Corrigendum to:“Phase transitions in the hexagonal tungsten bronze RbNbW₂O₉” [J. Solid State Chem. 286 (2020) 121275]

The authors regret that in our recent paper describing the phase transitions of the hexagonal tungsten bronze RbNbW2O9 based on analysis of variable temperature powder neutron diffraction data [1], we (erroneously) concluded that there is no direct continuous transition between the Cmc21 and P6mm structures and therefore a transition through the intermediate P63mc was proposed. While there is no reason to doubt that both the lower (Cmc21) and higher temperature (P6mm) structures are correct, the assignment of the intermediate P63mc phase at 673 K is unnecessary. This conclusion is based on two considerations: 1. On further inspection using the ISOTROPY software suite [2] we can now confirm that a direct continuous transition from P6mm to Cmc21 is allowed by introduction of the A6+ mode (in fact this tilt mode is labelled A6 in this transition from the non-centrosymmetric parent, P6mm, but we use A6+ for simplicity in comparing to the notation used for the equivalent tilt mode derived from the aristotype phase, P6/mmm, Table 1 – we thank our co-author CJH for bringing this to our attention). This pathway was not included in Fig. S2 of our original work, but a simplified version highlighting this additional transition and the transitions related to this work is now shown in Fig. 1.2. In re-considering our original refinement of the data at 673 K in the Cmc21 model (Table 1 of the original paper) we identified that this refinement had unfortunately resulted in a false minimum, whereby the sign of the A3+ mode (A3 relative to the P6mm parent) had effectively been reversed. This has now been corrected and the revised Cmc21 model results in a better fit to the data than P63mc at 673 K (Table 2, below).Consequently, the lattice parameters, bond lengths and distortion mode amplitudes (Figs. 4, 5 and 7 in the original article) were redetermined for all temperatures up to 673 K and updated Figs. 2–4 and Table 3 reflect these minor changes. The general trends observed in each parameter are consistent with those previously reported and support the assignment of Cmc21 rather than P63mc at 673 K, and then a continuous transition to P6mm. Extrapolation of the A6+ and Γ5⁻ modes using a weighted power law (Fig. 4b), similar to that carried out originally, indicates the transition from Cmc21 to P6mm corresponding to loss of octahedral tilting occurs at ~680 and ~690 K, respectively. The critical exponents are β ​= ​0.354 and 0.476 with fit constants of A ​= ​0.091 and 0.017, respectively. Considering this, the key point of this corrigendum is that the transition sequence can now be summarised (with no requirement for the intermediate P63mc phase) as: [Formula presented] The authors would like to apologise for any inconvenience caused.