Preparation of Single-Phase Films of CH 3 NH 3 Pb(I 1 − x Br x ) 3 with Sharp Optical Band Edges

: Organometallic lead-halide perovskite-based solar cells now approach 18% e ﬃ ciency. Introducing a mixture of bromide and iodide in the halide composition allows tuning of the optical bandgap. We prepare mixed bromide − iodide lead perovskite ﬁ lms CH 3 NH 3 Pb(I 1 − x Br x ) 3 (0 ≤ x ≤ 1) by spin-coating from solution and obtain ﬁ lms with monotonically varying bandgaps across the full composition range. Photothermal de ﬂ ection spectroscopy, photoluminescence, and X-ray di ﬀ raction show that following suitable fabrication protocols these mixed lead-halide perovskite ﬁ lms form a single phase. The optical absorption edge of the pure triiodide and tribromide perovskites is sharp with Urbach energies of 15 and 23 meV, respectively, and reaches a maximum of 90 meV for CH 3 NH 3 PbI 1.2 Br 1.8 . We demonstrate a bromide − iodide lead perovskite ﬁ lm (CH 3 NH 3 PbI 1.2 Br 1.8 ) with an optical bandgap of 1.94 eV, which is optimal for tandem cells of these materials with crystalline silicon devices.

S olution processable optoelectronic devices offer costeffective, large-scale manufacturing capability. Hybrid organic−inorganic perovskite (CH 3 NH 3 PbI 3−x Cl x ) solar cells now show photovoltaic (PV) performance 1−4 approaching 18%, 5,6 and high charge-carrier mobilities. 7 Perovskite films have also shown promising photoluminescence quantum efficiencies (PLQEs) of more than 70% and lasing, making them a promising candidate for cost-effective and efficient LEDs and lasing applications. 8,9 Despite the rapid progress in the past few years, relatively little is known about the properties required for efficient perovskite PV systems or why perovskite works so well in PVs while simultaneously demonstrating high PLQE. It has recently been demonstrated by Noh et al. that the bandgap of the perovskite can be tuned by varying the halide (i.e., bromide, iodide) composition of the perovskite precursor solution. 10,11 High open-circuit voltages have been demonstrated by using the higher bandgap tribromide-based perovskite. 12−14 The control of bandgap with mixed halide systems is attractive, but it is important to establish whether these form solid solutions, with monotonic variation of lattice constants and bandgap, or if there is compositional inhomogeneity, which may introduce a range of local bandgaps and reduce PV performance. To address this question, we use photothermal deflection spectroscopy (PDS) and photoluminescence measurements that are able to detect low levels of subgap states in the system CH 3 NH 3 Pb(I 1−x Br x ) 3 (0 ≤ x ≤ 1). We correlate our findings with the structural information obtained by X-ray diffraction (XRD) measurements. We find that films of mixed trihalide perovskites initially form a system with two absorption and luminescence features. Following our fabrication protocol, these convert to a homogeneous intermixed system with only one species within the detection limits of the techniques used in this work. The ability to engineer a single bandgap in these mixed compositions is important for the development of solar cells. We demonstrate a bromide−iodide lead perovskite film (CH 3 NH 3 PbI 1.2 Br 1.8 ) with an optical bandgap of 1.94 eV, which is optimal for tandem cell applications with c-Si cells. Our findings also show that pure trihalide perovskites have a remarkably low energetic disorder, which is, to our knowledge, by far the lowest reported for any low temperature solutionprocessable material. 15,16 PDS is an ultrasensitive absorption measurement technique that detects heating of the sample due to the nonradiative relaxation of absorbed light and is insensitive to reflection and scattering. PDS can be used to characterize energetic disorder as the exponential decay of the absorption below the bandgap with a characteristic energy, the Urbach energy (E u ), and has been used to study the perovskite materials system. 16 The Urbach energy is given by A(E) ∝ e E/(E u ) , where A is the absorbance and E is the excitation energy in electronvolts. In Figure S1 (SI), the linear fit to the Urbach tail in the PDS data for the pure triiodide perovskite is shown, which yields an Urbach energy of ∼15 meV that is similar to that obtained by De Wolf et al. 16 This value is about twice the typical value for single-crystal GaAs and half that of polycrystalline CIGS. 15,17 We prepared perovskite thin films by spin-coating from a precursor solution composed of 1:1 organic to inorganic ratios (CH 3 NH 3 I to PbI 2 and CH 3 NH 3 Br to PbBr 2 ) under a N 2 atmosphere in a glovebox (details in the SI). Solutions with varying iodide to bromide concentration were prepared by mixing the respective pure triiodide (CH 3 NH 3 PbI 3 ) and tribromide (CH 3 NH 3 PbBr 3 ) precursor solution in the respective ratios. Energy-dispersive X-ray spectroscopy (EDX) measurements show that the ratios of bromide to iodide in the film are the same as those in the solution (see Figure S4 and Table 1, SI). By increasing the concentration of the pure triiodide perovskite in the tribromide perovskite in solution from 0 to 100%, we tune the bandgap of the resulting perovskite between the extremes of pure triiodide and tribromide perovskites from 2.23 to 1.57 eV (Figure 1a,c). We detect the absorption edge of the pure triiodide lead perovskite CH 3 NH 3 PbI 3 down to the detection limit of the PDS setup without observing a leveling off, which indicates a very clean bandgap and sharp band edge. This very clean gap is preserved as bromide is added up to 20% bromide (80% iodide) content. For the pure tribromide perovskite, CH 3 NH 3 PbBr 3 , a sharp band edge with additional levels of sub-bandgap absorption is found, which indicates a population of sub-bandgap states. The absorption edge for films with 20% iodide shows a broad absorption tail below the band edge and significant sub-bandgap absorption below ∼1.8 eV. We interpret this as increased disorder in these films and a significant population of sub-bandgap states. Notably, for iodide−bromide mixtures with 40 and 60% iodide content, the films show a clear shoulder in the subgap absorption ( Figure 1b). We interpret this as the initial formation of two species in these films. Photoluminescence (PL) measurements of the bromide−iodide lead perovskite films with 40 and 60% iodide content show two emission peaks (Figure 1b inset). Time-resolved PL measurements (Figure 2a,b) show that the higher energy emission band decays more rapidly than the lower energy band. These findings suggest that the initially formed mixed-halide perovskite films, having 40 and 60% iodide content, comprise two species. We note, in particular, that for the 40% iodide films the second phase emits around 1.68 eV from a significantly lower energy than the 1.94 eV emissive feature, and this represents a potential loss mechanism for PV devices.
Surprisingly, we found that the PDS absorption and PL spectra changed after the films were stored under an inert nitrogen atmosphere at room temperature (these changes were complete within 3 weeks and showed no significant further evolution at 13 weeks) (Figure 1b). A similar effect was observed in fresh films prepared from solutions that had been kept at room temperature for several days (see Figure S2, SI). The most striking changes in the PDS spectra in these aged films, which we term here "single-phase" films, is the sharpening of the band edges and the loss of the extra absorption shoulder in the subgap.
In detail, in the as-formed films with low iodide contents (20 and 40%), made from fresh solutions, we find additional subbandgap absorption and broadened band edges. However, there is no longer a second phase detectable within the resolution of the PDS setup if the films are allowed to age or are made from aged solutions. For all other iodide contents (0, 80, 90, and 100%), the PDS absorption spectra show sharp band edges and little differences compared with the films measured directly after preparation. The steady-state PL spectra of the "single phase" films ( Figure 1d) show single transitions with a monotonic shift in peak position with increasing iodide content from pure tribromide perovskite (2.23 eV) to pure triiodide perovskite (1.57 eV). This finding supports our observation of a clean bandgap in these films. The photoluminescence kinetics (Figure 3d) show single bimolecular decays for the pure iodide and bromide films and also for iodide-rich films down to 60% iodide, as reported previously for the iodide material. 9 Films with 20 and 40% iodide content both show fast kinetics. These findings also point toward one photoactive species from which emission arises in the bromide−iodide lead perovskite films.
Time-resolved PL spectra of "single-phase" films also show one emissive species (Figure 2c,d). Figure 3c shows the full width at half-maximum (fwhm) of the PL peaks together with the Urbach energy calculated from PDS measurements. (For details about the fits used for calculating the Urbach energies, see Figure S7 in the SI.) Both parameters follow similar trends, peaking below 50% iodide content. We take this observation as an indication that absorption and emission arise from similar states. We find that the pure tribromide perovskite films and films with 100 to 80% of iodide content show the lowest fwhm and the lowest Urbach energies. This indicates very low disorder and sharp band edges in the films with higher iodide content compared with the ones with lower iodide content. Also, the larger fwhm and Urbach energies for films with 20 and 40% than for those with higher iodide content correlate with shorter lifetimes observed in transient PL (Figure 3d). However, when we compare the PL peak positions with the PDS absorption data, we find that the emission peaks for all films lie at the energy where the absorption has dropped to ∼15% of its maximum band edge value. (See Figure S5 in the SI.) First, this indicates that impurity levels at lower energy do not play a significant role in the radiative recombination of excited states. Therefore, they do not represent a major radiative form of energetic loss channel due to energy relaxation down to the lower lying states within the subgap. Moreover, the lower subgap absorption detected for "single-phase" films suggests a lower density of states that recombine through nonradiative relaxation, which again indicates a cleaner material system. Powder X-ray diffraction measurements were carried out on "single phase" films over the entire composition range ( Figure  3a). A systematic shift in the patterns to lower scattering angles is observed as the concentration of iodide increases, indicating an expansion of the unit cell with increasing iodide fraction. Because of preferential alignment in the films, a complete structural analysis was not possible. The symmetry and lattice parameters were modeled using a Le Bail analysis 18 in the Fullprof suite of programmes (see Figure S6, SI). 19 The characteristic perovskite peak at 2θ = 30° (Figure 3a, inset) shows a significant broadening of the (200) c reflection in all films containing iodide. We ascribe this to a tetragonal distortion and the emergence of two unique reflections, (002) t and (110) t ; a similar although less pronounced splitting is also observed in the (100) c reflection with iodide present (Figure 3a, inset). Therefore, with the exception of the pure bromide film, all films were modeled using a tetragonal unit cell with I4cm symmetry, as has previously been reported for CH 3 NH 3 PbI 3 . 20 The pure bromide film was modeled using a higher symmetry Pm3̅ m cubic structure. 20 In all films, unreacted starting materials of CH 3 NH 3 X and PbX 2 were detected in addition to the main perovskite phase. The unit cell of the tetragonal phase, a t and c t , can be related to that of the cubic cell, a c , by a t = √2a c and c t = 2a c . When the cell parameters of the tetragonal cell are normalized, a t /√2, c t /2, to allow for direct comparison to the cubic perovskite structure, a linear increase in the size of the normalized unit cell with increasing iodide fraction is observed (Figure 3b). This is consistent with both the larger ionic radii of iodide (r = 2.2 Å) with respect to bromide (r = 1.96 Å) and Vegard's law for a continuous solid solution. The degree of tetragonal distortion, ((c t − √2a t )/ ((a t /√2) + (c t /2))), also increases with iodide concentration reaching a maximum value of 0.9% in CH 3 NH 3 PbI 3 . An additional broadening of reflections in the mixed halide films is observed (Figure 3a, inset), which cannot be described as resulting only from changes in the symmetry. We attribute this to strain broadening. Strain of this type could be due to quenching of a higher temperature (cubic) symmetry phase or could arise from an inhomogeneous halide distribution beyond random site occupancy. Since submission, we became aware of a report by Park et al., 21 where changes in the crystallinity, microstructure and preferential alignment are observed in films of CH 3 NH 3 PbI 3−x Cl x and CH 3 NH 3 PbI 3 aged over five weeks.
In Figure 3b, we compare the extracted normalized lattice parameters with the optically determined PL peak positions. Both lattice parameters and PL peak energies follow a linear trend with iodide fraction. These observations support our finding that there is a single solid solution across the full composition range that can be obtained using a straightforward fabrication protocol. This protocol requires "aging" of either the starting solution or the as-prepared films. The underlying reason for the observed effect is currently under investigation.
Our demonstration of remarkably clean bromide/iodide solid solutions in this organometallic halide PV system brings this