is the signal to noise for the 1ppb chromatogram, especially the early peaks?

Best regards.

[Peter Apps]

Hi Celia

Is the deviation positive or negative at the high end of the calibration ? - please post an example of a calibration graph.

What kind of GC and MS are you using ?

Are you able to do liquid injections to the GC inlet ?

Peter

[Celiagg]

Hello!

I'm using a 5973N and a 6890CG from AGILENT .

The split-splitless port is conected to the transference line, but i think i can make direct injections on column. I have an ALS Tower too.
Colum flow rate is 1.2 mL/min (He ultrapure), and split ratio is 10:1. All time I'm running in split mode. EM volts are OK, about 1700, tomorrow I'll tell you exactly S/N, I'm at home now...Sorry....

This is an image of what it's happening to me:





Regards

[AICMM]

Celia,

At 1.2 ml/min and 10:1 split, you are only moving about 10-12 ml/min across the trap when you desorb. This is a very low value for a desorption flow. While I recognize the price you will pay with sensitivity, I would suggest trying the same curve with a split ratio of about 20:1 for the same column flow.

IF that doesn't work, there are a couple of other though less likely, possibilities.

Best regards.

[Celiagg]

Hello AICMM!!

I wrote wrong, I'm using 20:1 slit ratio ...SORRY!! I was at home and I could not remember well.

EM volts are 1200, and S/N rate for 1 ppb is of 37 for the smaller peak ; for my ISTD is of 160.


Do You or someone think that may be that the column is getting old? The stationary phase is of 0.5 um and 0.25 mm ID. Someone told me that are better to use columns of thin film that more thickker ones. DO they really age faster? I thought is was not a column problem because simmerty is great! And baseline is arround 1000. Yesterday I put my column at 240 ºC (It resists up to 260ºC) all the night.

The problem is that linear range is changeing!!!


Kind Regards

[Peter Apps]

Hi Celia

This is very strange. All the conditions that you give look reasonable, and yet you have a smooth decrease in response.

Can you replicate this calibration using a completely fresh set of standards (you do not need to include all the compounds) ?

How are you doing the dilutions for the standards, how much do you make up at each level, and how many runs have you made from the standards ?.

I presume that cuentas is peak area counts.

Are all the peaks equally affected ?, or some more than others ?

Look very carefully at the top of the peaks in the 15 and 20 ppb standards - do they look flat ?

Peter

[Celiagg]

Hello again..

I prepare each solution inmediatly before to inject it. Intermediate solutions (in methanol) of standars are at the freezer at -20º C, stored in capsuled vials without head space volume.


Calibration levels are : 1 , 2, 5, 10, 15, 20 ppb, at the aquous solutions. I take 5 mL of them, and inject at the purge-trap system. I purge this for 11 min, and after, i run the program of desorption at temperature and during the time recommended by the supplier of the trap .

I CHANGE THE TRAP AND PUT OTHER KIND OF FILLING and the problem persist ( I had TENAX nº 3 and change to VOCARB 3000).

I've been studing linear range with four calibration curves, and everything was OK. I know someone that works with the same system, and the same stationary phase, but less thikker INOWAX_60m*0.25mmID*0.25 um) and her calibration range is form 1 to 15 ppb without problems.

Yes, cuentas are "counts"; sorry it's Spanish way of speaking.

All the peaks are affected.


Some of them YES: LOOK FLAT.
The BFB (1-Bromo-4-Fluorobenceno) internal standard looks flat in some of the injections.

I've been thinking on it , and if it was a column problem i'd have tailing, and changing in retenction times and it is not so; If it was a saturation in the trap capacity, areas would be higher and are SMALLER.


DO I have leaks in the purge and trap? If it was a leak, I'd have less response, less area, but it would not affect the linear range.

I CAN'NT UNDERSTAND ANYTHING...

Thanks Peter, please, continue helping me...

Kind regards,

Celia

[danicrd]

Hi
I'm using a 5973N and a 6890CG like your..and my calibration range starts from 0.1ppb to 100.
With good linearity.
What are concentration of your IS? If bfb look flat the concentration is too high.
I'm working with 1ppb IS conc, but 10 is also a good concentration, especially in dirty matrix.
You can make another, simply test: try to low EM voltage to 1000V and rerun calibration again.
If the problem is detector saturation (I think so! In my sistem 100 ppb are the limit with split 25:1) you are able to see immediatly.
Regards

[Peter Apps]

Hi Celia

The flattened peaks are a vital clue, and they point to detector saturation. There might be a problem with the electrom multiplier, although the autotune should increase the voltage to cope with an overall loss in sensitivity I am not sure that it would compensate for a loss in dynamic range. One thing that you could try is to put all the MS tune settings back to the factory defaults, and then tune again (you may have to do a few cycles). You could also try using a different automatic tune option from the one that you are using now (s-tune instread of a-tune for instance).

Peter