retention behavior on a HILIC column (new to HILIC)

<p>Hi,</p><p></p><p>I'm currently developing a method using an Acquity BEH HILIC column (2.1 x 100mm) to seperate 4 basic compounds. While I have a number of years of experience with RP-HPLC, I seem completely turned around by the concept of HILIC. I have found a Waters aritcle by Neue et. al. (Sep. Sci. 2008, 31, 1551-1518) comparing pH and organic modifier and their effects on retention time. I'm using much of the mobile phase and diluent compositions as written directly out of the article and having some pretty OK separation. </p><p></p><p>Mobile phase A: ACN:10mM Ammonium Acetate, pH 3, 5, 6.8 or 9 (95:5 v/v)</p><p>Mobile phase B: ACN: 10mM ammonium acetate, pH 3, 5, 6.8, or 9 (50:50 v/v)</p><p>strong needle wash: 80:20 ACN:water</p><p>weak needle wash: 20:80 ACN:water </p><p>Sample solvent: 75:25 ACN:MeOH (no buffer)</p><p></p><p>I am trying various pH's to determine which may be the best for my basic analytes (main peak pKa>9)</p><p></p><p>While I am getting superior peak shape of my main peak (versus RP), I notice that there appears to be some drift in the retention time from injection to injection ( i.e. perhaps 0.2 minutes) on only one or two of the peaks. </p><p></p><p>I haven't made enough injections (nor do I have enough experience) to establish a pattern but I'm curious of the following and if it relates to my rt issue:</p><p>1. what is the typical column equilibration requirements of a HILIC column (I know the water layer plays an important role in seperating)</p><p>2. are my needle washes compatible?</p><p>3. should the sample solvent contain some (albeit small) amount of buffer at the same pH as the mobile phase?</p><p></p><p>Sorry for the long question. Any advice would be helpful.</p><p></p><p>Thank you in advance.</p>


  • lizh


    I know HILIC is weird. Doug will proabably reply, but with HILIC water is the strong solvent, I am not sure if 100% organic, would be the best to put on the column initially, but the rules say initial conditions, so I would have esentially one line equivament to your solvent bottles without the buffer.

    Can you share your gradient, injection mode and the nature of the compaound, polar I guess, how's is it solubility.

    Like you I am not comfortable, so will forward to better checmiss than I!


  • You bring up some important considerations for HILIC.

    As far as column equilibration is concerned, a brand new column out of the box should be equilibrated with 50 column volumes of 50:50 ACN:H2O with 10 mM buffer solution (the first buffer that you are screening your method with will be sufficient). If the column has been previously used, performing 2 blank gradients before sample injection, should be sufficient to yield stable retention times. 2 blank gradients is also recommended when changing from one pH to the next. If repeatedly injecting the same sample, 5 - 8 column volumes of reequilibration between gradient runs is recommended.

    For use with HILIC, both the strong and weak needle wash on the ACQUITY system, needs to be placed in a 90 - 95% ACN in water solution (or equivalent organic concentration as the starting mobile phase conditions) without the use of the buffer or additive. This is critical as the weak needle wash will come in contact with the ends of the sample plug.

    The sample diluent also plays a critical role in peak shape and retention. For the best combination of solubility and peak shape, the sample diluent should be between 50 and 95% acetonitrile if possible. Polar solvents such as 100% water or methanol are not recommended, as these are strong elution solvents in HILIC and the polarity of the sample solvent should be minimized. We have found a mixture of 75ACN:25MeOH(sometimes with 0.2% formic acid) will serve as a good all around HILIC diluent. However, it is always best to solubilize your sample in your starting mobile phase conditions (including buffer) if possible. Based on the information you have provided, you have adheared to these guidelines.

    If you are experiencing any other difficulties when developing your method, please post to this forum again, and I will gladly respond.

  • Thanks for the response, Liz,

    the gradient is 1%B to 5%B in 5 minutes. I am using PLNO as my injection mode. Unfortunately, I have not compiled a solubility index for this compound as of yet, but is polar and completely soluble in the diluent above at a concentration of 0.5mg/ml (from visual inspection).


  • Even the aquity uplc has reproducibility problems when running a gradient from 1% to 5%B.

    Please try to use a solvent b with higher conc of acetonitril so your gradient will run from 1% to 20% to 50% or something in that range and check again for rt drift. If the problem is solved it was the gradient otherwise you have a hilic depending problem.

  • lizh


    Running a 1-5% gradient with the BSM is no problem for reproducibility (although accuracy may suffer somewhat).

    The problem is inherent to chromatography not the UPLC specifically. If the gradient is shallow then the peak is eluted almost isocratically and isocratic peak elution time is significantly controlled by more variables than just the appropriate elution concentration. Temperature, sample diluent and wash solvents and other variables become more significant.

    The one area that could possible cause an issue, would be a small leak. (Note, Waters has always corrected for this with a small assumed leak rate. This is critical for example in nanoACQUITY systems). Assuming there is no major leaks (check using a BSM leak test also this could be determined with an overlay of the accumulator and primary pressure traces to determine if there is a pump problem).

    In addition to Eric's suggestions I would make sure that you are running with the column temp turned on and at least 5C above ambient. Overlays of the room temp from the SM would be useful to see if the ambient temperature is causing this. Significant ambient temperature changes (>5°C) can affect retention time (by changing solvent density) even if the column temperature is constant.

    Just a clarification.


  • Eric,

    thank you for your comments/suggestions. It appears that I was not alowing for proper equilibration between injections. Retetion time seems to be fine now.

    On another note about sample solvent: I switched my sample solvent from 75:25 ACN/MeOH to Mobile phase A (95:5 ACN: 10mM buffer, pH3). The API in my standards look beautiful from a peak shape perspective (from 3ul to 10 ul injection volumes), but when I use that same sample solvent to extract the API from the formulation, the peak is split (at 3ul injection volume).

    I have made solutions of all the excipients, and prepared placebo preparations exactly as real samples, but there does not appear to be any interference from any of the components in the formulation.

    I have also made various iterations of the diluent, but no real understanding of why the peak is split:

    75%ACN:25%MeOH diluent looks a little bit better than mobile phase A (but still split). So I tried 50:50 ACN:MeOH (this was worse) and I tried 95:5 ACN: MeOH (which still yielded a split peak). I went back and retried all three of these diluents but added 0.2% formic acid...still split peak in the sample.

    I am still using the gradient listed in the OP. Perhaps the solvent strength is not compatible with the solvent strength of the mobile phase at the time of elution? But that would not explain why the standard looks fine.

    Any thoughts..or is this API just not HILIC-able?


  • I believe what you are experiencing is not specific to HILIC and could be a classic loading capacity issue.

    It is possible that the split peak is due to sample overload, (too much mass or volume on column). This could happen in the formulation injection and not in the standards because there will be the addtional mass of the binding agents and other excipients in solution that will contribute to the mass on column.

    An easy way to test this theory is to try the following experiments:

    1) dilute your sample by 50% (use the conditions that yielded the best peak shape), inject that sample and observe peak shape. Is it still split?

    2) dilute your sample by 50% (use the conditions that yielded the best peak shape), and injet twice is much. That will double the volume of solvent injected on column, but the mass will be the same. Does the peak shape get worse or better?

    3) prepare a sampe that is 2X concentrated than the method you are running now and inject half of the volume you are injecting now. That will cut the volume you inject by 50%, but the mass on column will be the same. Does the peak shape get worse or better?

    Another possibility (less likely to be the case) is that the inlet frit of the column has become contaminated. Have you tried to run the standards again after injection of the formulation and achieve the same good peak shape? If not, you may have contaminated the frit with some of the excipients.

    Let me know the results.

  • Eric,

    thanks so much for your reply. I have been running my 0.1mg/ml standard with every sampleset as my "canary in the coal mine". The standard would let me know if the columnn is getting clogged...and it has not provided any hint of that.

    I carried out the steps as written and was very surprised by the results. Recall that my standard @ 0.1mg/ml (3ul injection, PLNO) is perfect and my sample @ 0.05mg/ml (3ul inejection, PLNO) was split

    1. sample diluted 50% (injecting nominal volume of 3ul) resulted in a much improved peak.

    2. sample diluted 50% (injecting 2X, 6ul) resulted in the peak to be split as worse original sample at 0.05mg/ml

    3. sample prepared at 2X concentration (injecting 1.5ul), yielded the worst peak shape

    I then went back and did a bit more thorough scan of concentration and injection volume. I prepared samples @ 0.005, 0.01, 0.015, 0.025, 0.1mg/ml and injected 3, 6, 8 and 10 ul of each solution to determine the point at which the peak would begin to split. I've attached some peak characteristics to share. I'm thinking that the yellow shaded concentration is a good starting concentration. of course, determination of DL and QL will let me know if this is true.

    I am bewildered at the fact that I can prepare a standard at 0.1mg/ml using mobile phase A, but I cannot prepare a sample at half the concentration. As mentioned, I prepared placebo's and individual excipient solutions to determine if they would interfere. And nothing suggests that it should happen from a chromatographic satndpoint.

    As I write this, I am realizing that the formulation is prepared at pH~5. The diluent is mobile phase A (95:5 ACN:10mM ammonium acetate bufffer, pH3). My sample is prepared by 'extracting" the API via sonication. Since the diluent only contains 5% buffer, perhaps there may be a conflict in pH in the flask of the sample causing changes in peak shape at higher concentration/injection volume which is not seen in the standard because the only pH the standard "sees" is the pH3 buffer exclusively?

    For now, I will work within the range I have indicated in the .xls file. Unfortunately, I may have to return for some guidance on increasing resolution (doh!).

    Thanks to all that have contributed.