The potential of mean force for uncharged macroparticles suspended in a fluid confined by a wall or a narrow pore is computed for solvent-wall and solvent-macroparticle interactions with attractive forces. Bridge functions taken from Rosenfeld's density-functional theory are used in the reference hypernetted chain closure of the Ornstein-Zernike integral equations. The quality of this closure is assessed by comparison with simulation. As an illustration, the role of solvation forces is investigated. When the "residual" attractive tails are given a range appropriate to "hard sphere-like" colloids, the unexpected role of solvation forces previously observed in bulk colloids is confirmed in the confinement situation.

The accuracy of the structural data obtained from the recently proposed generalization to non-additive hard-spheres (Schmidt 2004 J. Phys.: Condens. Matter 16 L351) of Rosenfeld's functional is investigated. The radial distribution functions computed from the direct correlation functions generated by the functional, through the Ornstein-Zernike equations, are compared with those obtained from the density profile equations in the test-particle limit, without and with test-particle consistency. The differences between these routes and the role of the optimization of the parameters of the reference system when the functional is used to obtain the reference bridge functional are discussed in the case of symmetric binary mixtures of non-additive hard-spheres. The case of highly asymmetric mixtures is finally briefly discussed.

We investigate the question of determining the bulk properties of liquids, required as input for practical applications of the density functional theory of inhomogeneous systems, using density functional theory itself. By considering the reference functional approach in the test particle limit, we derive an expression of the bulk free energy that is consistent with the closure of the Ornstein-Zernike equations in which the bridge functions are obtained from the reference system bridge functional. By examining the connection between the free energy functional and the formally exact bulk free energy, we obtain an improved expression of the corresponding non-local term in the standard reference hypernetted chain theory derived by Lado. In this way, we also clarify the meaning of the recently proposed criterion for determining the optimum hard-sphere diameter in the reference system. This leads to a theory in which the sole input is the reference system bridge functional both for the homogeneous system and the inhomogeneous one. The accuracy of this method is illustrated with the standard case of the Lennard-Jones fluid and with a Yukawa fluid with very short range attraction.

Fluid-fluid binodals of binary hard-sphere mixtures are computed from the recently proposed fundamental measure functional-mean spherical approximation closure of the two-component Ornstein-Zernike equation. The results, especially in the dense fluid region that was not accessible by previous theoretical methods, are compared with the corresponding ones for the one-component fluid of big spheres with effective potential obtained from the same closure. The general trends are those expected for hard-sphere potentials but small difference are detectable. The overall agreement found validates the equivalence of the two descriptions for size ratios R = 8.5 or greater.

A simple modification of the reference hypernetted chain (RHNC) closure of the multicomponent Ornstein-Zernike equations with bridge functions taken from Rosenfeld's hard-sphere bridge functional is proposed. Its main effect is to remedy the major limitation of the RHNC closure in the case of highly asymmetric mixtures--the wide domain of packing fractions in which it has no solution. The modified closure is also much faster, while being of similar complexity. This is achieved with a limited loss of accuracy, mainly for the contact value of the big sphere correlation functions. Comparison with simulation shows that inside the RHNC no-solution domain, it provides a good description of the structure, while being clearly superior to all the other closures used so far to study highly asymmetric mixtures. The generic nature of this closure and its good accuracy combined with a reduced no-solution domain open up the possibility to study the phase diagram of complex fluids beyond the hard-sphere model.