FTIR study of differences between D2O, DDW, and DI water

Stephen Myers from the NSMS department graciously helped Alex and I out with the next phase of the FTIR experiment. He was able to get a hold of a couple of quartz cuvettes from Spectrocell. On top of that I had ordered one from Amazon.

The first part of the experiment required us to “blank” the samples so we put a Spectrocell cuvette into the FTIR machine and took a reading. For fun we then did the same thing with the Amazon cuvette and then compared the two curves.

Amazon cuvette overlayed on Spectrocell cuvette

While the two graphs look identical we noticed that the sample from Amazon had a weird periodic feature in the graph. Stephen said in other experiments he attributes this pattern to some kind of film, whether or not that is what is going on here is debatable. Regardless I decided to use the Spectrocells for the remaining experiments.

Before we began I thoroughly rinsed each cuvette with DI water and dried with nitrogen. This did not remove all the liquid water, but most of it. I attempted to air dry the remainder. This is important information for later. I used 3mL of each water type (DI, DDW, D2O, and a D2O/DDW mix) for analysis in the FTIR spectrometer. For the D2O sample and for the D2O/DDW mixture I used the second Spectrocell cuvette (which was not blanked) so that the high D2O content wouldn’t contaminate the DDW samples. I used the same cuvette for the DI, and two DDW samples (I’ll explain this in a second).

We took spec readings in this order: (first) DI water, D2O, DDW, D2O/DDW mix, (last) old DDW. I chose this really random order because we weren’t sure we would be able to get any interesting data so I wanted to see DI water vs D2O first. When that gave us obvious results we tried the first DDW sample (taken from a bottle that was opened 11/30/11). Then I wanted to compare all these results with a close to 50/50 mix of D2O and DDW (I didn’t pipette exactly 1.5mL of each nor can I report on how close to that number it is, I was merely trying to determine if there would be a difference between DI, D2O, and a mix of those two with a larger percentage of D2O). Finally we measured a sample of DDW that was taken from a bottle that was opened 9/6/11.

Here are those results.

Comparison of DI, D2O, 2 samples of DDW, and a D2O/DDW mix sample

Wowie zowie! The data display isn’t ideal (done in excel quickly) but you can see a difference between each of the samples. There is one exception that is actually remarkable. The sample of DDW from 9/6/11 matches really well with the sample of DI water. The newer sample of DDW (11/30/11) is more distinguishable from these two than it appears above. And the mixed sample and that of pure D2O are very distinguishable both from each other and from the other samples!

If you notice that just above 1.5um (wavelength) there are two tiny humps in the data. I have blown that up below.

Zoomed in portion of the graph above from 1.5um to 2um. Note the sample of D2O is not in this graph.

Here there isn’t much different between the DDW samples and DI sample, but there is a difference between those and the DDW/D2O mixed sample. I didn’t include the D2O sample because the transmission in this range was much larger than these samples.

Now for the best part: Stephen was totally shocked that we were able to notice these differences. He wasn’t expecting much and frankly neither was I (I had been expecting to not notice anything because most papers I had read focused on the MIR range for FTIR). I was super stoked about these results, and happily, Alex was as excited about these results as I was/am!

I uploaded the graphs to Google Docs and all the data can be accessed here. I’ll upload the data to FigShare and link to that later.

Update: I forgot to mention that we are using a Nicolet 6700 FTIR spectrometer from Thermo.

Update (12/25/11): Fixed link to data. Thanks Koch

  • http://stevekochscience.blogspot.com Steve Koch

    Link to Google Docs data isn’t working for me?

    • http://stevekochscience.blogspot.com Steve Koch

      I think link is: https://docs.google.com/open?id=0Bwbdciapt4QZOGZkYmUzMGItNmNmYi00MWRlLWE5NzUtZWRlYjA1YzA2ZTNk

  • http://stevekochscience.blogspot.com Steve Koch

    Quick note: When you repeat the FTIR, you should do a spectrum on a mix of DI with a bit of D2O.  For example, 100 microliters DI + 1 microliter D2O.  So it has more D than DI water, but not a ton more.  The purpose of this would be to make sure that you have mostly HOH and some HOD, but not any DOD.  I haven’t thought this through, but pure DOD may (should?) have a different spectrum than HOD.  I am sure chemists know what all the vibrational peaks are in the FTIR, and hopefully we can find a reference.  It definitely seems that even DDW / DI has significant difference, which would imply that you’re seeing differences between HOH and HOD molecules.

    • http://www.iheartanthony.com Anthony Salvagno

      I agree with that. I still don’t understand exactly what we’re looking at in this spectra but the differences are obvious regardless what they are. Talk to Alex about setting this up, maybe you two can get together with Stephen this week or next to run another quick experiment (the FTIR studies take literally 2 min per sample).

  • http://stevekochscience.blogspot.com Steve Koch

    This is very cool!  I used R to plot your data compared with data from Seigelstein (1981)–see plot attached: your data is black, Seigelstein is red.  I guessed on a path length of 1 cm, which seemed to be correct.  It shows:
    * Very good match below 2 microns!
    * Absorption is too high at wavelength around 1.5 micron and higher.  That means the path length is too long.  Maybe your new cuvette will have shorter path length so information can be seen in longer wavelengths.

    Data are from here: http://omlc.ogi.edu/spectra/water/abs/index.html, search for Seigelstein.

    Good absorption spectrum of water (presumably with “natural” abundance of DOH) on wikipedia:  http://en.wikipedia.org/wiki/File:Water_absorption_spectrum.png  
    Wikipedia article on water absorption spectrum: http://en.wikipedia.org/wiki/Water_absorption

    • http://www.iheartanthony.com Anthony Salvagno

      Are you sure it’s path length and not the material of the cuvette?

      • http://stevekochscience.blogspot.com Steve Koch

        No, I’m not sure, but I’m assuming the cuvette is very IR transparent.  One clue is that the D2O has less absorption, in the same cuvette, I assume.

        • http://www.iheartanthony.com Anthony Salvagno

          Ah good observation! I didn’t think about that. This brings me back to the question I had in the beginning when I wasn’t sure what length was a good cuvette and I couldn’t get a straight answer from anything.

          • http://stevekochscience.blogspot.com Steve Koch

            To convert your data, I assumed the readings were transmission in percent:

            data = T / T0 *100

            So, I guess that pathlength was 1 cm and formula for transmission was:

            T/T0 = exp (-d * a)  where a = absorption coefficient in 1/cm

            so, a = – log (data / 100) / cm   (assuming d = 1 cm)

            After doing that, it matched the 1981 data very well, so I guessed that assumptions were correct.

    • http://stevekochscience.blogspot.com Steve Koch

      I updated the R code to include all of your water types.  Image is attached and available on github: https://github.com/stevekochscience/Deuterium-in-life-science/blob/master/Repeating%20Crumley/2011%20Dec%2029_all%20water%20abs%20compared%20seigelstein.png

  • http://stevekochscience.blogspot.com Steve Koch

    Images and code available on Github:

    * graph DI compared to Seig. https://github.com/stevekochscience/Deuterium-in-life-science/blob/master/Repeating%20Crumley/2011%20Dec%2028_di%20abs%20compared%20seigelstein.png

    * R code to produce plot (and others): https://github.com/stevekochscience/Deuterium-in-life-science/blob/master/Repeating%20Crumley/Water%20FTIR.R

  • http://stevekochscience.blogspot.com Steve Koch

    I updated the R code to plot DDW, DI, and D2O versus Seigelstein 1981.  Differences between DDW and DI seem to be more subtle than differences between your data and Seigelstein.  I don’t know what this means yet.  It appears that D2O is shifted to the right, as expected…

  • Pingback: Let the Ft-IR scans begin « Alex Haddad's Notebook()