Why can't I acheive RSD <30??

[shan820]

I am so frustrated about this. I am trying to implement EPA method 525.2 GCMS SVOC analysis for drinking water. I have a standard mix that includes 50+ analytes and my RSD is only under 30 for a handful of them. I have at least 0.995 (linear or quadratic) for most analytes. But the RF increases with increasing standard concentration. Yet my surrogate standards in the calibration have an RSD <10. What am I doing wrong?

Calibration range is 0.1 - 10.0 ppm in ethyl acetate (>80%)
Injection volume 1ul
Inlet Temp 250C
HP 6890N/5975B

[gpronger]

Without seeing the data itself, its hard to comment, but it sounds like you have fairly decent behavior, but not through the intercept. Assuming there is nothing amiss with your calibration standards, you are trying to calibrate over a range of 100-fold, which for some systems is difficult and meet the average response factor criteria.

Now, when I read 525.2, the paragraph (10.2.6) where it talks about needing to achieve 30% RSD for your calibration is followed by one (10.2.7) which says:

"As an alternative to calculating mean response factors, use the GC/MS data system software or other available software to generate a linear regression calibration by plotting A /A vs. Q ."

My interpretation is that if you can't achieve the average response factor criteria, it is acceptable to use a linear regression (which I believe you say works fine). I do not know if your internal protocol or local regulators require the use of an average response factor but otherwise my interpretation of the method is that you are free to use the linear fit.

[shan820]

Internal procedures in my lab typically require RSD < 20. But since EPA 525.2 states 30, then I can shoot for that. And yes, it does say I can use linear regression as an alternative, and I do, but in theory I know I SHOULD be able to achieve < 30 with my instrumentation so it frustrates me that I can't. There has to be a reason why each of my response factors increases with increasing standard concentration.

For Example: Prometryn from 0.1ug/ml - 10ug/ml

RF = 0.069 0.082 0.127 0.165 0.190 0.237 0.249
Mean RF = 0.160
RSD = 44.45

AS for standard prep, I add 1ul of IS to each vial, and the appropriate amount of analyte and ethyl acetate solvent to acheive the desired concentration at a final volume of 100ul.

[gpronger]

If I were to hazard a guess, you have active sites diminishing the response at the low-end leading to disproportionate drop in response at lower concentration. We almost always are replacing the liner and removing some of the front of the column on a daily basis. When you do you inlet maintenance have it down to ambient and don't start heating it for around 15 - 30 minutes to ensure any oxygen has been purged from the system.

For 525.2 we have gone with the Thermo system because Agilent would not guarantee the application (Thermo and Varian would) so you're not likely to get a lot of support from Agilent (unless you simply have a kindly and experienced service rep). We do have a 6890N-5973 system and what you're describing sounds pretty much identical to what we see due to inlet active sites.

[Rob Freeman]

For Example: Prometryn from 0.1ug/ml - 10ug/ml

RF = 0.069 0.082 0.127 0.165 0.190 0.237 0.249
Mean RF = 0.160
RSD = 44.45

AS for standard prep, I add 1ul of IS to each vial, and the appropriate amount of analyte and ethyl acetate solvent to acheive the desired concentration at a final volume of 100ul.

Did you have any luck after cleaning up the injector?

If not, what do the IS responses look like?

[tangaloomaflyer]

we use 6890/5975 and can usually get less than 20%dev for most analytes over the range 0.1 to 20ppm (except for the usual culprits)

Do you de-activate your liners?

[Carvone]

I would suggest making an IS standard solution in your ethyl acetate solvent acetate, and then pipetting this larger amount of this solvent+IS into each vial, since 1 ul is a very small amount to try to be reproducible with and slight variations will be magnified dramatically.

[mckrause]

I agree with Carvone - you need to be using a significantly larger volume for your IS, especially if you are transferring with a syringe. I'd suggest a minimum volume of 10 uL.

Prometryn is difficult if you have any activity in the system - not just the injector liner! You could have an active column as well. Do you use the standard semis column probe (penta/benzindine/DDT)? If not, you really need to use it to check you system activity.

[Ron]

The first two things I would check are the liner and the gold seal at the base of the inlet. I suspect that a liner with a better deactivation, a new gold seal, or both will make a significant difference. One other thing I have found over the years is that for compounds that degrade easily a column with a 0,5 micron phase thickness is more inert than the more customary 0.25 micron phase thickness. Downside to this is the run is 2 to 3 minutes longer.

[Bigbear]

How do you measure your ammounts? With a syringe? My Agilent rep says he got better curves using positive displacement micro pipettes. I'm waiting for him to send me the part numbers he used.
Have you ever solvent washed your injection port? Use a 9mm brass brush and a combination of solvents ( MeCl, Hexane, Acetone, and MeOH) . I have had good luck doing this. NOTE if you choose to do this disconnect the split vent line from the port ( 1/8" copper line on the left rear) so that the EPC stays dry!