Another Headspace Question

[GregK]

Seems like residual solvents / headspace questions have come up a lot lately. Is this because of the USP changing?

Anyway, I'm trying to run residual solvents <467> by headspace with our 6890 and 1888 and I'm having a couple problems.

First, some background: We are looking for specific solvents (methanol, acetonitrile, toluene, DMF) in pyrantel pamoate; dilute with DMSO. I'm trying to stay as true to the USP as possible, to avoid running a method validation (not sure that will hold true much longer). I recently had the whole system contaminated by shoddy workmanship on our new gas manifold (see the thread titled "Ghost Peaks"). Set-up: ZB-624, 30m x 0.53mm x 3um. 35cm/min He. Oven: 40C for 10min, 10C/min to 240C for 20min. Inlet @ 160C, FID @ 250C. Headspace: 45min equil. @ 85C, 1 min pressurization @ 10psi, 1mL loop @ 95C, transfer line @ 125C.

Now, the problems:
1. All my peaks are tiny. I know, they're residual solvents, they're not supposed to be huge. But only 3-4pA, 5-10 area counts? That seems a little too small.
2. I'm getting a lot of carryover on the DMF peak. I talked to Agilent about this, they suggested increasing inlet and transfer line temps (from 140 & 105 to 160 & 125). This is running over the weekend, but first glance tests aren't looking too good.
3. My peaks seem to be tailing. Not sure if this is just because of size? DMF tails the worst.
4. My work has a web block, so I can't post pics of chromatograms. Sorry.

Thanks for any help in advance.

[krickos]

Seems like residual solvents / headspace questions have come up a lot lately. Is this because of the USP changing?
Yes, that is likely one factor, "we europeans" have lived with a practical identical change in Ph Eur for a longer time. Another factor might be regulatory activity ie change in manufacturing causing updates (popular to move drug substance production to Asia).

Now, the problems:
1. All my peaks are tiny. I know, they're residual solvents, they're not supposed to be huge. But only 3-4pA, 5-10 area counts? That seems a little too small.
Not totally unexpected as a general comment. The USP/Ph Eur methods are made to fit all typical solvents used by manufacturing. As such they are not optimum for just a few solvents and generally only allowed for identification and limit test (qantification over 1000ppm?).
Therefore I always recommend users to validate a method based on the pharmacopiea methods (use the columns types). Sample prep is one thing that differs between USP/Ph Eur, default is similar but Ph Eur recognise that sample concentration might have to be changed to give suiteble response.

2. I'm getting a lot of carryover on the DMF peak. I talked to Agilent about this, they suggested increasing inlet and transfer line temps (from 140 & 105 to 160 & 125). This is running over the weekend, but first glance tests aren't looking too good.

Worst case you need to remove loop and clean it. Often forgotten issue that loop temp needs to be higher than higest bp of all solvents and diluent. So infact in this case loop should be 195°C (little over DMSO) and transfer line 10 degrees over loop temp. Inlet should also be 195 or so.
Residual DMSO in loop will/can cause carry over of DMF or other high boiler.

3. My peaks seem to be tailing. Not sure if this is just because of size? DMF tails the worst.
Can several contributing issues mentioned above. Temperature settings too low may cause a "bleed" in on column. Also GC program and column is not optimum (long start temp risk for diffussion). Again recommend a validation and a smaller column ID (0,32mm or even 0,18mm).
One thing though, as DMSO elutes last of all I guess, you will not get a solvent focus effect on the startband, but if method is improved the issue normally resolves/becomes acceptable.

[chromatographer1]

Unfortunately the USP method may be difficult for many to reproduce adequately.

In addition:

Bare metal in HS analysis is a NO NO. Metal = carryover and tailing

High temperature may alleviate the problems but usually doesn't solve them.

Fortunately, Varian, PE, and Tekmar and others offer near metal-free options. I do not think Agilent does in their HS analyzer. I might be wrong.

I was a alpha tester back in the 1990s for Tekmar's near metal-free modifications which worked extremely well.

I think it behoves a company to find a generic method which works for the solvents and matrices in their manufacturing processes.

Once you validate the method for one solvent and matrix then you can simply 'add on' additional solvents and sample matrices which further validates the original method.

This is not a pipe dream. I spent 5 years doing exactly this for a large pharmaceutical company. Three dissolution solvents were validated and over 20 different solvents. With this base of data it was easy to add a new drug matrix or formulation and perform an expansion of the validation. Essentially after you demonstrate linearity and recovery and time restraints for the HS system and sample preparation then all you have to show is that the new matrix does not affect these factors for any analysis of samples of the new matrix.

The company had a simple method: temperature, time, injection size, injection time, GC parameters, oven program and column selection were identical for scores of products. In orther words, even a high school graduate could perform the additional required validation for any new products. The technician just prepared the new matrix as usual.

If for some reason the generic method would not work then a chemist could review the problems and resolve them. Aspirin and acetaminophen(APAP) would work fine with the generic method but stannous stearate might not and could require a modification to the generic method.

But for most products after the matrix was validated with the generic method then only three preparations were required to quanitatively determine the residual solvent content of new samples of that matrix.

Can a generic method be accurate? Well when given a new drug, a peptide, I took two samples and determined that one sample contained 1730 ppm of ethyl acetate and the second contained 1745 ppm. This was a routine result variation and much of the variation was sample to sample variation, not method error.

While an initial cost in time and manpower will be spent in developing a generic method, the time and money saved in future validation and QC testing will be returned manifold.

To sum up: if a matrix was soluble in one of three dissolution solvents (one of them, of course, was water) and I could show linearity and accuracy of recovery of a residual solvent from that dis. solv., and then show that the matrix when dissolved did not affect the linearity and recovery of the res. solv. then my validation was completed and I could analyze samples for the FDA using the validated method within 24 hours.

If I had to make the USP method work, even as a limit test, for the different matrices I had to test, I would not have retired by now. I am sure I would have pulled out all my hair was well.

Oh, and by the way. I used DMF DMAc DMSO and other solvents 1N HCl etc and never had carryover problems or was forced to use temperatures over 120° C.

[krickos]

Hi

Rodneys advice with setting up a few generic methods that requires little "sample specific " additional validation is a really good advice, used be several companies.

The loop issue, well was perhaps a bit limited in feed back, but thyis is a common/known potential issue with Agilent 7694 and G1888 HS. Earlier you could get "tubing" (needle, loops etc )either in Ni or "silcosteel" material. As of G1888 I think silcosteel is standard. Nevertheless if enough DMSO for example is brought into loop you may still get residues in loop over time if temperature is too low.
Agilent even tend to include a "steam clean" method for the headspace sampler.
Also, the issue is usually related to polar higher boilers as DMF above(toulene usually do not as sensitive for example), but have seen carry over/loss of sensitivity even for methanol/ethanol related to residual DMAc, DMF, DMSO in loops.


And yes agree that other brands like PE HS 40 has a different injection system hence do not suffer as much with regard to potential contamination.

[GregK]

Well, the increased transfer and inlet temps did eliminate the DMF carryover. Now I don't have reproducible areas for DMF,but I do for the others:

• MeOH: 0.8%
  ACN: 11.1%
  Toluene: 4.8%
  DMF: 28.8%

ACN peak is only ~6 area counts, so a little higher RSD is kinda expected. The DMF peak averages ~400, with a low of 297 and a high of 601. And its not growing throughout the run, it bounces all over. Why would some peaks stay the same and others change?

And to add insult to injury, it seems everyone is backordered on headspace vials and I'm almost out. Can I do more than one injection from each vial? It seems like it wouldn't be reproducible?

[GregK]

(Meant to hit "Preview" not "Submit", sorry for the double post. )

I don't see a peak for DMSO. The water blank looks the same as the DMSO/water blank as the air blank. This isn't really a problem, as I'm not quantifying it anyway, just seems odd.

I think I've already stepped far enough away from the USP to require a validation, so I might as well go all out...

We have a DB-624 30 x 0.25mm x 1.4um. Smaller ID will help with tailing, yes?

Do I need to increase inlet and loop temps, even if carryover is gone?

Have I mentioned I'm under a time restraint here? Its really my first "big" project, and everything I touch with it goes to crap. Very frustrating. It will be nice when its all done though.

Thanks again for the help.
-Greg

[Schmitty]

You could always decrimp the vials and wash them. Or, purchase the screw-top type vials, which are much easier to open and close. I have had both good and bad luck with the screw-tops regarding reproducibility, however.

I usually get my HS supplies from Supelco. But if they are out/backordered, I will go with Microliter.

The issue with washing the vials is the same as any other glassware...you don't want something from the washing process to contribute to the issues you already face.

Good luck

[krickos]

Hi Greg

Made some progres, great

YES I still think you should increase loop, t-line, inlet to above bp of DMSO, as long temperature is lower interactions may occur and you may get a "pseudo phase" that interacts with yourresidual solvents.

Yes, that column, with your improved temperatues and more optimum GC oven program will improve peaks and give you a faster GC run.

DMSO peak gone that was a bit unexpected. Increase temperature as mentioned above, and a few questions:
DMSO/water mix is it like 1ml DMSO solution and 5ml water solution in a 20-22ml vial?

Could you please list the complete HS settings vent time, loop equilibration, injection time etcetera?
Could be something additional there.

NO, you can not make multiple injections from one vial. Well you can but then areas will not be reproducible (decreases with each injection) and that is called MHE (multiple Headspace extraction) and is a different quantification approach.

[GregK]

Progress made is due, in part, to the help I receive on this forum. So, again, thank you everyone.

Well, now that I have a nice clean blank injection, I need to go back and figure out which peak is which. Trying to ID peaks when you have massive carryover is not easy; I don't think I will try that again. In my weekend run, my "sample" prep also had a major peak at the RT I was calling DMF. Upon review of old chromatograms, I'm thinking this is actually the DMSO peak. Now I have to figure out where the DMF is actually coming out. That is tonight's agenda.

Complete HS details:
Vial Size: 20mL
Inject Time: 1.00min
Loop Equil: 0.15min
Loop Fill: 0.05 (manual says that lower value = more sensitivity, so this is decreased from the default)
Loop Temp: 95C (will increase to 190C for next run)
Oven Temp: 85C (should I increase this as well? Concerned that DMF boils at 153C, but MeOH is at 65C; manual says not to set within 10C of any solvent in sample?)
Transfer Line: 125C (will increase to 195C for next run)
Shake: Off (should I turn on?)
Vial Equil Time: 45min (from USP)
Inlet Temp: 175C will increase to 200C for next run)
Vial Press. Time: 1.00min (also from USP)
Vial Press: 10psi
Carrier Press: 5psi (He)
GC runtime: 65min (can probably shorten this time and the actual GC time, as I have no peaks in the last ten minutes of the injection...)

Standard: 150ppm MeOH, 20.5ppm ACN, 44.5ppm Toluene, 44ppm DMF (all 1/20th of USP concentration limits, with sample 0.5g to 10mL, gives standards at limit relative to sample). All are in DMSO.

Procedure: 1mL sample or standard into 5mL water in HS vial.

Agilent is shipping me more HS Vials on Thursday, till then I only have a dozen or so to work with. I'm gonna try IDing peaks by injecting a few uL of each solvent into previously used standard vials.

Thanks, yet again, for the help. I hope someday I can help someone on this forum... Or at least buy them a drink if they're ever in St. Louis.

-Greg

[Peter Apps]

Hi Greg

A couple of comments:

short loop fill times can give bigger peaks by leaving ahigher residual pressure inthe loop, but there is a price in repatability. I have connected a back=pressure regulator to the loop vent to keep the pressure high.

the limit on the oven temperature is set by the BP of the matrix (what the manual calls sovent) not by the BP of any of the analytes. BPs of analytes are something of a red herring since vapour pressures increase smoothly with temperature.

shaking might help, and cannot do any harm.

try setting the vial pressurization time to 0 (for some reason the 1988 still cycles the solenoid valves); there is plenty of pressure in the vial from thermal expansion of the headspace, and introducing extra gas simply dilutes the headspace.

Good luck

Peter

[krickos]

Hi

regarding Headspace settings:

Loop fill: yes Like Peter stated too low to get reproducible results, typically 0,15-0,20min is better precision wise.

Loop equilibration: Why 0,15min for a 1ml loop? it is like asking for interactions with surfaces. Equilibrum is reached more or less instantly, I never use a higher setting than 0,05min for 1 ml loop.

Vial oven temp, like Peter said, depends on diluent. As we talk USP/EP methods one can narrow it down to water/DMF/DMSO.
Your manual is "correct" when it comes to pure water as diluent, then it is usually a good idea to stay 10°C below boiling point of analyte.
But if DMF is used as diluent you can go higher than your analyte methanol boiling point.
In your case you have a mix of DMSO:Water so harder to predict a suiteble limit but as your RSD for methonal is OK this seems not like an issue.

[GregK]

Two steps forward, one step back... Sometimes three steps back...

The peak I had labeled as DMF is definitely DMSO. And I can't find a DMF peak. Doing an injection of DMF in an empty vial gave a RT of 25.3min, DMSO (in the matrix) has an RT of 27.9min. I think they are co-eluting.

Now I'm trying different GC oven programs to try and pull them apart, but I'm not having much luck. This is really my first time trying this, any tips?

I've tried different solvents, but pyrantel pamoate isn't really soluble in much. Soluble in DMSO, slightly in DMF (no help, as I'm testing for DMF), insoluble in water, alcohol. I know its soluble in ACN/dilute NaOH as we assay it in finished product form with that mix, but again, no help as I am testing for ACN...

I'm thinking it might be time to try a different column? The USP lists G16 (ZB-Wax) 30m x 0.53mm x 0.25um. We have a ZB-Wax that is 30m x 0.53mm x 1.00um that I'm going to try and run tonight. More coating = longer retentions, yes?

I tried the other suggestions (eliminating vial pres. and loop equil.), no big differences. I increased the headspace oven temp to 90C and again, no big change. I actually think my MeOH and ACN peaks are smaller now. It's hard to tell for sure, with everything else changing. I replaced the plastic vent outlet tube with a short length of CU tubing, smaller diameter, to try and keep pressures up.

I like doing this development work, it is fun (in a nerdy way). I just wish I didn't have a deadline of March 12th to get everything finalized. I'm afraid that's going to make for a hastily developed method, not necessarily the best method.

Alright, time to switch columns, start a run, and go home and drink some beer.

I'll try and keep updating and taking any advice offered.

-Greg

[Peter Apps]

HI Greg

Changing the column is the way to go.

How are you connecting the transfer line to the GC ?

Peter

[krickos]

Hi

Your observed RT differences for DMSO and DMF on a 624 type colum is consistant with those I have seen and listed in various litterature, a simple overlay of chromatograms should reveal if you have coelution.

Changing to a WaX column will approximately double the RT differences betwenn DMF/DMSO, downside is that RT difference between ACN and toluene will be smaller (less than a minute) but should be enough.

[qcChemist]

I have been running Residual Solvents for about 2 years now on the same system with a 6890N GC, and a G1888 Headspace autosampler with no major issues until recently. Recently in the past month or so, I have been getting a lot of carryover in my injections. I've tried everything I know to remedy this, and nothing has worked so far. I replaced the inlet liner, column, and cleaned the detector on the GC. I've set the sequence purge on the headspace autosampler to 20 minutes, and I am still getting a lot of carryover. Does the sample loop on the headspace need to be cleaned or replaced? I'm at my wit's end.