Quantum dot–induced phase stabilization of α-CsPbI3 perovskite for high-efficiency photovoltaics

See allHide authors and affiliations

Science  07 Oct 2016:
Vol. 354, Issue 6308, pp. 92-95
DOI: 10.1126/science.aag2700

Maintaining a stable phase

For solar cell applications, all-inorganic perovskite phases could be more stable than those containing organic cations. But the band gaps of the former, which determine the electrical conductivity of these materials, are not well matched to the solar spectrum. The cubic structure of CsPbI3 is an exception, but it is stable in bulk only at high temperatures. Swarnkar et al. show that surfactant-coated α-CsPbI3 quantum dots are stable at ambient conditions and have tunable band gaps in the visible range. Thin films of these materials can be made by spin coating with an antisolvent technique to minimize surfactant loss. When used in solar cells, these films have efficiencies exceeding 10%, making them promising for light harvesting or for LEDs.

Science, this issue p. 92


We show nanoscale phase stabilization of CsPbI3 quantum dots (QDs) to low temperatures that can be used as the active component of efficient optoelectronic devices. CsPbI3 is an all-inorganic analog to the hybrid organic cation halide perovskites, but the cubic phase of bulk CsPbI3 (α-CsPbI3)—the variant with desirable band gap—is only stable at high temperatures. We describe the formation of α-CsPbI3 QD films that are phase-stable for months in ambient air. The films exhibit long-range electronic transport and were used to fabricate colloidal perovskite QD photovoltaic cells with an open-circuit voltage of 1.23 volts and efficiency of 10.77%. These devices also function as light-emitting diodes with low turn-on voltage and tunable emission.

Hybrid organic-inorganic halide perovskites, with the common formulation ABX3 (where A is an organic cation, B is commonly Pb2+, and X is a halide), were first applied to photovoltaics (PVs) as methylammonium lead triiodide (CH3NH3PbI3) in 2009 (1). Perovskite PV devices processed from solution inks now convert >22% of incident sunlight into electricity, which is on par with the best thin-film chalcogenide and silicon devices, but durability of the semiconductor presents a major technical hurdle to commercialization. Under environmental stress, CH3NH3PbI3 dissociates into PbI2 and CH3NH3I, the latter of which is volatile (2).

Thus, an all-inorganic structure without a volatile organic component is highly desired. The all-inorganic Pb-halide perovskite with the most appropriate band gap Eg for PV applications is cubic (α) CsPbI3 (Eg = 1.73 eV) because geometrical constraints of the perovskite structure require a large +1 A-site cation, and Cs+ is the most feasible. However, below 320°C, the orthorhombic (δ) phase (Eg = 2.82 eV) is thermodynamically preferred (3). Nevertheless, groups have explored CsPbX3 compounds as PV materials, but films of α-CsPbI3 undergo immediate transformation to the orthorhombic phase when exposed to ambient conditions (4). Attempts to stabilize the cubic phase through alloying with Br have been explored because CsPbIBr2 shows a much reduced δ to α phase transition temperature of 100°C (3). However, the composition change leads to an undesired increase in the band gap. We show that nanocrystal surfaces can be used to stabilize α-CsPbI3 at room temperature, far below the phase transition temperature for thin film or bulk materials. We further show that we can control the electronic coupling of quantum dots (QDs) to produce air-stable, efficient PV cells (initial efficiency above 10%) based on this all-inorganic material.

Many physical properties differ between nanometer-sized and bulk crystalline materials of the same chemical compound. One such example is the structural phase in which the constituent atoms are arranged. For example, the semiconductors CdS and CdSe embody a rock salt structure at high pressure. However, the solid-solid phase transition point between the rock salt phase and the hexagonal wurtzite phase can vary greatly in temperature and pressure as a function of crystal size (5, 6). Manipulated size-dependent phase diagrams have been explored in a variety of material systems, with advantageous properties of the crystals emerging at reduced dimensions in oxides (such as TiO2), lanthanides (such as NaYF4) (7), metals (such as Ag) (8), and ferroelectrics (such as the perovskite BaTiO3) (9).

Synthetic protocols of colloidal halide perovskite QDs have recently been reported (1017). CsPbX3 QDs exhibit improved room-temperature cubic-phase stability and attractive optical properties for a wide range of applications (11, 1822). Experiments on size- and shape-dependent optical properties (11, 2325), surface chemistry (26), and other photophysics (27) are being explored for CsPbBr3 QDs. However, previous studies were unable to achieve α-CsPbI3 QDs that were stable enough for extensive characterization or to be used in PV cells.

We present an improved synthetic route and purification approach of CsPbI3 QDs. Once purified, the QDs retain the cubic phase for months in ambient air and even at cryogenic temperatures. A method for perovskite QD film assembly is described that allows for efficient dot-to-dot electronic transport while retaining the phase stability of the individual QDs. The PV cells produced from this approach have the highest power conversion efficiency (PCE) and stabilized power output (SPO) of any all-inorganic perovskite absorber, produce 1.23 V at open circuit (among the best of any perovskite PV cells), and also function as light-emitting diodes (LEDs), emitting visible red light with low turn-on voltage.

The tunability of the band gap via size control due to quantum confinement is shown in Fig. 1. The series of CsPbI3 QDs, with varied size (band gap), were synthesized with the addition of Cs-oleate to a flask containing PbI2 precursor, as first described by Protesescu et al. (11)—here, using injection temperatures between 60° and 185°C to control the size (28). This produces QDs solubilized by noncrystalline iodide and oleylammonium surface ligands (26). Unpurified QDs transform to the orthorhombic phase within several days (fig. S1) (28), as in previous reports (29, 30). However, we developed a process to purify the QDs by using methyl acetate (MeOAc), an antisolvent that removes excess unreacted precursors without inducing agglomeration. Using this extraction procedure, we found that the QDs are stable in the cubic phase for months with ambient storage.

Fig. 1 Characterization of CsPbI3 QDs.

(A) Normalized UV-visible absorption spectra and photographs of CsPbI3 QDs synthesized at (a) 60°C (3.4 nm), (b) 100°C (4.5 nm), (c) 130°C (5 nm), (d) 150°C (6.8 nm), (e) 170°C (8 nm), (f) 180°C (9 nm), and (g) 185°C (12.5 nm). The numbers in parentheses are the average size from TEM. (B) Normalized photoluminescence spectra and photographs under UV illumination of the QDs from (A). (C) High-resolution TEM of CsPbI3 QDs synthesized at 180°C. (D) XRD patterns of QDs synthesized at (from bottom to top) 60°, 100°, 170°, 180°, and 185°C, confirming that they crystallize in the cubic phase of CsPbI3.

The excitonic peak of CsPbI3 shifted between 585 and 670 nm, corresponding to QD sizes between 3 and 12.5 nm, respectively. The corresponding normalized photoluminescence (PL) spectra of the samples are shown in Fig. 1B, along with a photograph of the QDs in hexane. Upon ultraviolet (UV) excitation, emission was in the orange (600 nm) to red (680 nm) color range, corresponding to a band gap between 2.07 and 1.82 eV (photographs showing PL from dried QD powders are shown in fig. S2) (28). The full width at half-maximum of the PL for the smallest QDs was 83 meV and increased slightly for the larger sizes, whereas the PL quantum yield varied from 21 to 55% for different sizes (fig. S3) (28).

In contrast to the instability of the cubic phase of bulk CsPbI3 at room temperature, QDs have been reported to retain the cubic phase because of the large contribution of surface energy (Fig. 1D) (11, 31). The softer basic nature of I as compared with Br results in weaker acid-base interactions between the halide and the oleylammonium ligand (a hard acid) in the case of CsPbI3, compared with that of CsPbBr3 (30, 32). Therefore, the isolation of CsPbI3 QDs is more difficult than that of CsPbBr3 QDs because of the loss of ligand during extraction, causing agglomeration and conversion to the orthorhombic phase. Thus, we found that MeOAc, which isolates the QDs without full removal of the surface species, is critical to the phase-stable devices described below.

The high-resolution transmission electron micrograph (TEM) of the sample synthesized at 180°C (Fig. 1C) shows an interplanar distance of 0.62 nm, which is consistent with the (100) plane of cubic phase CsPbI3 (24, 31, 33). In Fig. 2, A and B, powder x-ray diffraction (XRD) patterns and UV-visible absorption spectra confirm the absence of diffraction peaks or the high-energy (~3 eV) sharp absorption characteristic of orthorhombic phase formation (31), even after 60 days of storage in ambient conditions. Additionally, the QDs remained in the cubic phase even after the solution was cooled to 77 K, further demonstrating the expanded temperature stability of the cubic phase.

Fig. 2 Phase stability of CsPbI3 QDs.

(A) Powder XRD patterns and (B) UV-visible absorption spectra, normalized at 370 nm, of CsPbI3 QDs synthesized at 170°C and stored in ambient conditions for a period of 60 days. (Inset) The slight blue shift that is seen in the excitonic peak with extended storage. (C) Rietveld refinement fitting of CsPbI3 QD XRD pattern, revealing pure cubic-phase CsPbI3.

Rietveld refinement of the XRD patterns (Fig. 2C) (28) allowed us to quantify the contribution from cubic and orthorhombic phases. No detectable orthorhombic phase was found. Additionally, lattice parameters of three different size CsPbI3 QD samples were estimated (Table 1). The lattice parameter values showed a size dependence and were lower than the previously measured experimental value (6.2894 Å at 634 K) of bulk cubic CsPbI3 (33). Our measurements were performed at 297 K, whereas high temperatures are required to characterize bulk cubic CsPbI3. A similar increase in lattice parameter with decreasing particle size has been reported in other systems and attributed to electrostatic relaxation with decreasing crystal size (34).

Table 1 Results of the Rietveld refinement.

a, lattice parameter; Rwp, weighted-profile R factor.

View this table:

In order to use these highly phase-stable α-CsPbI3 QDs in optoelectronic devices, we developed a method to cast electronically conductive QD films. The QDs were first spin-cast from octane then dipped in a saturated MeOAc solution of either Pb(OAc)2 or Pb(NO3)2 (neat MeOAc was used as a control). This process was repeated multiple times—typically, three to five—to produce QD films with thicknesses between 100 and 400 nm. The optical absorption and PL spectra (Fig. 3A, for three samples with indicated reaction temperature) show that in each case, the film absorbance and PL was red-shifted ~20 nm from that of the QDs in solution, whereas the tunable emission properties of the films indicate that quantum confinement is preserved. Fourier-transform infrared (FTIR) spectra show the removal of organic ligands from the film with exposure to neat MeOAc (Fig. 3B), given the near absence of C–H modes near 3000 cm−1 or below ~2000 cm−1 belonging to oleylammonium, oleate, or octadecene. We therefore attribute the preserved phase stability of the QDs in the films to the size of the crystals (given the quantum confined optical properties) independent of the surface species. However, we found that prolonged annealing at temperatures >200°C causes further grain growth and thus induces a phase transition to the orthorhombic phase (fig. S4 and table S1) (28). Additional strategies to preserve the phase in sintered QD films are being explored (35). We have observed cubic-phase CsPbI3 with edge length up to 50 nm using the solution-phase synthesis described here.

Fig. 3 CsPbI3 QD films.

(A) UV-visible absorption (solid lines) and PL spectra (dashed lines) of CsPbI3 QDs in solution (blue) and as-cast films (black) for QDs synthesized at 100°, 150°, and 180°C. (B) FTIR spectra showing the IR transmission of a CsPbI3 QD film as cast (black) and after treating with MeOAc (red).

We also probed the interaction of Pb2+ salts with QDs in solution and on films by monitoring the fluorescence (fig. S5) (28). Titration of a small amount of Pb(OAc)2 dissolved in MeOAc to the QD solution showed an enhancement in PL, suggesting improved surface passivation. The surface treatments increase the PL lifetime over that of neat QD films, which highlights the importance of surface chemistry in this QD system (fig.S6 and table S2) (28). Titrations with only MeOAc caused fast PL quenching. Similarly, dip-coating of the QD film in a saturated solution of Pb(OAc)2 in MeOAc resulted in a PL enhancement of ~350% compared with dip-coating in MeOAc alone.

We fabricated PV cells with CsPbI3 QD films asthe photoactive material. A schematic of the device architecture is shown in Fig. 4A, and a scanning electron micrograph (SEM) cross-section image of the reported device with 9 nm QDs is shown in Fig. 4B. The reverse-scan current density-voltage (JV) curves showed an open-circuit voltage (VOC) of 1.23 V, and 10.77% PCE for a 0.10 cm2 cell made and tested completely in ambient conditions (relative humidity ~15 to 25%) (Fig. 4C). The hysteresis along with SPO of a device scanned at various sweep rates is shown in fig. S7 (28). Furthermore, the PCE improved from its initial value over the course of 60 days storage in dry but ambient conditions (fig. S8) (28). In fig. S9 (28), we show the SPO of the cell by measuring the current density while the device is biased at 0.92 V. In Fig. 4D, the spectral response of the PV cell is shown, indicating a band gap of 1.75 eV for this film. We compare QD devices to thin-film CsPbX3 perovskite solar cells following literature reports, which have thus far reported at 9.8% PCE and SPO as high as 6.5% (4, 31, 36). The QD devices show improved JV-scan efficiency, operational stability, and tolerance to higher relative humidity levels (figs. S10 and S11 and table S3) (28). The VOC is remarkably higher than that of other QD solar cells (typically <0.7 V) and among the highest VOC in all perovskite PV cells for band gap values below 2 eV (fig. S12, stabilized VOC) (28). We have not optimized the device architecture or the QD film-treatment scheme. We found that dip-coating spin-cast films in neat MeOAc and MeOAc saturated with Pb(OAc)2 or Pb(NO3)2 all work reasonably well (JV-scanned PCE > 9%) in PV devices. Large diffusion lengths and mobility values have been measured in CsPbBr3 QDs by means of terahertz spectroscopy (37); however, a better understanding of the electronic coupling is critical to maximizing long-range transport in QD perovskite films.

Fig. 4 CsPbI3 optoelectronic devices.

(A) Schematic (with TEM image of QDs) and (B) SEM cross-section of the CsPbI3 PV cell. (C) Current density–voltage curves of a device measured in air over the course of 15 days. The black diamond represents the stabilized power output of the device at 0.92 V, as shown in fig. S9. (D) External quantum efficiency (black, left ordinate) and integrated current density (blue, right ordinate) of the device. (E) EL spectra of CsPbI3 PV cell (CsPbI3 QDs synthesized at 170°C) under forward bias. (Inset) A photograph of the luminescent device. (F) PL (dashed lines) and EL (solid lines) spectra of completed devices fabricated by using CsPbI3 QDs synthesized at 170° and 180°C, demonstrating size quantization effects in the completed devices.

Given the PL properties of these perovskite QDs, we explored their use as LEDs. The PV devices produced bright visible electroluminescence (EL) when biased above VOC (Fig. 4E, inset). The EL had a low turn-on voltage near the band gap of the CsPbI3, with increasing intensity at larger applied biases (Fig. 4E). These spectra provide direct evidence that quantum confinement is retained in the complete devices, which is critical to retaining the improved cubic-phase stability, as seen by the shift in both the EL and PL spectra of devices with different-size QDs (Fig. 4F). The synthesis of normally unstable material phases stabilized through colloidal QD synthesis provides another mechanism for material design for PVs, LEDs, and other applications.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S12

Tables S1 to S3

References (38, 39)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
Acknowledgments: We thank J. van de Lagemaat, W. Tumas, H. Choi, M. Beard, and J. Berry for helpful discussions and B. To for SEM imaging. We acknowledge support from the Center for Advanced Solar Photophysics, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences for quantum dot coupling and solar cell structures. Device durability and structural phase characterization was performed within the hybrid perovskite solar cell program of the National Center for Photovoltaics funded by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Solar Energy Technologies Office under contract DE-AC36-08GO28308DOE. The original conception and QD synthesis was performed under the Laboratory Directed Research and Development program at NREL. A.S. acknowledges the Bhaskara Advanced Solar Energy fellowship funded by the Department of Science and Technology, government of India, and Indo-U.S. Science and Technology Forum (IUSSTF). E.M.S. acknowledges a NASA Space Technology Research Fellowship. D.T.M. acknowledges the NREL Director’s Fellowship. All data in the paper and supplementary materials are available. An application has been made for a provisional patent (U.S. patent application no. 62/343,251).

Stay Connected to Science

Navigate This Article