Solid‐State Ligand‐Exchange Fabrication of CH3NH3PbI3 Capped PbS Quantum Dot Solar Cells

CH3NH3PbI3 capped PbS colloidal quantum dots have been successfully fabricated by solid-state ligand exchange from oleate and oleylamine capped PbS. The optimal solar cells made by layer-by-layer solution deposition give a high power conversion efficiency of 4.25% with an impressive short-circuit photocurrent density of 24.83 mA cm−2. Lead sulfide (PbS) colloidal quantum dots (CQDs), which possess tunable near-infrared absorption,1 multiple exciton generation effect,2 and solution processability,3 have emerged as attractive candidates for thin film photovoltaic applications.4 The surface ligands of CQDs play a crucial role in determining the optoelectronic properties of CQD solar cells.5 Generally, CQDs that bear long-chain organic ligands offer the solubility, yet those ligands unavoidably form insulating barriers of charge transport between CQDs. Therefore, short-chain ligands are employed to shrink the inter-spacing of crystals and reduce surface trap states, both of which favor charge transport.6 Currently, 1,2-ethanedithiol (EDT) and 3-mercaptopropionic acid (MPA) are two of most widely used chalcogenide-based ligands for PbS solar cells,7 which yielded power conversion efficiencies (PCEs) over 3%8 and 8%,9 respectively. Yet the PCEs of PbS-based solar cells seem difficult to further improve because of their several intrinsic obstacles. One is that traditional ligands such as EDT and MPA do not contribute to optical absorbance, thereby limiting short-circuit photocurrent density (Jsc). The other is that the thickness of CQD photovoltaic layers is limited by short carrier diffusion lengths between CQDs, which causes serious charge recombination.10 Thus, novel ligands, for instance, which possess a long carrier diffusion length and a complementary optical spectrum with PbS, are urgently needed to improve CQD solar cell performance. Recent years have witnessed unprecedented development of organohalide perovskites (CH3NH3PbX3, X = I, Br, Cl) in thin film solar cells,11-16 owing to their magnificent properties such as long hole and electron diffusion lengths,17, 18 broad visible absorption range,19 and high charge mobility.20, 21 More importantly, CH3NH3PbI3 shows perfect lattice matching with PbS CQDs.22 In this regard, CH3NH3PbX3 can be considered as a promising ligand for PbS. Very recently, CH3NH3PbI3 perovskites were successfully used as stable capping ligands for a range of CQDs—PbS, CdS, InP, and CdSe—via solution ligand-exchange reaction, resulting in efficient electronic passivation for highly luminescent CQDs.23 More recently, PbS-in-perovskite solids were fabricated through in situ epitaxial growth process in solution phase to attain excellent carrier transport between perovskites and PbS.22 Most recently, such PbS-in-perovskite solids were combined with EDT capped PbS (namely, PbS−EDT) CQDs and then applied in hybrid nanostructured solar cells, obtaining a high PCE of 8.95%.24 However, all these PbSCH3NH3PbI3 CQDs were synthesized through solution-phase ligand exchange and their solar cells involved the fabrication of multiple-layered device structures. Herein, we report a solid-state ligand exchange method to obtain PbS CQDs with CH3NH3PbI3 capping ligands (namely, PbSCH3NH3PbI3), followed by a layer-by-layer (LbL) solution deposition method to fabricate relatively thick film with controllable thickness as the active layer in solar cells. A highly volatile solvent, acetonitrile, is used for CH3NH3PbI3 in successive steps of ligand exchange, spin-coating, and solvent rinsing. Such solvent strategy of using acetonitrile is crucial to obtain smooth, continuous, and thick films, accompanied by complete replacement of original ligands and removal of extra CH3NH3PbI3. The resulting PbSCH3NH3PbI3 exhibits a broadened absorption spectrum in both visible and near-infrared wavelengths, and efficient charge transfer between quantum dots and perovskites. Moreover, cascade energy alignment can be achieved among PbS, CH3NH3PbI3, and electron-selective layer (TiO2) to facilitate charge transport. The optimal solar cell reaches a PCE of 4.25% under AM 1.5 G light irradiation with an impressive JSC of 24.83 mA cm−2. This PCE is remarkably higher than those reported values of traditional EDT-capped PbS solar cells (≈3%).7, 8, 25, 26 PbS CQDs were first synthesized by using lead(II) oleate with bis(trimethylsilyl) sulfide, and the original capping ligands are oleate and oleylamine.1 Prior to solid-state ligand exchange, the CQDs were dispersed in octane solvent. As-synthesized PbS CQDs were structurally characterized by transmission electron microscope (TEM) and ultraviolet-visible-near-infrared (UV–vis–NIR) spectrophotometer. As displayed in TEM image (Figure 1a), uniform-sized PbS are obtained with an average diameter of 2.5 nm and a lattice fringe of 3.5 Å. Moreover, energy dispersive X-ray spectroscopy (EDX) spectrum (Figure S1 in the Supporting Information) shows that the atomic percentage of lead to sulfur in PbS is nearly 1:1. Figure 1b displays the optical absorption spectrum of PbS in octane solution, which exhibits a near-infrared peak at 824 nm, giving an optical bandgap of ≈1.50 eV. Cyclic voltammetry (CV) was further applied to determine the energy levels of PbS. Figure S2 (Supporting Information) shows that the valence and conduction bands of PbS are −5.19 and −3.65 eV, respectively, yielding an electrochemical bandgap of 1.54 eV, which is very close to the optical bandgap. Next, perovskite CH3NH3PbI3 was introduced as a replacing ligand of PbS CQDs via solid-state ligand exchange. Scheme 1 outlines the fabrication process of the PbSCH3NH3PbI3. First, the PbS with mixed ligands of oleate and oleylamine was spin-coated from the octane solution (20 mg mL−1) onto glass substrate. Second, a saturated CH3NH3PbI3 solution in acetonitrile was drop-casted onto the entire PbS layer and remained for 1 min to achieve complete ligand exchange (Scheme 1a). Spin-coating was then applied to remove the excess solution (Scheme 1b). Lastly, two consecutive rinsing steps were applied with acetonitrile and octane to remove extra perovskites and organic ligands, each followed by spin-coating. These three steps were repeated several times through LbL methods to obtain the proper thickness of brown and smooth films. Note that we have initially attempted to use common solvents such as N,N-dimethylformamide (DMF) and N-methylformamide (MFA) for preparing CH3NH3PbI3 solution; yet owing to their low volatility and high solubility for CH3NH3PbI3, both solvents tend to redissolve the majority of PbSCH3NH3PbI3 in the fabrication process, yielding cracked and discontinuous films. Such ligand-exchange process of making PbSCH3NH3PbI3 CQDs was fully characterized, as shown in Figure 2. Fourier-transform infrared (FTIR) spectroscopy was first used to confirm the success of ligand exchange process. As shown in Figure 2a, oleate and oleylamine capped PbS CQDs display strong symmetrical and asymmetrical CO2− vibration bands at 1405 and 1548 cm−1, and C−H stretching peaks at 2854 and 2926 cm−1, respectively, both of which are characteristic of the carboxylate functional group in oleic acid.23 By contrast, these CO2− and CH peaks are absent in the PbSCH3NH3PbI3 when the loading time of PbS film in CH3NH3PbI3 solution is 1 min. This indicates that perovskite has successfully replaced the original mixed ligands. TEM image was further characterized to ensure the formation of PbSCH3NH3PbI3 CQDs as shown in Figure S3 in the Supporting Information. After ligand exchange, the size and lattice fringe of PbSCH3NH3PbI3 CQDs remain nearly the same. Meanwhile, with the aid of CH3NH3PbI3 ligands, PbS CQDs tend to inter-connect and form a continuous solid matrix. X-ray diffraction (XRD) measurement was then conducted to determine the crystalline structures of PbSCH3NH3PbI3. As shown in Figure 2b, neat PbS CQDs have no obvious diffraction peaks due to low crystallization of 2.5 nm CQDs, while neat CH3NH3PbI3 film spin-casted from acetonitrile solution shows strong peaks at 14.02°, 28.16°, 37.80°, and 44.02°, corresponding to the reflections from (110), (220), (310), and (330) lattice planes of typical CH3NH3PbI3 perovskite structure, respectively.27, 28 The PbSCH3NH3PbI3 CQDs display strong characteristic peaks of CH3NH3PbI3 at 37.80° and 44.02°, further confirming the existence of CH3NH3PbI3. However, the intensity ratio of (310) and (330) peaks in PbSCH3NH3PbI3 is lower than that of neat CH3NH3PbI3, implying different crystalline structures of perovskite caused by the interaction between CH3NH3PbI3 and PbS. Note that there is no diffraction peak assigned to PbI2 in both neat CH3NH3PbI3 and PbSCH3NH3PbI3 CQDs, which confirms the presence of CH3NH3PbI3 ligand. The XRD results are further supported by the grazing-incident wide-angle X-ray scattering (GIWAXS) measurement (Figure S4 in the Supporting Information), showing that all the characteristic peaks of both PbS and CH3NH3PbI3 retain in PbSCH3NH3PbI3 CQDs with enhanced crystallinity. X-ray photoelectron spectroscopy (XPS) was subsequently applied to further ensure the presence of both PbS and CH3NH3PbI3 in the final film after the solid-state ligand exchange (Figure S5 in the Supporting Information). Through the elemental analysis of XPS, it is estimated that the molar ratio of CH3NH3PbI3 ligand to PbS nanocrystals is about 1:3.7 after ligand exchange, in good agreement with that of solution-phase ligand exchange of halometallate-capped CQDs.23 The XPS results are also consistent with those of PbS-in-perovskite solids by in situ epitaxial growth process.22 Improved optical and electronic properties are anticipated in the PbSCH3NH3PbI3 CQDs. As shown in Figure 2c, PbS CQDs exhibit a wide but low absorption range of 300−1000 nm, while CH3NH3PbI3 displays a strong but narrow absorption range from 300 to 760 nm. As expected, PbSCH3NH3PbI3 indeed shows a broader and overall stronger absorption spectrum than neat films of CH3NH3PbI3 and PbS, respectively. Complementary optical absorption are thus achieved in such PbSCH3NH3PbI3 in which the absorbance of CH3NH3PbI3 mainly contributes in the visible wavelength while PbS in the near-infrared wavelength. Photoluminescence (PL) and electrical conductivity measurements were then combined to study the charge transport between the CH3NH3PbI3 ligand and PbS. As shown in Figure 2d, a high PL quenching efficiency of ≈94% is found in the PbS−CH3NH3PbI3 compared to neat CH3NH3PbI3 film, indicative of effective charge transfer between PbS and CH3NH3PbI3. In addition, the emission peak of PbSCH3NH3PbI3 blueshifts 6 nm than that of CH3NH3PbI3, suggesting the strong interaction between CH3NH3PbI3 and PbS, which is consistent with the XRD results. Moreover, electrical conductivity of neat PbS CQDs having mixed ligands of oleate and oleylamine is measured as only 5.33 × 10−11 S cm−1, while PbSCH3NH3PbI3 CQDs show an improved conductivity of 1.88 × 10−9 S cm−1 by two orders of magnitude (Figure S6 in the Supporting Information). This is presumably attributable to enhanced charge transport of perovskite ligands and decreased inter-particle spacing between PbS. Lastly, PbSCH3NH3PbI3 CQDs, which exhibit high conductivity and broad absorption ranging from UV to NIR wavelength, were investigated in CQDs solar cells as shown in Figure 3. The device configuration of SnO2:F (FTO)/compact TiO2/several layers of PbSCH3NH3PbI3/MoO3/Au and corresponding energy bandgap diagram are displayed in Figure 3a,b, respectively. Both the valence and conduction bands of CH3NH3PbI3 are lower than those of PbS, forming a suitable p–n heterojunction to facilitate charge separation. Additionally, it has proven that photoexcited electrons could transfer efficiently from quantum dots to TiO2 only for quantum-dot diameter below 4.3 nm.29 In our work, the average diameter of PbS CQDs is 2.5 nm, indicating that charge transfer occurs between PbS and TiO2. CH3NH3PbI3 ligands can also act as an energy relay to form the cascade energy alignment between PbS and TiO2, which assists in reducing energy loss and facilitating charge transport. The number of the PbSCH3NH3PbI3 layers on the TiO2 is anticipated to greatly influence the photovoltaic performance. Thin active layer shows poor light harvesting, while if the active layer is too thick, serious recombination would occur because of the limited carrier diffusion length of photogenerated charge carriers. The photocurrent density versus voltage characteristics (J−V) of the optimal PbSCH3NH3PbI3-based solar cells with different layer numbers under AM1.5G irradiation are shown in Figure 3c and summarized in Table 1. The average values of photovoltaic parameters with a standard deviation of 45 individual devices are presented in Table S1 in the Supporting Information. Note that the PCE of solar cells fabricated from either single layer of neat CH3NH3PbI3 or pure PbS CQDs approaches zero because of the low concentration of CH3NH3PbI3 in acetonitrile and the existence of insulating ligands in PbS, respectively. As shown in Figure 3c and Table 1, among the devices with active layer numbers of 3, 5, 7, and 10, respectively, the five-layer device exhibits an optimum PCE of 4.25% with a Jsc of 24.83 mA cm−2, an open-circuit voltage (Voc) of 0.45 V, and a fill factor (FF) of 38%. Note that this PCE is significantly higher than those reported values of traditional EDT-capped PbS solar cells (≈3%).7, 8, 24, 25 It is also notably higher than those of EDT capped PbS/CH3NH3PbI3 core/shell QDs sensitized solar cells (3.2%)30 and CH3NH3PbI3/oleic acid capped PbS bilayer heterojunction photovoltaic cells (3.6%),31 respectively. The thickness of 5-layer PbSCH3NH3PbI3 is measured as ≈350 nm from cross-sectional SEM image in Figure 3a. It is thus plausible that deposition of five layers is a reasonable compromise to satisfy both the light absorption and carrier collection. The Voc lies between 0.4 and 0.5 V, irrespective of the number of deposition layers, suggesting that the band structure of PbSCH3NH3PbI3 solar cells is nearly identical to that of traditional PbS CQD solar cells.7 Moreover, the Jsc is among the highest values of PbS CQD solar cells,5, 9 which is presumably caused by complementary optical absorption, improved electrical conductivity, and efficient charge separation as a result of the forming p–n junctions between PbS and CH3NH3PbI3. Such combined benefits overcome the issues of insulating ligands of EDT and oleic acid in literature reports,7, 8, 25, 26, 30, 31 which largely hindered charge transport between PbS and CH3NH3PbI3. Also, owing to the relatively low concentration of CH3NH3PbI3 ligand in PbSCH3NH3PbI3 CQDs as indicated by the XPS results, the resulting JSC is primarily contributed by PbS CQDs and further enhanced by CH3NH3PbI3 ligands in visible light wavelength. The remarkably high Jsc is further supported by the external quantum efficiency (EQE) curve in Figure S7 in the Supporting Information. The maximum EQE of the optimal five-layered device reaches 83% at 350 nm, and the wavelength covers broadly from 300 to 1000 nm. The photocurrent density calculated from the EQE spectrum is lower than JSC possibly due to the trap states caused by the defects in TiO2 layer, which can be however filled by additional light soaking during J−V measurement. Finally, we anticipate that the relatively low FF can be further improved by interfacial modification, substitution of transporting layer, and application of pyramid-patterned substrates/electrodes.5, 9 Recent studies showed an effective strategy of improving FF and PCE in the PbS CQD solar cells by inserting a thin top layer of PbS−EDT.24 We therefore added two layers of p-type PbS−EDT CQDs (≈50 nm)24 onto the optimal five layers of n-type PbS−CH3NH3PbI3 CQDs (≈300 nm) in solar cells to create graded band structures. As shown in Figure 3c and Table 1, the resulting bilayer devices reach a remarkably high PCE of 5.28% compared to single-layer PbSCH3NH3PbI3 cells. Such improvement is mainly reflected in enhanced VOC (0.6 V) and increased FF (40%), although slightly lowering the JSC. To interpret it, we attempted to investigate the influence of the PbS−EDT layer on transport dynamics by time-of-flight (ToF) measurements, as shown in Figure S8 in the Supporting Information. Without PbS−EDT layer, the hole and electron mobilities are 8.04 × 10−5 and 2.68 × 10−5 cm2 Vs−1, respectively. The low and imbalanced charge transport explains the low FF obtained in the single-layer device. After the addition of PbS−EDT layer, the electron mobility is increased by an order of magnitude to 1.04 × 10−4 cm2 Vs−1 while the hole mobility is slightly increased to 9.91 × 10−5 cm2 Vs−1. It can be thus seen that the PbS−EDT layer enables more balanced and efficient charge transport in the device. In conclusion, we have successfully fabricated CH3NH3PbI3 perovskite capped PbS CQDs by solid-state ligand exchange method. By using acetonitrile solvent in this process, original mixed ligands of oleate and oleylamine have been fully replaced by CH3NH3PbI3. Compared to original PbS CQDs, the PbSCH3NH3PbI3 CQDs exhibited complementary optical absorption spectra, facile charge separation, increased conductivities, and efficient charge transport. Through layer-by-layer process, PbSCH3NH3PbI3 CQDs-based inverted solar cells with various deposition layers were made. The best solar cell gave a high PCE of 4.25% with an impressive Jsc of 24.83 mA cm−2. Furthermore, by inserting a top layer of EDT-capped PbS CQDs, the resulting bilayer solar cells reached a significantly enhanced PCE of 5.28%, which is ascribed to more balanced and efficient charge transport in the device. This work was supported by National Natural Science Foundation of China (NSFC) under Grant No. 51473036. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.