1st injection differs from the suceeding four.

[Jumpshooter]

In conducting suitability assessment of set of chlorophenol analytes, I have observed that the first injection of the "STD" always has the lowes peak area in comparision to the succeeding four injections. Although there is some 'non-significant' rise in peak area values as the inject go on, if I exclude the first inject from data analysis, then the %RSD are all within 5% tolerance. If I include the peak areas for the first inject, then the %RSD is oft > 5% tolerance.
Are the following considered feasible solutions to this problem:
1) state n the Method that a first inject of STD is to be done, followed by a blank, then do 5 inject of the STD and include only the latter 5 values?
2) include the first inject and just state in the Method that the source of variance is due to "instrument equilibration"?
or
3) state in the Method that six injects are to be done---and to exclude the peak area values of the 1st inject?

I believe that I am certainly not the first investigator to have encountered this issue; however, this forum frequently enables the elucidation of problems.

[JGK]

I used to see this all the time especially in HPLC but sometimes in GC, I put it down to the injection system not being fully primed prior to the initial injection. The simplest option for SST injections was option 3, perform 6 injections but as per procedure only use data from the final 5.

[Jumpshooter]

Suggestion duly noted; I wondered how many chrom forum readers would indeed select option # 3. Although the other options are defensible, option # 3 appears to be the most parsimonious.

[chromatographer1]

Since the problem lies in active sites becoming 'filled' with chlorophenol / relateds, I would suggest you specify that one or more injections may be needed to stabilize the system and that as many as 12 injections may be needed to get stable data but only the data from the last 5 (or six) injections will be used for calibration and quantification. After every sample injection I would inject another std and if that standard gave values for the analytes that were within a certain amount specified in the method then another sample could be tested. A injection list might thus be:

7-12 std injections
sample
std (within range of first std set of 6)
sample
std (as above)
sample
std (as above)
sample
std (as above)

and if you wanted your method to be iron clad you would specify that a second set of stds (minimum of 3) be injected after all the samples and that their values would be within XX % of the first set of stds.

that would really give you a rock solid verification of your data.

Regulators would be satisfied. You just have to come up with the percent variances that your chemistry will allow and that will make the regulators content.

best wishes,

Rodney George
consultant

[Don_Hilton]

I have seen this kind of issue and the SOP stated that before calibration, one made a specified number of injections using leftover samples from the previous day - to "deactivate" the instrument. Then calibration checks and/or calibration was run and the lab was up and running more samples... I don't remember the number of injections, but it was determined during method development - and became the rule.

[chromatographer1]

Another comment:

If the amount of 'deactivation' could vary in the future, due to column or sample differences, then a variable deactivation step is preferred. Getting 5 or 6 injections of the std to be reproducible is the goal. Sometimes what worked in early research doesn't always work later. Making the conditions of accepting the hardware for the method flexible saves a world of issues later. Especially if you put into the method corrective actions to initiate if the method just won't work on a certain day due to injector or column problems. Write it in the method so you are NOT boxed into a corner later.

Rodney George
consultant

[Consumer Products Guy]

Suggestion duly noted; I wondered how many chrom forum readers would indeed select option # 3. Although the other options are defensible, option # 3 appears to be the most parsimonious.

OK, I don't know what "parsimonious" means. And I need to finish my taxes tonight, so I'm not going to look it up.

That said, we do #3 where I work. The test procedure actually states to make 6 injections, discard the results of the first, and use the last 5 for System Suitability. On the Sequence Table, our first line is the calibration standard vial but it's listed as a sample so the ChemStation doesn't include it in the SS calculations. Second line is calibration standard listed as "Replace" the response factor. Third line is the remaining four injections of calibration standard listed as "Average" the response factor. The average of the five is used to calculate amounts in samples, and SS is also calculated automatically.

We never trust the first injection in chromatography. NEVER means never.

[Peter Apps]

The most parsimonious and reliable way to solve this is to address the root cause of the problem - sorbtive activity in the system. Passing system QC is not what labs are there to do, passing system QC is a means to the end that labs are supposed to serve - delivering robust, reliable, accurate data.

Peter

[Ron]

It may take some work, and may not always work, but the way I like to address this kind of issue is to try to find modifications that eliminate or reduce the problem. Examples of things I would consider are using a liner with a different deactivation, changing the column stationary phase or deactivation, or going to a column with a thicker stationary phase.

I agree that injections of a high level standard can passivate the system temperarily, and in some cases may turn out to be the most realistic solution, but starting with a more robust method is better if feasible.

[BZK]

The most parsimonious and reliable way to solve this is to address the root cause of the problem - sorbtive activity in the system. Passing system QC is not what labs are there to do, passing system QC is a means to the end that labs are supposed to serve - delivering robust, reliable, accurate data.

Peter

Yes, this is the only way to solve the problems.

Surely the problem could be related to active site inside or outside column. Please don't miss to consider also cold spot outside of liner such as the tubes that feeding the injector. If you use splitless this phenomena could be possible into the first part of cold tubes. With the first run you can depose a small amount of substances into the cold line . Try to inject a pure solvent after first run and see if a little memory effect is present.

[Jumpshooter]

Notwithstanding that sorbitive issues, cold hardware/liners, and column thickness, and other well-reasoned explanations are the possible reasons for this phenomenon (i.e., "latent precision"); I am still confronted with the essential problem: 1) to keep doing injections until SST is attained, then to "bracket" the STDs and prove that SST was in-effect while the intermittent Sx's were assayed or 2) to delve into the issue of trying to "fix" the problem by changing columns and other asundry hardware. I must admit that I am leaning towards option 1 with the proviso that Method states to discard the initial values from the first 2 injections.
I appreciate all the well-reasoned opinions that were offered.

[Don_Hilton]

One other possibility crosses my mind . Add a compund into the mix in the GC vial that is polar and will compete for active sites. This would be something like a poly hydroxy compund that will elute after your analytes. - I heard this described at the Florida Pesticide Residues Workshop a few years ago and I think it was in a presented paper.

[Consumer Products Guy]

In our R&D lab, the columns, HPLCs, and GCs are used by multiple operators and for multiple assays. Say Operator #1 injects a liquid soap samples in methanol, assaying for D-limonene level by GC at 100C isothermal.

Next day, Operator #2 wants to assay liquid soap samples for glycerin, dissolves samples in DMF and makes trimethylsilyl derivatives. Excess derivatizing agent in the first injection will instantly derivatize any glycerin and other components still hanging on the column, and they'll elute in the first run. That's one reason we throw out the first run in chromatography.

[chromatographer1]

CPG has a point which also is pertinent to Jumpshooter's analysis.

Unless his samples are very consistent the amount of activity will vary from day to day. It makes it difficult to prove that any inflexible procedure will always work.

And when it doesn't work, then the bosses want explanations which chemists would rather not offer. Regulators reviewing the work are not kind to ad hoc solutions or when they discover that 'special' corrective actions were taken that were not documented in the original procedure.

Been there, done that, and would rather not have to deal with it ever again.

Good luck with your analysis and I hope it all works out well in the end, Jumpshooter.

Rodney George
consultant

[Jumpshooter]

[quote="chromatographer1"]
Unless his samples are very consistent the amount of activity will vary from day to day. It makes it difficult to prove that any inflexible procedure will always work.

And when it doesn't work, then the bosses want explanations which chemists would rather not offer. Regulators reviewing the work are not kind to ad hoc solutions or when they discover that 'special' corrective actions were taken that were not documented in the original procedure.

*Absolutely, this is what I [and other users of the SOP I'm developing] want to avoid. To wit: I post the draft for perusal by the chrom forum writers--it will be included in section 5.6 of the Method write up.

*5.6 Instrument Preparation
Prior to analyzing manufacturing samples, set the GC (System) up for eight consecutive injections of the working reference standard (w.STD). This procedure is conducted to verify that the System has stabilized because typically the peak area response of the first injection will be less than that of the succeeding injections. Inspect all eight peak area values and verify that the variance (%RSD) for replicates four through eight is within 5%. Exclude the values from the first three injections from data analysis.
5.6.a. If the criterion is met, then the System is considered equilibrated; proceed with five consecutive injections of the w.STD. Evaluate suitability and proceed with triplicate injections of the test samples. After the test samples, then conduct three consecutive injections of the w.STD. The percent agreement between the mean peak area responses of w.STD injects from the pre/post test sample injections should be within 95%.
5.6.b. If the criterion is not met, then proceed to Section 5.7 which explains corrective and preventative actions.