Efficient All-Inorganic CsPbBr3 Perovskite Solar Cells by Using CdS/CdSe/CdS Quantum Dots as Intermediate Layers

Highly efficient all-inorganic perovskite solar cells require a fast charge transfer from CsPbBr3 to TiO2 to reduce the recombination from trap states. Herein, we insert a CdS/CdSe/CdS quantum dot (QD) layer between the TiO2 and CsPbBr3 layers to fabricate allinorganic perovskite solar cells. By tuning the thicknesses of the CdSe layer of CdS/CdSe/CdS QDs, the conduction band (CB) levels can be adjusted to -3.72~-3.87 eV. After inserting the QD intermediate layer, the energy offset between the CB of TiO2 and CsPbBr3 is reduced, thus leading to a charge transfer rate boost from 0:040 × 109 to 0:059 × 109 s−1. The power conversion efficiency (PCE) of the solar cell with QD intermediate layer achieves 8.64%, which is 20% higher than its counterpart without QDs.


Introduction
Hybrid organic-inorganic lead halide perovskite solar cells (PSCs) are promising candidates for commercialization, owing to their extremely high power conversion efficiency (PCE over 22%) and low fabrication cost. Despite recent rapid progress in developing the novel perovskite materials or optimizing the structure of the hybrid PSCs, the unsatisfactory long-term stability of the PSCs and the high cost of the hole-transporting materials hinder their commercialization [1][2][3]. All-inorganic PSCs are regarded as one of the most hopeful platforms to tackle this issue [4]. As light absorbers, inorganic perovskites such as CsPbX 3 (X=I, Br) have high charge-carrier mobility, which is similar to hybrid perovskites [5][6][7]. Moreover, inorganic lead halides are capable of against humidity and heat, which guarantee their long stability [4,[8][9][10]. Besides, the light-induced halide segregation that commonly happens in hybrid perovskites does not occur in such all-inorganic PSCs because ions need to overcome higher energy barrier to migrate compared with hybrid ones. Therefore, growing efforts are devoted to developing all-inorganic PSCs, aiming at resolving the challenging faced by hybrid ones.
Hodes et al. first used CsPbBr 3 as the light absorber with a typical device structure of FTO/c-TiO 2 /m-TiO 2 /CsPbBr 3 /HT-M/Au, achieving a PCE of 5.95% and an open-circuit voltage (V oc ) of 1.28 V [11]. These parameters are comparable to those MA-containing hybrid PSCs. Later, Liu and his coworkers fabricated an all-inorganic PSC with architecture of FTO/cTiO 2 /m-TiO 2 /CsPbBr 3 /carbon in ambient environment without humidity control, reaching a PCE of 6.7% and good tolerance in humid air and extreme temperatures [5]. Tang's group further boosted the PCEs of CsPbBr 3 allinorganic PSCs from 6.7% to 10.6% by using interfacial or compositional engineering [12][13][14]. Despite these achievements, the charge-carrier trap states within the CsPbBr 3 film and the interface still impede the further improvements of the device performances [13,15]. Therefore, constructing an appropriate interface and enhancing the charge transfer may work in promoting the charge extraction efficiency and passivating the defects.
Quantum dots (QDs) are nanocrystals of several nanometers in diameter, and they are capable of tuning the optical and electronic properties by adjusting the size or composition [16][17][18]. Evidence has shown that modifying the energy level of QDs has significant impact on interface charge transfer [15,19,20]. Instead of pursuing proper intermediate buffer layers, we intend to design QDs with proper energy band structures to make better energy level alignment between CsPbBr 3 and TiO 2 in PSCs because of the featured energy level tunability of the QDs. Many work shows that reducing the energy offset between the donor and the acceptor can accelerate the charge transfer at the interface [21][22][23][24][25]. Herein, we design core/multishell QDs with the conduction band (CB) ranging in -3.72~-3.87 eV. By inserting the QD layer between the TiO 2 and CsPbBr 3 , we construct a cascade energy alignment, where photoexcited electrons can be extracted efficiently. Arising from the improved charge transfer rate, the optimized device exhibits a PCE of 8.64%, which is 20% higher than the device without QD intermediate layer.

Preparation of Precursors.
Cationic precursors (0.1 M cadmium oleate (Cd(OA) 2 ) and 0.5 M cadmium oleate (Cd(OA) 2 )) were prepared by dissolving CdO (4 mmol or 20 mmol) in a mixed solvent of OA (20 mL) and 1-ODE (20 mL) under N 2 atmosphere at 250°C for 30 min, degassing at 150°C for 30 min, and filling with N 2 . Anionic precursors (0.1 M TOP-Se) were prepared by dissolving the selenium power (1 mmol) in 10 mL of TOP under N 2 atmosphere at room temperature.
2.3. Synthesis of CdS Seeds. First, 38.5 mg CdO (0.3 mmol) was loaded in a three-neck flask containing 1 mL of OA and 8 mL of 1-ODE at 150°C under N 2 atmosphere. Then the mixture solution was heated to 270°C, and 0.25 M S-ODE (0.5 mL) was rapidly injected into the reaction flask and stayed for 8 min. The resultant nanocrystals were purified twice by the precipitation (ethanol) and redispersion (toluene) method.
2.4. Synthesis of CdS/CdSe/CdS Nanocrystals. For CdSe emissive layer growth, the desired amount of mixed Cd and Se precursors (0.1 M Cd(OA) 2 and 0.1 M TOP-Se) was injected into the three-neck flask containing CdS seeds (0.1 g) and 1-ODE (8 mL) at 300°C at a rate of 10 mL/hr. The obtained CdS/CdSe NCs do not require any purification. For CdS shell growth, the desired amount of 0.5 M Cd(OA) 2 and 1 M DDT-ODE were injected separately into the reaction solution at 300°C at a rate of 4 mL/hr. The reaction temperature was maintained at 300°C for the entire CdS shelling process and cooled down to room temperature to stop the reaction. The resultant CdS/CdSe/CdS NCs were purified twice.

Fabrication of All-Inorganic CsPbBr 3 Solar Cells.
Allinorganic CsPbBr 3 solar cells were prepared according to the previously reported method with some modifications [26]. Typically, 90 μL of CdS/CdSe/CdS solution in toluene was spin-coated onto the FTO/c-TiO 2 /m-TiO 2 layer at 2000 rpm for 30 s and heated at 80°C for 10 min. Subsequently, an N,N-dimethylformamide (DMF) solution of 1.0 M PbBr 2 was spin-coated onto the CdS/CdSe/CdS layer at 2000 rpm for 30 s, followed by drying at 80°C for 60 min. Then the 0.07 M CsBr methanol solution was spin-coated onto the PbBr 2 film at 2000 rpm for 30 s and continuingly heated at 250°C for 5 min. This process was repeated for several times. Finally, the carbon electrode served as both hole transport layer (HTL), and the counter electrode was deposited on the CsPbBr 3 layer by doctor-blade coating of conductive carbon ink and then heated at 70°C for 60 min. CdS is used as core and a monolayer or multilayer of CdSe grows on the CdS core, followed by growing a thick CdS shell. Both the CB and VB levels of the CdSe are contained between the CB and valence band (VB) of the CdS core and shell. Therefore, the CdS/CdSe/CdS QDs possess a spherical quantum well structure, where the electrons and holes are strongly confined in the CdSe layer. XRD results of the CdS, CdS/CdSe, and CdS/CdSe/CdS QDs are shown in Figure 1(b). The CdS exhibit characteristic peaks at 26.5°, 43.9°, and 51.9°, which are consistent with diffraction angles of the (111, 220, 311) of the standard zinc blende card (JCPDS 01-0647). After coated with CdSe shell, the resultant CdS/CdSe also shows three typical zinc blende diffraction peaks located at 25.61°, 42.89°, and 49.9°with a little shift to those of the CdS core. These XRD results confirm that CdSe shelling does not affect the zinc structure of the CdS just undergoing an epitaxial shell growth of CdSe. However, the growth of the second CdS shell on the zinc blende CdS/CdSe generates a wurtzite crystal structure featured with diffraction peaks at 24.6°, 26.3°, 28.4°, 43.6°, 47.0°, and 52.3°. The presence of the wurtzite structure in this QD is in accordance with some related core/shell systems containing both the wurtzite and zinc blende crystal structures [27]. Transmis Although changing the size of the CdS core or the thickness of CdS can affect the energy structure of the QDs, controlling the thickness of the CdSe offers much wider 2

Results and Discussion
Journal of Nanomaterials 3 Journal of Nanomaterials tenability on the energy levels. CdS has a higher CB and a lower VB than those of CdSe. Therefore, in the CdS/CdSe system, both the hole and electron charges are partially delocalized on the shell materials. Theoretical and experimental results have demonstrated that shell thickness has an influence on the band energy structure of this inverted type I system [28][29][30]. There is a competition between the kinetic energy and the potential energy in this system. With shell thickness increases, both the electron and the hole wave functions in the center of the particle are increased, while keeping the particle size constant [28]. Hence, the bandgap structure can be adjusted by changing the thickness of CdSe shell. Furthermore, type I and reverse type I QDs usually encounter the predicament of low PL QYs, and type II ones encounter low photooxidation stability. However, QDs with multilayer of shells exhibit both these two high properties [30,31]. Therefore, a CdS layer was introduced upon the CdS/CdSe QDs to realize the passivation of surface defects. In short, by controlling the thickness of CdSe layer, the CB of CdS/CdSe/CdS QDs can be tuned.
For core/shell QDs, shell usually have a nonnegligible impact on the energy structure, so as to the optical properties of the dots. Many II-IV and I-III-VI core/shell QDs have been synthesized to enhance the stability and the photoluminescence (PL) quantum yields (QYs) of pure cores or to restrain the dot-dot FRET that is beneficial to the improvement of device performance [32][33][34]. However, there exists a pronounced trade-off between the shell thickness and the PL QYs in the CdSe/CdS system, owing to the traps originated from the exceeded epitaxy of shells. Hence, appropriate shell layers are generally required.  Journal of Nanomaterials  [36]. This prompts us to employ QD as intermediate layer to reduce the "excess" energy offset between TiO 2 and CsPbBr 3 . Therefore, we performed CV tests to determine the energy level of the prepared QDs with respect to the vacuum level. Figure 2(d) shows obvious oxidation and reduction waves for all QD films,     Journal of Nanomaterials (-3.3 eV) and the CB of TiO 2 (-4.2 eV). We expect that inserting the QD layer between CsPbBr 3 and TiO 2 will reduce the excess energy offset and benefit the electron extraction.
To explore the influences of the QD layer on the charge transfer properties of the solar cell, steady state PL and time-resolved PL measurements were performed. Comparative analysis on CsPbBr 3 /QDs/TiO 2 against CsPbBr 3 /TiO 2 enables us identify the role of QDs that plays on the interface charge transfer. Figure 3(a) shows that the PL intensity of the CsPbBr 3 decreases notably when interfaced with TiO 2 , but inserting the QDs between CsPbBr 3 and TiO 2 can further quench the PL intensity. The degree of the PL intensity quenching is quite sensitive to the charge transfer from CsPbBr 3 to TiO 2 . Such a high degree of PL quenching indicates the strong charge transfer of the CsPbBr 3 /QDs/TiO 2 . To evaluate the exact charge transfer rate at the interface, time-resolved PL measurements were performed, and the results are shown in Figure 3(b). Interfaced with the TiO 2 layer, the emission decays dramatically. An additional intermediate QD layer results in a fastest decay as compared with the neat CsPbBr 3 and CsPbBr 3 /TiO 2 blend. Analyzing the slope of the decay curves, we found that the biexponential function is satisfactory to fit the decays [37]: where signal IðtÞ is the convolution of the biexponential function and the impulse response function (IRF) FðtÞ, a 1 and a 2 are the amplitudes, and τ 1 and τ 2 are lifetimes. Then the average lifetimes are calculated by Comparing the average lifetime of the CsPbBr 3 , CsPbBr 3 /TiO 2 , and CsPbBr 3 /QD/TiO 2 blend films, we can evaluate the interface charge transfer rate k ct by using All the kinetic parameters are listed in Table 2. The average decay time t of the neat CsPbBr 3 , CsPbBr 3 /TiO 2 , and CsPbBr 3 /QDs/TiO 2 is 12.4, 8.3, and 7.2 ns, respectively. Strikingly, the charge transfer rate of CsPbBr 3 /QDs/TiO 2 is 0:059 × 10 9 s −1 , which is over 40% faster than that of CsPbBr 3 /TiO 2 . Figure 4(a) illustrates the device architecture of the allinorganic PSC with FTO/c-TiO 2 /m-TiO 2 /QDs/CsPbBr 3 /carbon configuration. Unlike other PSCs that need holetransporting layers (HTLs), our all-inorganic PSCs are    Figure 4(b), the CB levels of the QDs range in -3.72~-3.87 eV, which locate above the CB of TiO 2 (-4.2 eV), implying that the QDs are electronically active for electron transfer. Inserting the QD layer also creates a cascade energy level architecture where the energy offset is lowered, which may accelerate the charge transfer process and direct the charge-carriers to circuit. Figure 5 shows the current density-voltage (J − V) curves of the all-inorganic PSCs under the illumination of the simulated sunlight (AM1.5G). The photovoltaic parameters of the PSC with different QDs are summarized in Table 3, including open-circuit voltage (V oc ), short-circuit current density (J sc ), and filling factor (FF). The pristine CsPbBr 3 PSC delivers a moderate PCE of 7.15% (V oc = 1:368 V, J sc = 6:69 mA cm −2 , and FF = 0:78), which is comparable to other HTL-free inorganic PSCs [8,15]. After inserting the QDs as intermediate layers, all the devices yield higher J sc , V oc , and PCE than that of the reference device. The underlying mechanism of this enhancement is that the cascade energy level is formed by inserting QDs with appropriate CB level between the CB of TiO 2 and CsPbBr 3 , thus offering fast charge transfer channel to gain efficient charge extraction. As a consequence, the champion PSC with QDs shows a PCE of 8.64%, which is 20% higher than that of pristine CsPbBr 3 PSC.

Conclusion
In summary, we have synthesized CdS/CdSe/CdS QDs with tunable energy levels and used them as intermediate layers to enhance the performances of the inorganic PSCs. Evidence show that the CB level lies between the CB of TiO 2 and the CB of CsPbBr 3 , which is suitable for transporting electrons. After inserting the QD intermediate layer, the CB energy offset between TiO 2 and CsPbBr 3 is reduced, thus leading to charge transfer rates boosting from 0:040 × 10 9 to 0:059 × 10 9 s −1 . The optimized PSC shows a PCE as high as 8.64%, which is over 20% higher than 7.15% for the pristine device. This work offers potential route to design HTL-free inorganic PSCs with efficient charge transfer.

Data Availability
Findings of this research work will be provided from corresponding author on reasonable demand.

Conflicts of Interest
The authors declare no conflicts of interest.