ADVANCES IN OPTICAL WATER ISOTOPE RATIO MEASUREMENTS
2. INFRARED SPECTROMETRY
The near and mid-infrared absorption spectrum of gas phase water shows a large number of highly characteristic rotational-vibrational transitions that are very sensitive to isotopic substitution (in fact, such is the case for most small molecules). Figure 1 shows the spectrum of water over a wide range in the infrared. At suffi ciently low vapour pressure and high instrumental resolution the individual ro-vibrational transitions are easily resolved, and can be uniquely assigned to a water molecule of particular isotopic composition (isotopologue).
Even at higher pressure, or in the liquid phase, the unresolved vibrational bands of the deuterium isotopologue may be suffi ciently displaced to be useful for a quantitative determination of the deuterium to hydrogen isotopologue ratio.
Since the intensity of the experimentally recorded spectral features (whether individually resolved ro-vibrational transitions, or an unresolved vibrational band) may be directly related to the abundance of the absorbing species, recording the spectrum containing abundant and rare isotopologue features, in both an unknown sample and an isotopically well-known reference material, enables one to relate the isotopic composition of the sample to that of the reference material in a manner as described in, e.g., [2].
In Figure 2 we illustrate the basic principal of an isotope ratio measurement by means of infrared spectroscopy. It shows the spectra as a function of the light frequency ν in a small spectral region near 2.73 µm (3664 cm–1) of two vapour- phase, water samples, one acting as the reference material (back-traceable to an international standard material such as Vienna Standard Mean Ocean Water, VSMOW), the other an ‘unknown’ sample. The absorption features correspond to ro-vibrational transitions in the most abundant isotopologue, H16OH, and the rare isotopologues H17OH, H18OH, and H16OD. The spectra were recorded as direct absorption spectra and converted to absorbance spectra by the applicationI of the Beer law of linear absorption:
0 0
0 0
I I l I I
e l Ln
I I
α α ¤ ³
¥ ´
¦ µ
$ $
(1) In the above, ∆I represents the absorbed intensity of the incoming (laser) beam with intensity I0 after traversing a length l of the absorbing medium, characterized by the absorption coeffi cient α.1
The relative deviation xδr(s) of the isotopic ratio of the sample (xRs), with respect to that of the reference (xRr) is given by:
1 Note that the defi nition of α here differs by a factor of l from the one in [2].
/
( ) 1 1
/
x a
x
x s s
r x x a
r r
n n s R
R n n
δ x (2)
The subscript s refers to the sample, r to the reference material. The superscripts a and x refer to the most abundant (H16OH) and the rare isotope species (H17OH, H18OH, or H16OD), respectively.
With a proper choice of experimental conditions, the δ-value follows directly from the intensities in the spectra:
/
( ) 1
/
x a
x s
r x a
r
s
α α
δ α α (3)
Here, α = α(ν0) = S × f(ν0)n represents the experimentally determined (maximum) absorption coeffi cient at centre-line frequency, provided that the line shape, given by the normalized line shape function f, is the same for all transitions. If not, one has to use the integrated line intensity. In Eq. 3 the, normally justifi ed, assumption has been made that the (effective) optical path FIG. 2. Principle of the isotope ratio determination. In this fi gure, the absorbance spectra of sample and reference have been scaled to give equal H16OH absorbances. The isotope ratio may then be “read” directly from the ratio of the corresponding line intensities. In this case, the reference material was a local (Groningen) standard (2δ = –41.0‰, 17δ = –3.36‰, and 18δ = –6.29‰) and Standard Light Antarctic Precipitation was used as "unknown"
sample (2δ = –428.0‰, 17δ ≈ –29.7‰, and 18δ = –55.5‰).
length l is the same for each isotopic species, or for the sample and reference spectra, or for both.
Since the line strength S depends on the number of molecules in the lower level of the transition, it is in general temperature-dependent: a change in temperature will redistribute the population over the rotational levels of the ground vibrational state (see, for example, [2]). Sample and reference spectra should therefore be measured at exactly the same temperature, and/or the isotopologue lines should be chosen such that their temperature coeffi cients are nearly equal. However, since to good approximation the measurement does not depend on the absolute temperature, but rather the difference between the sample and reference gas cells, passive stabilization with a good thermal contact between both gas cells is suffi cient to make the temperature induced error negligible (i.e., < 0.1‰). Long-term absolute temperature stability is, however, important for measurement schemes that measure the sample and reference spectra not simultaneously but rather sequentially (in the same gas cell).
A careful selection of the spectral features is also important to assure that no interference from other species (whether other isotopologues of the water molecule, or entirely different molecules) is present, while a similar absorption of the different isotopologues is desired in order to assure a comparable signal- to-noise ratio on all spectral lines of interest, and eventually the best precision on the isotope ratio determination. This is generally achieved by searching for a set of lines with similar absorption coeffi cients (taking into account the natural abundances of the isotopologues). The alternative of compensating a lower absorption coeffi cient (usually for the rare isotopologue) with a correspondingly longer optical path length has been used fi rst by Bergamaschi et al. [8], but so far not for water, as far as we are aware.
The search for the best spectral region is further complicated by the requirement that all isotopologue lines, including at least one belonging to the major isotopologue and one to (one of) the rare isotopologue(s), occur within a narrow wavelength range, limited by the tuning range of the laser. In fact, the simultaneous measurements of the three major isotopic ratios in water (δ18O, δ17O, and δ2H) was demonstrated to be possible, in both the fundamental band near 2.73 µm [9] and the overtone band near 1.39 µm [10], thanks to the fortuitous occurrence of suitable ro-vibrational lines of the H16OH, H17OH, H18OH, and HO2H isotopologues within a single laser frequency scan of about 1 cm–1 wide. In both spectral regions, very few other possibilities, if any, exist.
We have shown that this situation can be improved by the use of two lasers, each tuned to a different set of isotopologue lines [11]. In this scheme, the two lasers are wavelength modulated at different, incommensurate frequencies, and the two corresponding absorption spectra are retrieved from the detector signal using for each gas cell channel two phase-sensitive detectors, one for each
modulation frequency. An alternative to this dual-wavelength multiplexing technique would be to time-division multiplex the signals from each gas cell, resulting in a slightly less complex setup, but at the cost of an increased measurement time.