GPC Assistance

[shaun78]

I am running a GPC method for a client on a proprietary polymer. The column being used is a 7.8 mm x 300 mm DVB mixed bed column with a fractionation range of about 1,000 Da to 1,000,000 Da.

Samples should routinely be in the range of about 100,000-150,000 Da for a "good" distribution. Samples having a problem show a small distribution that generally starts somewhere around 400,000 Da. This information has been determined via the GPC module for Empower. For what it is worth, problematic samples show precipitation of the polymer in solution.

We have two control samples that we routinely run. One control represents a "good" distribution while the other represents a "bad" distribution.

GPC columns are calibrated using a 20,000 Da to 410,000 Da narrow commercially available standards (we use 4 of them).

Recently we have had to change columns because performance of the column we have been using for the past year is not the same. The column is still a DVB mixed bed, it's just new.

I've injected our narrow distribution standards to calibrate the new column. I am getting reproducable chromatography and calibration results.

Then there are my controls. The good distribution is now bad and the bad distribution is now good. I've checked for the obvious of vial error or sample preparation error. Unfortunately, a mistake was not made with respect to vial locations or sample preparation. Furthermore, each injection of these controls shows chromatographic shifting; general peak shape changes, shifts of about 0.1 min are observed (retention times are getting shorter), and peak intensities vary (peak intensity is increasing).

All throughout the injection of controls and all of the associated chromatographic changes, the narrow distribution standards are exactly the same --- all the time every time. This is also suggesting to me that there is something going on with our controls/samples that we don't know about...

What is going on here? The fact that the narrow standards come out the same every time rules out any possibility of the physical properties of the column changing (pores clogging, phase collapse, analyte binding/blocking pores, etc.).

I just don't get it. If anyone could provide some insight I would be very greatful!

[JMB]

Then there are my controls. The good distribution is now bad and the bad distribution is now good. I've checked for the obvious of vial error or sample preparation error. Unfortunately, a mistake was not made with respect to vial locations or sample preparation. Furthermore, each injection of these controls shows chromatographic shifting; general peak shape changes, shifts of about 0.1 min are observed (retention times are getting shorter), and peak intensities vary (peak intensity is increasing).

A decreasing RT value implies a higher MW, i.e. aggregation, and I can understand that for the GOOD distribution going to the BAD; however, for BAD to GOOD...........???

Do BOTH of the control samples (GOOD & BAD) show reduced RT values ?

Are the GOOD to BAD and BAD to GOOD changes both quantitative, i.e. not an equilibrium ?

Are the control samples now somewhat aged in solution ?; have you prepared fresh controls ?

For what it is worth, problematic samples show precipitation of the polymer in solution.

This may well be consistent with aggregation.

Recently we have had to change columns because performance of the column we have been using for the past year is not the same.

At the risk of stating the obvious, this suggests that your proprietary polymer is having some interaction with the column. Is this possible from the perspective of potential, residual functional groups, incomplete polymerization etc ?

[shaun78]

Do BOTH of the control samples (GOOD & BAD) show reduced RT values ?

Yes, samples on the new column show reduced RT values across the board. I assumed this was because there are more open pores in the new column (we know the polymer junks up the column).

Are the GOOD to BAD and BAD to GOOD changes both quantitative, i.e. not an equilibrium ?

No, this is not an equilibrium issue.

Are the control samples now somewhat aged in solution ?; have you prepared fresh controls ?

Yes to both. They all tested the same.

For what it is worth, problematic samples show precipitation of the polymer in solution.

This may well be consistent with aggregation.

Recently we have had to change columns because performance of the column we have been using for the past year is not the same.

At the risk of stating the obvious, this suggests that your proprietary polymer is having some interaction with the column. Is this possible from the perspective of potential, residual functional groups, incomplete polymerization etc ?

. Yes, we feel it does interact with the column.

To make matters more interesting, after several hours of running both the aged and freshly prepared standards switched back to what they were supposed to be (good standard was good and the bad standard was bad). All the while this was going on, I was injecting the 4 calibration standards. All during this time there was no change in chromatography as it relates to these standards.

The only changes observed were for our actual control samples.

Again, I am having a hard time with this.

Due to the nature of what we are testing, we know polymerization is not complete. We believe this is why we are having interaction with the column. Generally, this interaction is fairly minimal. Never before have we observed this type of a shift.

[JMB]

Yes, samples on the new column show reduced RT values across the board. I assumed this was because there are more open pores in the new column (we know the polymer junks up the column).

More open pores should give greater RT values, because of the opportunity for the analyte molecules to diffuse in/out of the pores, rather than just taking a path between the column packing spheres.

BUT, I still have NO idea what is happening with your system !!!

[JMB]

More questions.....but no answers

1. Do your four MW calibration standards exhibit exactly (and I do mean exactly) the same behavior on the old/new columns ?

2. Can you disclose the solvent used to prepare samples ?

JMB

[shaun78]

My apologies; you are correct... More pores would be longer retention!

No, the calibration standards are exceptionally broad peaks on the old column and are very sharp peaks on the new column. This was the true indication that the column is dead. I noticed a significant change in the chromatography during one run and convinced the client to allow us to purchase standards and a new column. After data was produced, I did not have to try very hard to convince the client that the old column had indeed gone bad during the last run.

Yes, solvent used to prepare samples is 38:62 H2O:ACN containing 0.1% Formic Acid. Samples and calibration standards are prepared in the same solvent.

Edit:

I don't know if I have mentioned this yet, but I should also point out that we are using a mixed bed column. It has been a very long time since I have done GPC work (late 90's, early 00's.). I do not know how reproducible mixed bed columns are either between columns or lots. I have not greatly concerned myself with the RT sifts between the columns (new and when the old column worked) because I have been operating under the assumption that it's a mixed bed column and there may be differences between columns/lots. Hence another reason to start using true molecular weight calibration as opposed to qualitative overlays if chromatograms, but that discussion is for another day

[JMB]

I need to summarize the situation, to help me get my head around it; please correct any of the following statements.

1. 2nd (Brand new) column:

GOOD control (MW 150k) runs as BAD control (MW 450k)
BAD control runs as GOOD control

2. 2nd (Slightly used) column:

GOOD control runs as GOOD control
BAD control runs as BAD control

This suggests to me that the problem IS with the DVB packing--washing with mobile phase possibly "conditions" the packing and/or elutes some residual unknown material (residual polymerization catalyst/residual monomer/oligomers etc ??) which triggers the GOOD/BAD interconversion. Interesting that there is no apparent equilibrium.

Question: was the scenario described in 1 & 2 observed with Column 1 ? Was Col. 1 perhaps initially washed with mobile phase longer, before controls and samples were first run ?

3. Peak width of four MW calibn. stds. deteriorates on col. 1 after prolonged use----this is consistent with interaction of proprietary polymer with packing (I know I have said that previously).

Your comments about lot-to-lot uniformity, esp. of mixed bed resin, are spot on. I think that you should contact the column manufacturer; describe the poor results obtained with col. 2 and request the manufacturer to supply another column packed with the same lot as Col. 1.

Please let us know any outcome to solving your problem.

[shaun78]

You have it just about right with the addition that sometimes this new column generates results that matches the old column, and sometimes not.

I would feel a lot more comfortable with all of this if I only knew what was going on. Right now it looks like this method works (or not) based upon planetary alignment; and that is not good science.

It is good to have my belief confirmed that there will be differences between mixed bed column lots.

Thank you for your help and I will try to keep everyone in the loop...

[shaun78]

Update:

It turns out that all of these problems are column related.

The client does not understand this polymer or it's distribution very well. After successfully destroying another column, this is what I noticed:

1. Standards and samples alike are binding to the column. Though gradual, sample RT shifts (0.05-0.1 min earlier) are observed after about every 200 injections. Pores in the column are being blocked or phase is slowly collapsing.

2. As samples/standards bind and/or phase is collapsing the column takes longer and longer to equilibrate. Large equilibration time increases are noticed about every 500-750 injections. Increase in time changed from about an hour (no injections) to upwards of 8 - 10 hours (right before the column became a piece of junk).

3. The reason for the long equilibration appears to be related to analyte possibly being released from the column and pores re-opening as retention times would trend from shorter to longer as the column equilibrated, but would never go all the way back to the when new RTs.

4. As all of this happened, the injections of the narrow distribution molecular weight standards started to look worse and worse. Peaks would start narrow when the column was new (~45 second peak width) and would become insanely broad when the column was junk (~18 minute peak width).

5. Installation of a 500A column behind the mixed bed showed the two control samples we have been testing are not what the client thought. The good control is indeed good. The bad control actually contains the distribution of the good control as well as a small portion (~20%) of an insanely high molecular weight distribution. When bad became good and good became bad the large molecular weight distribution of the bad was either totally excluded from the column or went below detection limits because the peak had become too broad; only the less intense lower molecular weight distribution was detected. The good control looked like the bad control because it had greater peak intensity due to the chromatography but because the peak had broadened, it now looked like the bad control. The bad control looked good because it was a less intense peak and the broadening was not as readily apparent.

Solution: guard column. Install and replace when chromatography starts to change, or every 500 injections.

Oh the frustration that this has caused! However, I am glad to see it sorted so I don't have to deal with it anymore.