Headspace Analysis for USP 467

[mbicking]

I know this is an old topic, but I'm an LC guy doing some GC for a while, and I would appreciate some feedback on instrument configurations for this method.

I have an Agilent 7694 HS unit attached to a 6890. The HS is connected to the GC via the transfer line, directly into the injection port (through the septum in this case). So, all the HS controls are manual and on the 7694. The only communication is the start signal. Using standard column for procecure A: 0.32 mm X 30 m ZB-624. In working through the HS parameters, I have looked at exact USP conditions as well as application note settings from Agilent, but I can't get the sensitivity I need for the Class 1 solvents.

My question relates to the settings for the carrier flow set on the HS unit. As originally installed (don't know by whom), the flow is about 50 mL/min. This is obviously very effective for quickly clearing the 1 mL sample loop. The problem is that the 6890 has no idea that this flow exists, and it ends up going out the split vent. With a column flow of 2 mL/min, and a split ratio of 5, the split vent flow is 60 mL/min. This produces a net split ratio of 30, rather than the desired value of 5. Obviously, sensitivity suffers, which is the big problem.

I can cut back the HS carrier flow to, say 16, set the column flow to 4, and the split ratio to 1, then I get a split ratio of about 5. Sensitvity is better, but the peaks are broader, due, I assume, to the slower HS carrier flow.

Does anyone have any other sets of conditions that might get a better response for the Class 1 solvents? CCl4 is only barely visible right now.

[Consumer Products Guy]

Apparently USP hasn't figured out that a GC-FID response to CCl4 is poor at best. FID isn't the right tool for that job. USP is full of quirky stuff.

[shaun78]

Agreed. The responce to CCl4 is poor at best. Running procedure A on a HS40/Clarus 500 I get a water soluble peak height of about 0.4 mV and a water insoluble peak height of about 1.0 mV for carbon tet.

If I remember correctly, the USP also gives giudelines for HS flow rates/pressures that you need to stay within ... as well as transfer line temperatures.

In my opinion, you are better off developing your own method that tests for only the solvents you need to look for.

[mbicking]

I expected the carbon tet response to be low, but it is not meeting the S/N = 3 limit. And the other components probably meet the limits, but if this method is to be run on a regular basis, it will need to be more robust.

Like most USP methods, it has its share of goofy settings and unnecessary restrictions. But this client is in a regulated environment, and it is easiest to simply do a "verifcation" on the method, rather than developing a new one, which requires "validation." So, I need to stay within the operating limits set by the method. But, there are some other instrument-specific settings that I would feel comfortable adjusting, while staying within the spirit of the method.

I'm hoping someone may have a better handle on this equipment. It obviously isn't the best choice for trace analysis.

[shaun78]

I understand your restirctions from the client coming from a regulated environment myself.

However, most of the time it seems to me that manufacturers only want to know if they are below the specified limit set in Q3C. Again, in my opinion, this is where people need to start taking advantage of the four validation types allowed by the USP.

If all your client wants is to know if the carbon tet is below the ICH limit of 4 ppm, then all you need to do is validate a limits test. I understand validation takes time in most cases. Limits tests do not however. The only APCs that need to be evaluated for a limit test (in accordance with the USP) is precision (in this case at your limit of 4 ppm), LOD, and specificity.

In the end it is a very short protocol and very little work needs to be done. It will allow you to change what you want and give you a more robust method at the same time. You even would have the option of getting rid of the HS all together.

[mbicking]

For whatever reason (I'm assuming an internal policy), they want to claim "USP." Fortunately, carbon tet is not on their list of potential solvents, so in the end we will probably just ignore it and only verify the few solvents that are actually required.

My interests were also technical, in addition to the regulatory stuff, just trying to get the most out of this equipment.

Thanks for the comments.

[Peter Apps]

A workaround, which will almost certainly fall foul of the regulators is to tell the 6890 that it has a narrower bore column than it actually has. For a given linear flow it then thinks that it needs to send less volume flow to the column, for a given split ratio it then uses less total flow, which results is less gas going out through the split, and a lower effective split ratio.

Whatever happened to GCs that you could actually set to do what you wanted ???, and how long is it going to be before there are no controls at all on lab instruments and the standard methods are all downloaded automatically over the internet ??

Peter

[mbicking]

I think the problem here is that this is an older instrument with a simple interface. It was easy to install and use.

A better way to do the job is to connect the HS directly in-line with the injection port flow, which then allows you to control it through the 6890. But this involves cutting and splicing, and is not for everyone.

I believe the new units are completely controlled by the GC/software, but not everyone has the budget to buy one.

Interesting idea about the column diameter.

[stevel001]

We have the same problem. We set the split ratio at the GC to 0.2:1 on our 6890 so there is almost no contribution from the GC to the split vent. We then use a tranfer line flow of 10 mL/min from the H/S. I also use a very narrow bore (1.5 mm) liner in the GC injection port. It makes a difference in the peak broadening and improves the sensitivity of the carbon tet.

I can't say that it solves the problem totally. Our s/n for the carbon tet is usually 3 plus or minus 1. The ChemStation calculates s/n with a worst-case noise calculation, so we can usually justify the analysis even when the s/n calculates out at less than 3.

The narrow-bore liner also helps with the resolution requirements of the Class 2 solvents.