Fluorescent Alloy CsPb x Mn1-x I3 Perovskite Nanocrystals with High Structural and Optical Stability.

CsPbI3 nanocrystals are still limited in their use because of their phase instability as they degrade into the yellow nonemitting δ-CsPbI3 phase within a few days. We show that alloyed CsPb x Mn1-x I3 nanocrystals have essentially the same optical features and crystal structure as the parent α-CsPbI3 system, but they are stable in films and in solution for periods over a month. The stabilization stems from a small decrease in the lattice parameters slightly increasing the Goldsmith tolerance factor, combined with an increase in the cohesive energy. Finally, hybrid density functional calculations confirm that the Mn2+ levels fall within the conduction band, thus not strongly altering the optical properties.

Nanocrystals (NCs) of cesium lead halide perovskites (CsPbX3 with X = Cl, Br, I) have recently become a topic of extensive research.[1−5] Among them, CsPbBr3 NCs are the most studied and understood, which have photoluminescence quantum yields (PLQYs) close to 100%[6] and a wide range of applications.[2−4,7] CsPbI3, NCs which have a band gap of around 1.8 eV and a larger exitonic Bohr radius (6 nm for CsPbI3 compared to 3.5 nm for CsPbBr3) are, in principle, more appealing than CsPbBr3 NCs for tunable LEDs, lasers, and solar cells, but they are significantly less stable.[8,9] In the cubic CsPbI3 phase (α-CsPbI3), the Cs+ ions are not large enough to stabilize the cubic framework of corner-sharing PbI64– octahedra, and a transition to the orthorhombic phase (δ-CsPbI3) is seen below 315 °C.[9] The band gap of δ-CsPbI3 is larger than that of α-CsPbI3 (2.25 vs 1.73 eV) and is indirect, which makes it less suitable for many optoelectronic applications. Various methods have been reported on how to stabilize the α phase both in NCs and in thin films, for instance, by partially replacing I– with Br– or Cl–, by incorporating large organic cations, or by surface treatments with different organic molecules.[8,10−13]

Although Mn2+ has mainly been studied for its Stokes shifted emission in CsPbCl3 NCs, leading to strongly red emissive CsPbCl3:Mn2+ NCs,[14−18] recent work has shown that Mn2+ alloying in CsPbX3 NCs can help to stabilize their cubic phase.[19] While previous works have mainly focused on the stabilizing effect in CsPbBr3 NCs, we demonstrate that the highly unstable α-CsPbI3 phase in NCs can be stabilized by alloying it with Mn2+ up to 10% (see Figure a), without any major changes in optical properties. This is confirmed by hybrid density functional theory (DFT) calculations, which show that the Mn2+ levels fall within the conduction band of α-CsPbI3; thus, they do not greatly alter the NCs band structure. The alloyed CsPbMn1–I3 NCs remained stable and brightly luminescent for up to over a month, both in solution and as films stored in air. Similar to the replacement of I– ions with Br–, the stabilization through the replacement of Pb2+ with Mn2+ stems from a small decrease in the lattice parameters, combined with an increase in the cohesive energy.

Figure 1

Overview of optical and structural properties of CsPbMn1–I3 NCs. (b) TEM and HRTEM image of CsPbMn1–I3 NCs. (c) Absorption and photoluminescence compared with pure CsPbI3 NCs. (d) Lifetimes and PLQYs of CsPbI3 and CsPbMn1–I3 NCs.

CsPbMn1–I3 NCs with sizes of 12.1 ± 1.7 nm (Figure b) were synthesized similarly to previous works,[1] with some minor changes (we used MnI2 to introduce Mn2+, see the Supporting Information). Although the Mn to Pb ratio was initially determined to be 0.44:0.56 (based on inductively coupled plasma optical emission spectroscopy), energy-dispersive X-ray spectroscopy (EDX) indicated a lower ratio of 0.09:0.91, matching with previous reports stating that excess Mn2+ precursor complexes are difficult to remove from NC solutions.[16] These relatively low levels of incorporation of Mn2+ compared to the high feed ratios match with those previously reported for CsPbCl3 and presumably stem from the fast nucleation of CsPbX3,[20] thus reducing the incorporation of Mn2+. Nonetheless, this high ratio of Mn:Pb indicates the formation of an alloy rather than Mn doping, as doping in NCs normally refers only to a few dopants per NCs, with doping percentages up to about 1%.[21] It must be noted that higher feed ratios of Mn lead to the precipitation of PbI2, as has been reported by others, thus precluding the preparation of alloy NCs with higher Mn:Pb ratios with the current synthesis method. Neither the absorption nor the photoluminescence (PL) of the CsPbMn1–I3 NCs differed much from those of the pure CsPbI3 NCs (Figure c), in analogy with previous reports by others on CsPbBr3.[14,15,19] Here, both the CsPbI3 and the CsPbMn1–I3 NCs were emitting at 680 nm with a PL full width at half-maximum (FWHM) of about 40 nm, similar to previous reports.[1] Furthermore, as shown in Figure d, both the PL quantum yield and the PL lifetime in the CsPbI3 sample and in the alloy sample were similar (74% for CsPbI3 and 82% for CsPbMn1–I3) and matched with those previously reported for CsPbI3 NCs.[22]

According to the X-ray diffraction (XRD) patterns of Figure a,b, the CsPbMn1–I3 NCs have a unit cell that is only slightly smaller than the that of pure CsPbI3 NCs, with no other structural differences. This is in accordance with previous reports on CsPbMn1–Cl3, where relatively high incorporations exhibit only a small change on the lattice parameters.[23] The long-term stability of CsPbMn1–I3 NCs is reported in Figure c,d: drop-cast films of CsPbI3 and CsPbMn1–I3 NCs were stored under ambient conditions (in a drawer, at about 25–30 degrees), and their XRD patterns were measured over the course of a month. The CsPbI3 film started to degrade to δ-CsPbI3 within a day, with a complete transition within 5 days. However, the CsPbMn1–I3 film did not show any visible degradation during the first 2 weeks, and it was still mainly in the α-phase after one month. Similarly, a solution of CsPbI3 NCs in toluene, stored in a nitrogen glovebox, completely degraded over 1 month, as it turned white-yellow and did not show any PL (Figure SI2). A solution of CsPbMn1–I3 NCs, stored under the same conditions, was still red and strongly fluorescent. Notably interesting, CsPbMn1–I3 NCs also were highly luminescent and stable without separation from the reaction solvent, as unwashed CsPbMn1–I3 NCs after 3 days still were brightly luminescent with a PLQY of 60 ± 5%.

Figure 2

(a, b) XRD patterns of CsPbI3 and CsPbMn1–I3 NCs. (c, d) Stability of CsPbI3 and CsPbMn1–I3 NCs. XRD reference patterns correspond to 98-018-1288 (α-CsPbI3) and 98-016-1480 (δ-CsPbI3) respectively.

To corroborate the stabilizing effect of Mn2+ and the unaltered optical properties of CsPbMn1–I3, we performed DFT calculations. Starting from the α-cubic phase of CsPbI3, a 2 × 2 × 2 supercell was built by fixing the cell parameter to the experimental value, i.e., 6.289 Å.[24] For CsPbMn1–I3, two configurations with a Pb:Mn ratio of 7:1 and 1:1 were obtained by replacing one Pb or four Pb atoms, respectively, with Mn atoms (see the Supporting Information and Figure ). Such configurations correspond to Mn-alloying of 12% and 50%, respectively. Two geometry optimizations were performed: (i) the relaxation of the ion positions by fixing the cell parameter to the experimental value and (ii) the relaxation of both the ion positions and the cubic cell parameter. All relaxations were performed by using the Perdew–Burke–Ernzerhof (PBE) exchange correlation functional.[25] The equilibrium lattice parameters, volume, cohesive energies, and band gaps of the three systems are shown in Table S1 for both ions and total relaxations. Upon variable cell relaxations, the cell parameter of the CsPbI3 supercell increased from 12.58 Å (twice the cubic lattice size) to 12.79 Å, while that of the alloyed CsPbMn1–I3 supercells decreased to 12.67 and 12.22 Å for the system with one and four Mn replacing Pb atoms, respectively. Hence, the calculations predict a 1% lattice contraction going from CsPbI3 to CsPb0.88Mn0.12I3, which corresponds with the observed lattice contraction with XRD (1% contraction for CsPb0.91Mn0.0.9I3). The cell contraction in the Pb/Mn alloy systems is due to the shortening of the metal–I bonds (from 3.14 Å for Pb–I to 2.97 Å for Mn–I). Most importantly, the ion relaxation preserves the octahedral coordination of the Mn ions in the Pb sites. The calculated cohesive energy shows a progressive stabilization of the CsPbI3 perovskite upon Mn-alloying toward a value of 0.13 eV/formula unit (f.u.) (0.09 eV/f.u.) in CsPb0.5Mn0.5I3 for the relaxed cell (fixed cell) optimizations. It should be noted that, besides the stabilization reported by the analysis of the cohesive energies, the configurational entropy further stabilizes the Pb/Mn alloy phase, while compensation effects by native defects on the other sub lattices are expected to play only a marginal role due to the isovalent substitutional doping.

Figure 3

Models of relaxed α-CsPbI3 (top panel) and CsPb0.88Mn0.12I3, and respective DFT projected densities of states calculated at the HSE06 (α = 0.43) level by including spin-orbit (bottom panel).

In order to investigate the effects of the partial replacement of Pb with Mn on the electronic structure, the projected density of states (PDOS) on the atomic orbitals of the pristine CsPbI3 and CsPb0.88Mn0.12I3 systems were calculated at the HSE06 (exchange fraction α = 0.43) level[26] by including spin–orbit corrections (SOC). In Figure , the PDOS of α-CsPbI3 and CsPb0.88Mn0.12I3 supercells are compared (see Figure SI3 for calculation at PBE level without SOC, and SI4 for Cs2PbMnI6 with SOC). The calculated band gap of CsPbI3 is 1.01 eV. The underestimation of the experimental band gap (∼1.70 eV) is possibly due to neglect of thermal fluctuations which were shown to raise the band gap of cubic MAPbI3 by ca. 0.5 eV.[27] In α-CsPbI3, the valence and conduction bands are dominated by the p and s orbitals of I and Pb, respectively. The replacement of Pb2+ with Mn2+ introduces localized electronic states at ∼6.5 eV below the Valence Band Maximum (VBM) and between 4 and 6 eV above the VBM. The PDOS projected on the atomic orbitals (see Figure SI 5) shows that such states arise from the d orbitals of Mn, while negligible contributions of the Mn orbitals to the band edges are reported. The limited variation of the electronic structures at the band edges is in good agreement with experiments, which report no changes in the PL of the alloy system compared to the pristine perovskite.

In summary, we show that alloy CsPbMn1–I3 NCs have the same phase and essentially the same optical features as α-CsPbI3 NCs, but the former are much more stable than the latter. These preliminary results require further investigation, but calculations indicate that significantly higher Mn:Pb ratios, almost up to 1:1 like that of Cs2PbMnI6, should be obtainable, with a strongly stabilized cubic perovskite phase, yet with properties similar to those of pure CsPbI3. The shrinking of the lattice parameter upon alloying with smaller divalent metals has also been reported with Sn2+ and Zn2+, but only in CsPbBr3 NCs, and could thus potentially be interesting in alloying with CsPbI3 NCs for increased stability.[28,29] Also, the replacement of Pb2+ with Mn2+ can be seen as a way to partially mitigate the toxicity of these materials. Finally, as the alloying with Mn2+ seems to be limited because of the fast nucleation of CsPbI3, novel synthesis approaches should be explored, like cation exchange, room temperature synthesis, or insertion reactions with MnI2 into Cs4PbI6 NCs.[17,28,30−33]