Pushing flash column chromatography loading limits

A question I hear a lot from chemists is “how much can I load”. The answer is always “it depends on your separation quality”.  At that point I begin asking about the TLC data and purification goals. Purification goal setting should be your first step and the question to answer is – what do I need this purification to achieve? Is the goal high purity, high yield, or some combination.  Remember, you will typically sacrifice purity for high yield and yield for high purity so optimization is an important consideration.

In this post I will discuss the results of a study I conducted where I continually increased the crude sample load until my target compound purity fell below specific targeted levels to understand the real loading limitation.

In previous posts I have addressed this loading topic from the perspective of method optimization. Besides optimizing the gradient, choosing the most effective sample loading method is critical.  Clearly, when a separation method and loading technique used do not generate a targeted yield and purity levels for your target molecule, they should not be used.

In my study I prepared a 5-component sample with each chemical accounting for 20% (by weight) of the total. I dissolved the five compounds in acetone to a final concentration of ~ 0.67 g/mL.  My purification goal was 90%+ for my targeted product. A secondary goal was to maximize yield and conserve solvent.

In order to determine the best separation conditions, I ran TLC with two different ethyl acetate in hexane blends (20% and 30%) and calculated the Rf values for each, Table 1. The reason I ran two TLC plates with different solvent strengths is because I know I can use this data to create a step gradient.  Why a step gradient?  Because step gradients provide higher sample loads, typically, compared to linear gradients and consume less solvent – more on that later.

I also opted for dry loading over liquid loading for the same reasons as above – higher loading capacity. Injection solvents can, and usually will, negatively impact the purification, so I usually recommend dry loading.

Table 1. TLC data

Compound20% EtOAc20% EtOAc30% EtOAc30% EtOAc
Migration (mm)RfMigration (mm)Rf
Compound 1450.83440.85
Compound 2280.52310.60
Compound 3220.41270.52
Compound 4130.24180.35
Compound 5100.19140.27
Solvent5452

Preparative chromatography does not need to be pretty to be effective, depending on your goals. Loading at the predicted mass amount or even slightly higher can reap benefits in product yield though with some loss in purity. Overloading to excess, though, has minimal benefit unless your targeted compounds exist in very small amounts – e.g. metabolites. This is why you need to set your purification goals first.

For my mix, a 90%+ purity target with a yield of >70 mg for compound 4 was needed. From my TLC data, a linear gradient based on 20% ethyl acetate predicted a load of 234 mg (Figure 1) while a step gradient based on 20% and 30% ethyl acetate predicted 421 mg (Figure 2).  So, based on my goals and the TLC data, I chose the step gradient and a 10 gram high performance silica cartridge (Biotage® SNAP Ultra).  To get the same load, and therefore the same yield and purity with the linear gradient, a 25 gram cartridge or two separate purifications on 10 gram cartridges are required.

TLC-based load prediction for a 10 gram high performance silica cartridge.
Figure 1. TLC-based load prediction for a linear gradient using a 10 gram high performance silica cartridge using a TLC solvent mixture of 20% EtOAc in hexanes. The load estimation based on the Rf values is 234 mg or approximately 2% of the silica amount.
TLC-based step gradient load estimation
Figure 2. TLC-based step gradient load estimation for a 10 gram high performance flash cartridge.  Two TLC plates were developed, one at 20% EtOAc in hexanes and the other at 30% EtOAc in hexanes.  Using this methodology the load capacity for the 10 gram cartridge is almost double that of the linear gradient.

However, for the sake of comparison I did run the linear gradient.  Using the information from the TLC to linear gradient model I purified my sample at a 2% load (~200 mg) on the 10 gram high-performance flash cartridge, Figure 3.  The separation is quite good but at half the load of what I needed. To achieve my yield goal I would have needed to purify another 200 mg consuming another 221 mL (not including equilibration) as well as another flash cartridge.

2% load (200 mg) purification using a linear 5-40% gradient
Figure 3. Results of 2% load (200 mg) of a 5-component mixture purification using a linear 5-40% gradient and a 10 gram high-performance flash cartridge. The separation is good but insufficient for the goals set for the purification.

At a 4% load (400 mg) using the step gradient the separation was even better than the linear gradient (even at twice the load!). My targeted compound eluted in fractions 10-12 and looked fully resolved from its nearest neighbors, Figure 4.

Step gradient separation of a 5-component mixture (400 mg) on a 10 gram high-performance flash cartridge (4% sample load).
Figure 4. Step gradient separation of a 5-component mixture (400 mg) on a 10 gram high-performance flash cartridge (4% sample load). The separation achieved is improved compared to the linear gradient even a twice the load.

To determine peak purity I analyzed the fractions by TLC, Figure 5.  The fractions of interest (10, 11, and 12) showed no impurities indicating a high purity level.  Each of the peaks’ fractions were pooled and evaporated into pre-tared 20 mL scintillation vials to determine yield, Table 2.

TLC analysis of 4% load.
Figure 5. TLC analysis of 4% load purification show the targeted product fractions (10, 11, 12) to be pure.

Table 2. Post-purification purity analysis data

 Peak 1
(Fraction 1)
Peak 2
(Fraction 4)
Peak 3
(Fractions 7, 8)
Peak 4
(Fractions 10-12)
Peak 5
(Fractions 14-17)
Weight (g)13.031712.921113.093712.919012.9418
Tare (g)12.989212.838113.024412.838412.8609
Net (g)0.04250.08300.06830.08060.0809

The results in Table 2 show that I was able to achieve my goals of 90%+ purity and a yield of >70 mg for peak 4.   By choosing a step gradient I pushed the load limits of a silica cartridge and achieved my goals.

Have you ever needed to push your flash chromatography to its limits?

 

Published by

Bob Bickler

Technical Specialist, Biotage

10 thoughts on “Pushing flash column chromatography loading limits”

  1. Thank you very much for your helpful advice. I still have another question. My compound is very polar and highly water-soluable. In HPLC(C-18), it comes out with 100% of solvent A(water:TFA=1000:1) at 10 min. When I tried to purify it on KP-C18-HS column, it comes out at 2 CV with many impurities with 100% of solvent A (water:TFA=1000:1). It seems that the purification condition does not translate into KP-C18-HS column well. It is said that using over 95% water as mobile phase will cause collapse in the C-18 column, but my compound is very water-soluable. What can I do to improve the performance? Thank you in advance.

    1. Hi Cherry,

      Highly polar compounds can be very difficult to retain and separate. You mention a 10 minute retention time by HPLC – what are the column size and flow rate used? Likewise, what size flash column did you use and what was the flow rate. Also, were the gradients identical or did you just run isocratically?

      Reversed-phase is the usual separation mode but there may be some alternatives. Before we go down that road what can you tell me about this compound? Is it ionizable/ionic? Is it stable in base? In which solvents is it soluble (other than water)?

      Best regards,

      Bob

      1. Dear Bob,
        The HPLC column is 150mm*4mm. The flow rate is 0.8 mL/min. Mobile phase is 0.1%TFA in water (solvent A) and water : acetonitrile : TFA = 90:10 :0.09(solvent B). The elution gradient firstly holds at 100% solvent A for 7.5 min and then increases the percentage of solvent B to 100% over 20 min and holds for 10 min. The retention time of an impurity is 3min and retention time of my compound is 10min. I use a 12g KP-C18-HS column with flow rate of 12mL/min. In running flash column, the gradient holds 100% solvent A for 12CV and then increases the percentage of solvent B to 100% over 12 CV. My compounds comes quickly mixed with the impurity as just one peak at 2 CV. My compound is ionizable with both acidic functional group and basic functional group. It is not very stable in base and is soluble in DMSO besides water, I think.

        1. Cherry,

          Thank you for this information, very helpful. From this information I see your compound elutes at about 6 CV on the HPLC column so I would expect something similar with the flash cartridge. There are two ideas I have that may improve your compound separation on the flash cartridge.

          1. Recondition the flash cartridge

        2. Cherry,

          Thank you for this information, very helpful. From your HPLC information I see your compound elutes in about 6 CV and I expect similar retention with reversed-phase flash. I am suspecting the flash column may not have been properly conditioned in the beginning so I suggest reconditioning it following the steps below.

          1. flush the column with 100% acetonitrile (5 CV)
          2. flush the column with 50/50 acetonitrile/water/TFA (5 CV)
          3. flush the column with 90/10 acetonitrile/water/TFA (5 CV)
          4. flush the column with 100% aqueous TFA (5 CV)

          Following this procedure should help improve compound retention.

          Another idea is to dissolve the sample in a minimal volume of DMSO. DMSO is much more polar than water and should also help if forcing your compound into the C18 phase.

          Bob

          1. Bob,
            Thank you so much for your suggestions. I will try to recondition the column. What percentage of TFA in the solvent do you suggest when reconditioning the column? Is it OK to flush the column with >95% water. Will this cause collapse in the stationary phase?
            Best wishes,
            Cherry

          2. Hi Cherry,

            I would use the percentage you would use in the HPLC method.
            The step-wise conditioning process will minimize the effects of phase collapse. I have been successful operating reversed-phase flash in 100% water. Buffers will help with retention as well as an increase in ionic strength makes the solvent more polar but since your HPLC method only uses TFA then I would stay with that.

            Bob

  2. I have been using the Biotage SNAP KP-C18-HS column(a newly bought one). In the first 2 runs, I can purify my compounds succesfully, but later I can not purify the same sample under the same condition succesfully anymore. Is there somthing wrong with the column? Thank you in advance.

    1. Hi Ling,

      Over time as the number of purifications on the same cartridge increases you will build up a layer highly lipophilic compounds. his contamination can have a negative impact on future separations and this may be occurring with your column.

      Some cleaning options include flushing the cartridge with stronger solvents (IPA, Acetone, etc.). To prevent, or at least minimize, this issue from occurring you should consider using KP-C18-HS Samplets, not necessarily for dry loading but for cartridge protection. The Samplet will act as a trap for the lipophilic materials and protect the main cartridge from the contamination. Depending on the injection volume, you may consider replacing the Samplet every other injection.

      Another suggestion is to use a better sample solvent, say DMSO, so you can inject less solvent at a higher concentration.

      Bob

Leave a Reply

Your email address will not be published. Required fields are marked *