Step gradients - column shock?

[Camisotro]

Ok, I reread your info. A delay volume of 1 mL is not bad at all, and we used to run very fast LC/MS on systems like this. You objected, based on the idea that your loose a lot of sensitivity when you do that. Unless you have a MS system from the stone age, this should not be the case. So let me ask you the question why you think that your sensitivity goes down when you run at 1 mL/min. There could be problem in the thought, rather than a real problem.

Peak area decreases 5-fold when you change the flow rate 5-fold, because the peaks elute 5 times faster. But this is not a loss in sensitivity, unless the detector noise increases drastically with flow. With the MS systems that I know, the noise goes up a little bit with increased flow, bot not 5-fold for a flow rate change from 0.2 to 1.0 mL/min. Only maybe 50% or so. So to run at 1 mL/min is not a real problem.

Uwe, it's not just about noise, it's about ionization. The increased solvent flow suppresses ionization in ESI. I see reduced peak heights at flow rates beyond about 300 µL/minute.

If ESI behaved "ideally" and did not have this suppression property, then LC peak area would be identical regardless of flow rate; at higher flow rates peak width would contract and height would increase. But this is not the reality - with increased flow rate on ESI, I see reduced peak width *and* reduced peak height.

[Camisotro]

You wrote you have 2 isocratic pumps. Try combining them together and build your own gradient pump. You need a low-volume mixer for your mobile phase. You will mix the 2 mobile phases from the pumps on the high pressure side. If your pumps have problems with controling low flows, add splitters (with some restriction capillaries) before the mixer. You can re-use the phase from each pump as you will splits the phase before the mixer.
I know this is a little crazy idea but maybe it could work

Hi Tomasz: The isocratic pumps are even older than the front-panel-controlled 1994 gradient pump, and have no computer control available whatsoever. So there is no way to create a gradient pump out of them.

As fun as it can sometimes be to come up with creative solutions within a severely limited budget, I'm really itching to finish grad school ASAP so I can actually hone my skills somewhere on proper modern equipment!

[Kostas Petritis]

From what you described the best solution would be to split your flow 1/5 or 1/10 and operate at analytical flow rates with your LC equipment. You mention a lot of "wasted liquid" but do you really care about that (most people don't)?

Some modern ion sources work really well with higher flow rates. What mass spectrometer are you using?

[Camisotro]

From what you described the best solution would be to split your flow 1/5 or 1/10 and operate at analytical flow rates with your LC equipment. You mention a lot of "wasted liquid" but do you really care about that (most people don't)?

Some modern ion sources work really well with higher flow rates. What mass spectrometer are you using?

It's an API 3200, running the Turbospray source for ESI. We also have the APCI and APPI sources but these particular analytes work best on ESI.

Anyway I settled on a variation of the switching idea which I'll test in the next few weeks. I load the analyte onto the guard column only at 5-10% methanol, wash the matrix to waste, and then backflush it out onto the analytical column at 45-55% methanol through to the MS. The guard column experiences a sudden solvent change but the analytical column always sees the higher-organic phase except for a short pulse after switching. Then I switch back to re-equilibrate the guard column, and repeat.

For what it's worth, I asked a rep at Supelco if they thought there would be any column shock issues with the abrupt solvent change and the response was, "This should not cause any problems with your guard column. The change really isn't as dramatic as it may seem."

[Uwe Neue]

I am not sure that I agree with you on the sensitivity issue. The real question is signal-to-noise. ESI instruments behave more like concentration sensitive detectors rather than mass-proportional detectors. Thus the peak area decreases with increasing flow. In addition, you will qalso get a small, maximally square-root decline in peak height from chromatography. But the noise remains rather constant. Thus your detection quality and integration quality barely declines with increasing flow. The numbers decrease, but the signal(height)-to-noise barely decreases.

I suggest to think this through. Of course, at some point at very high flow that ESI signal will drop drastically. But from what I remember, this is above 1 mL/min for reasonable instruments. Maybe Kostas can help with this.

[Tomasz]

Why don't you clean up your sample with SPE? If your matrix can be eluted with 5 or 10% of methanol and the analytes need 45 or 55% methanol you could clean up the sample without any problem. SPE will take some time but you will need no gradient elution and no column switching.

[Camisotro]

Why don't you clean up your sample with SPE? If your matrix can be eluted with 5 or 10% of methanol and the analytes need 45 or 55% methanol you could clean up the sample without any problem. SPE will take some time but you will need no gradient elution and no column switching.

SPE is what we'll go for if we can't get adequate sensitivity without. The amount of time required is the possible barrier. Of course once I establish my method I'll be able to get an undergrad to do the sample prep, but still if we can save the time spent on doing SPE with hundreds of samples, we will try to do so.

[sam.pedraglio]

First of all, good morning to everybody, I'm new to this forum.
Next the problem:

Anyway I'm still wondering how bad or not-so-harmful an abrupt switch might be for the column.

We normally work with short step gradient and we don't loose column life-time; our usual gradient are similar to this:

0min 95/5
0.4min 95/5
0.6min 0/100
2min 0/100
2.1min 95/5
3min 95/5

workin at 45 Celsius Deg., 0.6ml/min on 30x2.1mm x3um column

This is what I'd been hoping to do, but the time it takes to re-equilibrate after starting a gradient on our pump is quite long at flow rates that work well on ESI.

On your MS system (API3200) you can control an external divert valve (if you want to use the present one to divert the front of your run to the waste) and working with two columns in parallel: the on-line column feels the gradient from your in-line pump, the off-line column is re-equilibrate from a second pump.
We used this set-up when we're involved in an extensive QC process of parallel systesis: you can save a lot of time.

[Camisotro]

On your MS system (API3200) you can control an external divert valve (if you want to use the present one to divert the front of your run to the waste) and working with two columns in parallel: the on-line column feels the gradient from your in-line pump, the off-line column is re-equilibrate from a second pump.
We used this set-up when we're involved in an extensive QC process of parallel systesis: you can save a lot of time.

Hi Sam, I like the idea of 2 parallel columns; but for an analytical run it could be problematic couldn't it? I would think each column (even if as identical as possible) would have its own quirks and a calibration from one might not work exactly the same on the other.

Curious though, how do you ask Analyst to use alternating columns on a large batch of samples? The only way that comes to mind is to create 2 different method files, one that uses A and one that uses B, and run them in alternation. But this requires a lot of manual fiddling to set up the batches each time.

I've also noticed Analyst *always* insists on switching the valve to B at the end of a run, no matter what the method file specifies should be the final or initial valve position. So you always have to set up your valve such that the position you want it to end/begin at is B. For parallel columns I guess that's not such a big deal though.

[sam.pedraglio]

Hi Sam, I like the idea of 2 parallel columns; but for an analytical run it could be problematic couldn't it? I would think each column (even if as identical as possible) would have its own quirks and a calibration from one might not work exactly the same on the other.

At the moment we used this strategy only in not-regulated analysis and for QC tests.
In any case, I imagine, especially for large batches, it could be better to insert a cross-column validation and then use this way of work than analyse them on just one column.

Curious though, how do you ask Analyst to use alternating columns on a large batch of samples? The only way that comes to mind is to create 2 different method files, one that uses A and one that uses B, and run them in alternation. But this requires a lot of manual fiddling to set up the batches each time.

This is the reason why we work a lot with Excel to create the sample list: you can fill the first 2 lines of the sample list and than use a simple formula in the cells below.

I've also noticed Analyst *always* insists on switching the valve to B at the end of a run, no matter what the method file specifies should be the final or initial valve position. So you always have to set up your valve such that the position you want it to end/begin at is B.

That's right we always start and finish a method with the divert valve to waste.

[Uwe Neue]

With a second pump to flush the column, and the actual chromatography run on a single pump but two alternating columns, you do not need to run calibration runs on both columns. The peak areas are the same on both columns. A calibration run is checking the detector and the injector, but not the column... unless of course, the column has gone dead and gives you double peaks, but this is a separate story.

[bob]

Just a thought

If your MS sends a ready with a contact closure and the autosampler and the pump respond to a contact closure then I suggest that you get friendly with the Electronics department at the university They should be able to easily knock up a relay that can receive the ready from the MS and then be timed to deliver a contact closure to either the pump to start the gradient at time T-1 and then the sampler at time T-2. The MS still controls all.
I had this problem some 20 years ago and before computers and just inserted a resistor capacitor circuit that fired when charged up, Fiddled with it until res/cap were correct.

For what it is worth.....