By Anonymous on Monday, June 14, 2004 - 04:17 pm:

Iam working with an Applied Biosystems API 3000 using the ion source at a flow of 400 ul/min. My standards have poor linearity over the course of a run (about 50 hours running). I would like a correlation factor of about .995 or better but I am unable to achieve this. Anyone have any idea why? My mobile phase is A=water with .1% formic acid and B=MeoH with .1% Formic Acid. A gradient is run. Thanks

-------------------------------------------------------------------------------------------------------
By MG on Monday, June 14, 2004 - 05:40 pm:

What is your calibration range? Do I read correctly that you are running your curve interspersed throughout a 50 hour sample analysis? In that case, what type of samples are they? And, do you run a check standard of the same level throughout the run to see if the response is changing?

-------------------------------------------------------------------------------------------------------
By Anonymous on Tuesday, June 15, 2004 - 09:16 am:

My calibration range is .033 ppm (1xLOQ) to .33 ppm (10x LOQ) This is the concentration of most the pesticides in the standard mix. Some have concentrations of .17 ppm to 1.7 ppm and .34 ppm to 3.4 ppm. First I run a set of standards 1x,2x,6x,10x then some samples (about eight), another set of standards, then samples, etc. They are in either orange matrix or cucumber matrix made up with 70:30 MeOH:H2O. The concentration of the different pesticides doesn't seem to make a differenc. For instance, Oxamyl, it may have poor linearity for one set of orange data, but good linearity the next time I run another set. The standards are the same, made up about once a month and stored in a refrigerator until needed, in which they are freshly vialed from the stock bottle.

-------------------------------------------------------------------------------------------------------
By MG on Tuesday, June 15, 2004 - 09:28 am:

When you say the linearity is poor, do you mean that your response is nonlinear (approximating quadratic), or that your points "jump around" and no curve can be drawn through them?

Going above about 0.25 or 0.5 ppm, I would expect some nonlinearity (roll-off) at the top end, from past experience with that instrument. But you said the concentration doesn't seem to matter.

This sounds like it could be an issue of the ion source getting dirty. Do you notice that the linearity gets worse after running a large sample set?

-------------------------------------------------------------------------------------------------------
By Hyun Joo on Tuesday, June 15, 2004 - 01:39 pm:

I agree with MG's opinion. But first I need to make myself clear. Based upon your question, you run standards and samples with MS/MS mode for constructing calibtration curve and for quantitation. I do not know if your LC/MS is triple-quad or ion trap MS. Probably you use full-scan MS2 followed by reconstructed ion chromatogram for base peak or SRM for target ion.
And, it seems that you have been using external calibration instead of using internal or isotopic surrogate method. To compensate variations and to simulate real sample conditions, you probably spike standards into the orange or cucumber matrix.
I think the problem is coming from source contamination or matrix effect as the analysis time goes on. I experienced this kind of thing from highly organic matter dissolved surface water and soil extracts with LCQ Deca ion trap MS, even though I used internal standard method.
According to MG's suggestion, you can try to run a check standard before and after one set of standards and samples. If the response dramatically changed, the ion source got dirty. Cleaner sample is better. You could reduce the whole analysis time period.

I forgot one more thing to ask. Are you running samples with APCI (+,-) or ESI (+,-)?
Hyun

-------------------------------------------------------------------------------------------------------
By MG on Tuesday, June 15, 2004 - 02:41 pm:

Hyun Joo jogged my memory and made me think of something else. Besides a dirty source, it could also be strongly retained matrix components eluting in later runs and suppressing ionization. I had that problem recently, and there's a thread on this board where I was discussing it with other board members. The problem was solved by washing the column with a strong solvent (acetone in my case) at the end of each run.

-------------------------------------------------------------------------------------------------------
By Anonymous on Monday, June 21, 2004 - 02:28 am:

Try a lower flow rate. Clean the source.