3.3.1- Lamp configuration
To avoid the problem connected with energy feeding from the gas volume, the high voltage is introduced through an atmospheric pressure column feedthrough to the lamp. The lamp is also designed with a multi-cylindrical electrode geometry, and consists of half sealed quartz tubes containing a ground conductive material (iron filings) (figure 2.20). Each tube being assembled through a flange of the lamp chamber, a high voltage lead can be connected from the outside of the gas volume. See also pictures on the next page.
Flange Iron filings filling Flange
Vacuum seal (torr seal) Pyrex high voltage
electrode holder “
Vacuum seal (torr seal)
Discharge Gap
High voltage electrode Ground electrode
Figure 2 20 : Lamp set-up enabling the generation o f radiations o f wavelengths shorter below the suprasil wavelength cut-off (160nm)
Spectra of the rare gas continua were obtained for the cases of argon, krypton and xenon (figure 2.10). High pressure second continua of neon (8 6nm) and helium
(73nm) could also be obtained with this lamp, but this implies the absence of a window between lamp and reaction chambers, hence it limits their use to high pressure reactions. The need for high purity of the excimer gas also implies the absence of any other gas plasma. Such a light source can therefore be used for curing or etching, patterning and surface reactions, but are incompatible with photolytic
I
VUV Excimer lam p (top) and system (bottom).
This system allows radiations dow n to 126nm to be directed to a deposition chamber.
3.3.2- Optimisation for Ai2* continuum generation (126nm)
The device presented in figure 2.20 is mounted in a chamber where is introduced argon. The electrical microdischarges generated in the gap separating the three annular electrodes create Ar2* excimer species, and their radiative dissociation emits
126nm photons. This radiation correspond to a 9.8eV energy transition between the and states of the excimers to their stable ground states. It is clear that the higher the transition, the more it is likely that some impurities present energy levels located between the transient states. As such, the excimer generation of 126nm from a discharge in argon is a lot more exposed to purity problems than that of lower transition excimers (e.g., xenon at 7.2eV). Also, it has been shown that in the case of xenon most of the purity problems could be overcome by completely sealing the discharge gas in a quartz tube. Here, since the gas is directly in contact with the chamber volume, impurities trapped on the walls of the chamber may be released and alter severely the intensity of the spectrum. For instance, at the early stages of the development of this lamp, an unexplained peak at 300-305nm was measured in the spectrum of the lamp output. This peak was persistent with respect to lower background pressures, as well as higher cylinder purity. Eventually, it occurred that it was caused by the presence of a few tens of centimetres of PTFE piping in the argon feeding line, i.e. between the cylinder and the lamp chamber. As argon is know to be extremely non reactive at room temperature, and also since no degasing of the piping materials is likely to happen at high pressures (few atm), this shows well how dramatically anultra low concentration of impurities can alter the spectrum output of such lamps. Also, unlike for the xenon cylindrical lamp, the development of this device did not consider the possible implementation of water cooling facilities. As a result, not only the photo-emission decreases due to the temperature increase (phonon losses), but also the heat generated by the lamp promotes impurities to degas from the chamber walls and to affect the spectrum output.
To overcome those purity problems, one empirical solution is to allow a permanent flow of the discharge gas to be flown through the discharge volume. With the use of an optimised 1.51/min flow of argon, no such decrease of the output emission is
the gas has to be taken into consideration when further development is to be made. For instance, according to the excessive prices of xenon and krypton, it may be more appropriate to consider the use of a UHV bakeable chamber thus enabling higher gas purity and preventing the need for a flow.
3.3.3- Output power
Although the use of actinometry techniques was preferred for the measure of the xenon lamp intensities, less convenience is offered in the case of argon excimer emissions. In fact, a look at the absorption cross section data for oxygen shows that a varies from 0.3 to 1000 and down to 1.6 atm'^cm"^ at 121, 125, and 127nm respectively [Calvert] [Inn] [Marr] (as known, oxygen shows very little absorption in the region of the spectrum close to the Lyman line (121.6 nm)). These huge dispersions on a render quantum efficiency calculations almost impossible for the case of ozone formation. As a result, the present intensity measurements were limited to spectra and photo-detector evaluation, with reference to calibrations obtained with the xenon lamp devices. Using the device presented in § 2.2.1, in-vacuo, it comes that the lamp can emit intensities of 30 mW/cm^ at a narrow range of wavelengths centred around 126nm.