Low emissivity (low-E) coatings consisting of dielectric/silver/dielectric multi-layer stacks are applied to large-area architectural glazing to reduce heat losses from buildings. In this work TiO2/Ag/TiO2 stacks were deposited onto soda-lime glass by pulsed DC reactive magnetron sputtering. The coatings were annealed in the range 100−600 o C to study silver diffusion through neighbouring layers. Depth-profiling analysis was performed on these samples using time-of-flight secondary ion mass spectrometry and selected samples were also analysed by X-ray photoelectron spectroscopy and Rutherford backscattering spectrometry. Fick's second diffusion law

The power density delivered by particles to an electrically isolated substrate in an asymmetric bipolar pulsed dc unbalanced magnetron has been quantified. The plasma source was operated in argon with a titanium target, and measurements were made using both a calorimeter probe and time-resolved Langmuir probe incorporated into a specially made substrate holder. The main results from the calorimeter probe show clearly that with increased pulse frequency (from dc to 350kHz) and reduced duty cycle (90%–50%), the particle power density (from ions, electrons, sputtered Ti, and backscattered Ar) at the substrate increases significantly. For instance, at 350kHz and 60% duty cycle, the total power density is 83mW∕cm2, about 60% higher than in dc mode for the same time-average discharge power. However, from an inventory of the individual particle contributions to the total power density derived from time-resolved Langmuir measurements and a simple model of the substrate sheath and plasma internal processes, we predict values of power density much lower than those measured. The measured and calculated values are in close agreement for the results obtained in dc mode but diverge at high frequencies. It is believed that this is due to the Langmuir probe measurements being unable to observe the presence of high-energy ions, created during the transient peaks in the electron temperature at the transitions from on off and off on [J. W. Bradley et al., Plasma Sources Sci. Technol. 11, 165 (2002)] which subsequently bombard the substrate. This paper shows conclusively the benefit of pulsing the magnetron over and above dc operation for enhancing the ion power per depositing neutral in the ion assisted deposition process.

In this paper we demonstrate an improved technique for obtaining time-resolved ion energy distribution functions (IEDFs) in pulsed plasma discharges using a commercial quadrupole mass energy analyser. The method involves extracting ions from the plasma at selected times during the pulse cycle through the application of a synchronized electrical bias on a grid assembly built in the barrel of the instrument. Placing the triggered mesh in the barrel of the energy analyser makes the ‘transit’ gap length (between the chopping mesh and neighbour electrodes) ∼2 mm and that results in a significant increase of the time resolution. Due to the high vacuum conditions inside the barrel, no unwanted ‘parasitic’ discharge will exist between the extractor and orifice. The electrostatic chopping by the built-in mesh allows time-resolved IEDFs to be determined without reference to the inherent time delay caused by the ion flight time through the spectrometer. The technique has been applied to the low-pressure pulsed dc magnetron discharge operating in Ar gas and at a pulse frequency of 100 kHz. Time-resolved IEDFs were obtained with a 1 µs time resolution, and these revealed large variations in the form and width of the ion energy spectra at different times during the discharge cycle, governed by the temporal evolution of the plasma potential. A calculation based on ion dynamics in the grid assembly agrees well with experimentally obtained results and demonstrates that this method can provide time resolutions as small as 200 ns.

This paper investigates the influence of pulse frequency and duty on the deposition rate during the pulsed magnetron sputtering process. Whilst deposition rates increased with duty, they also showed a very marked decrease with pulse frequency. Detailed analysis of the data implies that there is a ‘dead time’ of the order of at least 500 ns at the beginning of each pulse‐on cycle, during which negligible sputtering takes place. As pulse frequency increases, the ‘dead time’ becomes a greater proportion of the total on‐time, thus leading to a fall in the deposition rate. Other factors also contributing to this effect include a reduction in the average power delivered to the target as pulse frequency is increased, and significant differences with frequency in the rate of change of target voltage at the beginning of each pulse on cycle.magnified image

No abstract