Prep. LC Questions (from a newbie)

[Rob Burgess]

I am very new to this topic, (in fact never have done it) but I'd like to learn about its capabilities.

Some questions:

1) What would be a good "target" amount to aim (mg) for id purposes and for calculating relative response factors when trying to isolate an unknown?
2) What are typical loading amounts (volume and mass) for the various column dimensions used in prep. and semi prep LC?
3) Can we just convert an existing (isocratic or gradient analytical method) and expect it to work assuming we have good resolution for the component of interest?
4) How long practically can one leave a prep. LC unattended to collect your unknown?
5) Say if I had an analytical impurity peak at 0.1% area/area level from a analytical separation loading of a main component of 0.1 mg. What would be the most sensible Prep. LC option to acquire enough material to answer Q. 1?

Any thoughts / viewpoints are very much appreciated... cheers!

[Rob Burgess]

PS. Does anyone know of any good websites / introductory books on this subject that may help answer these questions?

[Bruce Hamilton]

I haven't checked the FAQ here, but I'm sure it will point to some excellent resources.

There are good free resources awailable on the web, mainly at instrument and column suppliers like Waters, Agilent, Varian etc. Just search their literature databases for information on "preparative HPLC".

Agilent also has a free booklet, " Principles in Preparative HPLC - a primer " - March 2004 Pub 5989-0652EN ( might be available on their website as a pdf file - have a look ). I'm sure other suppliers have similar generic guides.

Basically, provided your analytical and preparative columns have the same material ( particle size might be larger for preparative ), then scaling is straightforward, and most column suppliers have the equation and data on their websites or in catalogues for their products.

I'm not sure I fully understand your questions, but here goes.
1. If you're using MS for detection and identification, then ug may be enough, but if you have to perform other tests, the amount will be constrained by those instruments, eg mgs may be needed for nmr.

2. Loading amounts depend on column capacity and resolution, as columns are often overloaded. Look at column supplier's catalogues, but as an example,

If selectivity is < 1.2 then Zorbax RPC columns with different diameters can take 2-3 mg ( 4.6mm ), 10-20 mm ( 9.4 mm ),
50 - 200 mg ( 21.1mm ).

If selectivity is > 1.5 then Zorbax RPC columns with different diameters can take 10 times as much, eg 20-30 mg ( 4.6mm ), 100-200mg ( 9.4 mm ), 500 - 2000 mg ( 21.1mm ),

Note that flows increase with column diameter, so ensure you system can deliver suitable flows and pressure for the prep column.

3. Use the equation provided by suppliers, and you can choose the equivalent prep conditions, but there may be other factors, like reducing solvent use, time on column, or elution column volumes to minimise downstream processing. Be aware that removing large volumes of solvents takes longer, so you may want to avoid sample degradation by optimising your analytical method first.

5. As long as you like, but consider the issues of 3.

6. Depends on your analytical toys - make sure you know what characterisation will be needed, nmr, physical property, or bioactivity testing could require significant amounts. Then scale up to provide that quantity. Note that sometimes it's economic and sensible to buy a smaller column and do multiple runs rather than one large single run on a large diameter column.

I hope this is helpful,

Bruce Hamilton

[Ary]

Hi Rob

Can't say I'm an expert in this but I know a man who is but here's my tuppence worth

1. Not 100% sure about the question here but if you want to identify an unknown I guess you will need enough for an NMR and an MS and probably a couplle of other tests so 10-20 mg would probably be more than enough (depends on what resource you have available) If you mena what can you hope to get from a single shot well it just depends on the separation. Worst case I've ever had I only managed 3mg on a 1" column for a particularly difficult separation but much larger 10-100 mg would be typical

2. See 1 it really depends on the separation and what you are trying to recover, an impurity or just cleaning up the main component. Usually the idea is to get as much material on column in the smallest volume without losing resolution (though sometimes this too is sacrificed)

3. Maybe. Depends on column chemistries. If scaling from an analytical to prep of the same material its usually fairly easy. Changing chemistries from one C18 to another can sometimes be far more difficult than you may expect.

4. Depends on your safety guys usually. If your running at 25ml/min how much waste will they let you accumulate and how will you bring in solvents etc become serious questions.

5. Ideally you want to do this in one shot but it again depends on your separation.

Sorry if some of this is a bit vague. If you want to drop me an e-mail I'll send you some more details and try and get someone you can talk to who does this every day (said I knew a man)[/quote]

[Veronika R. Meyer]

Dear Rob, a long time ago I wrote a paper on preparative HPLC, maybe it can help you: Veronika R. Meyer, Preparative high-performance liquid chromatography as an aid in organic synthesis, J. Chromatogr. 316 (1984) 113. (By then the J. Chromatogr. was not yet separated in an A and a B issue.) It's mainly about normal-phase separations of small organic molecules.
Veronika

[Albany-12303]

1) What would be a good "target" amount to aim (mg) for id purposes and for calculating relative response factors when trying to isolate an unknown?
2) What are typical loading amounts (volume and mass) for the various column dimensions used in prep. and semi prep LC?
3) Can we just convert an existing (isocratic or gradient analytical method) and expect it to work assuming we have good resolution for the component of interest?
4) How long practically can one leave a prep. LC unattended to collect your unknown?
5) Say if I had an analytical impurity peak at 0.1% area/area level from a analytical separation loading of a main component of 0.1 mg. What would be the most sensible Prep. LC option to acquire enough material to answer Q. 1?



Wow! A lot of good questions. I'll see if I can help.


1) I assume that you want to get a conclusive ID, which usually consists of 2 or more tests.

MS or LC/MS

Only a fraction of a Mg is needed.

Proton NMR and or Carbon NMR

A mg or more should be adequate for this (and the tests are non-destructive), so you can recover the impurity if you want to.

PDA

Its nice to have a PDA to help ID the peak in the future should it occur again. This can be done on the Relative Response Factor solution.

To calculate the Relative Response Factor, you will need about 5 mg (if you have a micro-balance). For a regular balance, the amount depends on your weighing SOP's. For instance, if we weigh less than 50mg on the analytical balance, we can only use 1 decimal place. So if I were to weigh 5.03mg, I can only report 5.0mg, so the error can be already over 1% - hence the use of a micro-balance, which allows me to report 5.03mg. Going as low as 3 mg on the micro balance may be OK

2) It depends on the separation. If the unknown is far away from the active and other impurities, over 1mg of the drug can be loaded on. In my case, years ago, I used a 5mg/mL solution and injected 1000ul. My column was, I believe, a 25cm by 30mm ) and my flow rate was 20mL/min.

3) For isocratic, yes! If you scale-up the flow rate properly and use the corresponding prep column, the chromatograms should be very close to the original method. The resolution will not be as good - the thickness of the column allows for more diffusion, and usually the column is overloaded to put on as much sample as possible.

I'm sure that gradient methods would work, but the scaling-up would be more tricky.

4) If you have a big enough mobile phase container, waste container, and collection flask(s) for your fractions, then you can let the system run overnight (or even over the week-end if you are doing semi-prep with a small column that needs more time)

5) Math time!

Lets assume that your flow is 1.0mL/min and the run time is 20 min, and your column is 4.6mm x 30cm.

A common sized prep column is 30mm x 30cm is I believe that this column would cost about 3000 US $.

If you want to run the prep system for 50 hours, thats 150 injections.

30mm/4.6mm = 6.5 mm. So the volume of the prep column is 42 times bigger than the analytical column (6.6 x 6.5 since volume increases by the square of the diameter.

Can you run the flow at 42mL/min? This may require a different pump and UV flow cell. We use an old waters system for this (the new Waters Alliance systems are not very good for this) I think that I could go as high as 35ml/min on my system. Lets assume that you can do 42ml/min (if you cannot go this high, the run time may have to be extended)

Now you can load at least 0.1mg x 4.2 mg onto the column. If the peak of interest is well separated, then you can probably load 10 mg onto the column (assuming that your drug will be that soluble).

10mg x 150 injections gives 1.5g of drug put onto the system, and assuming some loss in the collection and extraction from the Mobile phase, let say that you are able to get 1 mg.

1mg is good enough for identification, but not enough for getting the relative resonse factor. Having 4 mg would be nice - 1mg for ID and 3mg for the RF. This requires 600 injections or 200 hours (more than half a week). And you will need 600inj x 20 min x 42 ml/min = about 500L of mobile phase needed!

If you are collectiing the peak over 1 min, then the amount of mobile phase that contains your unknown will be 600 x 1 x 42 = 25 L!

Note:

The mobile phase should not contain buffer! This makes the extraction of the unknown almost impossible! If you need a low pH for the MP, use formic acid, which is easily driven off by rotavap.

In the above example, assume that we can change the analytical method (by increasing the organic portion in the mobile phase) to use a lower flow rate and lower run time? Say (in the above example) 0.5 mL/min and 10 min run time. This will decrease the collection time by a factor of 4 and then you will have only 6 L to extract from.

Long, tedious work!

Good Luck!

[Albany-12303]

By the way.

I was trying to ID an unknown at 0.1% and I was only able to collect, at the most, 2mg (time constraints). I never had to do a relative response factor, but it is a good idea. The FDA/ICH makes us ID any unknown over 0.10%. If the relative response factor shows the true amount as (for instance) 0.04% then an extensive qualification would not be required.

[Rob Burgess]

Hi Albany,
Many, many thanks for your detailed answer. It was exactly the answer I was looking for and helps my thinking a great deal! It sounds like you have done this kind of work a few times. Does the math usually work out as such especially with respect to time and solvent usage? It's difficult to get my head around these concepts coming from a strictly "analytical" perspective.

Two quick further questions for anybody:

1) It appears that running isocratic is preferable to gradient... is this the case always?
2) I presume we increase the timing constraints by using larger columns. What are the max. sizes available?

We are in a fortunate position that we know what some of unknowns are, but as you say it could be a lot of tedious work. Does it make more sense to actually get our synthetic guys to make the samples for us I wonder?

[Bruce Hamilton]

Unless you have to use gradient, isocratic is alway preferred, often because you can use a RI detector, and you can also use recycle facility with relative impunity. Desirable if you plan to recover and reuse solvent as well.

However, the critical aspect is that you have to resolve the impurity from other impurities, and you have to elute ( or backflush ) all the other material before the next injection. If gradient is required then use it,. Gradient also may reduce solvent use, and that's preferred.

Firstly, you have to determine optimum resolution and retention, then column size options become apparent.

The typical maximum column size is 50mm diameter, which usually is getting up towards 100 ml/min. If you want to purify large amounts, and you have plenty of resolution, consider using a Biotage column system, they go up to 400mm diameter, but don't usually have the resolution for closely-eluting compounds. Cheaper and faster than larger HPLCs, but absolutely no use if you can't get the separation.

In my limited experience, fewer larger runs are preferable to many smaller runs, especially if you've optimised the method for rapid elution and fast cycle time. If you are going to submit sample data to regulators, you may need to show that all runs and feeds were similar.

I'd also caution about response factors - yes, they can be used as a guide, but remember that balance uncertainy bites deep when weighing a mg or two on a 5 figure balance, and you still have to assay the impurity to derive a purity for the solid to generate the factor.

The choice of using synthetic chemists depends on your company, if it's an option, and the impurity is needed for future work, that would be an option I'd favour. Nothing beats heaps of crystals .

Bruce Hamilton