Choosing a flash cartridge, and why most bad runs are decided before you start

Most flash runs that fail were already lost before the pump started. The cartridge was the wrong size, or the sample was loaded in the wrong solvent, and no gradient afterwards can fix either.

Here is how I pick, and what I check first when a separation will not resolve.

Start at the TLC plate, not the catalogue

Run your TLC in the solvent system you plan to use on the column. You want your target compound at an Rf of about 0.25 to 0.35.

Too high and the compound runs with the front and everything co-elutes. Too low and it stays on the silica. The band spreads, and you use twice the solvent to elute it.

Then look at the gap between your compound and the nearest impurity. That difference in Rf is what decides everything else:

If ΔRf is near zero, no cartridge helps. Change the solvent system first. Buying a more expensive column to fix a selectivity problem is spending money on the wrong variable.

Size the cartridge from the sample, not the budget

Silica capacity depends on how hard the separation is, not just how much you are purifying.

• Easy separation: you can load roughly 5 to 10% of the silica mass.
• Difficult separation: drop to 1 to 2%.

Which gives:

The difference between the two columns in that table is the whole point. The same 12 g cartridge takes five times more of an easy mixture than a hard one.

The common error is sizing on sample mass alone and overloading a difficult separation. You get a broad band, tailing, and fractions that all need re-running. The cartridge did nothing wrong.

Irregular or spherical

Standard irregular silica, 40 to 63 µm, is the standard choice, and most runs should use it.

Spherical silica at 20 to 25 µm gives you more theoretical plates, sharper bands and better resolution, at higher cost and higher back pressure. It is worth the price in three cases: when ΔRf is below 0.1, when you have to load heavily, or when the compound is valuable enough that a second run costs more than the column.

For a routine synthesis workup it is usually money you did not need to spend.

Normal phase or reversed phase

Silica first, for most organic synthesis. It is cheaper, it takes more load, and the solvents evaporate easily.

Go to C18 in three cases. When the compound is too polar to move sensibly on silica. When it is unstable on acidic silica. When the sample is aqueous and you would otherwise have to dry it first. Cannabinoids, peptides and many natural product extracts are in this group.

C18 flash costs more per gram and the water in the mobile phase makes recovery slower. Use it when you need it, not by default.

The loading mistake, and it is the same one that ruins HPLC peaks

Do not dissolve your sample in a strong solvent and inject it onto a weak mobile phase.

Say you are running hexane and ethyl acetate. You dissolve the sample in dichloromethane or methanol, because that is what dissolved it. The sample plug then travels down the column in its own strong solvent, before the gradient reaches it. The band is spread before the separation starts. You see a fronting or split peak, and you probably blame the cartridge.

This is exactly the same error as injecting an HPLC sample in a diluent stronger than the mobile phase. I lost 2 days to that one at the bench years ago, chasing a fronting peak, swapping the column, re-prepping every sample. The column was fine. My injection solvent was stronger than my mobile phase.

If the compound is too polar to hold on silica in the first place, that is a different problem. Choosing a column for a polar analyte covers it.

The fix in flash is the same principle:

flowchart TD
    A["Sample ready to load"] --> B{"Does it dissolve in a small<br/>volume of your starting<br/>mobile phase, or weaker?"}
    B -->|"Yes"| C["Wet load"]
    B -->|"No"| D{"Does it need a strong solvent<br/>to dissolve at all?"}
    D -->|"Yes"| E["Dry load.<br/>Not optional here"]
    D -->|"No, borderline"| E
    E --> F["Adsorb onto silica or celite,<br/>dry it properly,<br/>use a solid load cartridge"]

Dry loading costs an extra step and it saves the run.

What I check before blaming the cartridge

1. What was the Rf, and in which solvent system.
2. What was the ΔRf to the nearest impurity.
3. How much did you load, as a percentage of the silica.
4. What solvent was the sample loaded in.
5. Was it wet loaded or dry loaded.

4 out of 5 times the answer is in that list, and a new cartridge would not have helped.

If you want a second opinion

Send me the TLC, the solvent system, the compound class and roughly how much you need to purify. I will tell you the phase, the size and the loading I would use, and whether the cartridge you have is already the right one.

The cartridges I supply are on the flash and preparative columns page.

If it is, that is the answer you will get. I would rather you keep the column and come back next time.