Peak retention

[DanielP]

Hi all, apologies if this question is simple but my experience is limited.

We are using a Thermo HPLC system attached to an LCQ MS/MS with a 2.1mm C18 column. We are running gradients of ACN (+0.01% formic acid) and water (+0.01% formic acid) with a flow rate of 0.3 ml/min. We are injecting 10ul of sample in 66% ACN.

We are trying to analyse rhodamine 123 via MS/MS (we have a method for HPLC fluorsecence analysis using a 4.6mm column that works fine with this molecule). The gradient we use is the same as the one for fluorescence, injection at 5% ACN then increase to 95% and hold for a few minutes before equilibrating for next injection. However we are having issues the compound retention on the column. Of a standard curve run 2 peaks came out with a retention time of 12 minutes as expected and 7 came out in the solvent front, there seemed to be no correlation with concentration.

We are having similar problems with loepramide and I am starting to worry about the condition of the column although it is new.

We have tried -
- methanol instead of ACN
- samller injection volumes
- injecting in just water ratehr than 66% ACN

This has really stumped me and I am grateful for ay advice on how to proceed and sort this out.

Dan

[Hollow]

Hi Dan

what are the lengths of the 2.1 and 4.6mm column?
what about the gradient delay (dwell) volume of the system(s)? (point of mixing - column head)

Beside this missing info:
- have you tried to increase the conc of the formic acid? maybe it is not enough to keep your molecule in one ionic form?
- reducing the ACN conc in your sample is in any case a good thing. as close to the initial mobile phase as possible (or below)
- be aware, that by going from 4.6 to 2.1mm, you're decreasing the volume of your column about 4 times, therefore decrease the inj volume by the same factor (-> 2-3 µL)
- this factor should also be applied to the flow rate (0.3 ml/min for a 2.1 mm column seems reasonable to me, corrsp. to ca. 1.4 ml/min on the 4.6 mm)
- if run on the same system but with only 1/4 of the flowrate, your isocratic step due to the gradient delay volume is increased 4 times (eg. 1.2 ml of VDwell with 1.2 ml/min introduces an iscratic step of 1 min to your method, but with 0.3 ml/min it will be 4 min (!) before your gradient reaches your column. This thing needs also to be considered when re-equlibrating after the run.)
Are these isocratic steps comparable on both systems?
- if the column has also different length, then you need to adjust the gradient table as well (keep the ratio of [the volume of each gradient segment / colum volume] constant)

[DanielP]

Hi Hollow, thanks for the ideas!

the 2.1mm column is 150mm and the 4.6mm column is 250mm.

Unfortunately we cannot decrease the conc of ACN in the sample as it is needed for protein precipiation, we could use alternative extraction techniques but these are not desirable.

I will try with an increased concentration of FA and see if this helps.

In the 4.6mm column we use a 25ul injection volume, I have tried a 5mm injection volume on the 2.1mm column but with no difference.

There is a 2min equilibration period after the run, I will try and increase this to 5 minutes and see if this helps.

I am not sure I understand your last point, could you please clarify?

Many thanks
Dan

[Klaus I.]

Many HPLC vendors (at least, as far as I know, Agilent, Phenomenex, Waters) provides calculation tools to transfer lc-methods to other column dimensions. Also AdvancedMaterialsTechnology provides a pdf-document with explanations of the neccessary calculations.

Flow and injection volume are directly proportional to the column volume. To inject 25 µL on a 250x4.6 mm column corresponds to a injection volume of 3 µL on a 150x2.1 mm column. Also the flow-rate of 0.3 mL/min with 150x2.1 mm is transfered too a 250x4.6 mm column 2.4 mL/min, if the same time for the gradient is used.

[HPLCaddict]

the 2.1mm column is 150mm and the 4.6mm column is 250mm.

There is a 2min equilibration period after the run, I will try and increase this to 5 minutes and see if this helps.
Dan

There you are - 2 mins of reequillibration is way too low at this low flow-rate. A rule of thumb is to use at least 10 column volumes for reequillibration. Your 150x2.1mm column has a column volume of roughly 0.3mL. 2 minutes reequillibration at 0.3mL/min is just 0.6mL corresponding to only 2 column volumes. At his low flow-rate, you should use at least 10 minutes of reequillibration. And at this point you've not considered the dwell volume of your HPLC-system. Is it known? Are you using a UHPLC-system or a standard HPLC system? With a standard HPLC you may have 1-2mL of dwell volume, meaning the gradient takes roughly between 3 and 7 minutes only to reach the column! With a reequillibration time of just 2 minutes, you've not even reached the starting conditions of your gradient. When you inject the next sample, you're still running with 95% ACN - no wonder your peaks are jumping directly off the column.
Even 5 minutes of reequillibration is too low - for a test, increase the reequillibration time to 20 minutes and see if retention times are constant then.

[DanielP]

Many thanks guys, I will have a play with those calculators and chage the equilibration times and set a run up for the weekend. Will post the results again on Monday.

Cheers for helping out a newbie.

Dan

[Hollow]

Flow and injection volume are directly proportional to the column volume.

disagree for flow rate!
(I know, PhEur states this as well and would even scale the flow rate of an isocratic method in relation to the column volume...)

Instead, the flow rate should be scaled in proportion to the cross-sectional area (=diameter) of the columns. -> in one model (w/o diffusion coefficients), the x-axis of the Van-Deemter is the the flow velocity u=L/t0 and is in mm/s.
t0 is determined by the column volume / flow, therefore u = L/t0 = L/(VCol/F) = L*F/VCol;
with VCol = dc^2*pi*e*L/4 >> the length will clear out and the linear flow velocity u = F*4/(dc^2*pi*e) >> u = F / dc^2
"e" = porosity = ca. 0.66) [make sure you use the correct units...)

Therefore 1 ml/min on a 4.6mm column would translate to 0.21 ml/min on a 2.1mm column, independent of the column length.
Otherwise you're moving along the Van-Deemters curve and maybe change your plate heights = column efficency.

For my last point, I try to explain it with an example:

Code: Select all

Column1                     Column2

dc1	4.6	mm		dc2	2.1	mm
lc1	250	mm		lc2	150	mm
e	0.65	-		e	0.65	-
dp1	5	um		dp2	3	um
F1	1	ml/min		F2	0.3	ml/min

L/dp1	50000	-		L/dp 2	50000	- (proportional to plate count)
VC1	2.70	ml		VC2	0.34	ml
v1	1.54	mm/s		v2	2.22	mm/s

tg1	Vg1	VCg1		VCg2 Vg2	tg2
min	ml	Vcol		Vcol	ml	min
0	0	0.00		0.00	0.00	0.00
10	10	3.70		3.70	1.25	4.17
15	15	5.55		5.55	1.88	6.25
30	30	11.11		11.11	3.75	12.50

- tg1 are the time steps of gradient1
- with the flow rate of 1, these steps will give the volume of mobile phase used at that time (Vg1)
- express the volume of mobile phase (Vg1) in units of column volumes (VCg1 = Vg1/VC1)
- hold this ratio constant for the new column (VCg1=VCg2)
- calculate the volume of mobile phase for the new column dimension (Vg2 = VCg2*VC2)
- calculate the time needed to deliver this volume with the new flow rate (Vg2/F2)

-- remarks:
a) no correction for the dwell volume is made yet
b) notice that the particle size of column 2 is reduced to 3 um to maintain a constant L/dp ratio of 50000. If the same particle size would be used, the L/dp and therefore the plate count would be reduced by about 40% (50000 vs. 30000 > resolution reduced by about 20%)
c) the flow of condition2 is not strictly scaled (see above)
d) "Klaus I." was effectively doing the same with his flow rate translation of 0.3 > 2.4 and holding the gradient steps constant, but due to the above explanation I would not do it this way...

Another nice translator is from University of Geneva:
"HPLC Calculator" (Excel)
http://www.unige.ch/sciences/pharm/fana ... gement.htm

[carls]

5 CV is usually sufficient for re-equilibration, however, as others have mentioned the delay volume/time of the system must be considered as well. For example, at 0.3mL/min flow rate with 1mL delay volume 3.3min must be added to the re-equlibration step. For a 150x2.1mm column 1CV ~370uL so 5 CV = 1.85mL or 6.2min. The delay time of 3.3min must be added to this so the minimum re-equilibration time is 9.5min.

By the same logic, the 1mL delay volume adds an additional 2.3min isocratic hold (you had 1min with at 1mL/min) to the beginning of your run so you must extend the end time of your run by at least this amount in order to record the late eluting peaks.

I have not tried any of the online calculators so I dont know if any of them account for the delay volume of the HPLC system.

[Klaus I.]

Many HPLC vendors (at least, as far as I know, Agilent, Phenomenex, Waters) provides calculation tools to transfer lc-methods to other column dimensions. Also AdvancedMaterialsTechnology provides a pdf-document with explanations of the neccessary calculations.

Flow and injection volume are directly proportional to the column volume. To inject 25 µL on a 250x4.6 mm column corresponds to a injection volume of 3 µL on a 150x2.1 mm column. Also the flow-rate of 0.3 mL/min with 150x2.1 mm is transfered too a 250x4.6 mm column 2.4 mL/min, if the same time for the gradient is used.

Flow and injection volume are directly proportional to the column volume.

disagree for flow rate!
(I know, PhEur states this as well and would even scale the flow rate of an isocratic method in relation to the column volume...)
...
d) "Klaus I." was effectively doing the same with his flow rate translation of 0.3 > 2.4 and holding the gradient steps constant, but due to the above explanation I would not do it this way...

Another nice translator is from University of Geneva:
"HPLC Calculator" (Excel)
http://www.unige.ch/sciences/pharm/fana ... gement.htm

I have never known so far this link, thank you. Besides this, also thanks for the very extensively answer to our new inexperienced member of this board. Maybe we should also tell him, that the Van-Deemter is only surely valid for isocratic methods.

Your example data looks something familiar to me, was this data generarted by Drylab or do you use a different software?

[Hollow]

Your example data looks something familiar to me, was this data generarted by Drylab or do you use a different software?

Thank you I'm glad to help.

To be honest, I've created my own "translator" in Excel, because most of the ones from the manufactures have some limitations on column dimensions and/or numbers of gradient steps.

But the above example was done shortly from a blank excel sheet without much effort (that's why I didn't correct for the dwell times).