The preference depends on the context. Mole fractions remain constant when a sample is compressed or heated (as long as no chemical reactions occur), and therefore a “good” variables to report whenever pressure or temperature are changed. When phase equilibria are studied, a common technique to determine the phase compositions is gas chromatography. This yields mole fractions. On the other, when phase compositions are determined by means of spectroscopy, they are obtained as concentrations.
Traditionally, chemical studies of solutions at ambient pressure and fixed temperature often report concentrations (mol/l). These, however, are often obtained by mixing a known mass of a component with a given volume of solvent—and neglect the volume change caused by the mixing process. Reporting concentrations in mol/kg is clearer.
In computations of phase equilibria the use of concentrations (mol/l) instead of mole fractions + volumes allows for a more elegant formulation of the conditions of phase equilibrium (→ so-called isochoric thermodynamics).
There is an even “older” reason for using mole fraction instead of molality.
Let us first ask ourselves why we are studying concentration measures for binary mixtures? The primary goal is to design and operate industrial separation apparatus (fractionating columns) based on these binary equilibrium data.
The thermodynamic simulation of rectifying columns (simply called ‘distillation’) is based on the so-called MESH equations (Mass balance, Equilibrium expressions, Summation equations, Heat balance). The individual equations of the MESH equations vary in complexity:
1): Mass balance: simple
2): Equilibrium expressions: simple, but requires e.g. measurement data for the equilibrium
3): Summation equations: very simple
4): Heat balance: complicated because the enthalpy of vaporisation depends, among other things, on the concentration of the liquid mixture on the individual column plates.
McCabe and Thiele:
W. L. McCabe and E. W. Thiele: „GraphicaI Design of Fractionating Columns“ Industrial and engineering chemistry, Vol. 17 (1925), No. 6. p. 605 – 611.
already used four preliminary assumptions for the simplified graphical design in their world-famous article from 1925. The most important simplification for the reduction of complexity of the design is the introduction of the
"constant molal overflow":
“The number of mols of vapor ascending the column, and hence the molal overflow except for the change at the feed plate, is constant from plate to plate. Since heat losses from the column can ordinarily be neglected either because of lagging or large ratio of volume to surface, and since the absolute boiling points of the components are not far apart (otherwise separation is easy in any case). Trouton's rule requires constant molal vaporization from plate to plate.”
In other words: If the enthalpy of vaporisation of the mixture could be regarded as constant, the calculation would be much easier to perform.
a):
If we consider the specific enthalpy of vaporisation r in J/kg, this does NOT remain constant, but changes strongly with the concentration and from substance to substance.
b):
However, if we take the molal enthalpy of vaporisation ∆Hvap = M*r in J/mol as a basis, then this varies only slightly with the concentration for the binary mixture in question and can be regarded as constant with a good approximation, and it does not vary much even from substance to substance (see Trouton’s rule): https://en.wikipedia.org/wiki/Trouton%27s_rule
Summa summarum:
==============
If the molal overflow = constant, then the following applies:
One mole of condensing vapour vaporises one mole of liquid mixture at every point (on every plate) of the rectifying column.
This justifies the use of molal concentrations, i.e. we relate all the physical quantities and properties to a mole as a basis instead of mass or volume.
So, we use: J/mol, mol/mol instead of J/kg, kg/kg, mol/kg or mol/litre.