Precise Composition Tailoring of Mixed-Cation Hybrid Perovskites for

Aug 23, 2017 - All that glitters is not gold: Recent progress of alternative counter electrodes for perovskite solar cells. Lusheng Liang , Yu Cai , X...
21 downloads 0 Views 2MB Size
Download PDF
Precise Composition Tailoring of Mixed-Cation Hybrid Perovskites for Efficient Solar Cells by Mixture Design Methods Liang Li,†,‡,∥ Na Liu,§,∥ Ziqi Xu,† Qi Chen,*,§ Xindong Wang,*,‡ and Huanping Zhou*,† †

Department of Materials Science and Engineering and Department of Energy and Resources Engineering, College of Engineering, Peking University, Beijing 100871, P. R. China ‡ Department of Physical Chemistry, University of Science and Technology Beijing, Beijing 100083, P. R. China § School of Materials Science and Engineering, Beijing Institute of Technology, 5 Zhongguancun South Street, Beijing 100081, P. R. China S Supporting Information *

ABSTRACT: Mixed anion/cation perovskites absorber has been recently implemented to construct highly efficient single junction solar cells and tandem devices. However, considerable efforts are still required to map the composition−property relationship of the mixed perovskites absorber, which is essential to facilitate device design. Here we report the intensive exploration of mixed-cation perovskites in their compositional space with the assistance of a rational mixture design (MD) methods. Different from the previous linear search of the cation ratios, it is found that by employing the MD methods, the ternary composition can be tuned simultaneously following simplex lattice designs or simplex-centroid designs, which enable significantly reduced experiment/sampling size to unveil the composition−property relationship for mixed perovskite materials and to boost the resultant device efficiency. We illustrated the composition−property relationship of the mixed perovskites in multidimension and achieved an optimized power conversion efficiency of 20.99% in the corresponding device. Moreover, the method is demonstrated to be feasible to help adjust the bandgap through rational materials design, which can be further extended to other materials systems, not limited in polycrystalline perovskites films for photovoltaic applications only. KEYWORDS: mixture design, mixed cation, composition, perovskite solar cells, bandgap, structure

O

and X is a halide. Each ion serves as an independent constituent to occupy the corresponding site, obeying the tolerance factor rule.13 Therefore, possible combinations of elements and molecules are ready to create a diverse family of perovskites with distinct and tailorable properties. In the early stage, efforts were witnessed to exploit X site occupants, where “Br” was found to effectively adjust the bandgap from 1.48 to 2.23 eV.14 Meanwhile, the myth of “Cl” has received quite a few hot discussions.1,15,16 Particularly, CH3NH3PbI3(Cl)15 (commonly denoted as CH3NH3PbI3−xClx) has been reported to exhibit a substantially longer carrier diffusion length of over 1 μm than that of its counterpart CH3NH3PbI3 (∼100 nm).1 Device performance has consequently been elevated dramatically, mostly due to the improvement in the resultant absorber quality and the relevant interface.15 On the other hand, B site

rganic/inorganic hybrid perovskites have been well recognized in both academia and industry thanks to their superior optoelectronic properties including long carrier lifetime,1 high absorption coefficient,2 and exceptional defect tolerance.3 To date, this family of semiconductors has been intensively exploited in the fields of solar cells,4−8 light-emitting diodes,9 lasers,10 and field-effect transistors (FET).11 In particular, perovskite solar cells have achieved power conversion efficiencies (PCE) of 22.1%.12 Due to their extremely low cost and scalable processability, hybrid perovskites are regarded as one of the most promising photovoltaic (PV) materials that are a potential to compete or integrate with c-silicon PVs. Moreover, the encouraging progress in device fabrication has been accompanied by a persistent pursuit of fundamental understandings in materials composition, crystal structure, and defect associated optoelectronic properties. Hybrid perovskites share the chemical formula of ABX3, where often A is a monovalent cation, B is a divalent metal ion, © 2017 American Chemical Society

Received: April 26, 2017 Accepted: August 23, 2017 Published: August 23, 2017 8804

DOI: 10.1021/acsnano.7b02867 ACS Nano 2017, 11, 8804−8813

Article

www.acsnano.org

Article

ACS Nano

Figure 1. Contour plot of the relationship between (Cs,MA,FA)Pb(Cl,Br,I)3 with different composition (MA = 0−0.4, FA = 0.6−1.0, Cs = 0− 0.4, with fixed Cl:Br:I) and outputs: (a) power conversion efficiency (PCE), (b) short circuit current density (JSC), (c) fill factor (FF), and (d) open circuit voltage (VOC).

candidates like ethylammonium (EA), guanidinium (GA) and rubidium have also been exploited.32−34 Indeed, material properties can be manipulated in a broad range through variable constituent combinations in the hybrid perovskites (ABX3), which enables a great compositional space for future development. However, it leads to further complexity in experimental design if taking account of various processing protocols. For instance, it is reported that a set of total 49 samples have been investigated to illustrate the effects of different substitution in certain range.35 The large sampling cost and time are not always affordable in most laboratories which hinders the research momentum. In addition, most relevant study in mixed perovskites is mainly based on independently tuning one site composition rather than the whole ternary composition plane. To our understanding, there is no study to exploit the composition−property relationship among the abundant mixed-cation perovskite materials (Cs,MA,FA)PbX3 to reveal the synergistic effects from multiple components (e.g., surface recombination, crystal lattice, bandgap, etc.) based on rational experimental design. Here we report the exploit of the compositional space of tertiary cation (FA/MA/Cs) perovskites and the relevant devices via mixture design methods. Particularly, we employ the simplex lattice designs36,37 and simplex-centroid designs36,38 by picking up only 16 conditions for three factors, which drastically decreases the number of experiments while keeping the uniformity through the entire compositional space of the mixed-cation perovskites. More importantly, we conducted detailed material characterizations to bridge the materials optoelectronic properties to the resultant device performance. It proves the scientific soundness of this method, which is efficient to predict the optimum point from the experimental data independent of most theoretical models. Insights are revealed in the context of device efficiency and the tunability of

occupants have been systematically explored with the major purpose to find lead-free perovskites. When replacing lead (Pb) with tin (Sn) in the perovskite, the optical absorption edge can be successfully red-shifted to over 1000 nm.17,18 While the stability is reported to be higher in the mixed Pb/Sn perovskites compared to the pure tin-based materials.19 Meanwhile, the evolution of hybrid perovskites has occurred in A site and received continuous attention among the community to improve PCE and stability in the resulted devices. Formamidinium (FA) has been first documented as a possible constituent to compose hybrid perovskites for PV cells.20,21 It was reported to modify Pb−I bonding length and/ or corresponding bond angles that extend absorption edge of the resultant materials to 850 nm,14 corresponding to the bandgap of 1.45 eV. However, pure FAPbI3 is prone to transform into a yellow polymorph, which is not suitable for PV applications.22 Subsequently, the mixed-cation perovskites were reported, where α phase of FAPbI3 can be achieved/stabilized at room temperature by partial replacement of FA with MA.23−25 One of the most successful recipes includes (FAPbI3)0.85(MAPbBr3)0.15, which has contributed to the device with certified PCE exceeding 20%.22,26 Compared to FA, pure cesium (Cs)-based perovskites solar cells showed limited PCE, because the desired α phase of CsPbX3 endures phase transition even at room temperature.5 Encouraging results have been reported recently, while the fabrication challenges remain.27−29 Interestingly, when cesium is partially incorporated to either FA- or MA-based perovskites, it endows astonishing phase stability and moisture resistance in the resultant mixed-cation perovskites.13,30 In particular, partial substitution of cesium successfully eliminates the halide segregation and creates hybrid perovskites with a wider bandgap for tandem devices.31 Besides, other A site occupant 8805

DOI: 10.1021/acsnano.7b02867 ACS Nano 2017, 11, 8804−8813

Article

ACS Nano

Figure 2. Photovoltaic properties of the best cell under condition 8 (MA = 26.7%, FA = 66.6%, Cs = 6.6% in precursor solution). (a) J−V curves obtained under AM1.5 illumination. (b) Photocurrent density and PCE as a function of time held at 0.90 V bias. (c) The reproducibility of the perovskite solar cells. (d) IPCE spectrum of the representative cell and its integrated current density.

fixed Cl:Br:I ratio) that is most likely to result in the best performance devices. When the compositional space of particular interest is identified, an ordinary Kriging interpolation method40,41 was used to predict their output function from the experimental data generated through MD methods. In this method, the input parameters are the concentrations of FA/MA/Cs in the precursor solutions, and the processing conditions are kept the same (Figure S2, Tables S2 and S3). The outputs of this experiment are set to be the most influential device parameters, namely, fill factor (FF), short circuit current density (JSC), open circuit voltage (VOC), and PCE. We collected the device data (as was summarized in Table S4) and applied a quadratic model to conduct computational optimization by minilab.42,43 The detailed precursor formula, device fabrication, and data processing can be found in the Supporting Information. The perovskite film was obtained via two-step process. Typically, a DMF solution containing PbI2, PbBr2, and CsI was deposited by spin coating and drying. A solution composed of HC(NH2)2I, CH3NH3I and CH3NH3Cl was deposited by spin coating in a nitrogen glovebox and dried in air with desired relative humidity. Figure 1 shows the output contour distribution of JSC, VOC, FF, and PCE. From these contour plots, we can approximately predict the reasonable region of input parameters to reach the highest output value. Figure 1a illustrates the PCE output in the mixed perovskite compositional space. To achieve a PCE higher than 19.52%, the molar concentration reaches 0.2−0.3 for MA, 0.65−0.7 for FA, and 0.1−0.15 for Cs in the precursor solution. To obtain a JSC higher than 22.5 mA/cm2, the region was approximated as 0.0−0.2 for MA, 0.6−0.8 for FA, and 0.2− 0.3 for Cs (Figure 1b). For a VOC higher than 1.11 V, the region was approximated as 0.35−0.4 for MA, 0.7−1.0 for FA, and

materials properties (e.g., carrier lifetime and bandgap) when exploring the compositional space of the mixed-cation perovskites via MD method. In the materials perspective, this method is feasible to design hybrid perovskites in the forms of not only polycrystalline thin films but also single crystals and nanostructures and in applications of solar cells and other optoelectronic devices.

RESULTS AND DISCUSSION The MD methods, one major design of experiment (DoE), are employed to decrease the number of experiments and have been proven to effectively reduce the cost and time for optimizing the process conditions.36 For instance, Bashiri et al. used DoE coupled with the response surface models to study optical properties of Cu and Ni-doped TiO2 in photocatalysts.39 Here, we employed the simplex lattice designs and simplexcentroid designs to pick up samples within the newly developed mixed-cation perovskite system in a solar cell application. The aim is to find the optimal composition and discover the relationship between component and PV characteristics. The compositional space is picked up from recent publications with high credibility (Table S1). In this study, we mainly focus on the impacts of materials composition on device performance by taking advantage of MD methods, whereas other parameters are decoupled intentionally, such as device configurations and processing protocols/conditions. As shown in Figure S1, the optimized efficiency is estimated to locate at the compositional space where MA = 0.16, FA = 0.79, and Cs = 0.05. In this case, we have screened out our target area (with FA cation prevailing) in the compositional space of mixed-cation perovskites (MA = 0−0.4, FA = 0.6−1.0, Cs = 0−0.4, with 8806

DOI: 10.1021/acsnano.7b02867 ACS Nano 2017, 11, 8804−8813

Article

ACS Nano

Figure 3. Optoelectronic properties of the cells/films. Response surfaces of the (a) VOC loss of the solar cells, (b) lifetime τ1 (fitted by the lifetime decay curves), and (c) lifetime τ2 (fitted by the lifetime decay curves) as a function of ternary methylamine, formamidinium, and cesium mixture composition.

0.0−0.05 for Cs (Figure 1c. With respect to FF, we define the local maximum area approaching 80% in red (Figure 1d). The composition for the best efficiency in the present study is based on the precursor solution with molar ratios of FA = 66.6%, MA = 26.7%, and Cs = 6.6%. This composition ratio is different from the aforementioned best compositional space supported by other published reports, where MA = 0.16, FA = 0.79, and Cs = 0.05. The difference could be associated with the fact that the final products of the perovskite materials are different from the precursor solution in a two-step process, due to the different diffusivity, reaction kinetics, and volatile characteristic between MA and FA. By the employment of MD methods, the number of sampling has been greatly reduced when exploring the compositional space of tertiary mixed-cation perovskites to identify the best recipe. Figure 2 shows the reproducibility of the best composition (condition 8 in Table S2, FA = 66.6%; MA = 26.7%; Cs = 6.6%). Over 43.9% of the cells show efficiencies above 18%, while ∼80% of the cells show efficiencies above 17%. The average PCE is 17.46%. Under the optimum conditions discussed in the experimental section, our best cell achieved a PCE of 20.99% with a VOC of 1.103 V, a FF of 79.27%, and a JSC of 24.02 mA/cm2 (Figure 2a). The stabilized efficiency is demonstrated by measuring current at the maximum power point at a slow scan, and the steady efficiency of our planar perovskite solar cell is reaching 18.57% (Figure 2b). It is emphasized here that the optimization process over the entire compositional space takes only 16 samples, which greatly improves the experimental efficacy. A typical incident photon-to-electron conversion efficiency (IPCE) spectrum of a representative cell is shown in Figure 2d, which reaches its maximum of 87% at 400−600 nm wavelength and gradually drops at longer wavelength corresponding to its

absorption spectrum. The integration of the IPCE spectra for the cell gives a current density of 22.56 mA/cm2, which is in good agreement with the current density obtained from J−V measurement. The relatively lower current density obtained from (EQE) measurement, compared to that from the J−V curve, probably indicates the presence of surface defects on the TiO2 transporting materials.44 In addition, we reveal the impacts of the single element (input parameters) on the device performance (output parameters) by carefully analyzing the results from MD methods (Figure 1). It is found that upon the continuous increase of cesium concentration in the FA/MA mixed perovskites devices, JSC, FF, and PCE show obviously enhancement at first, respectively. They reach the optimal concentration around 0.2 and show a decrease when cesium is further increased. Interestingly, VOC follows a different pattern, which decreases when cesium is incorporated. Similar trends are observed when MA is incorporated in the FA/cesium system. It suggests the further use of the information generated by MD to guide the rational materials design to meet requirements for various applications. We conducted further characterizations to bridge the device performance to materials properties with the assistance of MD methods. Time resolved photoluminescence (TRPL) measurement is often used to probe the carrier dynamics within the perovskite films, which predominantly determines device performance. We obtained the lifetime decay curves for each sample, which was subsequently fitted with a biexponential function containing both fast (τ1) and slow components (τ2), as is summarized in Figures 3 and S4 and Table S5. It is generally believed that the fast component of the decay profile refers to the carrier recombination dynamics at surface/grain boundaries and/or the associated carrier diffusion45−48 and that 8807

DOI: 10.1021/acsnano.7b02867 ACS Nano 2017, 11, 8804−8813

Article

ACS Nano

Figure 4. (a) Static photoluminescence spectra and absorption of sample 4 (Figure S2 and Table S2, MA = 30%, FA = 60%, Cs = 10%). (b) Response surfaces for the bandgap (which was fitted by the static photoluminescence spectra and absorption curves) as a function of ternary methylamine, formamidinium, and cesium mixture composition.

able to passivate the defects and enhance the device performance.50 Followed by intensive research efforts recently,quite decent PCEs have been achieved by adopting this strategy.51 Herein, we explore the effects of residual PbI2 in the compositional space of mixed perovskite systematically with the assistance of MD (Figure S5b). First, we observed that it exhibited a higher content of PbI2 mostly at the region with higher FA content, and the incorporation of other cations, for example, cesium decreased the PbI2 content in the as-prepared films. It is in consistent with the report that the addition of cesium will significantly decrease the transition temperature of the mixed perovskites, indicating the thermodynamics of crystallization changes due to the smaller size of cesium.33 Furthermore, we observed that the increase of PbI2 leads to the increase of τ1 in the resultant film. It is in line with our previous argument that residual PbI2 effectively passivates the defects to inhibit the carrier recombination at the grain boundaries in the polycrystalline perovskite films. Accordingly, the pattern of VOC loss is observed to follow a similar trend as that of the residual PbI2 distribution in the compositional space of mixed-cation perovskite (Figure S5b). It thus infers that PbI2 serves as a native species that is able to lead to well-passivated absorbers at the grain boundaries to boost the corresponding device performance. For the mixed perovskites in tandem applications, it is highly appealing to tune the bandgap in a controllable manner by rational incorporation of different occupants in either A or X sites. To date, binary mixed-cation perovskites have already been studied to understand the bandgap tuning mechanism in both theoretical and experimental realms,35 but systematic experiments on the tertiary materials systems are still limited due to the heavy sample loading. Therefore, we have thoroughly investigated the bandgap tunability of mixed-cation perovskites in the tertiary compositional space via MD method. Bandgaps of the 16 samples were determined by conducting the UV−vis and PL measurements to obtain the necessary data for the Kubellka−Munk equation, as summarized in Table S5 and Figure 4. It shows the bandgaps vary from 1.56 to 1.65 eV by simultaneously adjusting the compositions in the tertiary space. The lowest value of bandgap is found to locate in the area with 100% FA occupancy, while the highest one sits in the composition of 6.6% MA, 66.6% FA, and 26.7% Cs (sample 7). When Cs and/or MA cation with a small radius is incorporated in the FA-based perovskites in a certain range, the bandgap of the resultant mixed perovskites increases, respectively.

the slow one stands for that in bulk. The FA-based perovskites exhibit the largest τ1 when a small amount of either MA and/or cesium is incorporated to form the mixed-cation system (Figure 3b). This is in accordance with the variation in VOC within the compositional space (Figure 1c). VOC loss plot is also in line with the recent reported respectable VOC in FA-based perovskites with a small concentration of either cesium or FA (Table S6). Interestingly, the contour plot pattern of the slow component τ2 within the compositional space exhibits great discrepancy to that of τ1 (Figure 3). It is more likely to correlate the VOC loss in corresponding perovskite solar cells to the carrier recombination at the surfaces (τ1) rather than in bulk perovskites (τ2). It thus suggests improving the device performance based on current perovskite film growth technique, whereas efforts are appreciated to be focused on inhibiting the carrier recombination at the surfaces/interfaces. This also consistent with recent demonstrated work that surface recombination was more critical than bulk recombination in conventional perovskite.48 We attempt to reveal the insights of the carrier recombination profiles within the compositional space of mixed-cation perovskites in the perspective of crystal structure. When fabricating polycrystalline absorbers for PVs, larger crystal/grain size is often preferred. It is because a larger grain size is believed to possess better crystallinity and less grain boundaries that host more carrier recombination centers. However, decent PCEs have been observed in perovskite solar cells with grain sizes at both ends.49 X-ray diffraction (XRD) has thus been conducted to investigate the crystal properties (Figure S3). The grain size was calculated from XRD pattern by the Debye−Scherrer equation, which is summarized in Table S7 and plotted in Figure S5. To be noted, these are much smaller compared to that obtained from scanning electron microscopy (SEM), because the Debye−Scherrer equation is accurate for the film with a grain size