ReportSOLAR CELLS

Incorporation of rubidium cations into perovskite solar cells improves photovoltaic performance

See allHide authors and affiliations

Science  14 Oct 2016:
Vol. 354, Issue 6309, pp. 206-209
DOI: 10.1126/science.aah5557

Improving the stability of perovskite solar cells

Inorganic-organic perovskite solar cells have poor long-term stability because ultraviolet light and humidity degrade these materials. Bella et al. show that coating the cells with a water-proof fluorinated polymer that contains pigments to absorb ultraviolet light and re-emit it in the visible range can boost cell efficiency and limit photodegradation. The performance and stability of inorganic-organic perovskite solar cells are also limited by the size of the cations required for forming a correct lattice. Saliba et al. show that the rubidium cation, which is too small to form a perovskite by itself, can form a lattice with cesium and organic cations. Solar cells based on these materials have efficiencies exceeding 20% for over 500 hours if given environmental protection by a polymer coating.

Science, this issue pp. 203 and 206

Abstract

All of the cations currently used in perovskite solar cells abide by the tolerance factor for incorporation into the lattice. We show that the small and oxidation-stable rubidium cation (Rb+) can be embedded into a “cation cascade” to create perovskite materials with excellent material properties. We achieved stabilized efficiencies of up to 21.6% (average value, 20.2%) on small areas (and a stabilized 19.0% on a cell 0.5 square centimeters in area) as well as an electroluminescence of 3.8%. The open-circuit voltage of 1.24 volts at a band gap of 1.63 electron volts leads to a loss in potential of 0.39 volts, versus 0.4 volts for commercial silicon cells. Polymer-coated cells maintained 95% of their initial performance at 85°C for 500 hours under full illumination and maximum power point tracking.

Low-cost perovskite solar cells (PSCs) have achieved certified power conversion efficiencies (PCEs) of 22.1% (1). The organic-inorganic perovskites used for photovoltaics (PV) have an AMX3 formula that comprises a monovalent cation, A [cesium Cs+, methylammonium (MA) CH3NH3+, or formamidinium (FA) CH3(NH2)2+]; a divalent metal, M (Pb2+ or Sn2+); and an anion, X (Cl, Br, or I). The highest-efficiency perovskites are Pb-based with mixed MA/FA cations and Br/I halides (24). Recently, Cs was used to explore more complex cation combinations: Cs/MA, Cs/FA, and Cs/MA/FA (59). These perovskite formulations exhibit unexpected properties. For example, Cs/FA mixtures suppress halide segregation, enabling band gaps for perovskite/silicon tandems (10). The Cs/MA/FA-based solar cells are more reproducible and thermally stable than MA/FA mixtures (9).

In general, increasing the perovskite complexity is motivated by the need to improve stability by adding more inorganic elements and increasing the entropy of mixing, which can stabilize ordinarily unstable materials (such as the “yellow,” nonphotoactive phase of FAPbI3 that can be avoided by using small amounts of the otherwise unstable CsPbI3) (6, 7). However, all combinations of Cs, MA, and FA cations were selected because each individually forms a photoactive perovskite “black” phase (1113).

Further progress requires exploration of a wider circle of cations. Unfortunately most monovalent cations are mismatched to sustain a photoactive APbI3 perovskite with an appropriate Goldschmidt tolerance factor [Embedded Image, where r is ionic radius] between 0.8 and 1.0 (14, 15), rendering almost all elemental cations too small for consideration. We illustrate this point in Fig. 1A, which shows tolerance factor calculations for the alkali metals (Li, Na, K, Rb, Cs) as well as MA and FA (see table S1 for numeric values and ionic radii). We selected specifically the alkali metals that are oxidation-stable monovalent cations, as these would have a stability advantage over oxidation-prone Pb/Sn mixtures that have distorted material electronic properties (16).

Fig. 1 Tolerance factor and perovskites at different temperatures.

(A) Tolerance factor of APbI3 perovskite with the oxidation-stable A (Li, Na, K, Rb, or Cs) and MA or FA (see table S1 for detailed calculations and ionic radii). Empirically, perovskites with a tolerance factor between 0.8 and 1.0 (dashed lines) show a photoactive black phase (solid circles) as opposed to nonphotoactive phases (open circles). Rb (red open circle) is very close to this limit, making it a candidate for integration into the perovskite lattice. (B) CsPbI3 [(a) to (c)] and RbPbI3 [(d) to (f)] at 28° , 380°, or 460°C. Irreversible melting for both compounds occurs at 460°C. RbPbI3 never shows a black phase.

The tolerance factor shows that only CsPbI3, MAPbI3, and FAPbI3 fall into the range of “established perovskites” with a black phase. Li, Na, and K are clearly outside of the range, whereas RbPbI3 only misses by a small margin. The ionic radii of Cs and Rb are 167 pm and 152 pm, respectively. This small difference still has a large impact, with RbPbI3 and CsPbI3 drawing the demarcation line between photoactive black perovskite and photoinactive yellow nonperovskite phases. As shown by heating CsPbI3 and RbPbI3 films at different temperatures (Fig. 1B), both films are yellow at 28°C; upon heating to 380°C, only CsPbI3 turns black, whereas RbPbI3 remains yellow. At 460°C, both films start melting irreversibly, without RbPbI3 ever showing a black phase; this is consistent with the observations of Trots and Myagkota (17). Thus, only CsPbI3 has a black phase, which explains why Rb has so far not been used for PSCs despite its desirable oxidation stability.

In this work, we propose embedding Rb+, only slightly smaller than Cs+, into a photoactive perovskite phase using multiple A-cation formulations. We retain FA as the majority cation because of the beneficial, red-shifted band gap. We identify four previously unexplored combinations: RbFA, RbCsFA, RbMAFA, and RbCsMAFA. In (18) and figs. S1 to S3, following the antisolvent approach pioneered by Jeon et al. (2), we present device data on a glass/fluorine-doped tin oxide/compact TiO2/mesoporous TiO2/perovskite/spiro-OMeTAD [2,2′,7,7′-tetrakis(N,N-di-p-methoxyphenylamine)-9,9′-spirobifluorene]/Au architecture. [See fig. S4A for a cross-sectional scanning electron microscopy (SEM) image and fig. S4C for an image of typical devices.] All preparation details are given in (18). We use the nomenclature of RbFA, RbCsFA, RbMAFA, and RbCsMAFA to denote the entire perovskite compounds at the optimized values found in (18) (usually achieved with an addition of 5 to 10% Rb).

Reasonable device performances were reached with RbFA (14%), RbCsFA (19.3%), RbMAFA (19.2%) [comparable to CsFA (20%)], and CsMAFA (19.2%), as shown in figs. S1 to S3 (measured on a device area of 0.16 cm2). Thus, Rb can stabilize the black phase of FA perovskite and be integrated into PSCs, despite not being suitable as a pure RbPbI3 compound. Surprisingly, RbCsMAFA (with 5% Rb; fig. S3) resulted in PCEs of 20.6%, with an open-circuit voltage Voc of 1186 mV (18). Hence, we focus below on RbCsMAFA to substantiate the impact of the Rb+ integration approach for PSCs.

We investigated the starting condition of the crystallization process for the RbCsMAFA compound upon annealing at 100°C, which is needed to fully crystallize the perovskite films. In Fig. 2A, we present the ultraviolet-visible (UV-vis) and photoluminescence (PL) data of the unannealed MAFA and RbCsMAFA films. Whereas MAFA showed several PL peaks with maxima ranging from 670 to 790 nm, the RbCsMAFA film had a narrow peak at 770 nm attributable to perovskite. The insets in Fig. 2A are fluorescence microscopy maps of the surface of the unannealed films, showing that the MAFA films comprise various emissive species that force the preannealed film to crystallize with inhomogeneous starting conditions. However, the RbCsMAFA films were emissive in a narrow range and began to crystallize from more homogeneous conditions. Thus, the addition of the inorganic cations enforced a crystallization that starts with a photoactive perovskite phase (near the final emission after annealing) instead of a mixture of varying emissions that need to converge toward the final emission (see Fig. 2C). These results are consistent with the high reproducibility and lack of yellow phase in the RbCsMAFA films.

Fig. 2 Characterization of unannealed and annealed films.

(A) UV-vis (dashed lines) and PL (solid lines) of unannealed MAFA (black) and RbCsMAFA (red) films. The inset images show fluorescence microscopy measurements (image size ~26 μm × 26 μm) of MAFA and RbCsMAFA films. Each image is an overlay of three emission ranges sampled from 640 to 650 nm (assigned as green), 680 to 690 nm (blue), and 725 to 735 nm (red). The colors were chosen to ensure easily discernible features. (B) XRD data of the unannealed MAFA and RbCsMAFA films. (C) UV-vis (dashed lines) and PL (solid lines) of MAFA (black) and RbCsMAFA (red) films annealed at 100°C for 1 hour. (D) XRD data of the annealed MAFA and RbCsMAFA films. The PbI2 and yellow-phase peaks are denoted as # and δ, respectively.

Furthermore, we collected the corresponding x-ray diffraction (XRD) data of the unannealed films (Fig. 2B) that showed a pronounced perovskite peak for RbCsMAFA as compared to MAFA films. In Fig. 2, C and D, we show analogous data after annealing, including UV-vis, PL, and XRD data, that reveal a RbCsMAFA band gap of ~1.63 eV (slightly blue-shifted relative to MAFA at ~1.62 eV) containing neither a PbI2 nor a yellow-phase peak. The low-angle perovskite peaks for MAFA and RbCsMAFA occur at 14.17° and 14.25°, respectively, revealing that Rb indeed modified the crystal lattice. In figs. S5 and S6, we show XRD data of RbMAFA perovskite where we increased the concentration of Rb. We observed, similar to CsMAFA (9), that the Pb excess and the yellow-phase impurities of MAFA perovskite disappeared when Rb was added. For Rb5MAFA, there was a shift to wider angles for the perovskite peak. Moreover, in figs. S7 and S8, we show a series of RbCsMAFA perovskite with an increased amount of Rb together with a RbPbI3 reference. We observed that the perovskite peak shifted to wider angles for Rb5CsMAFA as well as further suppression of the residual PbI2 (12.7°) and yellow-phase peak (11.7°) of FA-based perovskite. As more Rb was added, we noted the appearance of a second peak at 13.4° and a double peak at 10.1° that coincide with the peaks for the pure yellow-phase RbPbI3, indicating phase segregation at higher Rb concentrations. This is in good agreement with previous work where a phase segregation was also observed as more and more Cs was added to FA-based perovskite (8).

In addition, top-view scanning electron microscopy (SEM) images revealed large crystals in the RbCsMAFA devices (fig. S9), which have been shown to be beneficial for the PV metrics (19). Energy-dispersive x-ray spectroscopy measurements (fig. S10) indicated the presence and distribution of Cs and Rb within the perovskite layer.

We collected statistical data on RbCsMAFA devices (with 12 CsMAFA and 17 RbCsMAFA devices measured at a scan rate of 10 mV s−1, without preconditioning such as light soaking or long-term forward voltage biasing; see fig. S11) and observed superior performance relative to CsMAFA. Remarkably, the average Voc increased from 1120 to 1158 mV and the fill factor (FF) increased from 0.75 to 0.78. In Fig. 3A, we show that the best-performing RbCsMAFA cell reached a stabilized power output of 21.6% with FF of 81% and Voc of 1180 mV. The measured short-circuit current density Jsc matched the incident photon to current efficiency (IPCE) measurement in fig. S12. We also achieved a stabilized PCE of 19.0% on a large-area 0.5-cm2 device (see fig. S13).

Fig. 3 Efficiency, open-circuit voltage, electroluminescence, and high-temperature stability of the best-performing RbCsMAFA solar cell.

(A) Current density–voltage (J-V) curve, taken at 10 mV s−1 scan rate, of the solar cell with 21.8% efficiency (Voc = 1180 mV, Jsc = 22.8 mA cm−2, and FF = 81%). The forward and reverse scan is shown in table S2. The inset shows the scan rate–independent MPP tracking for 60 s, resulting in a stabilized efficiency of 21.6% at 977 mV and 22.1 mA cm−2 (displayed as triangles in the J-V and MPP scans). (B) J-V curve of the highest-Voc device. The inset shows the Voc over 120 s, resulting in 1240 mV (displayed as the red triangles in the J-V and Voc scans). (C) EQE electroluminescence (EL) as a function of voltage. The left inset shows the corresponding EL spectrum over wavelength. The right inset shows a solar cell (device size ~1.4 cm × 2.8 cm) with two active areas. The left area is operated as an LED displaying a clearly visible red emission even under ambient light. At the same time, the right area can be operated as a solar cell or a photodetector. (D) Thermal stability test of a perovskite solar cell. The device was aged for 500 hours at 85°C under continuous full illumination and MPP tracking in a nitrogen atmosphere (red curve, circles). This aging routine exceeds industry norms. During the light soaking at 85°C, the device retained 95% (dashed line) of its initial performance.

To correctly determine the value of Voc, we investigated RbCsMAFA devices with the active area being fully illuminated, held at room temperature, and under an inert nitrogen atmosphere. This setup permitted an accurate Voc value without heating or degradation effects (from moisture, for example). In Fig. 3B, for one of our highest-performing devices, we measured an outstanding Voc of 1240 mV, as confirmed by the inset tracking Voc over time. The “loss in potential” (difference between Voc and band gap) is only ~0.39 V, which is one of the lowest recorded for any PV material, implying very small nonradiation recombination losses (20). The high Voc is particularly intriguing because this is the major parameter preventing PSCs from reaching their thermodynamic limit (Jsc and FF are already approaching their maximal values). Theoretically, in very pure, defect-free materials with only radiative recombination, the loss in potential can be as small as 0.23 V (band gap of 1 eV) to 0.3 V (band gap of 2 eV). In particular, silicon, the main industrial PV material, cannot approach this limit because of its indirect band gap and Auger recombination, exhibiting a loss in potential of ~0.4 V for the most efficient devices (20).

The nonradiative recombination losses were quantified by measuring the external electroluminescence quantum efficiency (EQEEL), which is > 1% at a driving current that is equal to the short- circuit current (see Fig. 3C). This value is in the same order of magnitude and thus consistent with a measured external PL quantum yield of 3.6% for RbCsMAFA (and 2.4% for CsMAFA). Following the approach in (2125) [see also fig. S14 and (18)], we used the EQEEL and the emission spectrum to predict a Voc value of 1240 mV, confirming independently the value measured from the J-V curve.

Furthermore, for higher driving currents, the EQEEL in Fig. 3C reaches 3.8%, making the solar cell one of the most efficient perovskite LEDs as well, emitting in the near-infrared/red spectral range (Fig. 3C, inset) (2628). Movie S1 shows a RbCsMAFA device mounted in our custom-made device holder. As we tuned toward maximum emission and back, we observed bright EL in daylight. For comparison, for commercially available Si solar cells, EQEEL ≈ 0.5% (20). These values for our PSC devices indicate that all major sources of nonradiative recombination were strongly suppressed and that the material has very low bulk and surface defect density. We also investigated transport behavior by means of intensity-modulated photocurrent spectroscopy (IMPS); the findings suggest that the charge transport within the RbCsMAFA perovskite layer is substantially faster than in CsMAFA, which is already much more defect-free than MAFA (19) [see also fig. S15 and (18)].

Despite the high efficiencies and an outstanding EL, this Rb-containing perovskite material must be able to achieve high stability. This task is not trivial given the hygroscopic nature of perovskite films, phase instabilities, and light sensitivity (29). Interestingly, the Achilles’ heel of PSCs is not necessarily the perovskite itself, but rather the commonly used spiro-OMeTAD hole transporter material that becomes permeable (at elevated temperature) to metal electrode diffusion into the perovskite, causing irreversible degradation (30, 31). This effect can be mitigated with buffer layers or by avoiding the use of metal electrodes (3032). Alternatively, for the combined heat-light stress tests in this work, we found a thin layer of polytriarylamine polymer (PTAA) (see SEM image in fig. S4B) to work equally well (33). We imposed the above protocols simultaneously and aged devices for 500 hours at 85°C under continuous illumination with full intensity and maximum power point (MPP) tracking in a nitrogen atmosphere. This compounded stress test exceeds industrial standards (34). We show the result in Fig. 3D (red curve). The device started with >17% efficiency at room temperature before the aging protocol was applied (see fig. S16 for non-normalized values of PCE, FF, Jsc, Voc, JMPP, and VMPP). During the 85°C step (in which Voc is inevitably lowered), the device retained 95% of its initial performance.

Supplementary Materials

www.sciencemag.org/content/354/6309/206/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S16

Tables S1 and S2

Movie S1

References (3541)

References and Notes

  1. See supplementary materials on Science Online.
  2. Acknowledgments: M.S. conceived, designed, and led the overall project; M.S., J.-Y.S., A.U., and J.-P.C.-B. conducted SEM, PL, and XRD experiments on the perovskite films; M.S. and W.R.T. performed EL and PL quantum yield experiments; A.U. conducted confocal laser scanning fluorescence microscopy for PL mapping; M.S., K.D., and W.R.T. conducted long-term aging tests on the devices; M.S., T.M., J.-P.C.-B., and A.A. prepared and characterized PV devices; A.H. participated in the supervision of the work; M.G. directed and supervised the research; M.S. wrote the first draft of the paper; and all authors contributed to the discussion and writing of the paper. Supported by the co-funded Marie Skłodowska Curie fellowship, H2020 grant agreement no. 665667 (M.S.); the European Union’s Seventh Framework Programme for research, technological development, and demonstration under grant agreement no. 291771 (A.A.); the Swiss National Science Foundation, funding from the framework of Umbrella project (grant agreement nos. 407040-153952, 407040-153990, and 200021-157135/1); the NRP 70 “Energy Turnaround”; the 9th call proposal 906: CONNECT PV; and SNF-NanoTera and the Swiss Federal Office of Energy (SYNERGY). We also acknowledge funding from the European Union's Horizon 2020 program, through a FET-Open Research and Innovation Action under grant agreement no. 687008. A.A. conducted IMPS experiments at the Adolphe Merkle Institute, Fribourg, Switzerland. M.G. and S.M.Z. thank the King Abdulaziz City for Science and Technology for financial support under a joint research project. All data are available in the main paper and supplement. M.S., T.M., K.D., J.-Y.S., S.M.Z., W.R.T., and M.G. are inventors on European Patent Application 1618056.7 submitted by École Polytechnique Fédérale de Lausanne and Panasonic Corporation that covers the perovskite compounds in this work.

Stay Connected to Science

Navigate This Article