Experimental

4.2.1 Experimental set-up

The experiments were performed using the methods outlined in Chapter 2, with the liquids described in section 2.4, and the single-point detection set-up detailed in Section 2.7.1. The probe-laser beams was set to provide ~200 μJ pulse-1_{, using a} combination of a polarizer and a λ/2 waveplate. The valve was placed to generate a molecular beam at normal angle of incidence with respect to the surface, intersected by the probe beam at right angles at a distance, d, of 230 ± 1 mm. The surface itself was placed at 10 ± 1 mm from the probe beam, thus the overall distance between the valve and the liquid surface was determined to be 240.0 ± 1.4 mm.

The experiments reported in this chapter were carried out prior to the comprehensive molecular beam optimization tests reported in Chapter 3, and without monitoring the current draw of the discharge device. As such, the properties of the molecular beam were optimized by monitoring the stability and reproducibility of the LIF signal over a lengthy period of time with the ‘tune-up’ program, and via acquisition of preliminary TOF profiles and excitation spectra. For each carrier gas used, the conditions were optimised to provide the highest OD number density and the most rotationally cold beam, by varying the voltage of the HV pulse and the backing pressure of the gas. All other in situ conditions were set arbitrarily based on previous conditions determined from experiments where OH was produced via photolysis (Section 2.5.3)

A high pressure of carrier gas (3 bar of He or 2 bar of Ne) was bubbled through the D2O reservoir, creating molecular beam mixtures of ≈1% and ≈1.5 % D2O in He and Ne, respectively. The mixture was fed to the pulsed valve, which had an aperture of 1.0 mm. The nominal pulse length of the pulsed valve was set at 500 μs. The discharge device attached to the valve was the original un-modified design provided by the Greaves group, described in section 2.5.4. The fast switch was triggered to apply an optimized voltage of -1400 V for the He-seeded beam and -1150 V for the Ne seeded beam for a duration of 10 μs, timed to fire typically 100 μs into the gas pulse. To further stabilize the discharge, a tungsten filament was placed in the ‘source’ part of the main chamber (see below), and a current of between 0.4 and 2 A was applied, adjusted regularly throughout the day to achieve the optimal discharge conditions.

The free-jet expansion of the beam was collimated by a 5 mm diameter conical skimmer, located at 167.0 mm from the valve nozzle and 160.5 mm from the last constraining element of the discharge device, i.e. the aperture of the front electrode. This was chosen to ensure that the molecular beam would dose a ~ 1 cm2 _{area on the} liquid surface. A 30 x 30 cm gas baffle (as described in section 3.6.2) was attached to the skimmer holder plate, effectively dividing the main chamber into source and scattering chambers. The position of the skimmer holder allowed the user to easily accommodate both the hot filament on the ‘source’ side of the main chamber, and the large gas baffle itself.

4.2.2 Monte Carlo Simulations

To aid in the interpretation of the experimental profiles obtained from the single-point detection set-up, OD TOF profiles were simulated with a variant of the LabVIEW™ Monte Carlo simulation previously developed in the McKendrick group [59].

The simulated TOF profiles were obtained by calculation of a large number of individual OD trajectories with correctly weighted Monte Carlo sampling over the experimental parameters. The LabVIEW™ program allowed the user to set parameters, such as the pulsed valve-to-surface distance, the valve-to-skimmer distance, the diameter of the skimmer, the angle of incidence with respect to the surface, length of the HV pulse, the probe beam size and shape (round beam or rectangular sheet), and the dimensions of the liquid surface to match the experiments. The trajectories of the incident OD radicals were selected by checking whether they passed through the skimmer. If the incident OD trajectories then intersected the detection volume (i.e. the fraction of the probe beam viewed by the single-point detection set-up), the simulation program would calculate the range of delays in which the OD radicals were within that volume, based on the velocity of OD in that trajectory. The intensity of the simulated TOF profile was then appropriately incremented at the relevant delays.

The velocities of the OD radicals with incident trajectories were picked at random from a distribution of speeds which was assumed to follow a sum of individual Gaussian distributions. The average speed and full width half maximum (FWHM) of each Gaussian distributions, as well as the relative weighting of each individual distribution with respect to the overall simulated profile, were varied by the user until the simulated

profile matched an experimentally obtained incident beam profile imported into the simulation program. To accurately determine the incident beam parameters, the experimental profile was obtained from the set of incident OD profiles acquired at the lowest N level probed.

The trajectories and velocities obtained from the simulated incident beam were then used to determine the trajectories of the scattered OD from the surface. Once an incident OD trajectory that had successfully passed through the skimmer was determined to have intersected the liquid surface, the impact point and flight time to the surface were used as the starting point for an OD recoil trajectory. The IS or TD character (see Chapter 1) of these trajectories was assigned by sampling from a binary distribution given by a user-set IS/TD ratio. TOF profiles of both the IS and TD component were then simulated separately. As with the incident beam, when the trajectories intersected the probe volume, the range of delays when OD was in the probe volume were calculated, and the intensity of the simulated profiles was incremented at the relevant delays. A representative simulation of incident and scattered OD profiles with typical experimental parameters and a 1:1 IS/TD ratio is presented in Figure 4.2.

To obtain the TD profile, the direction of the simulated TD trajectories was sampled from a cosine-weighted distribution around the surface normal, whereas the recoil speed was selected at random from a Maxwell-Boltzmann distribution characterized by a user-defined surface temperature [4, 88-90]. With regards to the IS profile, the speed and direction of the IS trajectories was determined by the program from a user-defined angular distribution of the IS trajectories, a final-to-incidence angle ratio, and an effective surface mass of the liquid. Unlike the assumed TD distribution, those parameters are unknown a priori, as they are dependent on the mass, angle of incidence and collision energy of the incident projectile [88-91]. There are no known results in the literature that provide those parameters for experimental conditions comparable to the ones used in the scattering experiments reported in this chapter. Therefore, only the simulated TD profiles will be used for comparison with the experimental TOF profiles shown below.

100 120 140 160 180 200 220 0.0 0.2 0.4 0.6 0.8 1.0

Simul

ated LIF

Intensity

Discharge-probe delay / s

Figure 4.2: Representative Monte Carlo simulated TOF profiles. 107 _{trajectories were} simulated in these profiles, with an IS/TD ratio of 0.5, a TD temperature of 300 K, and detection time step of 1 µs. The parameters of the simulated incident beam (black line) were manipulated to match an experimental TOF profile (open circles) on the Q1(2) transition. This gave a simulated scattered OD profile (red line), where the IS and TD

contributions are represented by the cyan and magenta lines, respectively. For demonstrating purposes, the parameters of the simulated IS profile were arbitrarily set

to match the results obtained from scattering O(3_{P) from squalane at high collision} energies, as reported in references [4, 88-90]