flush all the lines (I have 4 going into the quaternary pump but 2 are never used and even thou they are in solvent they still need to be flushed weekly), specially if you are not using them because bubble within them will cause weird things to happen. Normally I flush those with IPA, but of course you can use methanol or water as well. we could tell something was odd by following the pressure when mixing. We also discovered that our mixing compartment wasn't set up properly so we fixed that too. Hope that helps.

Dear Jalice, thanks for your answer.

We usually flush three of the four lines at the end of each working day. The fourth line is in solvent but it's not used and then it is not periodically flushed. I will do it.

Actually, I'm wondering too whether something is wrong in our mixing chamber. We did some measurements to check the proportioning system: we made a binary gradient A/B using 0,1% acetone as tracer of the solvent in B (abs readings at 265nm). In the first run we put water in A and water/acetone in B. In the second run we changed the solvent in B (methanol/acetone in place of water/acetone)

The gradient program was:

0 equil 100 0

1 0-25 98 2

2 25-50 96 4

3 50-75 94 6

4 75-100 90 10

5 100-125 85 15

The fifth step was carried out only with methanol in B

The results are in the figure (pink line = water/water; blu line = water/methanol). It seems to me that when we use methanol the mixing of the mobile phase is definitely unacceptable. Why does this happen? Is there any other test that we can do? Also to be considered is that we’re using a mixer of 675ul...

Lots of basic information missing. Can you contact someone experienced using HPLC systems to assist you troubleshoot the problems and answer some questions?

• Column Dimensions are needed (Very important to any HPLC topic on mixing).
• Exactly what brand, model and type of HPLC pump and system are you using?
• Are you properly degassing your liquids (both) using modern inline vacuum degassing and/or helium sparging systems? The liquids must be degassed.
• Proper Mixer size depends on the flow rate, column volume and size. Normally, the standard mixer will work well for most methods, but in some cases, it should be optimized (not enough info in your post to know if this should be addressed as it sounds like your issue may be lack of maintenance on the gradient valve).
• Any HPLC system which uses a proportioning valve (aka: "GPV", "MCGC" etc) for mixing 2,3 or 4 liquids together must be properly maintained for it to function properly. Have you run any scientific tests to determine if the pump is operating correctly?- You need to. If you do not use ALL of the channels in these valves on a regular basis, one or more of the seals can fail (stuck or leak) resulting in poor mixing and/or cross-flow contamination.

• "HPLC Gradient Valve / Proportioning Valve / MCGV Leaks. How to Identify Them."; https://hplctips.blogspot.com/2021/08/hplc-gradient-valve-proportioning-valve.html

• "Appropriate Mixer Volume for HPLC and UHPLC Applications"; https://hplctips.blogspot.com/2014/10/appropriate-mixer-volume-for-hplc-and.html

BTW: This statement that the HPLC "specialist" told you: "that the problem is the composition of the mobile phase, which is "not compatible" with a low-pressure mixing system."

- - probably 99% chance this statement is false and misleading. With proper degassing, no problems in general. There are compositions and methods where this may be true, but they are very few. Your mobile phase presents no issue at all. Your flow rate poses no issue at all. Low pressure mixing on most MODERN HPLC systems works great. However, you have not provided any information on the HPLC system used or the HPLC column dimensions or back-pressure, so 99% incorrect for now...

William Letter

Hi William,

please find hereafter my answers to your questions:

- Column Dimensions are needed (Very important to any HPLC topic on mixing).

We're using a Luna Omega C18 column, 150x4.6mm, 5um, 100A

In the gradient test reported earlier a union was used instead of the column.

- Exactly what brand, model and type of HPLC pump and system are you using?

Perkin Elmer Flexar LC system, equipped with: inline 3-channel degassing system, quaternary pump, autosampler, column oven, PDA and fluorescence detectors.

- Are you properly degassing your liquids (both) using modern inline vacuum degassing and/or helium sparging systems? The liquids must be degassed.

Yes (see the point above). However, only the A and B lines were degassed, the third degassing channel being connected to the autosampler solvent. The degassing system has been the first system we checked when we started experiencing casual shifts in the retention times. The degassing system is working fine, at least according to the HPLC specialist (I mean the technician who assisted us troubleshooting the problem). We also checked the flow rate, and found it ok. Please note that by using the pre-mixed mobile phase (line A) the ritention times were very stable

- Proper Mixer size depends on the flow rate, column volume and size. Normally, the standard mixer will work well for most methods, must in some case it should be optimized (not enough info in your post to know if this should be addressed as it sounds like your issue may be lack of maintenance on the gradient valve).

The specialist gave us 3 mixers of different volumes and brands (100, 250 and 675ul) but none of them fixed the problem.

- Any HPLC system which uses a proportioning valve (aka: "GPV", "MCGC" etc) for mixing 2,3 or 4 liquids together must be properly maintained for it to function properly. Have you run any scientific tests to determine if the pump is operating correctly?- You need to.

Some tests were initially carried out by replacing the column with a union and checking the mobile phase abs at 265 nm over a four-step gradient: 100% A, 100% B, 96% A plus 4% B, 4% A plus 96 % B, with A = water and B = 0,1% acetone in water. The results were acceptable. Later, I did another test by carrying out a five-step gradient (0 = equil, 2, 4, 6, 10, 15% of B) with A = water and B = 0,1% acetone in either WATER OR METHANOL (see my previous post). By using methanol, as in our real analytical conditions, the results were really poor (the blue line in the above figure). Replacing water with buffer in A did not change the results.

- If you do not use ALL of the channels in these valves on a regular basis, one or more of the seals can fail (stuck or leak) resulting in poor mixing and/or cross-flow contamination.

Usually we use: the line A for buffers (or for pre-mixed mobile phases), the line B for organic solvents (usually methanol or acetonitrile), and the line C for the washing solution (methanol or acetonitrile diluted in water, depending on the analytical method in use). A-C lines are flushed every working day with their corresponding solvent.

The line D is never used and flushed very rarely (though it is always dipped in 5% MeOH in water).

- This statement that the HPLC "specialist" told you: "that the problem is the composition of the mobile phase, which is "not compatible" with a low-pressure mixing system." probably 99% chance this statement is false and misleading.

Actually, this conclusion sounded rather strange to me too and that's why I posted here my question. I know that, before to give up, he did some "mechanical operations" on the pump. Unfortunately, a written report of the work he did wasn't given. In the next days, I will wash the line D with IPA and I'll check the mixing chamber for cross-flow leakages (as reported in the link you posted above). Should you indicate other tests, please let me known.

Your "specialist" does not sound like they have any formal HPLC training either. HPLC systems are complex analytical tools. Is someone else available to help you? You need to start at the beginning and use valid tests to check the different instrument operations, esp the pump. NOTE: The largest error I see is that you stated you replaced the column with a union for your gradient test. Doing so will invalidate most of your HPLC tests as the HPLC pump can not operate properly unless it is under a reasonable amount of backpressure. Operating it under very low pressure, as with just a union in place, results in the pump no longer functioning properly. The data gathered has no value as the pump is not functioning properly. Without backpressure, the pump will cavitate, and loose prime, resulting in very poor baselines (unusable). The column normally provides the needed backpressure, but replacing it with an empty union makes the system unusable (basic HPLC fundamental). Please see this article link:

• "The HPLC Restriction Capillary; Troubleshooting, Qualification and Running Without A Column"; https://hplctips.blogspot.com/2018/04/the-hplc-restriction-capillary.html

Thank you for some of the missing info. A few comments:

Based on the column size of 4.6 x 150 mm, with 5u particles, the HPLC pump's standard mixer should be fine (250 ul). At 0.6mL/min (which is too low a flow rate and non-linear for that column size. Should be 1.00 mL/min. Why did you use 0.6 flow rate? It does not follow good chromatography practices. You really need to work with someone who has HPLC training).

You wrote; "The line D is never used and flushed very rarely (though it is always dipped in 5% MeOH in water)"; - - ALL lines need to be used/flushed on a regular basis. NOT using a line (liquid flowing through) leads to proportioning valve problems, where this may have all started (see my earlier article). *Quat system = 4 channels = Use all 4-channels. ***Needle flush lines (on A/I) and pump seal flush lines do not need to be degassed, but should still be flushed out regularly.

You wrote: "I'll check the mixing chamber for cross-flow leakages"- - No, not the "mixer" which is a static device. Leave the mixer alone, inline. The multi-channel valve is the part that must be checked, but only AFTER you learn how to run the pump while under back-pressure so the results will be valid (use a restriction line).

You need to start with a properly functioning HPLC pump before moving on. Verify pump operation, under normal back-pressures, all channels individually, then in combinations before moving on. Stick with the std mixer (no need to mess with things like that, not relevant here). The fact that poor results may have been obtained on the 'B' channel, with Methanol, implies that the 'B' channel seal (in grad valve) may be leaking. Mixtures of methanol and water are challenging for an HPLC pump so a good indicator of their condition (IOW: Service the pump and test).

• "Diagnosing & Troubleshooting HPLC Pressure Fluctuation Problems (Unstable Baseline)"; https://hplctips.blogspot.com/2014/01/diagnosing-troubleshooting-hplc.html

William Letter

Hi William,

I got the point of back-pressure, it is clear. Honestly, I'm quite sure this point was clear also to our specialist when he did the first gradient test, and that he adopted the correct procedure. Most probably I reported an inaccurate information on this regard, my fault. I'm facing now with gradients and mobile phase mixing issues. What I was wondering were the potential reasons underpinning our poor results with the binary method and my question was posted to get some hints to address the problem. All your suggestions have been very useful to me and are warmly aknowledged. We'll put them in good use.

As noted earlier, (1) Verify proper pump operation (check/clean and/or replace: seals, pistons, inlet/outlet valves, pickup frits, lines etc); (2) Verify proper multi-channel valve operation to insure no cross-flow leaks; (3) PRIME the system properly with FRESH solutions and operate at appropriate pressures. These basic steps will identify where the problem is.

• Many more free HPLC troubleshooting articles on the website provided; https://hplctips.blogspot.com/

Using such a small fraction of methanol (3.5 %) against a large majority of acetate buffer pH5.25 puts a high demand on your pumping system, as William Letter explained in detail. I suggest you remain using a premixed mobile phase

Jos Wielders

Hi Jos, thanks for your answer.

Yes, it is what we'll do for the moment.

I hope that after the final checks problems will be fixed and then the binary method can be carefully re-evaluated