Step-by-Step Tutorial

Important: Before going through this tutorial, we recommend that you read the Retention Predictor Overview.

The following are step-by-step instructions on how to predict retention times with this software using the built-in isocratic retention database:

Step #1 - Select Standard Column

We collected our data on the Waters Acquity BEH C18 (2.1 x 100 mm, 1.7 um) column, so make sure it is selected.

In the "Standard column:" box, select "Waters Acquity BEH C18 (2.1 x 100 mm, 1.7 um)"

(At this point, there is only one standard column for which we have any data.)

Step #2 - Enter Approximate Gradient Conditions

Enter the following values in the text boxes:

In the "Initial solvent composition:" box, enter "10"

In the "Final solvent composition:" box, enter "100"

In the "Gradient time:" box, enter "20"

In the "Flow rate:" box, enter "0.2"

These are the values that were programmed into the HPLC for the run. These values are used only as starting points for optimization, so they do not need to be very accurate.

Step #3 - Select Gradient Calibration Solutes

The first list box (the one right under "Step #3: Select Gradient Calibration Solutes") contains a list of all 35 compounds in the retention database. Checking the box next to a compound indicates that you wish to use it as a gradient calibration solute. As mentioned in the overview, there is nothing special about gradient calibration solutes. However, you must select a set of compounds to be used as gradient calibration solutes that elute at evenly-spaced intervals over a wide range of retention time.

Of the 35 compounds in the retention database, we chose
1) adenosine,
2) N,N-dimethylacetamide,
3) p-toluenesulfonic acid,
4) N,N-diethylacetamide,
5) indole-3-acetic acid,
6) dimethyl phthalate,
7) indole,
8) diethyl phthalate,
9) diallyl phthalate,
10) di-n-propyl phthalate,
11) di-n-butyl phthalate,
12) di-n-pentyl phthalate,
13) di-n-hexyl phthalate,
14) di-n-heptyl phthalate,
and 15) di-n-octyl phthalate

Check the box next to each of the above compounds to indicate that they should be used as gradient calibration solutes

Step #4 - Enter Their Measured Retention Times

Now that you've selected your gradient calibration solutes, you must enter each of their measured retention times.

Double-click the box next to the each compound in the table and enter the following retention times:

(If you get tired of entering all of these numbers manually, skip ahead a couple paragraphs to the section labeled, "Shortcut:")

adenosine: 1.407

N,N-dimethylacetamide: 2.018

p-toluenesulfonic acid: 3.684

N,N-diethylacetamide: 4.701

indole-3-acetic acid: 7.302

dimethyl phthalate: 9.475

indole: 10.243

diethyl phthalate: 12.276

diallyl phthalate: 13.642

di-n-propyl phthalate: 15.001

di-n-butyl phthalate: 17.337

di-n-pentyl phthalate: 19.383

di-n-hexyl phthalate: 21.22

di-n-heptyl phthalate: 22.961

di-n-octyl phthalate: 25.051

There are actually 10 sets of measured retention times preloaded into this software. To automatically load these data sets, click on the button at the bottom of the window called "Use Preloaded Values...". This brings up a window that lets you choose one of 5 different experiments, each run on two different HPLC systems. After selecting one of them, you may click the "OK" button to have the gradient calibration solutes selected and their measured retention times entered in for you.

If, by trying this out, you accidently replaced all the data you worked so hard to enter, you can quickly get it back by doing the following:

Shortcut:

Click the "Use Preloaded Values..." button

Under "HPLC Instrument:", choose "Primary Instrument: Thermo Fisher Scientific Inc. Accela UHPLC Pump"

Under "Gradient and flow rate:", choose "10% B to 100% B in 20 min at 200 uL/min"

Click the "OK" button

Once you've entered in the retention times of each of the gradient calibration solutes (either manually or with the "Use Preloaded Values..." dialog), click on the "Next Step" button in the lower right corner of the window.

Step #5 - Back-Calculate Gradient And Flow Profiles

In this step, we back-calculate the effective gradient and flow rate profiles based on the measured retention times of the gradient calibration solutes.

The table on the top right-hand side shows the gradient calibration solutes and their measured retention times. The graph on the top-left shows the ideal gradient profile (black line) and the "working" gradient profile (red line). At this point, the working gradient profile mirrors the ideal gradient profile because it has not yet been optimized. The graph on the bottom-left shows the ideal flow rate profile (black line) and the "working" flow rate profile (red line)

Click the "Back-Calculate Profiles" button to begin the back-calculation process

Back-calculation takes approximately 5 min. During this time, the gradient and flow rate profiles are adjusted to minimize the difference between the measured retention times of the gradient calibration solutes and their calculated retention times (which are calculated from their isocratic retention vs. solvent composition relationships in the database along with the gradient and flow profiles).

After each gradient or flow rate profile adjustment, the variance in the retention calculations is determined (look for the "Variance (σ2):" indicator). If an adjustment is made that lowers the variance, the change is accepted, if an adjustment is made that increases the variance, the change is rejected.

Watch how the variance drops as adjustments are made to gradient and flow rate profiles. You'll also notice that in the table, the calculated retention times are getting closer and closer to the measured retention times.

Optimization of the gradient and flow rate profiles is actually performed in two phases. In Phase I, only the gradient profile is optimized. In Phase II, both the gradient profile and flow rate profiles are optimized. This two phase approach makes it less likely that the profiles will get stuck in a local minimum.

Notice two other indicators:

"Last iteration variance:" shows the variance in retention calculations after the last iteration

"% improvement:" shows the percent decrease in variance after the last iteration of optimization.

A large % improvement indicates that the variance dropped substantially during the last iteration. This also indicates that it could probably still be improved a great deal. On the other hand, a very small % improvement indicates that there probably isn't much more to gain from further optimization of the profiles

Back-calculation begins in Phase I. After the % improvement in the variance drops below 10% (not much more is gained from further optimization of the gradient profile alone), the process moves on to Phase II. In Phase II,  when the % improvement in variance drops below 1%, the back-calculation process stops.

The "Status" indicator will then read, "Optimization complete! Continue to next step." At that point, the gradient and flow rate profiles shown in the two graphs are the back-calculated profiles. They may then be applied to predict retention of other compounds in the retention database.

If you wish to see the back-calculation process again, you can click on the "Previous Step" button, then click on the "Next Step" button, and then click again on the "Back-Calculate Profiles" button.

Click on the "Next Step" button to continue

Step #6 - Predict Retention of Other Database Compounds

The back-calculated profiles are still visible on the left-hand side of the window, but now the table holds a different set of compounds. These are all the compounds in the database that were not used as gradient calibration solutes - there will be 20 of them (35 - 15 = 20).

Click the "Predict Retention Times" button

This calculates their retention times using the newly determined, back-calculated profiles along with their isocratic retention vs. solvent composition relationships in the retention database. One they're calculated (it should only take about a second), take a moment to compare the predicted retention times with those that were actually measured:

The measured retention times of these 20 compounds were as follows:

Compound

Measured Retention Time (tR, min)

acetophenone

8.697

propiophenone

11.004

butyrophenone

12.772

valerophenone

14.364

hexanophenone

15.82

heptanophenone

17.161

octanophenone

18.407

nonanophenone

19.565

decanophenone

20.645

undecanophenone

21.668

dodecanophenone

22.663

1-pentanesulfonic acid

3.795

p-coumaric acid

5.52

4-n-hexylaniline

13.214

indole-3-propionic acid

8.556

indole-3-butyric acid

9.481

nortriptyline

9.657

amitriptyline

9.834

chlorogenic acid

3.628

cortisone

8.409



You'll find that the standard deviation of your retention predictions is approximately ±4 s among all 20 solutes.

Feel free to check the accuracy of the system with the other 9 preloaded conditions. You can find the measured retention times of the 20 solutes under each condition here.

If you're really feeling ambitious (and I mean really ambitious), you could even run the same 35 solutes on your own HPLC system with your own Waters Acquity BEH C18 (2.1 x 100 mm, 1.7 um) column and check the accuracy of retention predictions with this tool.