Appendices

A1. Electronics

The infrared laser is run from a Fluke 415B high voltage supply through a 120 kW ballast chain. The operating current is chosen, largely, according to the optical power requirements and the condition of the discharge tube.

The discharge in the tube is struck by means of a piezo-electric device which puts a negative voltage spike (-5 kV) onto the cathode thus increasing the potential difference across the discharge tube momentarily.

Fine tuning of the cavity length is achieved by means of the piezo-electric stack which is controlled manually with an in-house piezo driver running off a Brandenburg 475R power supply.

A2. Low Pressure Gas Cell and Filling Rig

A low pressure gas cell may be used to obtain direct measurements of the line shapes of lower resolution instruments, using a variety of gases, but has primarily been used to validate laser measurements of the line shape of high resolution instruments such as the Bruker IFS 120M (chapter 6). A portable filling rig was constructed to enable the cell to be filled as required during field trials.

A2.1 Low Pressure Gas Cell

The cell body is 12 cm long and fabricated in Pyrex with two side arms fitted with PTFE taps (Young's of Acton). A Baratron pressure gauge may be attached to one arm for accurate measurement of gas pressure within the cell both during the evacuation and filling process, when the filling rig is attached to the other arm, and while in use in a spectrometer.

The potassium bromide cell windows have 2º wedge and are attached to the body with epoxy resin. The clear path through the cell is 95 mm in diameter enabling it to be inserted into the the solar beam in a Bruker 120M without obscuring any of the solar or laser beams.

A2.2 Cell Filling Rig

The cell filling rig is shown, schematically, in figure A2-1. The sample bottle may be charged to 20 bar with hydrogen bromide (appendix A3) prior to field trials to ensure an adequate supply.

Figure A2-1. Schematic diagram of low pressure gas cell filling rig with double armed cell attached (appendix A2.2).

A2.3 Cell Filling Procedure

The gas cell and the cell filling system (excluding the sample bottle) are pumped down to the ultimate vacuum (20 mbar) of the rotary pump. Needle valve N2 is then shut and N1 opened, to permit a small amount of the high pressure gas to pass out of the sample bottle, and then shut. The quantity of gas trapped between the needle valves is insufficient to present any threat to the structural integrity of the cell, should it all be released into the system. Ball valve B3 is then shut and N2 used to raise the pressure in the system and cell to 0.1 bar. The system is pumped down once more, this time with N2 shut. This procedure is carried out to ensure that any residual gas in the system is that from the sample bottle and not air or a previously used sample gas.

When the system is evacuated, B3 is again shut and N2 opened to raise the pressure in the system to a suitable pressure for the gas cell (less than one millibar). Valves Y2 and B1 may then be shut and the cell removed by disconnecting a joint between these valves.

Figure A2-1 shows the system configured for a double-armed cell but the Baratron may be removed from the cell at tap Y1 and attached to the system at B3 if the system is to be used for filling single-armed cells.

A3. Selection of Gas for Low Pressure Cell

The criteria used to select a gas for the low pressure cell were that it should have a simple spectrum consisting of narrow, isolated lines and that these should be in one of the routinely measured spectral windows, preferably close to the laser frequency of 2947.9 cm^-1. It should also be a substance not found in the atmosphere in significant quantities and should have some lines clear of atmospheric absorptions. These resulted in the choice of hydrogen bromide which has a Doppler line width at 293 K of approximately 0.003 cm^-1. A survey of the P and R branches of the hydrogen bromide spectrum, recorded with the sun as a source, was conducted to establish the best lines to use for this instrument line shape work (table A3-1) and the doublet at 2412.7 cm^-1 and 2413.0 cm^-1 was selected.

Table A3-1. Survey of atmospheric features interfering with 19 lines from P and R branches for each of H⁷⁹Br and H⁸¹Br . '?' denotes a feature not identifiable from the HITRAN '96 database (appendix A3).

A3.1 Simulation of the Hyperfine Spectrum of HBr

While the high mass of hydrogen bromide ensures that it has a relatively narrow Doppler width, its large nuclear quadrupole moment causes significant broadening of the spectral lines in the 1-0 vibration-rotation band (Duxbury (private communication), Coffey [1998]). Broadening of the lines arises from nuclear hyperfine splitting of the energy levels in the J®J+1 and J®J-1 transitions between the ground (u=0) and first excited (u=1) vibrational states. The broadening effect is known to be greatest for low values of J (Duxbury (private communication), Coffey [1998]). Figure A3-1 illustrates the effect by showing the measured widths of the 38 HBr lines visible in a cell spectrum recorded with a glow-bar source.

Figure A3-1. Widths of 38 HBr absorption lines from spectrum taken on Bruker IFS 120M using 257 cm optical path difference and 0.65 mm diameter aperture using a black body source. Lines are shown for both isotopes and in order of ascending wavenumber (appendix A3.1).

Due to the lack of useful data in the literature concerning the hyperfine spectra of the P8 vibration-rotation lines of H⁷⁹Br and H⁸¹Br it was necessary to simulate these in order to ascertain their shapes and widths. The shifted frequency of the hyperfine line relative to the unperturbed frequency can be shown (Gordy & Cook [1970]) to be

s = so - eQq [Y(J+1, I, F') - Y(J, I, F'')] ,

(A3.1)

where so is the unperturbed frequency and s the new, shifted frequency. Values for the functions Y(J,I,F'), Y(J+1,I,F'') and the relative intensities for the lines can be obtained from tables (Gordy & Cook [1970]). The hyperfine spectra of hydrogen bromide lines in the P and R branches of the spectrum were simulated using the nuclear quadrupole coupling coupling constants, eQq, for deuterium bromide reported by van Dijk & Dymanus [1974].

Equation (A3.1) provides values for the frequency shift of each hyperfine component, which was assumed to have a Gaussian profile of width (Ds) equal to the Doppler width calculated (section 4.1) to be 0.00333 cm^-1 for H⁷⁹Br and 0.00329 for H⁸¹Br at 293 K.

The hyperfine lines so generated were summed for each observed line, and the resultant convolved with a theoretical instrument line shape (calculated assuming 257 cm optical path difference and 0.65 mm diameter aperture, and perfect alignment). This was carried out the for eight P-branch lines and eleven R-branch lines of each isotope and the widths of these are shown in figure A3-2. Hyperfine spectra and simulations of corresponding absorbtance measurements are shown for the P1 and P8 lines of H⁷⁹Br and H⁸¹Br in figure A3-3.

Figure A3-2. Widths of simulations of the 38 HBr lines shown in figure A3-1 (appendix A3.1).

Figure A3-3. Hyperfine spectra (left) and simulations of corresponding absorbtance measurements (right), for the P1 and P8 lines of H⁷⁹Br and H⁸¹Br (appendix A3.1).

The results from the simulation of the hyperfine spectra (figure A3-2) show reasonable qualitative agreement with the observations (figure A3-1) across the P and R branches (with the exception of the R1 line) which supports the belief that nuclear quadrupole coupling was causing hyperfine splitting to broaden the observed lines, and also the choice of the doublet near 2412.7 and 2413.0 cm^-1 for instrument line shape validation work.

Following this work the hyperfine structure of HBr was published by Coffey et al. [1998]. Figure A3-4 shows simulated hyperfine spectra of the P1 (top) and P8 (bottom) lines of H⁷⁹Br constructed, as described above, using parameters from van Dijk & Dymanus [1974], Gordy & Cook [1970] and the HITRAN 1996 database (left) alongside those constructed using the new parameters reported by Coffey et al. (right). There is obviously good agreement between the P1 lines and little difference between the P8 lines with the simulation based on Coffey's results being only 3% broader than the van Dijk/Gordy/HITRAN simulation with no discernible change in shape. There is a frequency shift (not shown here) between the two sets; Coffey et al., who calculated line positions using the spectroscopic constants of Braun and Bernath, (Braun & Bernath [1994]) note that their line positions "are consistently lower by 0.001 - 0.003 cm^-1 from the 1996 HITRAN" values used here. This frequency shift is of no practical relevance to the work described herein as it does not impinge to any significant extent on the hyperfine structure of the P8 lines.

Figure A3-4. Comparison of simulations of P1 and P8 lines of H⁷⁹Br based on data from van Dijk & Dymanus/Gordy & Cook/HITRAN '96 (left) and Coffey et al. (right). Hyperfine spectra are shown in dashes; solid line depicts simulated measurement; wavenumber scale is relative to the centre of the simulation (appendix A3.1).

Summary
Acknowlegements
Contents
Chapters: 1, 2, 3, 4, 5, 6, 7, 8
Appendices
References