The mysterious 21 μm emission feature seen in sixteen C‐rich proto‐planetary nebulae (PPNe) remains unidentified since its discovery in 1989. Over a dozen of materials are suggested as the carrier candidates. In this work, we quantitatively investigate eight inorganic and one organic carrier candidates in terms of elemental abundance constraints, while previous studies mostly focus on their spectral profiles (which could be largely affected by grain size, shape and clustering effects). It is found that: (1) five candidates (TiC nanoclusters, fullerenes coordinated with Ti atoms, SiS2, doped‐SiC and SiO2‐coated SiC dust) violate the abundance constraints (i.e. they require too much Ti, S or Si to account for the emission power of the 21 μm band, (2) three candidates (carbon and silicon mixtures, Fe2O3 and Fe3O4), while satisfying the abundance constraints, exhibit secondary features which are not detected in the 21 μm sources and (3) nano FeO, neither exceeding the abundance budget nor producing undetected secondary features, seems to be a viable candidate, supporting the suggestions of Posch, Mutschke & Andersen.