Weak Near-Infrared Aborption Lines in Water Vapour


Global warming is a "hot topic" these days, but "traditional spectroscopy" is seen by some as "old-fashioned" or "boring".  What is the connection?  An accurate spectroscopic model of the atmosphere is vital to measuring the earth's radiation balance (how much heat is coming in and how much heat is going out).  The model relies on a database of spectral absorption lines (such as HITRAN), each with a measured wavelength, absorption strength and line shape.  This is where spectroscopy comes in - to populate the database requires detailed, painstaking and very careful experiments followed by very sophisticated analysis to determine the quantum transitions that give rise to each observed absorption. 

Measurements are required for each atmospheric consitituent, and it was the case until recently that weak water vapour absorption features in the near infra-red region were relatively poorly characterised.  Water is an important greenhouse gas not only because of its absorption spectrum (especially in the long-wave infrared) but because its concentration in the atmosphere will rise (by evaporation from the oceans) if the planet warms.  At the same time it became clear that there was an unexplained discrepancy between the measured radiation flux coming through the atmosphere and that predicted on the basis of models. 

It was clearly time for some high-quality experimental work!


The practical team comprised Roland Schermaul, Dick Learner, Jim Brault from the USA and myself.

Measurements were made at the Molecular Spectroscopy Facility at the Rutherford Appleton Laboratory using a Bruker IFS120 spectrometer and a long-path White cell. 

The White cell is a chamber about 9 metres in length which can be evacuated and then filled with water vapour at any desired pressure and temperature.  The cell contains three mirrors which, with very careful alignment can pass a beam of light many times up and down the cell so that a very long effective path is achieved.  This permits even very weak absorptions to be detected.  It also gives the operator eye-strain and a headache because one hase to count many small light spots and the alignment is very sensitive at high path lengths: we had the cell set for about 900 metres at one point, and routinely used about 50 passes.

The cell is illuminated by light from a high-resolution Fourier-transform spectrometer - a commercial Bruker IFS120 instrument.  The moving mirror on this instrument has a travel of about 3 metres, so extremely fine spectral resolution can be achieved. The instrument is really designed for operation in the infrared, so we were pushing it quite hard to operate in the near infrared / visible region!

The standard detectors and electronic filters on the IFS120 were  judged not to be adequate for the task, so I designed and built custom replacement units and fitted them to the instrument - directly to the signal digitisers.  Special optical bandpass filters were procured and fitted to optimise the signal quality within each waveband of interest, and we also carried out a very careful end-to-end re-alignment of the interferometer and all auxiliary optics to ensure peak performance.  We are grateful to the MSF staff who put up with our "disruptive behaviour" and allowed us to "mess about" with their very expensive instrument - they showed considerable restraint at times!

An extremely pure sampl of water was prepared by Roland Schermaul and freeze-distilled to remove all traces of air.  The sample, in a tiny flask, was connecetd to the evacuated White cell and the valve bwtween the two was opened. The water level dropped slowly as the vapour diffused into the White cell, and measurements started...

Considerable care was taken to operate the instrument at the peak of its capabilities: we avregaed data over many hours, being careful of ageing effects in the light source (which was operated very gently and burned in for each run).  We did not opt for maximum resolution but instead selected the best resolution that we could achieve given the shape of the spectrum and the prevailing noise floor; the data we collected were therefore minimally redundant, and we were able to use the available observing time to best advantage to reduce noise by averaging many scans.  This strategy paid off: our final signal to noise performance was excellent!

Analysis and Results

The final data were analysed by Roland Schermaul, Dick Learner, Jim Brault and a group headed by Prof Tennyson at University College, London who are experts in the quantum mechanics of water molecules. The process comprised very careful phase correction of the raw interferograms followed by careful calibration of the wavelength scale (which depends on the instrument alignment and optical beam solid angles), determination of the background (zero absorption) level and then extraction of each line with fitting to try to determine its strength and shape.  Finally the quantum mechanics tried to assign each line to a transition of the molecule.  The results can be seen in two papers here and here .

The study was very successful in that many new weak absprption features were identified and transitions were assigned to some of them.  It was also possible to make some corrections to the established database.

In a small but hopefully valuable way our models of the atmosphere have been improved.  Who says spectroscopy is boring or useless?!