is there a optimum dp2/l for prep scale separation

[jiangds06]

and what is optimized for?

[tom jupille]

Not really. Prep scale separations are run under gross overload conditions. In that situation, peak width has a lot more to do with the linearity of the distribution isotherm than it does with particle size.

[jiangds06]

So for the same amount of stationary phase, we want to pack it into short and fat column rather than thin, long column if we want to optimize for production rate.

Not really. Prep scale separations are run under gross overload conditions. In that situation, peak width has a lot more to do with the linearity of the distribution isotherm than it does with particle size.

[tom jupille]

Yes.

For "semi-prep" separation (not true large-scale "process scale"), the usual procedure is to start with an analytical column of reasonable length and optimize the conditions to give the greatest alpha between the product and the closest preceding impurity. To get the greatest recovery from a single injection, try for a k' between 5 and 10. To get the greatest throughput (albeit with multiple injections), you try for a k' between 1 and 2.

With those conditions, a loading study, increasing the mass-on-column in geometric progression, gives the loading. As the loading is increased, the product peak will broaden and tail, and the peak maximum will shift to shorter time. When the leading edge of the product peak just touches the closest impurity, you're there.

The next step is generally to a wider column (with appropriately higher flow) and scale the injection to the square of the diameter.

For true process-scale, you would push the loading even higher than that and do "heart-cutting", possibly followed by reprocessing of the early part of the peak, but that's outside my experience (and area of expertise!)