Abnormal response on standard and samples

Chromatography Forum: GC Archives: Abnormal response on standard and samples
Top of pagePrevious messageNext messageBottom of pageLink to this message  By Uberto on Friday, July 26, 2002 - 12:29 am:

Dear all members,

I have some queries regarding the analysis of food samples. Initially I prepare a series of standards and sample spiked with known amount of standard and analyse by GCMSD. The sequence is set to inject standards first and then spiked samples and finally the standards again. I find that in the presence of sample matrices, the MS response for the analyte increases abnormally c.f. the standards. After injection of spiked samples, the responses of re-injected standards also increase to about 30-50%. The solvent used to dissolve the standards and spiked samples are iso-octane and thus I can exclude the possibility of concentration of standards due to solvent evaporation. Would someone have the similar experience and have the idea how to solve it?
Thanks in advance.


Top of pagePrevious messageNext messageBottom of pageLink to this message  By Anonymous on Friday, July 26, 2002 - 08:04 am:

Make up the standards in a clean matrix which is similar to the samples. It is not uncommon for compounds in the sample matrix to bind to active sites in the inlet and increase the response.


Top of pagePrevious messageNext messageBottom of pageLink to this message  By Scott Fredrickson on Thursday, August 1, 2002 - 11:13 am:

Learn as much about LC and LC/MS as you can, and migrate your analyses to that platform. It will take several years to do it, but that is the only real solution to the problem I have seen, and I've been looking 30 years. LC generally does not have this problem, although it has others.

In the meantime, standards in matrix, above, will help. At some point the system may 'stabilize' and give you acceptable results, but don't hold your breath. This approach is frowned on by some regulatory agencies, but sometimes there is no alternative, and they accept that--sometimes.

Other things you can try include:

Cooling the injector and column when not running samples. When/If you get the active sites filled, don't burn the coating off with high temperatures!

Extensive clean-up of samples. Ideally, they would eventualy be as clean as standards. But then, just injecting standards over and over may also demonstrate the same behavior.

Restrict your standard curve levels as much as practical. This helps because some of the 'potentiation' problem is compound specific. You will soon learn which compounds are the best and worst. Choose control/surrogate compounds with this in mind.

Try to maintain your system in a relatively constant state. If it seems to run better with fresh insert and pre-column, then keep it clean. If it works better 'dirty' then leave it dirty.

Megabore columns will give you better results than smaller diameter columns, as will thicker film coatings, usually. You may be able to use this info on other systems, but your MSD won't work with large bore columns, or thick films.

Inserts, injector types and pre-columns can play a very large role. Track system responses to see what the effects are, and adjust your game plan accordingly.

Accept that your recovery ranges are going to be larger than you would like--perhaps 50-150%. Depending on the final use of the data, this may not be as bad as it looks. Be SURE to include this type of variability data in any decisions about the products.

Be comforted by the fact that you are not alone in working on this problem, and enjoy those days when the data looks really good!!!


Top of pagePrevious messageNext messageBottom of pageLink to this message  By Sue Coward on Thursday, August 8, 2002 - 12:30 am:

At last I find something similar to my problem! For ages now I have been modifying a method for organophosphorous pesticides in soil/water matrices on a gc-ms system. Just when everything had started to go well, good response and reproducibility, I started working full steam ahead and hey presto retention times start dropping and peaks start disappearing - the column had died.
So new column, fiddling with temps etc to get back to decent separation, recalibration with new standards - beautiful peaks! Textbook stuff.
Now I am actually analyzing real samples. After a calibration check in the morning of 2-3 standards in neat solvent, I run a series of 4 samples then another standard, more samples then std etc. Checking back on the very first batches there was none of what I now see ... the standards injected after samples have a response of anything from 3-5 times the morning response. This is not due to a drift in instrument response - that was checked during initial calibration, no more than 25% difference, morning to night. It is ameliorated if I just do a blank solvent run but is still not within the acceptable calibration range. I have read of matrix enhancing effects on samples injected but not for standards injected after samples.
Standards in sample matrix could be a solution, but would be a major pain as I effectively have four matrices. Also how do you take into account the concentration of the matrix? Additionally I imagine that there will be a strong possibility of std deterioration in such matrices. Is further cleanup the only real solution? I have read that even that may reduce colour etc but not the matrix enhancing effect. Any thoughts would be greatly appreciated.


Top of pagePrevious messageNext messageBottom of pageLink to this message  By Scott Fredrickson on Monday, August 12, 2002 - 05:21 pm:

Sue--
Yes, stds in matrix are a pain. Yes, you have to take into account the matrix concentration -- somehow. I group them by dilution, sometimes by how dirty they look, or by matrix if experience says that works. And yes, it is wise to include stability of the extracts in your validation, if the analysis will not be immediate.

Limiting the number of samples between standards, if you can do nothing else, is the best way of tracking the changing response, and controlling the maximum error. Averaging standards on either side of samples will help, or calculating the samples either side of the standards (2 samples, stds, 2 samples) for your std/4 sample/std routine.

You might be able to use internal standards, but be careful! The internal standard must display the SAME behavior as your compound(s)of interest. If it doesn't, your errors will be even worse. The only way to be sure of this is by validating very carefully.

Soil is probably the worst matrix, especially if it has a high organic content. I believe it is because of the organic acids generated from decomposition. Soil extracts will also destroy LC columns, which indicates how bad they can be.

One of our labs uses standards in matrix for their water samples for the reasons mentioned. They also use megabore columns and FPD's for OP's whenever possible. They can do work down to the 0.01-0.05 ppb range with these systems. Recently, we have been using LC/MS for confirmation of some of the OP's such as dimethoate and diazinon, but not all OP's have good LC/MS response. We're still learning.

The frustrating thing about it all is that there may be no, or very little, indication from your system that anything is wrong. No extraneous peaks, no change in background signal, nothing. The response just goes up, up, and up. Or down, down, down. That's because the 'stuff' is adsorbing to the active sites, never getting to the detector, or not seen by the detector. And bleeding off so slowly the change is not noticeable over a short time.


Top of pagePrevious messageNext messageBottom of pageLink to this message  By sue coward on Tuesday, August 13, 2002 - 03:49 am:

Thanks for your thoughts Scott. The major problem is, as you imply, that the response can be as unpredictable as hell. It seems to effect lower concentration standards more than higher conc. (which sort of seems logical somehow). With the old column I had stabler results with slowly ramped temps during the gc runtime for the compounds of interest, followed by a rapid rise to a higher temp held for a short time to "burn" off what I didn't want on the column. Killing that high ramp part with the new column seems to have dropped the amplification effect of the matrix somewhat. Whether in the long term that's a good idea ...
I will spend the next week or so looking into whether matrix conc. has any relevance or not for my samples (life will be easier if not, but I won't be holding my breath!) and on the stability of the standards within such a matrix. Further cleanup is obviously an option too.
It's all part of the great learning experience I guess, even if at the end of the day you only have a 2 page article to show for all the months/years of work behind those ppt or ppm results. That's research for you! Always expect the unexpected!


Top of pagePrevious messageNext messageBottom of pageLink to this message  By Ron on Tuesday, August 13, 2002 - 02:54 pm:

Sue,

One thing that you may want to consider is a change in the liner, column, or both, that you are using. There are liners available from both SGE and Restek that are deactivated using a polymeric process, and in addition Restek has steel liners that are Silcosteel coated. Some labs doing pesticide analysis report better lifetimes and enhanced response, other labs see no difference. The liners are more expensive. A thicker phase column may also help, e.g. 0.5 instead 0f 0.25.


Posting is currently disabled in this topic. Contact your discussion moderator for more information.