Low Pressure vs High Pressure Mixing HPLC's

[Jimi]

Hey Folks

I'ld like to ask a question about Low vs High Pressure Mixing HPLCs. When I used to work in the lab - about 10 years ago - I recall the situation being as follows.

Most HPLCs were low pressure mixing systems, and the benefit of this was that you could use two, or even four, mobile phases in your gradient, but you only needed one pump. High pressure mixing systems were used only on Ultra-High Pressure LC's, and the reason there being the lower dwell volume (which is important in UHPLC).

But I'm realizing that things have changed. In our lab, for example, we have several systems with high pressure mixing, but on LC systems that are not UHPLC. One didn't used to see this combination very often. So I'm curious to ask why this trend has changed.

I can postulate one reason, which is that smaller ID columns (e.g. 2 mm or 3 mm, instead of 4.6 mm) are becoming more common. And for these smaller ID columns, dwell volume can be more of an issue, even when working at conventional pressures. Is that all there is to it? Or is there anything else I may be missing?

Thanks in advance for sharing any insight.

[mattmullaney]

Well, I sort of agree with you...one thing for certain, the less the dwell volume, also the quicker the gradient reaches the column (if you're doing lots of gradient work), so the quicker (in principal) the separation may be completed. Also, the lower system volume is, overall, the lower the dispersion of the resulting peaks...for sure, but these days there are many options for low-pressure mixing (e.g. the Acquity-H and the Agilent 1260) but with UHPLC pressures after the mixing (1000 and 600 bar for those two systems in particular). Of course, for systems such as one flavor of Acquity-I and the Agilent 1290, these are high-pressure mixing as well as high pressure after the mixing (1000 and 1200 bar, respectively, if I properly recall). There's likely a number of other examples from other vendors, and I mean those folks no slight.

Don't know that the trend has changed over from low-pressure to high-pressure mixing for HPLC applications at, say, less than 400 bar (at least not where I work)...but I speak only from my own experience. Others working at other places easily may observe exactly the same thing that you do.

[DR]

High pressure mixing systems can provide lower dwell volumes and can be simpler in some respects but they generally provide a bit more baseline noise, especially at the extreme ends of gradients with very large percentage swings between mobile phases.

[uzman]

Low pressure mixing systems may produce air bubbles if the solvents are not properly degassed so a degasser is a must.

Also gradient proportioning valve malfunctions more probable than high pressure pump malfunctions .

[Klaus I.]

In my opinion most of applications are well served by both systems. The differences in dwell volume are not so big for modern systems but the statement from DR is surely correct. With high-pressure systems are lower dwell volumes possible.

Beside of the lower costs, the biggest advantage of low pressure system is obviously the possible mixing of more than two mobile phases. This might be very helpful for method development when you want to examine the influence of an additive concentration. For example
Phase A: Water
Phase B: Acetonitrile
Phase C: Additive stock solution (e.g. TFA 0.5 %)
Now it’s easy to make several runs with different ratios of water and buffer for the desired additive concentration (e.g. TFA 0.05%, 0.10% and 0.15%).

I have to agree to uzman statement, low pressure systems are less robust. Due to the volume contraction during the mixing of water and organic on the low pressure part, bubbles can occur. When buffers are used and mixed with organic it is possible that a local insolubility occur and the salt might be partially precipitate and clogging the GPV. Therefore it is recommended to premix the mobile phases.

For real fast gradients high pressure systems are better than low-pressure systems since the stroke volume will limit the real gradient resolution. Also the method transfer between high-pressure systems of different manufacturers is less complicated and the retention time precision is usually slightly better.

Problems are often observed when a gradient program starts with one of the two pump is driven below a specific flow rate e.g. gradient starts with 0%B. The premixing of mobile phase has sometimes disadvantages und should be avoided in favor of a meaningful flow rate of the second pump e.g. 5%B.

There are also some additional differences due to the volume contraction when mixing water with organic (concerning flow rate correctness and gradient accuracy). But usually these effects are negligible and have no influence to the run-to-run repeatability.

Best regards
Klaus

[lmh]

(1) a lot of the issues that originally differentiated low-pressure mixing systems from high-pressure are a lot less relevant now, because the manufacturers have improved the design of both. Low-pressure mixing systems nowadays have much smaller dead-volumes than they used to.

(2) You have to differentiate between post-column and pre-column system volume. Post-column system-volume is a problem because it causes peak-dispersion and it's an unavoidable delay. Pre-column is less serious because there is no peak to disperse at that point! It delays the arrival of the gradient at the column (which is something for which some software can compensate, by starting the gradient early) which makes runs longer (bad). It also acts as an extra mixing-volume, potentially smoothing sudden changes in solvent composition - but that is probably desirable!

(3) The change in volume on mixing methanol and water is pretty small. I've seen 4% bandied about, but only in unreliable forums. I suppose it's also possible that the change in temperature is relevant if you're worried about bubbles appearing in the mixing chamber of a low-pressure system (increased temperature in MeOH/water mixing should reduce solubility of dissolved air). The real issue about bubbles forming in a low-pressure mixer is what happens in the pump heads after they've formed. Some systems tend to accumulate bubbles in the pump-head, which is disastrous. Others are more tolerant, and flush them straight through into the high-pressure part. If this happens, the bubble will vanish and redissolve as soon as the pump piston starts moving, and the overall effect will be a 4% loss of flow, and possibly some pressure fluctuation depending on how well the pulse dampener is doing its job. Note that a high-pressure mixing system would automatically have a 4% reduction in flow when creating the same % mix, because the mixing and reduction happen after the pistons that do the measuring. A low pressure system starts off with 0% reduction because the pistons are working with pre-mixed, "pre-shrunk" mixture. As Klaus I. said, this is probably irrelevant in practice. Degassing is, in any case, a jolly good idea.

(4) Pumping low percentages in high-pressure mixing systems. I don't think this is (usually) a problem. What doesn't work is pumping a gradient from 0-3%, particularly at low flow. Basically it's problematic because the step-size of the pump delivering the low-percentage solvent becomes large compared to the overall gradient.

What you have to think about is this: a high-pressure pump can deliver a range of flow-rates in definite steps. The steps are usually very small, but imagine a (rather bad!) pump that delivers 0-1mL/min in 1uL/min steps. If you use it in a high-pressure mixing system, and pump an overall flow of 100uL/min, and try to pump a gradient from 0-10%, the pump doing the low-side of the gradient has to start at 0uL/min and progress linearly to 10uL/min, but it can only do this in 1uL/min steps, so the gradient isn't linear; it's a staircase of 10 steps, and if you do it over 20 minutes, it means you're not really running a gradient, you're doing 10 successive 2-minute isocratic runs!

It's OK to get your pump to start at 0% provided you are working at a flow-rate, and with a gradient, where the lowest flow the pump can actually deliver, and the step-size by which it sets its flow, are too small to matter. You can't start at 0% and run a gradient if that means that your pump is operating at its minimum flow of, say, 0.1uL/min, and the elution of your favourite analyte is critically dependent on this being 0.1uL/min and not 0.2uL/min! But if that is the case, you are in any case stuck, because starting at a higher percentage won't help.

I suppose what I'm getting at is that you can pump a gradient from 0%, but it will actually be, maybe, 0% +/- 0.2%. You could also start at 1% +/- 0.2%, which would look like a 20% variability on the actual pumped mixture (proportionally speaking), which looks bad. You could start at 5% +/- 0.2%, which is a 4% variability in what's actually pumped, which looks a lot better. But in any properly designed gradient run, an actual, absolute variation of 0.2% in the absolute mix is irrelevant.

Hope this makes some sense???

[lmh]

And I agree with everything everyone's said about reliability. Gradient proportioning valves go wrong, and are harder to diagnose than binary high-pressure mixing systems.

[yazmer]

In general, high pressure mixing tends to be better due to lower gradient delay and dwell volumes. however, recent advances with LPGE (low pressure gradient elution) valves have closed the gap between high pressure mixing and low pressure mixing systems if you're flowing at higher flow rates > 1 ml/min.

As for being able to use multiple solvents, you can still get 2 pumps with LPGE valves in them and get the best of both worlds - high pressure mixing and the ability to use many solvents. Some systems even allow you to "blend" solvents with each pump so you don't have to premake mobile phases.

As for reliability, LPGE valves do tend to have issues especially if you're not careful with buffers.

[Jimi]

A lot of good comments here. Thanks for that.

Just to close out this discussion, does the group agree with me about the reason for the trend, i.e. it used to be that non-UHPLC systems were almost always low pressure - but this isn't the case anymore.

My suggestion was that this trend was due to the more common use of smaller ID columns. Which means lower flow rates and larger dwell times, both of which make high pressure systems more attractive.

Fair statement?

[tom jupille]

Just to close out this discussion, does the group agree with me about the reason for the trend, i.e. it used to be that non-UHPLC systems were almost always low pressure - but this isn't the case anymore.

No. Both types have been around for as long as I've been doing LC (1977) as has the debate about their relative advantages/disadvantages. I've seen good and bad implementations of both.

[Klaus I.]

Just to close out this discussion, does the group agree with me about the reason for the trend, i.e. it used to be that non-UHPLC systems were almost always low pressure - but this isn't the case anymore.

I’m not sure that your observation applies generally. Many laboratories are intentional limited to a few device types or manufacturers. I personally have also observed an accumulation of a specific instrument type in some laboratories. But I think the decision for a certain type was usually not made by the question of high- or low-pressure mixing. I was not able to observe any trend in the last 15 years.

My suggestion was that this trend was due to the more common use of smaller ID columns. Which means lower flow rates and larger dwell times, both of which make high pressure systems more attractive.

About which concrete instruments you are thinking here exactly? I would assume that the dwell volume for most of the modern conventional hplc systems (e.g. Alliance, 1100 Bin and Quat, Ultimate) would be quite comparable when standard configuration is used. Of course the instruments of some manufacturers are different.

If I remember correctly an article about the usage of HPLC columns (LCGC some years ago), the conventional columns with 4.6 mm diameter) were still the most popular dimensions and 2 mm columns were mainly used for LC/MS analysis.