Synthesizing compounds predicted to have exceptional properties that are close to or below the ground-state convex hull has proven to be very challenging because avoiding the formation of more thermodynamically or kinetically stable mixtures of known compounds is often required. A homogeneous amorphous phase has been suggested as a very general reaction intermediate. However, the preparation of amorphous phases with controlled composition is also very challenging. We postulated that amorphous intermediates with controlled composition can be made by avoiding the formation of regions with compositions close to those of the known compounds. Specifically, we demonstrated that we could avoid the formation of PbSe and MoSe 2 by sequentially depositing nonstoichiometric ultrathin submonolayer thickness layers on a nominally room-temperature substrate. The substrate temperature needs to be low enough to minimize surface diffusion, preventing the agglomeration of elements and resultant concentration gradients. The amount of diffusion required to form nucleation embryos can be controlled by changing the difference between the composition of the film and the stoichiometry of the compound in question. Large enough differences should result in amorphous intermediates in most systems. The presence of more than two elements will further suppress the nucleation of binary compounds, making this approach particularly useful to prepare amorphous precursors for the synthesis of metastable ternary and quaternary compounds.U nderstanding how to control reactions between elements is critical to advancing technology and synthesizing new predicted compounds. For example, the development of metal silicides as contacts in silicon integrated circuits 1 was accelerated by the development of empirical rules to predict what compound forms first at the reacting interface between metal and silicon. Walser and Bene's first phase rule provided insight into the species formed at reacting interfaces. 2 It suggested that the deepest eutectic was the most stable amorphous composition which likely formed at the reacting interface from the initial interdiffusion of the elements. 2 The congruently melting compound with the highest melting point adjacent to the eutectic was suggested to be the compound that formed first because it was the compound closest in composition to the amorphous phase, with the largest driving force to nucleate. In the synthesis of predicted new compounds, a key challenge is avoiding the formation of thermodynamically stable compounds at the interface between the reacting elements. For example, the compound FeSb 3 was predicted to have interesting thermoelectric properties but could not be made by direct reaction of the elements as it is thermodynamically unstable relative to a mixture of FeSb 2 and Sb. The synthesis of FeSb 3 was only possible through using an Sb-rich amorphous precursor as a reaction intermediate to favor nucleation and growth of FeSb 3 . 3 More recently, Persson