Have you observed differences in UV response for some compound solutions in HPLC vs UPLC comparisons

hagermi1

<p> While trying to validate a method for choice of HPLC and UPLC instrumentation we found a bias in response for one compound out of the four of interest.  Three behave similarly on HPLC and UPLC, giving assay results that agree between the two techniques.  The fourth compound is the last of those in elution order by reverse phase, therefore the more hydrophobic.  The UPLC conditions employ same binary mobile phase but gradient is steeper and is shorter run time.  Same UV detection wavelength selected, 270nm for both instruments.  Consistently, the UPLC assays run 1.5% -2.0% higher, relative, than on the HPLC for this fourth component.  Chromatography looks equivalent and specificity has been tested and is equivalent.</p><p></p><p>I should add that the standard solution and sample solution have slightly different solvent composition out of necessity.  Normally we make these the same in methods where possible but cannot when standard includes, say a pure acid form of analyte and the formulation is with salt form of differing solubility.  Otherwise, the final dilution step of standards and samples are as close in composition as we can design.</p><p></p><p>Any ideas on sources of bias to investigate?</p><p></p><p>Thanks,</p><p></p><p>Mike</p>

lizh

Hello Mike

I am not sure that we can pinpoint the issue without you HPLC and UPLC method and sample prep details. We need full detail of the HPLC and UPLC methods - injection mode and needlewashes included and the nature of the HPLC/UPLC detectors and their settings for comparion.

Also, can you explain why the 1.5 to 2 % difference is of concern, is it because it is only the single component that is being discriminated? What would you expect?

I am assuming this is a RP method and as such the last compound probably is the most non-polar/hydrophobic also will have the widest peak width. Therefore, there are multiple things that could be at play

• Are you close to sensitivity limits with the last peak?
  
• Have you check for carryover or extended the run time?
  
• Could this be a recovery/solubility issue (hence I asked for the injection mode) as you mentioned there was a sample prep deviation.
  

At this stage method and chromatogram, as a picture in these cases is worth a x1,000 words, and would help most.

BR

Liz

hagermi1

Hi Liz,

I'll attempt to attach chromatograms and injector - detector conditions.  Please note, though, that at time of initial posting I saw a discussion on PLUNO injection settings that I thought might be relevant because of symptoms expressed.  Our initial tests (varying flush settings) do give some promise.

We had been using PLUNO with the automatic settings in the "Advanced" feature set.  However, per the instruction in earlier discussion initiated by "Bart J.", I think(?), I decided to experiment with flush settings in "Advanced" settings: 

quote from a previous discussion:  "....I talked yesterday to Ilse Hoes and she came up with the idea of migration of one peak in the mobile phase/WNW (I'm not sure anymore).

She suggested to increase the needle overfill flush to 30µl so that it will be comparable to FL..."

=============================

I would like to see more details on this overfill flush, an explanation of what it means.  What does the default, automatic, setting actually give you?  Our preliminary test data for either a 3 or 30ul setting gave results comparable to HPLC data, whereas the automatic setting gives assay differences.

Thanks for any suggestions or clarification on how PLUNO works, specifically - how to optimize settings.

lizh

Hello

Thank you and if you are working with IIse, you are in good hands.

If the increase of overfill makes a difference than the sample would rather come out of solution than stay solubilized. In my experience this often indicates that the sample diluent needs adjustment to solubilize that component.

Depending on the stage of the method development, I would try a change in the sample diluent.

I attach an explanation of the three modes which I hope anwers your question. In FL and by increaing the PLNO overfill volume more of the surface see the sample and diluent - it' like pre-rinsing a pipette.

BR

Liz

The ACQUITY UPLC Sample Manager uses a fixed loop injection scheme.  It provides three modes of injection:  (1) full loop for maximum precision and column efficiency for a specific injection volume; (2)  partial loop for maximum speed and minimum consumption of sample; and (3)  partial loop uses needle overfill (a.k.a. PLNO) for best recovery and efficiency with partial loop injections.  It must be noted that all mechanically driven injections result in a dispersal of the sample as a result of viscous flow.  In full loop mode, the sample loop is overfilled several times to ensure that the diluted portions of the volume drawn from the sample are not injected and that the sample loop is filled with a “heart cut” of the sample solution.  Because the internal surfaces of the sampling needle and sample loop are thoroughly rinsed with the sample itself, the effects of solute adsorption and solvent exchange with the needle wash and mobile phase are reduced.  It should be noted, that volumetric accuracy in full loop mode is ultimately limited by the tolerances of the tubing used to construct the sample loop.

Here is the entire explanation.

hagermi1

Testing confirms that the advanced PLNO setting for overfill volume affects recovery for our more hydrophobic compound in the four compond mix.  However the test gives one surprise in that 3µl and 30µl overfill gave recoveries comparable to conventional HPLC data.  In contrast, 15µl overfill gave poor recovery, much like we were observing with default, automatic settings.  We have some indication that 15µl happens to be the default...Can you confirm?  The surprise is in the 3µl performance matching the good recovery of the 30µl  overfill.  This doesn't make sense to me.  Are there any engineering differences in the way 3, 15, and 30µl overfill settings are executed?  I'm looking for additional information and guidance on use of this setting option.

Further comment.  It seems odd that this "overfill volume" can test to be of such importance and yet is an almost hidden controllable setting.

By the way, I have not had correspondence with IIse... I was merely quoting from a different posting earlier.

lizh

Hello

I have passed along to the experts, and will get back to you for additional thoughts.

However, I must stress that there is no engineering difference between the settings, and it is very important that you accept that this is not an engineering issue - this is chemistry, there is a mismatch between solubility of the compound, the diluent, the WW in the loop that the sample diluent replaces and/or possibly an effect of the gradient initial conditions. We need to be given some of the method’s critical parameters to explain what you are seeing.

But to answer your question Waters determined experimentally that in general 15 ?L is usually a good choice, we conducted a wide number of experiment using samples with wide polarities. However, the overfill is less useful if the other method parameters are not optimal. And this explains why we placed it in an area where the user would not use this setting frequently and would consider other choices before making a change here.

I am speculating that the effects you are seeing remain a sample to sample diluent effect or sample to WW effect or sample to gradient.

To help, think about it in this way. If I assume that you have a 10 ?L loop and are using a 2 ?L injection volume.

With a 10 ?L loop and a 2 uL injection a 3 uL overfill makes the total sample slug 6 ?L out of a possible 10 ?L of total loop volume. Before the loop is filled with the sample it is full of the weak wash. Here the WW is 50/50 ACN/Water and some will still remain in parts of the loop.

For a 30 ?L and a 2 uL injection, the sample slug is now a total of 33 ?L and the weak wash of 50/50 gets entirely replaced, with almost 3x the volume of the loop.

The parameters that will help us explain this phenomenon, are at a minimum the following:

• What is the WW composition?
  
• What is the sample diluent?
  
• What is the initial condition of your gradient table?
  
• What is the injection size and the loop size that you are using?
  

What happens if you make the WW more like your sample diluent or is it the initial gradient composition is affecting the outcome, is it too hydrophilic for your compound?

BR

Liz

hagermi1

Answers to your method condition questions follows:

• What is the WW composition?  50% acetonitrile/ 50% water
  
• What is the sample diluent? Samples in 50% acetonitrile/ 50% aqueous, modified with pH 10 buffer
  
• • Standards in 60% acetonitrile/ 40% water, modified with formic acid
    
• What is the initial condition of your gradient table?  60% aqueous, modified with 0.1% formic acid/ 40% acetonitrile; gradient goes to 75% acetonitrile at 0.7 minutes.
  
• What is the injection size and the loop size that you are using?  2 µl Injection on a 10 µl loop.
  

What happens if you make the WW more like your sample diluent or is it the initial gradient composition is affecting the outcome, is it too hydrophilic for your compound?  We have a mixture of hydrophilic and hydrophobic compounds…The compound with recovery issue is hydrophobic.  The WW is close to sample diluents in composition.  Initial gradient is 60% aqueous compared with 50% sample diluent or 40% aqueous standard diluent.  There are no red flags based on peak shape or chromatography for any of the compounds of interest.