Synthesis and Optical Properties of Lead-Free Cesium Tin Halide

Nov 21, 2016 - ABSTRACT: Herein, the fabrication of a lead-free cesium tin halide perovskite ... continues to raise research in low-cost and high-effi...
1 downloads 0 Views 2MB Size
Letter pubs.acs.org/JPCL

Synthesis and Optical Properties of Lead-Free Cesium Tin Halide Perovskite Quantum Rods with High-Performance Solar Cell Application Lin-Jer Chen,*,† Chia-Rong Lee,† Yu-Ju Chuang,† Zhao-Han Wu,‡ and Chienyi Chen§ †

Department of Photonics, ‡Department of Materials Science and Engineering, and §Department of Chemistry, National Cheng Kung University, Tainan 70101, Taiwan

J. Phys. Chem. Lett. 2016.7:5028-5035. Downloaded from pubs.acs.org by UNIV OF GOTHENBURG on 01/29/19. For personal use only.

S Supporting Information *

ABSTRACT: Herein, the fabrication of a lead-free cesium tin halide perovskite produced via a simple solvothermal process is reported for the first time. The resulting CsSnX3 (X = Cl, Br, and I) quantum rods show composition-tunable photoluminescence (PL) emissions over the entire visible spectral window (from 625 to 709 nm), as well as significant tunability of the optical properties. In this study, we demonstrate that through hybrid materials (CsSnX3) with different halides, the system can be tunable in terms of PL. By replacing the halide of the CsSnX3 quantum rods, a power conversion efficiency of 12.96% under AM 1.5 G has been achieved. This lead-free quantum rod replacement has demonstrated to be an effective method to create an absorber layer that increases light harvesting and charge collection for photovoltaic applications in its perovskite phase.

T

optical density of electrodes covered by up to a single layer of nanoparticles and therefore improve light harvesting. QRs are of great interest for fundamental research and are highly promising novel materials for new technology.15,16 However, QR properties will also depend on the quality of their internal crystal structure because may also exhibit decreased Auger recombination rates compared to those of QDs,17 which makes them potentially useful for advanced optoelectronic devices. Many modern methods have been developed for the synthesis of onedimensional QRs.18,19 However, these preparation processes often need a relatively high temperature, vacuum, complexity, and so forth; solvothermal synthesis offers great advantages of low cost and maneuverability.20−22 In this research, we demonstrate a simple, fast, and “green” lead-free high-quality perovskite-type CsSnX3 with a uniform QR-shaped structure for enhancing photovoltaic efficiency. To our knowledge, this is the first example of solvothermal synthesis of high-quality CsSnX3 QRs with finely tunable photoluminescence (PL) and uniform diameters and high-performance solar cells up to 13%. Figure 1a,b shows that the optical absorbance and PL of all of the inorganic perovskite QRs are continuously red shifted to longer wavelengths because of the confinement of the halide.23 The QRs exhibit typically sharp absorption onsets, similar to the bulk dispersion. The slight shift of the first transition to longer wavelengths, when comparing CsSnCl3 to CsSnI3, might be

he global need for renewable and green energy alternatives continues to raise research in low-cost and high-efficiency photovoltaic devices. As one kind of perovskite material, allinorganic perovskites CsPbX3 (X = I, Br, Cl) were recognized as one probable substitute for organic−inorganic perovskites.1,2 The hybrid, all-inorganic lead halide perovskites have generated enormous attention in the photovoltaic community due to their surprisingly rapid improvements in power conversion efficiency.3,4 While recent research has been mostly focused on hybrid organic− inorganic compounds, the study of their inorganic analogues, like ABX3 (A = Rb, Cs; B = Ge, Sn, Pb; X = Cl, Br, I), is limited.5 However, the lead component is a high human health concern, toxic, and subject to environmental damage, particularly chemical stability.6,7 By replacing the lead component with tin at the B-site, a lead-free perovskite can be formed, obviating the problems of pollution and environmental friendliness. Toward wider application of these perovskites, replacement of unsafe lead is essential. As a result, there has been a great deal of interest in replacing lead with nontoxic metals such as tin, bismuth, and germanium.8−10 Over the past several years, considerable advances have been made toward the synthesis of colloidal semiconductor nanorods or nanofibers with diameters sufficiently small to produce quantum confinement of charge carriers. Semiconductor quantum rods (QRs) are visible at the transition between zerodimensional quantum dots (QDs) and one-dimensional quantum wires.11−13 QR volumes are larger than those of QDs, and therefore, they have significantly larger per-particle absorbance cross sections.14 This is expected to increase the © 2016 American Chemical Society

Received: October 11, 2016 Accepted: November 21, 2016 Published: November 21, 2016 5028

DOI: 10.1021/acs.jpclett.6b02344 J. Phys. Chem. Lett. 2016, 7, 5028−5035

Letter

The Journal of Physical Chemistry Letters

Figure 1. (a) Optical absorption and (b) PL spectra for different compositions of CsSnX3 (X = Cl, Br, and I).

file no. 70-1645), and CsSnCl3 monoclinic (JCPDS file no. 712023) at room temperature.24,25 This reversible anion exchange (CsSnCl3 to CsSnBr3 to CsSnI3) experiment confirms that the size and size distribution of the QRs remained intact throughout all anion-exchange processes. Therefore, the CsSnX3 QRs obtained after anion exchange, despite having a quantum-scale distribution, did not exhibit any excitonic features. The atom ratio of Cs/Sn/X (X = Cl, Br, and I) is around 1:1:3, which further verifies that the resulting products are CsSnCl3, CsSnBr3, and CsSnI3. Accordingly, because larger I atoms replace Br and Cl, the size of CsSnX3 QRs slightly increases. Similar to other semiconductors, the PL peak shifts toward longer wavelengths as the CsSnX3 QR size increases.26 Both the absorption and PL emission peaks of the CsSnX3 QRs continuously red shift along with the increase of Br or I content, reflecting the concomitant change of the band gap. These results are in good agreement with recent investigations on perovskite hybrids, where the singlecrystalline structures possess a narrower band gap and a subsequent red-shifted PL emission, suggesting lower trap densities due to coherent structural units that are tighter in the single crystal.27 As shown in Figure 3a−c, the typical XPS survey spectra for both sets in the entire binding energy region indicate that Cs, Sn, and I exist, which demonstrates the high purity of the assynthesized CsSnI3 QRs. Two distinct Cs 3d peaks appear at 726.6 (2p3/2) and 740.4 eV (2p1/2) with a binding energy splitting of 13.8 eV, thereby verifying the presence of Cs.28 The peaks of Sn 3d located at 485.3 (3d5/2) and 493.7 (3d3/2) eV confirm the Sn state.29 The peaks of I 3d5/2 located at 619.9 and 630.9 eV,30 are in good agreement with values reported for the oxidation state of halides. As such, these results demonstrate that CsSnI3 QRs have been successfully synthesized. Figure 3d shows the time-resolved PL decay curve of the CsSnX3 QRs. PL decay dynamics were recorded with 50 μW of 3.0 eV photons produced by a 35 fs, 2 kHz Ti:sapphire laser. The mean maximum-intensity lifetimes for samples CsSnCl3, CsSnBr3, and CsSnI3 are 16.3, 21.4, and 25.8 ns respectively, and correspond to the previously reported radiative relaxation times of the excitons associated with hybrid perovskite nanocrystals.31 For QRs with CsSnI3, the longest components are over 25.8 ns, which are significantly longer than the lifetimes typically observed for the other perovskite hybrid QRs. We suggest that this results from halide atom inhomogeneity (i.e., atomic weight) and exciton radiative recombination in the small nanocrystals.32 Figures 4a and S1 display the transmission electron microscopy (TEM) images of the perovskite CsSnX3 QRs and show the microscopic morphology and structural information on

caused by a slight change in the lattice constant of the surrounding atoms. Although a red shift in the absorption band was found for CsSnX3 (starting from Cl to Br to I), a 80 nm shift in the absorption onset (from 588 to 668 nm) was observed when halide (chloride or bromide or iodide) was used in the formation of CsSnX3 under similar conditions, which confirms the formation of mixed halide perovskite. Moreover, the overall intensity of the absorption band was altered. As such, the addition of different amounts of Br and I seems to affect the optical properties of the perovskite composites. With increased ratios of Br and I, the light absorption of the perovskite QRs in the visible light range was gradually enhanced. This good agreement indicates that the halide improvement is primarily attributed to the enhanced absorption due to the lattice constant with larger atoms. The PL (excited by light with λ = 532 nm) emission wavelength could be tuned from 625 to 709 nm with a full width at half-maximum (fwhm) at around 32 nm. The purity and crystallinity of the CsSnX3 QRs with different weight addition ratios of Cl, Br, and I were confirmed by XRD, as shown in Figure 2. QRs of CsSnI3 were identified to be orthorhombic (JCPDS file no. 43-1162), CsSnBr3 cubic (JCPDS

Figure 2. XRD patterns for varying halide composition of perovskites (CsSnI3, CsSnBr3, and CsSnCl3). 5029

DOI: 10.1021/acs.jpclett.6b02344 J. Phys. Chem. Lett. 2016, 7, 5028−5035

Letter

The Journal of Physical Chemistry Letters

Figure 3. (a) XPS Cs 3d spectra for CsSnI3, (b) XPS Sn 3d spectra for CsSnI3, (c) XPS I 3d spectra for CsSnI3, and (d) time-dependent PL decays of various CsSnX3 QRs prepared with different halides (CsSnCl3, CsSnBr3, and CsSnI3) in toluene, averaged across measurements from five samples.

weight increases from Cl to I, the intensity of the PL emission is enhancemed with the various halide ratio QRs. Further, the position of the PL emission band is slightly red shifted, indicating that the size distribution of the CsSnX3 QRs increased during the solvothermal process. On the other hand, the CsSnI3 QRs have very good crystallinity, which is preferred for improving the PL performance. However, compositional analysis by XPS showed that the average composition of the QRs in the sample had a molar Cs/Sn/X (X = Cl, Br, and I) ratio of 1:1:3 and the composition of individual particles measured by EDS were 1:1:3, with a variation from particle to particle of less than the experimental error of approximately ±2 atom % (see Table 1). Figure 4. (a) Typical high-resolution TEM image of the perovskite CsSnI3 QRs. (b−e) Selected zoom-in TEM images of several QRs: (b) TEM image of the obtained perovskite QRs and corresponding energy dispersive spectroscopy (EDS) elemental mapping images, which confirms that the (c) Cs, (d) Sn, and (e) I are sequentially and uniformly deposited onto the entire surface of the CsSnI3 QRs on a copper grid. The scale bars are 5 nm.

Table 1. Measured CsSnX3 QR Composition

Cs/ Sn/CI Cs/ Sn/Br Cs/Sn/I

precursor atomic ratio

measured by EDSa

measured by XPSb

1:1.2:3.5 1:1.2:3.4 1:1.1:3.2

1:0.82:2.85 1:0.81:2.82 1:0.85:2.94

1:0.85:2.92 1:0.92:2.94 1:0.93:2.97

a EDS measurements have an error of ∼±2 atom %. bXPS measurement have an error of ±0.2 atom % for Cs, ±0.1 atom % for Sn, and ±0.5 atom % for CI, Br, and I.

the composite. It can be seen from Figure 4a that the perovskite sample has the rod-like structure of CsSnI3. As seen in Figure 4a, inset, the HRTEM image of an individual CsSnI3 QR confirms the crystalline structure with observed lattice spacings of 0.62 and 0.64 nm, corresponding to the (220) and (202) planes of perovskite CsSnI3, respectively. In the PL spectra of all perovskite QRs, the PL emission bands are significantly enhanced compared to that of the CsSnX3 QRs, which demonstrates that increasing the atomic weight has effectively passivated the defects on the surface of the CsSnX3.33 Moreover, the PL intensity increases and red shifts with increasing atomic weight. As the atomic

Accordingly, the continuity of the fiber suggests that our CsSnX3 QRs are of high quality. The samples have diameters of about 5 nm (150 QRs), with a diameter deviation of about 15−18%, which indicates the high quality of our samples. Additionally, the typical samples have lengths of several tens of nanometers. Figure 4b−e shows elemental mapping of Cs, Sn, and I in CsSnI3 QRs obtained via TEM energy-dispersive X-ray spectroscopy (EDS). These images suggest the formation of ternary perovskite CsSnI3 5030

DOI: 10.1021/acs.jpclett.6b02344 J. Phys. Chem. Lett. 2016, 7, 5028−5035

Letter

The Journal of Physical Chemistry Letters

Figure 5. (a) J−V characteristics of CsSnI3 (blue), CsSnBr3 (red), and CsSnCl3 (black) solar cells. (b) EQE of CsSnX3 perovskite devices with different halide ratios. The photovoltaic results are summarized in Table 2.

was fabricated in order to further explain the relationship between device performance and halide ratio. From these studies, we conclude that a definite amount of I precursor enhances all of the photovoltaic parameters. In addition, it is well-known that the crystallinity of the perovskite decides the ultimate performance of the constructed devices because defects in the crystals will form shorting and trapping sites for charge recombination. Crystallinity will also very influence the ability of transport, charge separation, and diffusion length.34 The high FF further demonstrates the active charge transfer and extraction of such a planar p−i−n heterojunction.35 This resulted in greater efficiency of the device formed on QRs, which motivated us to explore our solvothermal influence on this facile system. The external quantum efficiency (EQE) plotted in Figure 5b confirms the increased Jsc of these devices, in which the merged Jsc values for the devices derived from lead-free perovskite CsSnI3, CsSnBr3, and CsSnCl3 are 12.96, 10.46, and 9.66%, respectively. As can be seen, the maximum EQE peaks of the top-performing devices can reach over 63% (CsSnCl3), 65% (CsSnBr3), and 70% (CsSnI3) EQE. We study the use of the lead-free all-inorganic QRs as absorber materials in the perovskite solar cells considering that they may improve the cell characteristics due to several reasons: low cost, nontoxic compounds, and inorganic hybrids possessing better chemical effect, photostabilities, and thermal effects than organic composites.36 This is expected to increase the optical density of the active layer and therefore improve light harvesting. Furthermore, the surface areas of QRs are larger compared to bulk, which may allow a larger contact area and hence improve the conversion efficiency. This can further improve the charge separation efficiency, leading to higher performance of the perovskite solar cells. To further investigate charge transport and recombination in perovskite solar cells, impedance spectroscopy and transient photocurrent decay were undertaken. The devices were measured at forward bias between 0.5 and 1.0 V in the dark, and Figure S3 gives the Nyquist curves at 1.0 V bias as an example and the recombination resistance results fitted with an equivalent circuit model. The recombination between electrons in the perovskite layer and hole-transporting material (HTM) has been suggested to play a key role in determining the open-circuit voltage. Transient photocurrent decay is another powerful tool and has also been used to investigate the electron-transport process. The fitted processes are provided in the Supporting Information, and the fitted results are shown in Figure S4. The pore character and the large specific surface area are believed to

QRs consisting of Cs (blue), Sn (green), and I (red) distributed throughout the QR. The quantitative EDS analysis shows a Cs/ Sn/I atomic ratio of around 1:1.1:3.2 in the CsSnI3 QR, which is quite close to the nominal 1:1:3 ratios given by stoichiometry of the precursor solutions. This result makes it clear that the QR is mainly composed of CsSnI3. Similar quantitative results were also obtained from the EDS measurements (SEM) and X-ray photoelectron spectroscopy (XPS) analysis (Table 1). The exemplary morphological and cross-sectional images after spincoating and annealing the samples are shown in Figure S2. According to these SEM images, it should be noted that the grain size increased with annealing process. Figure 5 a shows typical current density−voltage (J−V) curves for lead-free all-inorganic perovskite solar cells with these different halides, which were obtained in the presence of different concentrations of halide under Air mass (AM) 1.5G solar irradiation, and the photovoltaic parameters are summarized in Table 2. The best solar cell performance was measured Table 2. Summarized Photovoltaic Parameters Derived from Current−Voltage Curves Recorded within 24 h (after device fabrication) for the Different Perovskite Devices with Different Halide Ratios Cs/Sn/CI Cs/Sn/Br Cs/Sn/I

atomic ratio

Voc (V)

Jsc (mA cm−2)

FF

PCE (%)

1:1.2:3.5 1:1.2:3.4 1:1.1:3.2

0.87 0.85 0.86

19.82 21.23 23.2

0.56 0.58 0.65

9.66 10.46 12.96

when using the CsSnI3 QR perovskite, and the photovoltaic parameters determined were the short-circuit current density (Jsc) = 23.21 mA cm−2, open-circuit voltage (Voc) = 0.86 V, fill factor (FF) = 0.65, and power conversion efficiency (PCE) = 12.96%. The best solar cell using the CsSnBr3 QR perovskite shows Jsc = 21.12 mA cm−2, Voc = 0.85 V, FF = 0.58, and PCE = 10.46%. The best solar cell using CsSnCl3 QRs shows Jsc = 19.82 mA cm−2, Voc = 0.87 V, FF = 0.52, and PCE = 9.66%. The PCE decreases to 9.66% for the perovskite solar cells with a decrease in the atomic volume by Cl. This decrease is primarily the result of reduced Jsc, Voc, and FF. The IV and PCE values of the CsSnI3 are better than the values measured for the CsSnBr3 and CsSnCl3, demonstrating the benefits of CsSnI3 as an absorber material. Note that the Cs/Sn/I = 1:1.1:3.2 molar ratio in the device was found to be optimum for device performance (see Table 2). A group of devices, with 10−15 devices for each composition case, 5031

DOI: 10.1021/acs.jpclett.6b02344 J. Phys. Chem. Lett. 2016, 7, 5028−5035

Letter

The Journal of Physical Chemistry Letters

Figure 6. (a) Normalized fluorescence decay of reference (MAPbI3) and CsSnI3 QR devices stored in the dark. (b) Normalized fluorescence decay of reference and CsSnI3 QR devices illuminated at certain time intervals. Excitation wavelength at 532 nm using a Nd:YAG ns laser. (c) PL spectra of the CsSnI3 QRs and CH3NH3PbI3 with storage times (1 day and 60 days). (d) TGA curves of CsSnI3 and MAPbI3 under a nitrogen and oxygen atmosphere. Temporal evolution of photovoltaic parameter values (e) VOC, (f) JSC, (g) fill factor (FF), and (h) efficiency for the reference and CsSnI3 QR-based devices.

parameters of the perovskite CsSnI3 as a function of the hybrid composition. MAPbI 3 (for details, see the Supporting Information) and CsSnI3 QR devices were submitted to a long-term degradation process in air, and the decay of their

benefit the deposition of perovskite QRs and therefore highly efficient perovskite solar cells. In order to rule out the performance decay process, we have studied the PL intensity, thermal stability, and photovoltaic 5032

DOI: 10.1021/acs.jpclett.6b02344 J. Phys. Chem. Lett. 2016, 7, 5028−5035

Letter

The Journal of Physical Chemistry Letters fluorescence signal was recorded both under dark storage and during illumination (three pairs of samples for each case). The collected fluorescence spectra are presented in Figure 6a,b, which compares the normalized fluorescence intensity calculated using the peak values of such spectra at 709 nm, following laser excitation at 532 nm, as a function of the dark storage and illumination times. As a result, the fluorescence signal of the MAPbI3 device degrades rapidly upon illumination following a first-order exponential decay. Upon 30 min of illumination, MAPbI3 only emitted very weakly, and the PL almost disappeared upon 120 min of exposure in air. After 2 months of storage in the dark in air, no shift in the PL position was observed for the CsSnI3 QRs, as shown in Figure 6c. This approach led to colloidal stability of the QRs in air for more than 2 months. To compare the thermal stability of CsSnI3 and MAPbI3 quantitatively, we carried out thermogravimetric analysis (TGA) under a nitrogen and oxygen atmosphere (Figure 6d), which shows that the onset of the decomposition temperature of CsSnI3 is significantly higher than that of MAPbI3. In the case of CsSnI3 (onset ∼450 °C), CsSnI3 is more stable (with respect to evaporation or decomposition) than MAPbI3, and in this case, the latter is lost first. In fact, CsSnI3 is somewhat more thermally stable than MAPbI3. The stability of the sealed cells is remarkably improved, as can be seen in Figure 6e−h, which shows the temporal evolution of J−V curves and the extracted photovoltaic parameter (Voc, Jsc, FF, and PCEs included) values from long-term stability tests for more than 240 h. It may also be that the polar organic MA cation makes MAPbI3 more hydrophilic in character than CsSnI3 and thus allows water molecules to permeate faster through the edges of the devices, increasing the decomposition rate. Here, we report such a comparison between the Cs-based and MA-based lead halide perovskites in terms of thermal and optical properties and the corresponding photovoltaic device performance and stability. CsSnX3 (X = Cl, Br, and I) with tunable emission wavelength ranging from 625 to 709 nm has been prepared through a versatile and relatively inexpensive technique of the composite QRs under different halide salt conditions. This work finds that the characteristics of the optical properties in lead-free allinorganic perovskites can be altered through solvothermal without an annealing process, which makes them attractive for novel optoelectronic devices. High-quality lead-free perovskite QRs with composition-tunable emission have been synthesized using a simple solvothermal synthetic approach. Emission wavelengths covering the visible spectrum window can be facilely tuned by variation of the composition of halide salts. Finally, remarkable enhancement in the photovoltaic performance of the CsSnX3 devices to a PCE close to 13% was realized. We strongly believe that this work would contribute toward the fundamental understanding of structural, morphological, and optical properties of lead-free halide perovskite QRs and would be useful for inexpensive nontoxic optical materials, light absorbers for solar cells, or other electric devices, LEDs, and biosensor materials.

Aesar, 99.8%), toluene (Sigma-Aldrich, 99.8%), and hydrochloric acid (Alfa Aesar, 36.5%) were also used. Titanium diisopropoxide bis(acetylacetonate) was purchased from Alfa Aesar. In a typical synthesis, 0.7 g of SnBr2 (2.3 mmol) (or 0.45/0.86 g of SnCl2/SnI2, respectively), 60 mL of ODE, 6 mL of OA, and 6 mL of OLAM were loaded in a 100 mL four-neck flask and dried under vacuum for 1 h at 120 °C. Then, a mixture of 10 mL of ODE with 5 mL of previously synthesized Csoleate (2 g of Cs2CO3 degassed in 60 mL of ODE and 5 mL of OA at 120 °C for 1 h), hydrochloric acid (1 mg), TOPO, and DETA was heated with stirring at 80 °C while purging continuously with nitrogen for 1 h. The reagents were loaded into a 50 mL Teflon-lined autoclave, which was then filled with anhydrous DETA up to 80% of the total volume. The autoclave was sealed and was maintained at 180 °C for 6 h, and then, the QR solution was quickly cooled down to room temperature with an ice bath, and the products were transferred to a glovebox. After this step, the solution was filtrated and washed by excess acetone, distilled water, and toluene by centrifugation. Finally, it was dried in vacuum at 80 °C for 3 h, and the prepared products were redispersed in toluene for further use. To fabricate the all-inorganic CsSnX3 (X = Cl, Br, and I) perovskite hybrid solar cell (indium tin oxide (ITO) /TiO2/ CsSnX3/HTM/Au), ITO glass (15 Ω sq−1) was cleaned using a three-step ultrasonic process of ethanol, distilled water, and acetone for 10 min, respectively. After drying under a nitrogen stream, substrates were further cleaned by UV ozone for 6 min. Subsequently, TiO2 precursor (titanium diisopropoxide bis(acetylacetonate)) was spin-coated onto the ITO substrate at 3500 rpm for 70 s to produce a thickness between 250 and 300 nm and annealed at 350 °C for 1 h in air. To avoid oxygen and moisture, the substrates were transferred into a nitrogen-filled glovebox. Sequentially, precursor solutions of CsSnX3 perovskites (160 mg/mL) in toluene were spin-coated at 6000 rpm for 80 s and dried on a hotplate at 110 °C for 10 min. The spiroOMeTAD HTM solution (10 mg/mL) in toluene was spincoated at 2500 rpm for 40 s on top of the perovskite layer. Finally, gold electrodes were deposited under vacuum (