Validation Issue - VOC's in gc/ms Agilent 7890A/5975C

[TiagoBele]

Greetings!
Need a little help in validation issue: Our laboratory is passing through a iso 17025 accreditation, and so i'm trying to validate several volatile organic coumpounds in the gc/ms. What am i facing it's a lack of linearity. We just not getting a linear curve, what we get is a R^2 of 0,80 at max. The analysis for the quantitation is with the SIM (selected ion monitoring), the scan it was just for confirmation of the spectra and retention time.

The standard have a [] (concentration) of 200ug/ml in methanol. I've made dilutions for [] of 1 ppb, 4, 10, 20, 40, 100, 140 and 200 ppb. And for example, to make an 1ppb dilution, i've made another dilution using water as matrix.
To solve this, my thinking is to make a working solution with methanol instead of water, and/or make the solution with a larger amount of the matrix to minimize the error in precision.

Any help?

Another question is referent the internal standards and surrogates. I'm trying to validate the method without the surrogate (is this possible?) : the surrogates like chlorobenzene-d5 are eluting with the respective non deuterated compound: chlorobenzene-d5 with the chlorobenzene, and toluene-d8 woth toluene. If it's not possible, can i validate the method without separating those?

The column it's a HP-VOC (agilent) with 30 meters.

Thanks to all! And apologies any bad writing.


Regards,

T.B. - gc/ms analyst - Brazil

[dblux_]

What matrix are you determining VOC's in ?
What method of extraction and concentration VOC's from samples ?

[TiagoBele]

The matrix it's treated and ground water. The samples are always clean.

We're basing our method in the EPA 8260 with extraction and concentration using static Headspace.

I'm injecting about 500ul of the headspace (autosampler (CTC-Pal)). Heating at 80 degrees (celcius) for 15 minutes.

[Peter Apps]

The concentrations that you are using are very low for equilibrium headspace - can you confirm that the lowest standard contains 1ng/ml ?

Although you are apparently diluting your methanol solution at least 1000: 1 in water (i.e. 1 ul per ml) the lack of linearity may be due to matrix effects as the amount of methanol from the stock standard is higher for the more concentrated standards, which will affect the partition of the analytes into the headspace.

How good is the repeatability - is the lack of linearity becuase you have a curved response, or because you have a varibale response ?

Peter

[dblux_]

...
We're basing our method in the EPA 8260 with extraction and concentration using static Headspace.
...

Can't help. Have no experience in HS.

[TiagoBele]

The concentrations that you are using are very low for equilibrium headspace - can you confirm that the lowest standard contains 1ng/ml ?

Although you are apparently diluting your methanol solution at least 1000: 1 in water (i.e. 1 ul per ml) the lack of linearity may be due to matrix effects as the amount of methanol from the stock standard is higher for the more concentrated standards, which will affect the partition of the analytes into the headspace.

How good is the repeatability - is the lack of linearity becuase you have a curved response, or because you have a varibale response ?

Peter

Hello Peter,

I'm making the dilutions using microsyringes of 10, 100 and 1000 uL. All of them calibrated. The volumetric flask of 10 ml it was calibrated too.
I review the calculations over and over and i'm pretty sure they're correct.

The lack of linearity varies of compound to compound. And that's what i'm not getting. I'm using a standard with 16 VOC's, some of them are presenting a curved response, some a linear response and some a variable response. but if i'm injecting all of them at the same time, the responses should be equal right? If the response varies, all of them should have the same variance. Or don't?

I've been researching, and one thing that many analysts do it is to make two curves. One with low concentrations and another with higher concentrations. Probably this is conected with what you said about the effect of partition. What do you think about the two curves?

thanks!

t.b.

[Bigbear]

You may be suffering from the water issue. It's been discussed at length on this forum. Search for EPA 524.
At first glance I would suggest that you go to a 20M X 0.18 column which would enable to cut the carrier flow rate.
Are you splitting at the injection port? If so how much? What liner? I would suggest a 1 or 2 mm straight ( as your sample is in the gas phase when it gets to the injector).

[TiagoBele]

You may be suffering from the water issue. It's been discussed at length on this forum. Search for EPA 524.
At first glance I would suggest that you go to a 20M X 0.18 column which would enable to cut the carrier flow rate.
Are you splitting at the injection port? If so how much? What liner? I would suggest a 1 or 2 mm straight ( as your sample is in the gas phase when it gets to the injector).

Thanks BB,

but if i use a 20M column, this could increase the problem of compound separation? I'm trying to separate the deuterated compounds (used as surrogates) of the the respective non deuterated compounds, like toluene-d8 and toluene as they're eluting together.

I'm splitting at a ratio of 10:1.
I'm using a sge focus liner with quartz glass wool, two tapered sections, ( i'm using it just for safety, as if a septum particle passes trought the liner, contaminating the system. I've faced this before and as long i used a liner with a glass wool, it never happened again.). The glass wool helps to increase vaporization of the injected sample right? I think that, if i'm just injecting a gas phase, this would make no difference, and i'm preventing some contamination.

[Bigbear]

Co elution shouldn't be an issue, that's why the IS's are duterated.
What compounds are non linear? The "active ones? Where on the curve are they non linear? Too much at the top and not enough at the bottum?
Water activates the source in the 73/75 series and the only solution is to clean. How long ago have you cleaned?
Are you doing a manual headspace injection? I don't recognise the system you mentioned. 500 ul injection, are your peaks quite broad? For grins try your expirement splitting 40-50:1 to see if things get better.
In mass spec more isn't always better. I run EPA 524 and my results are better now I split 150:1 than when I split 70:1!

[TiagoBele]

Co elution shouldn't be an issue, that's why the IS's are duterated.
What compounds are non linear? The "active ones? Where on the curve are they non linear? Too much at the top and not enough at the bottum?
Water activates the source in the 73/75 series and the only solution is to clean. How long ago have you cleaned?
Are you doing a manual headspace injection? I don't recognise the system you mentioned. 500 ul injection, are your peaks quite broad? For grins try your expirement splitting 40-50:1 to see if things get better.
In mass spec more isn't always better. I run EPA 524 and my results are better now I split 150:1 than when I split 70:1!

Thanks BB, i'll try it with a 40-50:1 split. One question, do you run the EPA 524 with a purge-and-trap system or do you use a static headspace? I'm doing automated injections.

regards, T.B.

[Bigbear]

I use a Tekmar Stratum purge and trap.

[Peter Apps]

The concentrations that you are using are very low for equilibrium headspace - can you confirm that the lowest standard contains 1ng/ml ?

Although you are apparently diluting your methanol solution at least 1000: 1 in water (i.e. 1 ul per ml) the lack of linearity may be due to matrix effects as the amount of methanol from the stock standard is higher for the more concentrated standards, which will affect the partition of the analytes into the headspace.

How good is the repeatability - is the lack of linearity becuase you have a curved response, or because you have a varibale response ?

Peter

Hello Peter,

I'm making the dilutions using microsyringes of 10, 100 and 1000 uL. All of them calibrated. The volumetric flask of 10 ml it was calibrated too.
I review the calculations over and over and i'm pretty sure they're correct. Using this size syringes and 10 ml volumetrics does not really fit with the dilutions that you are aiming at. Can you post the actual procedure, step by step, that you use to make up standards ?

The lack of linearity varies of compound to compound. And that's what i'm not getting. I'm using a standard with 16 VOC's, some of them are presenting a curved response, some a linear response and some a variable response. but if i'm injecting all of them at the same time, the responses should be equal right? If the response varies, all of them should have the same variance. Or don't? This suggests a partition effect, or you have active sites in the inlet or column

I've been researching, and one thing that many analysts do it is to make two curves. One with low concentrations and another with higher concentrations. Probably this is conected with what you said about the effect of partition. What do you think about the two curves? first you need to understand why the response is not linear


thanks!

t.b.

Peter

[TiagoBele]

Hello Peter and BB!

I think it worked. What i do : From de stock standard, i've made a work standard with methanol instead of water, and from that solution i've made the dilutions for all the other concentrations : 1ppb, 20, 50, 100, 140 and 200 ppb.

All of them has reached a r^2 of 0,98 at least! =)

all of them with some exceptions =( :

the vinyl chloride ( it's the first to elute, and doesn't have it any linearity at all, has a r^2 of 0,35!

And the 1,2 Dichlorobenzene; 1,2,3 ; 1,2,4 and 1,3,5 Trichlorobenzenes ( they make a curve that fits perfectly when i use quadratic regression.. can i use it? =) )

[Steve Reimer]

Trichlorobenzenes are frequently quadratic in P&T. It is allowed in 8260.
Vinyl chloride has issues with reactivity.

[TiagoBele]

Trichlorobenzenes are frequently quadratic in P&T. It is allowed in 8260.
Vinyl chloride has issues with reactivity.

Thank you so much Steve!

Do you know if the use of surrogates and internal standards are a recommendation or a necessary step?

What can i do to resolve the Vinyl chloride?