The effects of incident ion/metal flux ratio J(i)/J(Me) and ion energy E(i) on the microstructure, texture, and phase composition of polycrystalline metastable Ti0.5Al0.5N films produced by reactive magnetron sputtering have been investigated using x-ray diffraction (XRD), plan-view and cross-sectional transmission electron microscopy, and Rutherford backscattering spectroscopy. The films, typically congruent-to 1 mum thick, were deposited at a pressure of 20 mTorr (2.67 Pa) in pure N2 on thermally oxidized Si(001) substrates at 250+/-25-degrees-C. The N2+ ion flux to the substrate was controlled by means of a variable axial magnetic field superimposed on the permanent magnetic field of the magnetron. Films deposited at E(i)=20 eV (congruent-to 10 eV per incident accelerated N) with J(i)/J(Me)=1 exhibited a complete (111) texture with a porous columnar microstructure and an average column size of congruent-to 30 nm. Increasing E(i) from 20 to 85 eV, while maintaining J(i)/J(Me) constant at 1, resulted in a small change in texture as the XRD intensity ratio I002/(I111 + I002) increased from congruent-to 0 to 0.14, a decrease in average column size to 25 nm, and a reduction in intracolumn porosity. The stoichiometric ratio N/(Ti+Al) increased from 1 at E(i)=20 eV with J(i)/J(Me)=1 to 1.23 at E(i)=85 eV indicating trapping of excess N while the lattice constant a0 increased from 0.4157 to 0.4188 nm due to compressive stress. E(i) values greater-than-or-equal-to 100 eV led to alloy phase separation. In contrast, maintaining E(i) at congruent-to 20 eV and increasing J(i)/J(Me), from 1 to greater-than-or-equal-to 5.2 resulted in a change from a porous (111) texture to a dense completely (002)-oriented microstructure with an increase in the average column size to 35 nm. N/(Ti+Al) and a0 remained essentially constant and the alloy remained single phase. Mechanistic pathways leading to microstructure and texture changes through variations in E(i) at constant J(i)/J(Me) and in J(i)/J(Me) at constant E(i) were found to be quite different. The average energy deposited per metal atom, [E(d)]=E(i)(J(i)/J(Me)), is therefore not a universal parameter, as has been previously proposed, for describing film growth.