Mixed-halide perovskites are ideal mid- and wide-gap absorbers for multi-junction solar cells, but stable photovoltaic performance is severely compromised by halide segregation. Herein, we utilise a fine-grained compositional approach to investigate differences in the initial formation and photoinduced segregation of CH₃NH₃Pb(I_1-xBr_x)₃.^[1] Our multimodal spectroscopic approach, combining in-situ X-ray diffraction (XRD) and photoluminescence (PL) tracking, further allows the applicability of PL spectroscopy as a measure of halide segregation to be evaluated – a pertinent step given the near-ubiquitous use of this probe technique in the development of photostable mixed-halide absorbers.

X-ray diffractometry across stoichiometries spanning twenty-two bromide fractions demonstrates that central compositions near x = 0.5 form two macrostrained phases (in the dark), which may compensate for the detrimental difference in the iodide and bromide ionic radii and cause the absence of a theoretically predicted miscibility gap.^[2] Via in-situ XRD under illumination, we find that such macrostrained phases exhibit halide segregation at different rates, highlighting how initial strain engineering can modulate the speed of segregation. Alongside material strain, we provide experimental verification that halide ordering also influences segregation: CH₃NH₃PbIBr₂ is found to exhibit exceptional segregation resistance, as theoretically predicted.^[2] Combining structural and optical probes, we further demonstrate that halide segregation does still occur below the widely-quoted threshold of x equal to 0.2.

Strikingly, we find that whilst the structurally-determined segregation rate is broadly constant across the entire compositional space examined, suggesting that the mobility of halide ions dictates the rate of segregation, the rate extracted from in situ photoluminescence measurements rises over four orders of magnitude with increasing bromide fraction. We resolve this disparity by analysis of the energetics and luminescence efficiency of recombining charge carriers, evidencing how such underlying (photo)physics can lead to artificially accelerated segregation rates determined from PL measurements.

Together, our results provide a multitude of new evaluation benchmarks for proposed models of phase segregation, while also highlighting new routes for developing segregation resistant absorber layers via the engineering of macrostrained phases and local atomistic ordering. Furthermore, our multimodal spectroscopic approach demonstrates that photoluminescence, the most widely used spectroscopic tool in such investigations, is not an appropriate probe of halide segregation, and should be supplemented by average-structure-sensitive techniques.