We propose to use first principle molecular dynamics (MD) simulations using semi-local and hybrid functionals to compute vibrational properties of liquid mixtures and liquid/solid interfaces, with the goal of comparing directly with experimental data becoming available on specific systems, that is water/methane/carbon dioxide mixtures and water in contact with oxide and graphitic-like surfaces. These calculations will play a key role in understanding the physical and chemical problem of processes involved in the carbon cycle on Earth, in particular methane dissociation under upper mantle conditions, as well as in understanding structure and dissociation of water in contact with solid hydrophobic and hydrophilic surfaces. In particular we will be able to separate effects responsible for observed phenomena (e.g. pressure and temperature effect, density dependence and interface structure dependence on vibrational properties of liquids and liquid mixtures) that cannot be easily separated in any experiment. The problems addressed here are part of two funded projects: one by DOE/ASCR and a complementary one by Shell Corporation. The request per year represents an increase of a factor of ~ 2 with respect to our use of resources on BG/P per year, in the last two years. Although not dramatic, this factor will substantially increase our ability to understand complex liquids and liquid mixtures and will position us well to attack computational issues involved in the planned transition from petascale to higher performance computations. In particular it will allow us to fully test and optimize space decomposition techniques for hybrid functional calculations, that we expect will play a key role in efficiently performing ab-initio MD simulations on next generation architectures.