Fluorescence Spectra of Speleothems: Methods and Interpretations

There has been a recent interest in the possibilities of paleoclimate reconstruction from data obtained from speleothem banding. One of the methods being explored for this purpose involves examining the fluorescence spectra of speleothems to determine shifts in organic acids. Paleoclimate shifts can then be interpolated from the shifts in organic acids and correlated with ages obtained by stable isotope dating. My own research focused on developing methods and getting a general sense of what things looked like in our speleothems.

Goals

Define consistent and appropriate methods
Explore practical limits of non-destructive methods
Determine significance of different aspects of fluorescent spectra

Methods

Sample locations
Types of samples
    Solid
    Dissolvedósolution components
    Dripwater
Machinery used
Settings at which samples were run and rationale
Total Organic Carbon

Samples from Cold Water Cave, Spring Valley Cave, and Mystery Cave were used for my research. Only non-destructive fluorescent spectra were obtained for Cold Water Cave. Some small samples (@100mg) from SVC 99-1 were dissolved using 1mL of 2N HCl and brought back up to a pH near 10 with _ mL of 0.1M NaOH (van Beynen, et al., 2001). The amounts and concentrations of HCl and NaOH used were calculated based on the sample size. Spectra were obtained from the resulting solutions. Non-destructive spectra were obtained from SVC-983C, although the validity of these particular spectra are highly suspect because they were obtained while the instrument was covered with a black plastic trashbag and not the machineís lid. The lid would not close due to the length of the speleothem sample.  Spectra were also obtained for dripwater from Garden of the Gods in Mystery Cave. Additionally, spectra were obtained for deionized water and for solutions with different concentrations (1.0, 4.5, and 10mg) of an organic reference, Nordic Reservoir Normal Organic Matter from the Internation Humic Substances Society.

We hoped to obtain as fine a resolution as possible in analyzing the solid speleothem samples, as the timeframe over which we were analyzing would be reduced along with the sample size. However, we found that we were only able to narrow the resolution down to a 5mm X 2mm size. We estimated that this sample size encompassed approximately a century-worth of growth and therefore we still expected to see some sort of differences between the samples.

A Shimadzu RF5000U Spectrofluorometer was used in obtaining the fluorescence data. We ran excitation-emission-matrix (EEM) scans, synchronous scans, and some emission scans. The EEM scans produce a three-dimensional map of all of the various combinations of excitation and emission wavelengths, with the relative intensities comprising the values for the height on the map. We ran the EEM scans at wavelength separations from ­50 to 400nm with increase increments of 5nm once a positive wavelength separation value was reached. A 5nm bandwidth was used for solution samples, while for solid samples we were able to close down the light further to a 1.5nm bandwidth (BW). By looking at the EEM scans, we were able to choose the optimal wavelength separations to run the synchronous scans at. These wavelength separations varied from 50nm to 110nm. Additionally, we ran a few emission scans based on what we saw in the EEM scans.

The programs Peakfit and Winsurf were used to organize and interpret the data obtained on the Shimadzu. Peakfit was used on synchronous scans and on emission scans; whereas Winsurf was used for EEM scans.
I had some difficulties with consistency of results using the Peakfit program. I had difficulties at times finding a proper and consistent balance between peak height, placement, width and base width. For example, sometimes one peak would get stretched out laterally too much and steal height from surrounding peaks, which could then present a misleading interpretation of the spectra. However, my consistency increased with my familiarity with the program.

The other program that we used, Winsurf, was fairly straightforward and much less user dependent for interpretation than the Peakfit program. I used the volume feature to obtain the positive volume (volume under the scan and above the xy-plane). However, this necessarily included the volume under the features produced by Raman scattering, etc. I also used the grid editor to pinpoint the elevation (relative intensity) and location (emission and excitation values) of the tip of the main peak.

We also thought it would be useful to get some total organic carbon (TOC) numbers for use in calibrating the relative intensity values produced by the Shimadzu. For this purpose, we drilled out seven 100mg samples. Four of these samples were taken from CW3-807, from spots that had previously had EEM scans run on them. The other three samples were taken from different points along the same deposition layer (as far as I could tell) on the SVC 99-1 speleothem. These three samples had not had any fluorescence scans run on them. We wanted to use these three samples to see if the TOC was reproducible along the same band.

The method we used involved determining the total inorganic carbon (TIC) and the total carbon (TC) contents using two different machines and taking the difference to find TOC. Each machine had an error margin of ±1%. Additionally, the TIC machine tends to produce higher values and the TC machine tends to produce slightly lower values. These factors, combined with extremely low concentrations of organics in speleothems, resulted in mostly negative TOC values!

Realizing that there was no way to get actual values for the TOC, I still hoped to be able to get some sort of relative measure. Towards this end I normalized both the TIC and TC values using standards that I had run and thereby obtained normalized TOC values. However, I still found no correlation between my normalized TOC values and the relative fluorescent intensities of the CW3-807 samples (I had no relative fluorescent intensities for the SVC 99-1 samples). Due to these results, I must assume that the method I used was not resolved or precise enough to be an accurate measure of the extremely low levels of TOC present in speleothems.

Results

Initially, we had hoped to be able to compare our data with the work of earlier researchers, but we discovered that most previous synchronous scans had been run at a wavelength separation of 18nm (Miano et al.,1988; Senesi et al., 1991; Ramseyer et al., 1997), whereas we had run ours at 75nm. Realizing this and also recognizing the limited timeframe of my research I decided to focus instead on a narrower goal. We focused on simply determining whether there were differences in the spectra and where those differences occurred.

We found that the most significant differences within the same speleothem sample were between the relative fluorescence intensity values. The values observed from EEM scans for the wavelength separation at which the peaks fell did not vary much. Also, the position of the peaks didnít shift significantly on the spectra from the synchronous scan runs.

The excitation wavelengths corresoponding to the main peak varied from 325nm (Garden of the Gods) to around 370-380nm (SVC-983C and CW3-807). The corresponding emission wavelengths varied from 395 to 460nm. One interesting note is the appearance of a second peak on some of the EEM scans, appearing in close proximity to the lower left of the main peak. In all of my results in which a second peak appeared, the second peak fell at excitation wavelengths of between 340 and 350nm and emission wavelengths of between 390 and 400nm. This second peak may explain the shift of fluorescence emission wavelength noted by McGarry and Baker (2000). The second peak adds height to the primary peak, but unevenly. Thus the location of the maximum relative intensity of the primary peak shifts towards the second peak. Also notable is the significant shoulder observable in many of the scans. The humic acids produce this shoulder, whereas the main peak is a signature of the fulvic acids present in the speleothems.

A large part of my time was also spent in determining proper, consistent and reproducable methods. Hopefully, the results that were obtained will be of use in further research.