Same Column Different LC

[syx]

Dear Members,
We have used determination method for folic acid in dissolution medium (water). The columns are Waters Atlantis dC18 and Phenomenex Synergy 4u Fusion. The mobile phase contains buffer solution and acetonitrile (93:7). We tried the method using both columns in 2 LCs which have different brand. In LC A, we’ve got good resolution between peaks of medium and folic acid in both columns. In LC B, both peaks could not be separated.
We have checked the performance of both LC and found no problem. Could anyone please give me explanation for this phenomenon?

Best regards,
Siswanto Tanuatmojo

[tom jupille]

Depends on exactly what you mean by "could not be separated".

What was the resolution between the peaks on the "good" instrument, versus on the "bad" instrument?

Was the loss in resolution caused by:
- a significant shift in retention both peaks?
- a significant shift in retention of one of the peaks?
- excessively wide or tailing peaks?
- complete disappearance of one peak?

Are you premixing your mobile phase or are you mixing "on-line"?

If premixing, have you tried using the same batch of mobile phase on both instruments?

[syx]

Depends on exactly what you mean by "could not be separated".

What was the resolution between the peaks on the "good" instrument, versus on the "bad" instrument?

On “normal condition” we found a peak when medium dissolution (water) was injected. So, we called it the ‘peak of water’ – Peak A. When we injected folic acid solution in medium dissolution, we found 2 peaks: Peak A and Peak B (folic acid). On LC A, we could find these 2 peaks, but not in LC B.

Was the loss in resolution caused by:
- a significant shift in retention both peaks?
- a significant shift in retention of one of the peaks?
- excessively wide or tailing peaks?
- complete disappearance of one peak?

In LC B, there is significant shift in retention of Peak B and slight shift of Peak A compared to LC A. The peaks in LC A are in good shape: sharp and without tail. The appearance in LC B is shouldering peak in front slope of Peak B. When we injected medium dissolution, we found that Peak A was retained right on front slope of Peak B. So, the shouldering shape on Peak B was Peak A.

Are you premixing your mobile phase or are you mixing "on-line"?

If premixing, have you tried using the same batch of mobile phase on both instruments?

We use “on-line” mixing.

Note: Before the LC run for this research, the tube between column outlet and detector inlet was cracked and a hole was created on the part near the column outlet (+/- 3 cm from the column outlet). I cut the tube and then fixed it again to the column outlet. Could this length reduction be as the reason of the incident?

[Chris Pohl]

syx,

Since it appears that you're getting considerably different retention time in an isocratic separation it would seem that your problem is probably due to one of three below:

1). You are using separate columns for each instrument and there something wrong with one of the columns. You can check for this by moving the column from LC A to LC B. If the good separation follows the column, you need to replace the column your using on LC B.

2). If you are using a low pressure gradient pump, there's probably something wrong with the proportioning valve on one of your LCs. You can check this by premixing the eluent to see if the problem is fixed on LC B (although, I guess, it's possible that this one is giving the correct result and the other instrument has a defective proportioning valve, in which case using the premixed eluent will, instead, ruin the separation on LC A).

3). If you are using a high pressure gradient pump, one of your two pumps used to form the gradient either has a priming problem or a faulty check valve.

The other possibility is that your problem is system dead volume related. A significant disparity in peak shape suggests you have a system dead volume problem so perhaps the tube you had to re-cut wasn't cut square. Do peak A and peak B in the same retention time on both instruments? If so, your problem is almost certainly plumbing related. Do you know anything about the relative cell volumes for the two detectors? Another possible culprit could be the heat exchanger leading to the flow cell. You could have a serious loss of efficiency due to a poorly designed flow cell or heat exchanger or both which is compromising the separation in LC B.

[syx]

Mr. Pohl,
1) We use both Phenomenex Synergy and Water Atlantis columns on LC A, we found no problem. We use both column on LC B, we've got problem.

2) Let us see LC B in this case. Honestly, we did not try to make premixing mobile phase. What we have done was to change reservoir to other inlet tube, i.e., initial state: 1) ACN; 2) buffer solution --> changed to 3) ACN; 4) buffer solution…
The problem was still occurred.

Peak A and peak B from different LC in different retention time each. Peak A and B could be separated in LC A (resolution greater than 2.0), but created shouldering one peak in LC B.

[DR]

I suspect that if you premix a batch of mobile phase for LC B, the problem will go away. Some LCs are better at mixing on the fly than others are. Even with a decent LC, 93:7 is pushing a point for online mixing, especially in a 2 pump, high pressure mixing system. Switching from lines A&B to C&D should have made no difference unless you actually have 4 pumps. I would guess that you have 2 pumps, each having 2 lines coming in on the B system...

[syx]

I have tried to premix the mobile phase today. The result was the same with online-mixed mobile phase.
What is the effect of column outlet-detector inlet tube’s length to the substances those have been separated by the column packing?
Could the position of detector module give the influence too? Before I cut the tube, the position of detector module was beside the column-oven module. After the tube was cut, the tube was not long enough in that position. Then I put the detector above the column oven module. The last formation needs shorter length of tube. See picture for details.

[DR]

But I can tell you that the length of tubing between column & detector can make a difference. Shorter is better as extra column effects (diffusion and band broadening associated with it) are a function of total extra column volume. It is important to minimize system volume wherever you can. Pre column reductions help reduce "gradient lag" while post column volume reductions lead to taller, narrower peaks. In addition to minimizing the length of tubing between column & detector, check the width too. Switching from 0.020" to 0.007" ID (or even smaller if your flow isn't too high) can be helpful.

[syx]

It means reduction of tube lenght I have done is not the source of problem...
I really do not have any idea about the source of problem

[ym3142]

it is really interesting. I will believe you have right bottle of ACN and right bottle of Buffer

[unmgvar]

Syx,

are your HPLC from different brands?
do you use the same flow cell path size for both HPLC's or do you have strong differences (one 5mm the other 10mm, something like that)?
could it be that you have different UV setting for time constant and sampling period between the two methods?

[syx]

are your HPLC from different brands?

Yes

do you use the same flow cell path size for both HPLC's or do you have strong differences (one 5mm the other 10mm, something like that)?

Length of light path on both = 10 mm

could it be that you have different UV setting for time constant and sampling period between the two methods?

LC A:
- Sampling periode: 400 ms
- Slitwidth: coarse

LC B:
- Time constant: 2.00 s
- Sampling period: 2.00 s

[unmgvar]

Syx
for what my 2 cent are worth it seems to me that you have 2 reasons for your problem:

the first one is dwell volume, since these are 2 different HPLC brands, the difference in dwell volume can be the cause of your difference in RT. this provided of course that you checked the same column with the same mobile phase in both HPLC. If each time you also made a new batch of mobile then this would also add to the inconsistency.

second you have very different UV settings for both detectors.
detector A is a PDA, detector B is a regular UV.

you should change the setting of detector B to 1s time constant and at least 400 ms sampling period, i would suggest moving to 200ms on both detectors.

your settings for detector B are very "non sensitive" and are generally used for very very late eluting peaks.

at what RT are your peak eluting?
maybe thi is the reason that you are not able to differientiate between both peaks with HPLC B

[syx]

LC A:
Main peak: approx 4.8'
Water's peak: 4.3' + 4.4' (the latest as negative peak)

LC B:
Main peak: approx 4.55'
Water's peak: 4.5' + 4.7' (the latest as negative peak)

[unmgvar]

Ok the RT you gave here seem to fit with my assumptions, very early eluting relatively. width is small apparently and i am guessing that peak height is also small.

you should have you detector B set for:
time constant 1sec
sampling period 200 ms.

on those settings your detector can safely see 0.2 min width peaks and this will probably give you the necessary separation and resolution you were missing.

the settings on detector B before that were making it work "lazy"