HPLC problem

[AnneP]

Hi,

I've got a strange problem with my HPLC.
I've got to analyse a polar molecule. I use a C18 column: Kinetex from Phenomenex 2.6µm 100*4.6mm and the mobile phase is KH2PO4 buffer with ion pair agent (Sodium Hexane Sulfonate).
My problem is the following: when I quantify my analyte in a product with an HPLC from Agilent (1100); I only get half of the quantity I was supposed to find. When I repeat the same analysis, with the same vials, but on a Waters' system (Alliance 2695), I find the good quantity of my analyte.
The peak of my analyte is the first peak of the run, and to disolve my sample I have to add CH2Cl2, which makes a peak that goes after my analyte.

I hope you understood the problem; have you ever seen that?
Could it be a problem with the dead volume of the Agilent's apparatus?? Half of the analyte would be lost... I know I have no problem with the injector because I build a calibration table with differents injection volumes.

Thanks by advance for your answers.

[tom jupille]

A bit more detail would be helpful here:

Are you running your calibration standards on *both* instruments?
Do you add the CH2Cl2 to both the samples and the standards?
What is the k' value of your analyte peak (and do you get the same k' value on both systems)?

[Gerhard Kratz]

In addition to Tom's questions I would like to ask you, do you use the SAME column on both instruments? Or do you have 2 identical columns. It would be helpful to know more about the history of your columns. Ion pair reagent on a column will modify the surface. If you use two identical columns, but with different history with regards to ion pair reagent, that might be the problem. In my opinion it is not a problem which manufacturer built the instrument, or the column.

[AnneP]

Thank you very much for your quick answers.

I run my calibration standards on both instruments.
I add CH2Cl2 only on my samples.
The k' value of my analyte peak is 1,83.
I use the same column on both instruments...

[Bluesman75]

Do both instruments have the same type of detector e.g. diode array or VWD?
And do you have the same detector settings on both detectore, for example bandwidths and reference wavelengths?

[carls]

CH2Cl2 has very low water solubility and is not recommended as a sample solvent in reversed phase LC. The likelihood of sample components precipitating and depositing at the head of the column is very high if CH2Cl2 is required to dissolve the sample. I suggest it is time to revisit your method as there will likely be robustness problems.

Is the sample matrix soluble in 100% methanol, ACN, DMSO or THF? These solvents are acceptable for reversed phase LC and will reduce the likelihood of problems. They are also generally strong RPLC solvents so the volume of sample you can inject before seeing solvent effects is small. Precipitation could still be a problem is the mobile phase has a low conc of organic.

However, if your analyte is very polar (I assume this is why you're using an ion pair reagent) and the matrix non-polar (I assume this is why CH2Cl2 is used) then a HILIC separation could be ideal. In HILIC the inital mobile phase contains a high %ACN (80-90 v/v%) and polar compounds are retained while non-polar compounds elute un-retained. This separation mode gives the best performance when the sample solvent contains a high conc of organic solvent, i.e. 90% ACN, which should also dissolve a non-polar matrix.

[AnneP]

My detectors are both diode arrays. For Agilent there is a reference wavelenght (we choose 360nm in general). The detector of Waters is a 2998, we don't need to choose a reference wavelenght. My wavelenght for the assay is 210nm.

I will try to revisit my method
Maybe with SPE I will be able to get rid of the CH2Cl2 I added to disolve my sample.

I've already tried the Hilic mode but I had other problems (a signal less intense)...

Thanks again for all these answers very interesting.

[Bluesman75]

Hi AnneP

my understanding is that you get a low response from the Agilent, but a normal response from the Waters LC using the same column, vials mobiles phase etc. Correct me if I'm wrong.
This would suggest to me that it is an instrumental or instrument setting difference resulting in a response difference, for either the sample or standard, between the two instruments. Having absorbtion in just the sample at the reference wavelength (possibly unresolved impurity?) could result in a low response. Could you rerun with the reference wavelength set at 500nm or turned off?

[AnneP]

Tahnks, I will try to rerun with the reference wavelength set at 500nm

[HPLCCONSULT]

AnneP: I just started reading this thread and see that some good advice has been given. I would like to add (or second) the recommendation to TURN OFF THE REFERENCE WAVELENGTH on your Agilent DAD. Do not use a generic reference wavelength, or any reference wavelength (not 500nm as this introduces more heat and noise from the tungsten lamp), without a sound scientific reason to do so. Turn it OFF. *Please refer to this web tip [http://www.lc-ms.com/Tip%20102%20Reference%20Wavelengths%20HPLC%20UV%20VIS%20DAD.htm]. *As a consultant for major pharmaceutical companies, I see the incorrect use of the Reference Wavelength screw up the results of at least 25% of all the systems I am called in to troubleshoot method development problems. No one seems to take the time to educate themselves about how this feature works and why it was developed. Just because the manufacturer placed it in the software does NOT mean that it should be used.

When comparing the "same" method on two different HPLC systems you must insure that the two systems are configured as close as possible to each other. There are many things to take into account. A few examples: You must compare the flow cell volumes, flow cell path lengths (responsible for the most common reason for a change in the area count), delay volume (if gradient), sampling rate and ALL detector settings (signal wavelength & bandwidth). These must all be the SAME for a comparison to be valid. *BTW: Many instrument manufacturer's take advantage of this lack of knowledge to make one system appear 'better' than another in their demos. Education is your best defense against such biased "comparisons".

As carls said, the removal of the CH2CL2 is of course highly recommended from your sample clean up and method. Especially when running at 210nm!

[HW Mueller]

AnneP, did you say that you used the calibration standard on both systems, that is, you calibrated both systems? Now, unless the ref. wavelength does something I don´t understand, or unless some mistake was made, I don´t see how calibrated systems can yield different values.

[tom jupille]

Hans, that's where I was going with my initial questions.

If when the analyte is quantitated on system A using a calibration plot generated on system A the results are lower than the quantitation on system B using the calibration plot generated on system B, then the problem has to be tied to some difference between the chromatography of the samples and that of the standards.

I can speculate that a difference in peak shape between samples and standards is being handled differently by the Agilent and Waters systems. It would be useful to see chromatograms.

Anne, here's what I would add as suggestions / questions for now:
1. Look at the chromatograms and see if there are any obvious peak shape issues
2. Go back and print out the chromatograms and do a manual quantitation using peak height. If those results look better, then the place to look is at the data system settings.
3. Add CH2Cl2 to the calibrators as well as to the samples. My experience has been that I get the best results when the calibrators are as close as possible to the samples.
4. A k' of 1.83 is not horrible, but it is a bit on the low side. With a "disruptive" sample containing CH2Cl2, Anne may still be seeing "t0" problems. A bit more retention couldn't hurt.