musty odor in drinking water

[Peter Apps]

Reading through the question and responses, is there a reason you couldn't manually sample the headspace with a syringe and run a splitless injection?

We've already done that but we couldn't find any peaks different from the blank.

The MeOH extract has a musty odor so my unknown is in it.
I'll try some different techniques to get it analysed. I have very little experience with SSL liq injections (I mainly do 50µl CH2Cl2 PTV injections) but I'll have a go.

thanks again for all the input it is much appreciated!!!

You can do a PTV with the methanol as well - start just below the boiling point and ramp quickly to 240C or so (I sometimes do this since it seems to lead to less crud in the inlet). Start with a 1 ul injection and no solvent purge (in effect a splitless injection) and increase the volume with a solvent purge if you think that you need more sensitivity.

Peter

[GOM]

Hi Bart

I keep looking at this problem with interest.

I feel the need for a summary of what has worked and what hasn't

With regards to Peter's comment

Two possibilities are that the C18 does not catch the musty smell - in which case the sample that passes through will still smell musty - or that the C18 extracts it from the water but CH2Cl2 does not elute it from the cartridge

I have re-read the posts and am I missing a reply that says the eluate of 1 L passed through the SPE has a lower odour, indicating that the C18 has trapped the odour. (?)

Your last post mentioned a MeOH extract - is that something that you have now tried with the SPE rather than DCM since Peter suggested it?

With regards to your other queries,

The chance that it is a compound with a mass <50 is rather small I guess. (?) - yes but I would prefer to scan from 40amu (and use a polar column . Something like a sulphide or other molecules in that <50 MW range would have a distinctive odour and not the distinctive musty odour.

Edit
Note in a previous reply "wet" iron and "wet" concrete

To me the "wet" comment is significant

The wet iron comment does not indicate an inorganic compound - Iron is essential for the metabolism of many micro-organisms -the odour is from the metabolites of micro-organisms using the iron from wet surface rust.

A bit like the "smell of summer rain" on paving or from the ground, which is the release of metabolites from micro-organisms released from the ground after rain and the human nose has an increased sensitivity in a humid atmosphere.

Apart from that, inorganic compounds in the main will not have an odour (odor ) - those are usually associated impurities that will have an odour

Kind regards

Ralph

[BMU_VMW]

Hi Bart

I keep looking at this problem with interest.

I feel the need for a summary of what has worked and what hasn't

With regards to Peter's comment

Two possibilities are that the C18 does not catch the musty smell - in which case the sample that passes through will still smell musty - or that the C18 extracts it from the water but CH2Cl2 does not elute it from the cartridge

I have re-read the posts and am I missing a reply that says the eluate of 1 L passed through the SPE has a lower odour, indicating that the C18 has trapped the odour. (?)
I was unable to check since it is done on vacuüm and the eluate is gone after extraction. But the SPE itselff and the MeOH extract do smell so there is at least some degree of concentration.

Your last post mentioned a MeOH extract - is that something that you have now tried with the SPE rather than DCM since Peter suggested it?
Yes it is, I redid (is that correct English??) the test with another sample. The MeOH had a stronger smell than the DCM. So the compound does elute better with MeOH

With regards to your other queries,

The chance that it is a compound with a mass <50 is rather small I guess. (?) - yes but I would prefer to scan from 40amu (and use a polar column . Something like a sulphide or other molecules in that <50 MW range would have a distinctive odour and not the distinctive musty odour.

For now I can only use a Rxi5sil_ms since:
1) besides a DB_VRX for HS, I don't have another type of column.
2) the GC's are in full use for routine analysis and after the technical problems 2 weeks ago we are still trying to catch up. I can't put one out of routine and start playing with it. For now it al has to be done in the routine configuration
If this doesn't work I guess I'll have to start building a GC-snif with a spare GC (without detector).That way I can (or better might be able to) determine wich column is suited and at what retetion-time I can expect the unknown compound.

[BMU_VMW]

There might be a small succes in our search.

I extracted 1l of sample on a C18 cartridge and eluted this with 500µl of MeOH, then spiked 100µl of this extract in an empty HS vial for SPME FS analysis.
In both samples there is a peak at 15.7min that is not present in the blanks.
(both samples are from the same street so I still have to check if other locations with odor give the same peak, in order to exclude a 'local' effect that is not related to the odor).
Problem is that I cannot identify this peak with NIST. It does give a hit, but I'm not convinced it truly is this compound.
Below you can find some screenshots from the results.

full chromatogram:
zoom of the peak:
NIST result peak 1:
NIST result peak 2:

Does any of you have the possibility to analyse this FS agilent data in e.g. a library with odor compounds? And if so would be willing to help me out

[GOM]

This is interesting progress.

If I am interpreting the screen shots correctly I agree that your background subtracted spectrum of the extra peak doesn't look like a good NIST match.

I will look at the 414 molecular ion ( or is it 415) and the fragments to see if I can see anything of value - rb6banjo would enjoy this one

It seems a high MW for a musty odour (not a very scientific thought I know ) - but you are now seeing a difference - perhaps it is a precursor for a lower MW odour compound.



Regards

Ralph

[BMU_VMW]

Hi everyone,

It has been silent for quit a while, but we have been busy.
Time for a big update:

The odor problem has been identified as 246_TCA at very low concentration.
Because the concentration was below 1ng/l we discarded the first results as non significant and concluded that our target-analytes were not causing the problem. Al further analyses were focused on finding an unknown compound.

Only when we re-analysed some samples from the different production steps with the routine method we found a 7ng/l 246_TCA peak that indicated that there was something going wrong.

It is in fact the first time that TCA's are found in our waters, all other problems were related to MIB or Geosmin.

We than spiked 246TCA at very low level and determined the odor treshold. It proved to be between 0.05 and 0.1 ng/l !!!!

In the meanwhile I have been working on the SPME methode to be able to measure these low concentrations and so far it is looking good.
Still have to run some more tests to see if all is reproducible but it looks like I'm able to see 0.05ng/l and start my calibration curve at 0.1ng/l highest concentration will be 10 or 25ng/l, there is no need to go higher.

We re-analysed the samples and the 246-TCA concentrations in the actual tap-water did rise a little over time (time between the first complainds and the last samples collected). Concentrations found were between 0.5 and 2ng/l causing a lot of smell for a very small peak.

Toghether with the 246TCA we found some other TCA's and 246-TBA but I'm still waiting for the standard to arrive in order to quantify the TBA.

We even managed to find what is causing the formation of the 246_TCA, now we still need to get it solved.

Thanks a lot for all the help. I'm sorry that I've put you all on the wrong track to start. But lets call it a hard lesson learned

[GOM]

Hi

Thanks for the update - it is always good when posters take the time to do that.

I am kicking myself for not considering 246 TCA. It is discernible at extremely low concentrations and is the well known cork taint in wines

On looking back Rb6banjo originally suggested it

Regards

Ralph

[rb6banjo]

I'm glad you figured it out! Those things are buggers. Fortunately, they do give very unique mass spectra which makes it much easier to detect them at very low concentrations. SPME is great and pretty much what I use but we have also used the "twister" (pdms) stir bars to get really low concentrations in our samples. As I posted early on, TCA is the root of nearly every musty problem I've ever encountered. We worry about it a lot!

[GOM]

@ Rb6

You suggested this molecule originally

Regards

Ralph

[BMU_VMW]

The research is still in progress but so far it seems to be formed when our sand filters are back-flushed. This is done with water from the drinking-water reservoir, so this water contains NaOCl.

According to: Karlsson at al. " Formation of 2.4.6-trichlorophenol and 2.4.6 trichloroanisole during treatment and distribution of drinking water" Wat.Sci.Tech. vol 31, N°11 pp 99-103, 1995 :
""Organic matter (humic substances) on the sand filters can react with the chlorine to form chlorophenol (which we are unable to measure for now), this is the methylated by micro-organisms to form chloroanisole.""

Big question now is why this is a problem now when this production plant is in use for many years. ???????

[Peter Apps]

Big question now is why this is a problem now when this production plant is in use for many years. ???????

Possibly a change in the microbial population, possibly driven by the selection pressures of being exposed periodically to chlorine compounds that they need to detoxify ?

Peter