Category: research

One of my major problems with the SPME-GC/MS measurements was the amount of water that entered the GC-MS via the SPME fiber. After dipping the fiber into the sample solution, the amount of water remaining on the fiber after the adsorption process was finished was quite substantial causing a trail of water in the chromatogram and increased column bleeding.

Since the very beginning I have removed the drop that usually hangs on the protective needle, which covers the fiber during transfer and storage – removal with a lint-free tissue was fine. In order to reduce the water content further I have experimented with drying the fiber in a dry N₂ stream. For that purpose I have used a glass vial with a septum port (for the fiber) and gas-tight in-/outlets for N₂.

While exposing the fiber for drying I noticed that a second drop of water is always hidden in the needle and only removed by evaporation in the injector port (thus increasing the water load on the column). As a result I am now removing this second drop too. I have also dropped the idea of drying the fiber, because it does not remove any additional water (once the second drop is removed) and I am too afraid anyway to loose compounds through volatilisation in the N₂-stream.

By removing both drops of water, my baseline is considerably better and column bleeding was significantly reduced. I am surprised that I have never read about this problem in any of the SPME papers that deal with water analysis, although the problem is a faulty construction of the fiber and the protective needle.

I was off sick for most of the week last week, but I came in Friday to have a look at my ¹³C malonic acid spiked snow samples. Approx. a week ago, after my new batch of malonic acid had arrived, I have increased the concentration in order to make it sensitive enough for the NMR.

Yesterday I ran my first sample – and it worked. A nice ¹³C₂ malonic acid peak at approx 45 ppm (right next to the DMSO solvent). I am very confident that I will find the same signal in the other two samples that I have spiked the same way (a second snow sample and an ultra-pure water blank). There are two more tiny spikes in the spectrum, but I have to discuss this with the NMR-Admin, if these are artifacts or “potentially emerging” signals. Again, everything was done under sterile conditions in order to avoid any outside contamination.

I have also registered for this year’s AGU Fall Meeting in San Francisco – I am looking forwared to going there again and present some new aspects of my research. I also want to talk to a few people about my planned trip to the Arctic next spring and know about their experience.

Well, I still have not found any ¹³C metabolites in my samples (analysed with GC-MS), but the composition of my samples is changing quite signifcantly due to microbiological activity. E.g. butanol, which I have detected in large quantities in the initial samples has disappeared and a range of other compounds (a lot of them aromtics) have appeared in my mass spectra. Among them are chlorobenzene, increased amounts of toluene and several others. So there is definitely something going on here.

Still, my goal is to find at least one ¹³C metabolite. Some of the spectra I could not assign to a compound despite a distinct MS signature. Maybe the NIST database and my own experience (or rather the lack thereof) misses the compound due to the changed isotopic composition. I guess that I have to keep on hunting.

Not much luck with the NMR measurements either. As feared, the concentration of labelled malonic acid is too low – so that means overnight runs (with fingers crossed) or increasing the amount of malonic acid and buffering the lowered pH with diluted NaOH, to keep the microbiology happy (and even more important: alive and productive).

Well, well – I have now analysed my GC-MS data from the first batch of ¹³C malonic acid doped snow samples that I have incubated for one week now. No luck so far. Because of their heterogeneous and complex nature, the number of peaks was huge (160 in a 21 min run). I have tried to browse through the large peaks as well as the tiny humps in the baseline in order not to miss anything.

What I had hoped for was a compound spectrum that could still be correctly identified by the NIST Mass Spectral Database or that I know from past runs with one of the masses shifted from n to n+1, due to the ¹³C doping. However, this implies that the metabolite originates from a defined biochemical process and that it is an end-product of this pathway. This is pretty steep to begin with and probably I should dig more into the literature, although so far I have not found anything of relevance (especially for real life samples with unknown end-products).

Should more than processes metabolise the ¹³C, or should there be more than one end-product, then things get tricky, because the characteristical fragment that I am looking out for will be a lot less intense.

Anyway – more work and more measurements ahead. Wish me luck and patience 😉

Well, right during one of my training sessions, the first measurement problem surfaced. I want to monitor malonic acid degradation in aqeous solution and therefore I have prepared a couple of standards in D₂O. Strangely, the singlet at 45 ppm always showed up as a quintett, pointing towards coupling with some other species other than H (the decoupler was working fine and the ¹H spectrum was good too. All other ¹³C spectra that I recorded were fine too. The only other NMR active species present was deuterium, so I speculated that some exchange reactions must be going on, replacing the H in the molecule with D. The most acidic H was the one on the carboxylic acid group, but after discussion with the NMR facilities manager, we ruled out this process, because it takes place too quickly.

So the only other possibility was the C₂ and its 2 (less) acidic H atoms. And this was indeed the case. Assuming this exchange, the resulting quintet makes sense and in diluted D₂O (only 10%, with the rest being H₂O, this rather weak effect was not observed. During the day of measurements, I have also posted the issue on sci.chem and got my thoughts and experimental results confirmed. Nice!

I have completed my NMR training today. This means that I can schedule and run my own experiments and I do not have to ask the manager of NMR facility to do it for me. That is essential for monitoring my degradation and identify metabolites using ¹³C-NMR.

Training was good and now after several sessions under supervision (and a final test ;)) I am comfortable using the instrument. It is a Varian Mercury 300 (running with a 300 MHz magnet; a pretty standard configuration for routine measurements). The software is VNMR running under Solaris. My UNIX knowledge comes in handy here. I have no problem using the software and manipulating my data. The challenge was to control the settings for measurements properly, namely tuning and shiming in order to obtain nice symmetric peaks.

This is going to be fun!

Well, there is some more work to be done in the realm of microbiology and I do my best applying my meagre microbiology knowledge and skills for some experiments that investigate the degradation of ¹³C malonic acid by microbiology (bacteria, fungi) in snow samples that I have collected in the past.

I have prepared sterile solutions, added and doped the samples and now they are sitting in the fridge for a while. Hopefully, bacteria are already happily feasting on the provided nutrient. Let’s see, what the GC-MS analysis will reveal in a couple of days.

… since my last entry, but this is also due to the fact that I was on vacation, spending some excellent time in Newfoundland, Labrador and the Quebec Cote Nord. Now that I am back to work – and dealing with a couple of administrative issues such as improving backups of GC-MS data by adding an Ethernet card to the data PC, I am having a close look at my recent data.

And it looks good, indeed. Data analysis is dead slow. There is a large number of compounds to compare and after my inital analysis, I have missed compounds in one run, but found them in the duplicate run, but after re-analysing the data it turns out that the compound is present in both runs (as it should be), but signals are quite weak, which is delaying my analysis.

However, once I am done with a sample, I am getting good results and even the weaker signals match well in duplicate runs. I have now analysed samples from three locations collected in 2005 (Mont St. Hilaire, McGill campus and Resolute [sampled 2004 for comparison]).

In the past few weeks I have run samples that I have collected in spring 2005 on McGill campus and at the McGill research station in Mont Saint Hilaire. Samples were surface snow from the first 10 cm of the snow pack and I have taken grab samples with pre-cleaned and sterilised equipment.

Because of the low concentrations present in the sample I am resorting to large sample volumes that I have collected in sterile HDPE bottles. So far I have had no cross-contamination from plasticisers in the container material, which is fine. The samples from these containers allow me to obtain a total sample volume of about 130 mL and still retain about the same amount for later experiments.

I melt the samples before my measurements and once melted, I keep them on ice and covered in tin foil to avoid any exposure to UV light and minimise volatilisation of compounds. I do not (yet) have tight containers of that size, so I am using a similar set-up described by Petterson (Chemosphere, 2004) with a 125 mL Erlenmeyer and a tinfoil cover after sterile transfer of the sample.

I am still crunching my data, but from I first glance I get good signals for aromatic compounds and a few halogenated substances (mostly Chlorobenzene, tetrachloroethylene in one sample). There are surprisingly many aldehydes and alcohols present (mostly with aliphatic chains attached), which I have not seen earlier in this abundance. The absence of organo-halogens is still surprising and remains to be investigated – although a lot of explanations are possible here (age of snow pack, precipitation, …)

It seems that the hard work in the last week paid off. After a good look into my data I have decided to make some (rather simple, but nevertheless effective) changes to my method:

* I have increased the sample size by a factor of 6
* I have increased the adsorption time from 40 min to 120 min

The latter had only a minor effect on the results (so I will stick with the shorter adsorption time), but the increase in sample size did the trick. The tiny peaks with only a very small number of fragments for some aromatic compounds are now clearly separated from the noise and the underlying mass spectrum passes my and the NIST library search.

I have analysed samples from the McGill campus lawn and from a site at the McGill Research Station in Mont Saint Hilaire (east of Montreal) and the main compound groups that are have detected were aromatic compounds (toluene, xylenes, benzaldehyde), aldehydes and some aliphatic alcohols.

However, I am still wondering, if halogenated compounds a present (no ever so small isotope signatures from chlorine or bromine so far) and it seems that benzene is still buried in the water peak at the beginning of the run (although I have modified my method so that it elutes later in a less contaminated section of the chromatogram – this works fine for toluene, but not for benzene). Another explanation, of course, is that these compounds are not present or only at very low levels. Well, there is more to look at.