Evaluation of self packed MPLC column

[Hollow]

Hello

I'm not very experienced in preparative chrom and would like to ask you some questions.
(I think some questions/doubts are arising cause of too much HPLC thinking... )


We've packed a MPLC column (26x230 + precolum 11ml) with silica gel Si60, 15-40µm (Merck), total packing:

64g silica, slury packing, compacted up to 38bar.

Now we would like to evaluate the quality of the column.
We prepared a test mix with naphthalene/anthracene (5/0.5 mg/ml hexane) and run it with hexane 100% @10 and 20 ml/min

The result was that the resolution and peak shape was "poor" with injection of 1 or 2 ml. Plate count was about N=900 which would lead to a plate height of H=255 µm (corrs. h=6.4 (for 40µm particles))

Injecting anthracene for its own in a 1:10 dilution gave better peakshapes and about 2000 plates (H=115µm, h=2.88)

Beside of this the k' values don't correspond to those obtained by short TLC evaluation (Merck HPTLC plates, Si60).
Don't have the values by hand right now (think the Rf were about 0.5 and 0.7). On MPLC the first peak came out virtually with t0, followed by anthracene.

We then tried another test mix with phenone homologues (aceto-, propio-, butyro-) and hexane/t-butyl methyl ether 95/5.

Propio- and Butyro couldn't be separated good enough, so we prepared a solution of only aceto- and butyrophenone, with some toluene as t0 marker, dissolved in hexan 100%. b(phenones)= about 12mg/ml total.

Injecting of 1 ml gave a satisfying resolution, the k values were about 0.29 and 0.75, t0 was 4.8min, flow 21 ml/min. N was about 1200 (for the second peak.)
Injection of 0.5 ml the plate count rises to about 1800.
Injection of 2 ml, the plate count droped to about 800.

Beside of the plate counts there are these small k values. Evaluated by TLC we would have excpected k's of 1 and 1.6...

Now I wonder if someone can unravel these mysteries...

What effects do I see?
- Volume-overload, even if sample is disolved in eluent? Already with 1 ml or 2 ml?
- Bandspreading before and after column? (bandspreading volume was evaluated to be about 12 ml (w4.4% method))

Which plate count is "true", should be used for the estimation if a separation can be done with this colume?

How do/can I get these significant deviations in k values for silica (TLC vs MPLC)?

Where should I further "shoot troubles", column or flowpath?

Are these observations "normal" for MPLC?

What are some rule of thumbs for MPLC?
- reduced plate heights?
- injection volumes related to the column volume?
- mass load /g silica
- tests for evaluation, if the column can be used or bette should be repacked?

Thank you for your replies

[Uwe Neue]

What a looooong post....

First question: did you pack the column dry or from a slurry? If you packed from a slurry, waht was the slurry solvent?

Independent of this, equilibration can be lengthy process in hexane, and I would not bother trying to measure plates on a prep column in hexane with a retained peak. If you want to get plates in hexane, use an unretained peak to count the plates. Your better plate count with anthracene after the dilution could simply be a better equilibration, and a h of 2.8 is not bad at all...

You need to be careful also about volume overload. You can estimate the peak volume for a 2000 plate column, and then stay far away from this value in terms of the injection volume. Actually, your extra-column bandspreading is so large that it is not a good idea to measure the plates with an unretained peak. You can estimate the peak width as a function of retention, and you can then estimate how much retention you need to get over the large extra-column bandspreading.

In silica, the first suspect causing low retention is water: water on the silica, water in the mobile phase. You did not specify, if the silica is of a good grade, or if there is some effort to control the water on the silica, or if you have paid attention to this at all. Check with your supplier what they do.

To your last questions:

In principle, th plate height expecation depends on where you are on the van Deemter curve. Your test conditions are close to the minmum of the vD for this packing, so a reduced plate height of 2.8 is good.

Your injection volumn, or better you extra-column bandspreading, should be at least smaller than 1/5th of the peak volume.

You need longer retention times.

[Hollow]

Hi Uwe
thank you for your reply.

Water(s)!
This can really point to the center.

We didn't pay any attention to the water content of the stationary phase.

We prepared the slurry with water (!) @ 0.25g/ml.
Used this for packing and compression, about 20-30 column volumes.
Afterwards we changed to isopropanol (water content unknown) and again passed about 15 column volumes through.
Then we changed to hexane and after about another 10 column volumes we began with our tests.

The silica for its own isn't the newest one, as noone was doing prep LC for a long time... Brand is MercK KGaA.

What do you think? Were the euqilibration volumes too low for proper conditioning?

Since the column bed seems not to be so bad, how can we bring this column to proper working conditions (lower the water content)?

Should we go back, and use dry aclohol e.g. MeOH, (maybe we can even insert a drying column before the MPLC column, "packed" with rods of molecular sieve 3A), then again to working eluent >IPA or THF > Hexan? Maybe do additional washing steps before?
Or should we just be patient and circulate with hexane?

How many column volumes would you recommend for conditioning a MPLC column?


Second, the injection volume:
Assuming 2000 plates I would need a k of >5.7 so that the peak volume would be >5*bandspreading, right? (V inj = 20% Vpeak = V bandspreading), t0=4.8 min, F=21ml/min

If I chromatograph substances with lower k's, I will loose some performance to the system and therefore I will not get the "real" plate count of the column but a reduced one for the whole system, right?

for keeping in mind:
Could I say, that, if one injects less than the bandspreading volume, the "real" injection volume becomes the bandspreading volume, and so one can run into volume overload issues even with small samples dissolved in eluents (for small k's)? (Vpeak^2 = (Vinj^2+Vband^2) )

[Uwe Neue]

Alright... equilibration in pure hexane may take half a century and an ocean of solvent (of course, you should recycle).

The silica was definitely wet to start, since you packed in water.

I think the best and most practical thing to do is to go to the target mobile phase conditions for your target compounds, inject your target compounds, and check for retention and plates. (I doubt that you want to separate naphthalene from anthracene.) You want to have reasonable retention for your real target compounds, partially due to the large extra-column bandspreading. You can make an estimate of the plate count that you should get (which is what I did for your naphthalene, anthracene condition). I can advise you how to do that, if necessary. If your plate count is off by a factor of 2 from your estimate, something is definitely wrong in your chromatography.

Next, since the use of a prep column means that you want to do prep, you want to overload the column and get wide and tailing peaks. Under these circumstances it is probably not terribly relevant, if you pack your column to a plate height of 2 dp or 5 dp.

Yes, you need to work at a reasonable k. ~ 5 as suggested, to get a reasonable plate count (this is nearly a liter of solvent for each test !!!)

At lower k, you will not be able to measure the true performance of the column, but you get a number for the "system".

Yes, the "real injection volume" becomes the bandspreading volume, and you will run into volume overload issues even for a small sample volume injected.

Don't forget that real prep chromatography is not analytical chromatography. Wide peaks are good, as long as your separation is OK. You can use many tricks to get a higher load. For example, you can inject your sample in a more retentive solvent (if soluble).

[Hollow]

No, we aren't really go to separate naphtalen/anthracen.

We just wanted to have some quality indices to check if the column will be working and which we could trace for the future. Just simple standard test we can do when packing other columns or to check whether repacking is needed, or for the estimation if the column should be able to do the separation or if a longer column is needed...

For future packing, should we better use IPA to avoid the water saturation of the silica?

The working mobile phase will be somewhat of MeOH/IPA/Hexan 1:1:2 (MeOH gave good separation but to small k values on TLC).
How can one just reduce the solvent strength of MeOH without affecting the separation much? EtOH or IPA for its own didn't do the job.

"You can make an estimate of the plate count that you should get (which is what I did for your naphthalene, anthracene condition). I can advise you how to do that, if necessary. If your plate count is off by a factor of 2 from your estimate, something is definitely wrong in your chromatography."

Yes please, advise me, since I don't clearly understand from which parameters I should estimate the plate count. How can I estimate the peak width?
Maybe you can email me, if you have some files/tables to send.

Thank you for your instructional assistance.

[Uwe Neue]

I sent you a spreadsheet that you cna use to estimate the plates.

I would back silica in IPA, instead of water. Solvent cost could even be lower, since you need less sovents to purge out the water.

IPA and methanol work similarly, but methanol is the stronger solvent. I would try to stick with IPA. However, the prediction of solvent selectivity effects is likely to be as problematic in NP as it is in RP.

With any of the polar modifiers, you can reduce their concentration to get more retention. But since MeOH is a stronger solvent, you may need to go to very low concentrations to get retention. This is why IPA is preferred.

For other selectivities, you can consider acetonitrile as the polar modifier.

Finally, I would pack an analytical-size column with the same silica as you have used for your prep column and do the method development on this column. It is convenient, and it saves you a ton of solvents...

[Hollow]

Thank you Uwe,

received your spreadsheet.

Finally, I would pack an analytical-size column with the same silica as you have used for your prep column and do the method development on this column. It is convenient, and it saves you a ton of solvents...

I allready intented to do so.
Transfering from TLC doesn't convince me, too much uncertainities, foremost not exactly the same silica.