[en] The relation between stress evolution and surface morphology during deposition of sputtered films is examined by combining kinetic Monte Carlo simulations and stress measurements. We find that the surface morphology is susceptible to an instability, which transforms from layer-by-layer growth to the formation of pillarlike columns. The gaps between these columns prevent complete densification and can lead to a network of pores in the layer. We propose that the formation of this structure changes the stress in the growing layers from compressive to tensile.

[en] Materials synthesized by deposition techniques are often plagued by high levels of residual stress. While the origin and control of this stress in thin (sub-micron) films has been an active area of research, it is not clear how the results extrapolate with thickness. In the present work, in situ residual stress measurements are performed during the sputter deposition of beryllium, spanning the transition from thin to thick. Variables including sputtering gas pressure and substrate biasing are shown to strongly affect both the average and instantaneous stress levels measured during film growth. Detailed microstructural characterization is performed to assess the grain structure, surface morphology, and crystallographic growth texture of representative specimens. The microstructure is correlated with theoretical models of stress generation to interpret experimental measurements. A stress map is also constructed, generalizing the effects of processing and material parameters on stress state

[en] We present a general approach to prepare metal/aerogel nanocomposites via template directed atomic layer deposition (ALD). In particular, we used a Ru ALD process consisting of alternating exposures to bis(cyclopentadienyl)ruthenium (RuCp₂) and air at 350 deg. C to deposit metallic Ru nanoparticles on the internal surfaces of carbon and silica aerogels. The technique does not affect the morphology of the aerogel template and offers excellent control over metal loading by simply adjusting the number of ALD cycles. We also discuss the limitations of our ALD approach and suggest ways to overcome these