Ok, so, it is NOT the mass spec - ref gas injections are perfectly stable over long periods. Rebuilt the couple of tee pieces I messed with most recently, in case there was some sort of partial obstruction getting bounced around by the backflush switching on and off, but that didn't change anything either. Now I'm seeing improved stability, but still a massive isotopic offset (i.e. the standards are now ALWAYS very light, instead of only most of the time). And no, there's been no change to the ref gas or it's plumbing that would create a "well you need to update your ref gas value" situation.

I'm thinking at this point the best thing to do is install a new reactor. If I zoom in far enough to the chromatogram, I'm seeing elevated baselines after each peak that take so long to go away that you can't even observe a slope. I'm thinking the reactor may be partially blocked with graphite and is producing both a kinetic fractionation by retaining a hydrogen memory. Perhaps I over-conditioned?

Have any of you experienced a situation where the reactor induced a fractionation, but without the presence of leaks or the obvious symptoms of retention time/k-factor changes?

On Fri, Sep 21, 2018 at 5:24 PM mwolhowe <[log in to unmask]> wrote:
Several on-offs of the ref gas DO occur within one run, and they are consistent within the run (i.e. it's not just problems with certain individual ref injections, it's on a scale at LEAST as long as a run. As per Nils' advice, I an running an overnight ref-gas-only stability tests now to tell if the problem is with the MS or the GC side of things (no idea why I didn't think of that myself, I'm a bit embarrassed).

I DID just re-plumb the reactor-output capillary into the first t-piece inside the GC-C interface (where the BF helium comes in) - I was hunting down a flow obstruction after a reactor change, and pulled this out to check flow before re-attaching it. I can imagine accidentally creating a fractionation there with a bad capillary install, but I'd expect it to be constant, not intermittent.

I'll see what the overnight ref injects look like and get back to you both.

On Fri, Sep 21, 2018 at 4:00 AM Dyckmans, Jens <[log in to unmask]> wrote:
Hi Mattew,

I understand that the peaks within one run are consistent, but you find offsets of 50 permil between runs. If so, I would check if the reference gas (or whatever you want to call it) is biased (sometimes). Did you try to have several on-offs within one run (where no peaks elute) to check for ref gas stability over time?

Best,
Jens

Von: Stable Isotope Geochemistry <[log in to unmask]> Im Auftrag von mwolhowe
Gesendet: Donnerstag, 20. September 2018 22:25
An: [log in to unmask]
Betreff: [ISOGEOCHEM] Multi-hour-scale instability/offset in GC-TC-IRMS

Greetings all. I'm spinning our D/H GC-TC-IRMS back up after a month of disuse, and I've been running into a perplexing problem when checking standards.

I'm seeing massive offsets in raw delta values (the ones calculated vs the ref gas injections) from few-runs to few-runs. For example: If I run a sequence of standard injections (range of Schimmelmann alkanes) for 24 hours, about half the runs come out at around where they should be, and half the runs come out up to 50(!) permil too negative. If I plot offsets from known values versus retention time, each compound is split into two clusters of 'high' and 'low' delta values. The two run 'types' seem to alternate randomly every 1 to 5 runs (a run takes about 70 minutes).

Now, I've seen instances where things like this happened with a visible pattern (regular, quasi-sinusoidal offsets in a GC-C-IRMS over the course of the day that appeared to track with the room's air conditioning, for example), but nothing completely random and abrupt like this. Does anyone have any experience with anything like this, or thoughts for what to dig into?

Context: The chromatography is fine; the retention times are stable; peak height and area are stable; H3+ factor is nice and low; source has recently been tuned; air leaks as per argon are negligible; reactor is newly installed and conditioned; GC column just had a meter trimmed off the end of it and the inlet was re-assembled (new liner, new ferrules, etc); K factor for the GC is about where one would expect; ref gas values are stable and reproducible with a run; column bleed/background is where it's always been.

Setup: Thermo Trace GC Ultra -> Thermo GC-TC interface (furnace/reactor) -> Thermo GC-C III (open splits and backflush plumbing) -> Delta V Plus IRMS.

Thoughts anyone? Thanks!

Dr. Matthew Wolhowe
Research Scientist/Engineer III
Lab Manager, Oceanographic and Geochemical Stable Isotope Facility
School of Oceanography
College of the Environment
University of Washington