Help Needed: Potential Stationary Phase Interaction?

[Dreu]

Hi All,

I'm hoping that someone can help me with this very perplexing issue. I am running an Agilent 7890 GC utilizing a Cool On-Column inlet. My column is a DB-5/ZB-5 30m x 0.25mm x 1.00um (yes, I know this is narrow for CoC but our test mixes give beautiful, reproducible and linear chromatography). Our sample is mPEG6 mesylate in ACN.

Problem:

It appears that the act of injecting my sample on the column creates active sites, as evidenced by the loss of my alcohol peaks. Injection of subsequent blanks causes the resolution and peak response to deteriorate until my peaks are nothing but ugly lumps in the baseline. Baking does not solve the issue and the column appears to be dead at that point. I've also tried rinsing the column with ACN and MeOH (both compatible with the column phase according to the manufacturer) and that only seems to make things worse.

Upon visual inspection after the samples have been injected, there are "particulates" within the column. The obvious conclusion is that there is some non-volatile component within my sample that is depositing onto the column. However, I have only injected the sample 8 times! This is with a brand new column. I am using a retention gap/guard column and there is little to no evidence of anything on the guard column. All of the junk is in the center of the column itself. At 1uL injections of a 1ug/mL solution, I cannot fathom that I would be able to see residual particulates with the naked eye. Especially in multiple places throughout the column.

My thought is that our sample is actually reacting with the column itself, however everyone I've spoken with disagrees. But honestly I can't imagine what else would cause such rapid destruction of the column. I have killed 3 columns in 2 weeks, one of which died within 24 hours.

We've run an analagous method for months using a split injection and have not observed this problem. My thought is that somehow injecting the liquid directly onto the column is causing the column to deteriorate much more rapidly than using a volatilized sample. But I just don't know. I've spoken to Phenomenex, Restek and Agilent and no one seems to have any clue what's going on.

I have added chromatograms to the link below. I can add any other images requested as well.
http://www.flickr.com/photos/104228113@N02/

Note that our Impurity Profile Solution is a test mix containing 29 impurities and intermediates in our process. The linearity chromatogram included contains a subset of those peaks that demonstrate the formation of active sites most clearly.

Thanks for your help!

[lynoguchi]

Did you tried to lower the injection volume? Or cut the beggining of the column?

[Dreu]

Yes, we dropped our injection volume down to 0.1uL. We have cut the column but this is happening with 3 different columns, 2 of which were brand new out of the box. I also fully conditioned them before use.

[lynoguchi]

Dreu,

Did you tried to reinstall the column changing the connections between the inlet and detector?

[Dreu]

Dreu,

Did you tried to reinstall the column changing the connections between the inlet and detector?

I'm not sure I know what you're asking. I did verify that the column is properly installed. All of our chromatography looks perfect until we start injecting our solutions. We experience the same phenomenon every time we try the experiment.

[lynoguchi]

That I want to say, is maybe the beggining of the column was overloaded with your sample, you're using a narrow bore to inject a great amount of sample. You said that with a split injection this didn't happen, right? In this case, what was the split?
In the first injection in the column you can see this problem in results?
I suggested to you to connect the column exit in the inlet and the entrance in the dectetor, to try to backflush the contaminants in the column by conditioning.

[Peter Apps]

The 5% phenyl column that you are using is the most common of all the stationary phases, and so if you are getting some chemical attack by the sample on the stationary phase there must be something very unusual about your samples. What is MPEG6 mesylate, and how do you prepare your samples ?.

You mention a retention gas in passing. What are its dimensions and how is it deactivated ?. How do you connect it to the column ?.

With a 1 ul cool injection to a 0.25 mm column you will get a liquid plug that blows along the column at high velocity. Couple this with stationary phase swelling and you can get mechanical disruption of the stationary phase with bits coming loose and depositing further down. Can you confirm that the columns you are using have a cross linked and immobilised stationary phase ? what is the column temperature when you inject, and what is the temperature programme of the oven and inlet ?

Peter

[Dreu]

The 5% phenyl column that you are using is the most common of all the stationary phases, and so if you are getting some chemical attack by the sample on the stationary phase there must be something very unusual about your samples. What is MPEG6 mesylate, and how do you prepare your samples ?.

You mention a retention gas in passing. What are its dimensions and how is it deactivated ?. How do you connect it to the column ?.

With a 1 ul cool injection to a 0.25 mm column you will get a liquid plug that blows along the column at high velocity. Couple this with stationary phase swelling and you can get mechanical disruption of the stationary phase with bits coming loose and depositing further down. Can you confirm that the columns you are using have a cross linked and immobilised stationary phase ? what is the column temperature when you inject, and what is the temperature programme of the oven and inlet ?

Peter

Peter, you are the first person to tell me something that seem extremely plausible (considering I've already eliminated most of the possiblities that others have mentioned). Originally we were using a 0.1 uL injection (long story how we decided on that) and we didn't see this problem. We started doing the 1 uL injections and had zero issues with our test mixes.....until the sample injections. So what you're saying about the mechanical disruptions and deposits further down the column makes perfect sense. I think that's exactly what I'm seeing.

m6 Mesylate is a very reactive intermediate in our process. Our process development chemists say that it shouldn't react with the phenyl groups though.

We've used 2 retention gaps, the Rxi (http://www.restek.com/Landing-Pages/Content/gen_B004) and the Hydroguard (http://www.restek.com/catalog/view/52). We haven't seen any noticeable difference between the two. Both are 0.25mm x 5m. We originally used the press tight fittings to attach them to the column but we were getting active sites because there was a gap between the guard and the column. I ordered zero dead volume Siltek micro unions but we're having issues with the solvent peak tailing (even at 0.1uL). I'm still trying to work that out.

The intial oven temperature is 60C and ramps to 250C at 15C/min where it holds for 10 minutes. Then there's another ramp to 300, then 325. We need the high temperatures to elute all of our high boilers. The inlet temperature program is Oven Track.

Do you think that reducing the injection volume would solve the problem? I just ordered new columns and they're all 0.25mm.

[Peter Apps]

If you are using 0.25 mm i.d. retention gaps then to do a genuine on-column injection you have to use very specialised thin needles. Life would be much easier of you switched to 0.53 mm retention gaps so that you can use ordinary syringe needles. This will have the added advantage that the liquid plug will be much shorter and will spread only a little way down the retention gap. I'm surprised actually that with a 5 m gap you were still getting liquid onto the separating column.

Increasing your oven start temperature will improve solvent evaporation - if you can programme at 15C/min you have room to play with temperatures.

Have you thought about using a Uniliner instead of on-column injection ? What problem with split injections prompted you to go for the trickier on-column injections ?

You could try Siltek deactivated pressfit connectors. They are expensive but the deactivation is as tough as old boots.

Peter

[Dreu]

We are using the specialized narrow bore needles. Since the system is basically dedicated this is not an issue at all. I've already done some pre-validation work and we're getting extremely reproducible results with %RSDs in the 2-3% range for most peaks.

I tried a 0.53mm column on the split method and we're not getting resolution of our critical pairs. This method has a LOT of peaks to separate and the chromatography is extremely similar with several of them. Hence why we're using the narrow bore column. Unfortunately, our company has a limited budget so I did most of my development with what was available at the time. Since the chromatography worked (at the time) we went ahead with it.

We went to the on-column injections because we were seeing thermal degradation of our main peak using the split method. We tried lowering the temperature but lost sensitivity because we weren't volatilizing our high boilers.

I actually was using the siltek pressfit connectors. We're looking for extremely low levels of certain analytes and discovered that we were getting active sites at our union. After a lot of time trying to figure it out, we determined that the most likely source of activity was actually at the edge of the column where it was scored, not the union itself. With higher concentrations of our solutions the effect was not noticeable but at low levels our peaks disappeared completely.

This method has been an absolute bear. Especially since our timelines are super short and none of our analysts have any experience with on-column injection. I'm learning on the fly, to say the least.

I very much appreciate your advice. So few people I've talked to have been able to point me in the right direction.

[Peter Apps]

It does sound as if your samples are tricky, to say the least.

You can use a 0.53 mm retention gap to tame the solvent effects, and connect it to a 0.25 mm column to get the resolution.

The activity around the edges of the cut column will not be solved by a different connector - you still have two ends. Have you looked into the columns with built in retention gaps ? i.e. they have an uncoated section at one end, but are one continuous length of silica with no joins.

Reducing the injection volume will certainly help if the problem is due to solvent effects, but the repeatability of injection volume at 0.1 ul is about ten times worse than at 1 ul. It might be OK if you use an internal standard.

Peter

[Dreu]

Yes, I actually just received the combination columns yesterday afternoon and I've been tinkering with them. I agree about the unions, with these compounds it's critical that we don't have activity because of the sensitivity we need. And the new "zero dead volume" unions I'm using are not working at all, which is frustrating.

Unfortunately, I can't use a different size column and retention gap due to the union issues. I'll order a larger bore column/guard combination column. Hopefully we can make them work. In the interim I lowered the injection volume and initial flow rate and then increased the initial temperature so the ACN would volatilize more quickly. When you told me the likely problem I took a look at our method. I think we had the perfect storm of conditions to cause the problems we were seeing.

[Peter Apps]

Check with the production chemists what might be in the samples that could attack siloxanes - something as simple as high or low pH in the presence of water will do it.

Peter

[Dreu]

I've already done that. They said that they didn't think that it would react at all. I also checked the pH and it's between 6-7. I think that the column load is the primary issue. I ordered a 0.32mm column/guard combo. I couldn't find a 0.53 in the phase I need and I don't have the time to have one custom made. We have a couple of weeks to make this work. There is already talk of just going with another analytical technique, which would be disappointing since I've worked so hard on this one.