Postsynthesis Transformation of Insulating Cs4PbBr6 Nanocrystals into Bright Perovskite CsPbBr3 through Physical and Chemical Extraction of CsBr.

Perovskite-related Cs4PbBr6 nanocrystals present a "zero-dimensional" crystalline structure where adjacent [PbBr6]4- octahedra do not share any corners. We show in this work that these nanocrystals can be converted into "three-dimensional" CsPbBr3 perovskites by extraction of CsBr. This conversion drastically changes the optoelectronic properties of the nanocrystals that become highly photoluminescent. The extraction of CsBr can be achieved either by thermal annealing (physical approach) or by chemical reaction with Prussian Blue (chemical approach). The former approach can be simply carried out on a dried film without addition of any chemicals but does not yield a full transformation. Instead, reaction with Prussian Blue in solution achieves a full transformation into the perovskite phase. This transformation was also verified on the iodide counterpart (Cs4PbI6).

Halide perovskites have emerged as very interesting materials for a wide range of optoelectronic devices ranging from photovoltaics to light-emission applications.[1−3] Fully inorganic cesium lead halide colloidal nanocrystals (NCs) have attracted considerable attention in this context owing, among others, to their remarkable chemical and structural versatility. Several syntheses have been reported yielding different chemical compositions, sizes, and shapes.[4−9] Although for NCs the main focus has been on the perovskite CsPbX3 (X being mostly Br but also I and Cl) structure, recent works have demonstrated the synthesis of CsPb2X5[10−12] and Cs4PbX6 NCs.[13−20] These three structures can be named “3D” (CsPbBr3), “2D” (CsPb2Br5), and “0D” (Cs4PbBr6), based on the way the [PbBr6]4– octahedra are networked (see Scheme S1). Clear proof of the aforementioned versatility in these ternary compounds is that they can be interconverted by postsynthesis physical and chemical treatments. It has been shown, for instance, that thermal annealing of a dried NC film triggers a 3D to 2D phase transformation (3D → 2D), presumably through temperature-induced extraction of CsBr.[21] Furthermore, postsynthesis 0D → 3D conversion as well as the 3D → 2D one has been demonstrated through insertion of PbBr2,[13,22] while amine- and thiol-mediated extraction of PbBr2 from 3D NCs has been reported to yield 0D NCs (3D → 0D).[23,24] During the review process of this Letter, it was demonstrated that the 0D to 3D conversion can be carried out by extraction of CsX in an immiscible polar solvent (water).[25] In the present work, we study the postsynthesis conversion of insulating 0D NCs into bright photoluminescent 3D perovskite NCs by extraction of CsBr either through thermal annealing or by reaction with Prussian Blue (PB). Also, the 3D CsPbBr3 NCs remain stable in PB, a finding that can be used to enhance the stability of these materials in devices.

0D Cs4PbBr6 NCs were synthesized as previously reported.[13] The structural evolution of these NCs upon annealing in vacuum was monitored by in situ X-ray diffraction (XRD), as shown in Figure a. The diffractogram of the pristine NCs at room temperature (Figure a, bottom) matches that of bulk Cs4PbBr6 (ICSD 162158). As the sample is annealed above 150 °C, the new CsPbBr3 phase appears, with additional peaks matching the reference ICSD 97851 pattern for bulk CsPbBr3. Both phases coexist from 150 to 350 °C (in Figure a, the diffractograms where both phases are present are shown in green). Then, upon further annealing at 400 °C, the peaks corresponding to the 3D perovskite vanish and only the original Cs4PbBr6 phase remains. In order to gain more insight into these transformations, we also performed thermogravimetric (TG) analysis coupled with mass spectrometry (MS), as presented in Figure . From these data, we can infer that, from RT to 150 °C, a slight weight loss is due to the evaporation of residual solvent (toluene), as evidenced by MS. Around 150–250 °C, as we have seen by XRD, CsPbBr3 is formed. This transformation implies the extraction of Cs and Br from the original Cs4PbBr6. Indeed, it has recently been shown that CsBr can easily depart from Cs4PbBr6 NCs to form CsPbBr3 NCs.[25] In our case, we did not observe the appearance of CsBr XRD peaks upon annealing. Nonetheless, it is unlikely that it vaporizes at this low temperature (the melting point of CsBr is over 600 °C). Indeed, TG shows only a very low weight loss, and no signal for Cs or Br is seen by MS (Figure ). Furthermore, at these temperatures, the long-chain molecules used for the synthesis of the NCs (oleylamine, octadecene, and oleic acid) are stable (boiling points are above 300 °C). Therefore, the most likely explanation is that the extracted Cs+ and Br– ions are complexed by residual organics, or form very small clusters hardly detectable by XRD. Upon further annealing, a small step in TG translating a weight loss can be seen at around 300–400 °C. This weight loss can be attributed to desorption/decomposition of the aforementioned ligands. It is known that these species are difficult to identify unambiguously by MS as they undergo degradation and strong interaction with the stationary phase in the column.[26] Therefore, the signal attributed to “ligands” in Figure is derived from the total ion current (TIC) measured in MS. This desorption of ligands is concomitant with the reverse transformation of CsPbBr3 to Cs4PbBr6, as seen by XRD. Two different reaction pathways can be contemplated for this transformation: (i) the reinsertion of CsBr into CsPbBr3 to form Cs4PbBr6 or (ii) further extraction of PbBr2 from CsPbBr3 following the reaction 4CsPbBr3 → Cs4PbBr6 + 3PbBr2. It must be noted that the melting point of PbBr2 is around 357 °C, yet no signal for the melting of PbBr2 is seen by differential thermal analysis (DTA) around these temperatures (see Figure S1). Thus, we can rationalize the XRD (Figure a) and TG-DTA-MS (Figures and S1) data as follows: thermal annealing above 150 °C leads to the complete desorption of residual solvent (toluene) and partial extraction of Cs+ and Br–, which are complexed by residual long-chain ligands (stable up to ca. 300 °C), and a fraction of the Cs4PbBr6 NCs is transformed to CsPbBr3. Upon further annealing at around 250–350 °C, ligands are desorbed from the film, freeing the previously extracted Cs+ and Br– ions, which are inserted back into CsPbBr3 to form Cs4PbBr6, which is the only visible phase at 400 °C. This mechanism explains the evidence that the weight loss in the RT to 450 °C range is very moderate (6.7% as determined from TG) and no inorganic material is detected by MS. Melting of Cs4PbBr6 is detected by DTA at 507 °C (Figure S1),[27] and significant weight losses above 500 °C can be attributed to the vaporization of PbBr2 followed by vaporization of CsBr above 600 °C.[28]

Figure 1

(a) XRD diffractograms of NCs upon annealing in vacuum and reference patterns of bulk CsPbBr3 (ICSD 97851) and Cs4PbBr6 (ICSD 162158). (b) Photoluminescence spectra and photographs under a UV lamp (inset) of the reference sample and sample annealed at 200 °C. SEM images of samples at RT (c), 200 °C (d), and 400 °C (e). Scale bars are 100 nm.

Figure 2

TG (black) curve and evolved gas analysis of Cs4PbBr6 NCs; the signals are related to toluene (red, m/z = 91) and organic ligands (blue, TIC). Note: the blue intensity is five times magnified.

These chemical transformations upon annealing have a direct consequence on the photoluminescence (PL) of the film. Indeed, when thermal annealing is stopped before 400 °C (e.g., at 200 °C) and the sample is cooled down to room-temperature, the mixed-composition sample (0D/3D) exhibits bright PL at 520 nm (and a respective drop of transmittance at 500 nm; see Figure S2), typical of the perovskite phase and clearly seen by the eye under a UV lamp (see Figure b with inset). Indeed, several reports have shown that even a small fraction of CsPbBr3 in a CsPbBr3/Cs4PbBr6 mixture yields bright PL.[13] The bright PL of the mixed film annealed at 200 °C is especially striking because annealing a film of pristine CsPbBr3 NCs up to a similar temperature is known to almost fully quench its PL.[29−31] Eventually, it must be noted that the evolution of samples annealed ex situ in nitrogen under ambient pressure (see Figure S3) is similar to that observed in situ under vacuum. Thermal annealing also leads to NC sintering, as can be seen by the narrowing of the XRD peaks (Figures a and S4) as well as by direct observation under scanning electron microscopy (Figure c–e).

As the above transformation of Cs4PbBr6 NCs into CsPbBr3 is only partial and causes sintering and ripening of the particles, we have developed an alternative path to this transformation, involving chemical extraction of CsBr, which achieves full conversion and prevents particle sintering. The reaction employs FeIII4[FeII(CN)6]3, also known as Prussian blue (PB), a chemical known to intercalate Cs+ ions interstitially.[32] PB powder was added to a colloidal dispersion of Cs4PbBr6 NCs in toluene and left to react for 1 day at room temperature. PB is immiscible in toluene, but extraction of CsBr can be achieved through interfacial reactions, as recently shown.[25] After 24 h, the NCs were isolated by centrifugation and filtration. Figure b reports the evolution of the optical properties from a sharp absorption at 314 nm and no PL, as is characteristic of the 0D phase, to an absorption onset at around 500 nm with the corresponding Stokes-shifted PL at 515 nm of the resulting 3D phase. This transformation is further verified by TEM (Figure c,d), which evidences a shape evolution and corresponding XRD peaks (Figure a) matching the bulk patterns of Cs4PbBr6 (before) and CsPbBr3 (after 24 h). These data highlight the successful conversion of the initial 0D NCs into 3D ones. The shrinkage in size of the NCs following this treatment supports the hypothesis that CsBr was extracted from the NCs rather than the overall sample undergoing a dissolution-recrystallization. On the other hand, the CsPbBr3 NCs are smaller than expected and slightly more polydisperse than the starting Cs4PbBr6 NCs (average size goes from 9.8 ± 1.1 to 6.3 ± 1.4 nm; see Figure S5), suggesting that the transformation is accompanied by etching of the sample. This is not surprising as there is considerable reorganization of the lattice when going from Cs4PbBr6 to CsPbBr3. Evidently, the extraction of Cs+ ions is accompanied by that of Br– ions to maintain charge neutrality both in the NCs and in PB. It is worth noting that the resulting 3D NCs did not react further with PB, even after several days in solution (see Figure S6), meaning that PB is unable to extract additional CsBr from the 3D perovskite to yield the CsPb2Br5 phase.

Figure 3

XRD (a), optical (b) and TEM (c,d) characterization of 0D NCs transformed to 3D by PB. Black and green bars in (a) are reference patterns for 0D and 3D phases. Scale bars in (c,d) are 50 nm.

The same conversion could be achieved on iodide NCs, transforming Cs4PbI6 NCs into red-emitting CsPbI3 NCs (see Figure S7). This is seen both by the change in morphology evidenced by TEM (Figure S7a,b) and by the evolution of absorption and PL spectra (Figure S7c). Indeed, the initial sample shows a sharp absorption peak at 368 nm and no PL, while the resulting NCs after reaction with PB for 20 min at 80 °C show an absorption onset at around 647 nm with a Stokes-shifted PL peak at 667 nm. These features are typical of Cs4PbI6 (before reaction) and CsPbI3 (after reaction).[4,13]

In summary, we have shown that Cs4PbBr6 NCs can be transformed either by thermal annealing or by reaction with PB into bright-emitting perovskite CsPbBr3 NCs. These postsynthesis reactions represent a new tool to tailor the stoichiometry and optical properties of cesium lead halide NCs. Especially the use of PB as an additive in 3D perovskite films may help to stabilize the 3D phase by preventing its transformation to other phases.