**3. Results**

Figure 1a shows the schematic of the PSCs with a standard inverted structure of ITO/PEDOT:PSS/Perovskite/PCBM/BCP/Ag. The Pb(Ac)2-processed perovskite films (with and without cesium doping) were sandwiched between a poly(3,4- ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) hole transport layer and a 6,6-phenyl-C61- butyric acid methyl ester (PC61BM) electron transport layer. The J–V curves of the PSCs with 0–7.5 mol% cesium doping are exhibited in Figure 1b, with the photovoltaic parameters listed in Table 1. The reference PSCs without cesium doping had an average PCE of 14.1%, which is a standard value for Pb(Ac)2-processed inverted PSCs. When the perovskite was doped with 2.5 mol% cesium, the PCE increased to 15.04%. With 5 mol% cesium doping, the PCE further increased to 15.57%, which was mainly induced by the significant improvements in Jsc (from 20.16 to 21.08 mA cm<sup>−</sup>2) and FF (from 0.69 to 0.75). Doping with 7.5 mol% cesium degraded the PCE to 15.37%, indicating that 5% is the optimum cesium doping concentration for maximizing the PCE. As shown in Figure S1, we measured the J–V curves scanned in the reverse and forward directions at a scan rate of 200 mV s<sup>−</sup>1. The J–V curve of the reverse scan was almost the same as that of forward scan, indicating a negligible hysteresis of the device. To investigate the hysteresis deeply, dynamic J–V scans with calculation of the hysteresis index [17] are required, which is beyond the scope of this study.

**Figure 1.** (**a**) Perovskite solar cell structure. (**b**) Current density–voltage (J–V) characteristics of the MAPbI3 PSCs with different cesium doping concentration.


**Table 1.** Average photovoltaic parameters of the MAPbI3 PSCs based on perovskite precursors with 0 mol%, 2.5 mol%, 5.0 mol%, and 7.5 mol% cesium doping.

To identify the origin of the improved PCE by cesium doping, we investigated the morphology of the 5%-Cs-doped and undoped perovskite films. As is evident in the surface SEM images in Figure 2a, the undoped perovskite film had poor surface coverage with many pinholes. A perovskite layer processed from a Pb(Ac)2-based precursor showed a similar surface morphology with some flaws, which might be caused by MA and halide deficiencies as shown in [13]. Figure 2a,d show that with increasing cesium doping concentration, the coverage of the perovskite layer onto PEDOT:PSS increased. As shown in Figure 2c, the 5 mol%-cesium-doped perovskite film had a dense and uniform morphology with full surface coverage. AFM images, shown in Figure S2, further confirm the increased surface uniformity with cesium doping; the root mean square roughness of the cesiumdoped perovskite film is 8.6 nm, which is much lower than that of the pristine perovskite (14.6 nm). Furthermore, we found that the cesium-doped perovskite film showed less lateral grain boundaries compared to the pristine perovskite film. As discussed in previous studies, the pinholes in the perovskite film trap carriers, which further increase the charge recombination in the PSCs [18]. The SEM and AFM measurements indicate an improved morphology of the perovskite film upon cesium doping, which explains the PCE improvement, where improved perovskite seeding may be induced by the cesium addition [16]. These seeds might later turn into nucleation sites for further growth of perovskite during crystallization, which results in denser grains [16]. A similar process was found by Li et al. where MAI-modified PbS nanoparticles behaved as growth seeds for highly compact perovskite films [19]. To prove this mechanism, we characterized the crystallinity of the pristine and cesium-doped perovskite films.

**Figure 2.** Top-view SEM images of perovskite films with (**a**) 0, (**b**) 2.5, (**c**) 5, and (**d**) 7.5 mol% cesium doping concentration.

Figure 3a compares the XRD spectra of the cesium-doped and pristine perovskite films. All peaks in the XRD patterns show the presence of the CH3NH3PbI3 tetragonal crystal structure [20]. It can be seen that the intensity of the (110) peak at 14◦ of the cesiumdoped perovskite film is higher than that of the undoped one. Moreover, the peak at about 12◦, which relates to the (001) lattice planes of hexagonal PbI2, is dramatically reduced with cesium doping. This indicates that the decomposition of the perovskite to PbI2 was suppressed by cesium doping [21,22]. In the UV–vis light absorption (Figure 3b), a small blue shift can be observed with cesium doping, indicating a slightly increased optical bandgap, in good agreemen<sup>t</sup> with previous studies [23].

**Figure 3.** (**a**) XRD patterns. (**b**) Ultraviolet–visible absorption spectra of the perovskite films with different cesium doping concentrations. (**c**) Nyquist plots of the PSCs without and with 5% cesium doping with a bias of 0.8 V. (**d**) Photoluminescence (PL) spectra of the perovskite films without and with 5 mol% cesium doping.

We also conducted EIS for the PSCs under one sun illumination to obtain the resistance information upon cesium doping. Figure 3c shows the Nyquist plots that are fitted with the equivalent circuit, which is shown in the inset. After fitting, the series resistance (Rs), charge recombination resistance (Rct), and chemical capacitance (Cct) of the films could be obtained and the values of them are listed in Table S1. The Rs value for the case with 5 mol% Cs (60.5 Ω) is 33.0% lower than that without Cs doping (90.3 Ω), which contributes to the enhancement of Jsc and FF. The Cct values, which are associated with the densities of space charges at the interfaces, are similar for the cases with and without 5% Cs doping (2.9 × 10−<sup>9</sup> F and 3.0 × 10−<sup>9</sup> F, respectively). The Rct of 5% Cs-doped sample (3198 Ω) is lower than that of undoped sample (6653 Ω). Because the lower Rct indicates the larger electron recombination at the interfaces, the Rct values predict the higher leakage current and the lower Jsc for the Cs-doped samples. However, our experimental results show that the doping of 5% Cs reduces the leakage current, as explained below, and increases the Jsc (see Figure 4b). The experimental results of previous studies are also controversial; some studies reported that Jsc increases with Rct [24–26], while other studies reported the

increase in Rct reduced Jsc [27–29]. We speculate that Rct in our circuit model may not correctly represent the recombination resistance; Rct in the circuit model was extracted from a high frequency impedance semicircle, whereas some previous studies insisted that Rct is related to both high and low frequency semicircles [30–32]. Further studies using more sophisticated circuit models are required to obtain the more accurate Rct. Figure 3d shows the PL spectra for the 5 mol%-Cs-doped and undoped perovskite films on glass substrates. Evidently, the PL peak of the cesium-doped perovskite film was slightly blueshifted to 756 nm (the PL peak of the pristine perovskite film is at 760 nm), which is consistent with the UV–vis absorption results in Figure 3d. The intensity of the PL peak of the cesium-doped perovskite film is 24% higher than that of the pristine perovskite film, which indicates decreased surface-trap states (related to non-radiative PL recombination) and increased perovskite crystallinity (consistent with the SEM results) [33].

**Figure 4.** (**a**) Dark J−V characteristics of the PSCs with and without 5 mol% cesium doping. SCLC of the PSCs (**b**) without and (**c**) with cesium doping. (**d**) PCE distribution box chart of the PSCs without and with 5% cesium doping.

Figure 4a showed the dark J–V characteristics of the 5 mol%-Cs-doped and undoped PSCs. The cesium-doped PSC shows smaller leakage current than the reference PSC without cesium doping across the voltage range 0 to −1.0 V. To analyze the trap density of the perovskite films with cesium doping, we measured the SCLC of the hole-only devices described above [34,35]. As shown in Figure 4b,c, the J−V curve can be divided into three regions.

The first segmen<sup>t</sup> at low bias (<0.4 V) is the ohmic region, in which the current density shows the almost linear increase with the voltage [36]. The second segmen<sup>t</sup> is called the trap-filled limit (TFL) region, in which the current density has rapid nonlinear growth, indicating the TFL in which the injected carriers deactivate available trap states [36]. At high voltages, the current density increases slowly, which is referred to as the Child's regime. The TFL voltage ( *VTFL*) is the voltage where the ohmic and TFL current curves intersect. The trap density (*ntrap*) can be calculated from VTFL using the following equation [37];

$$V\_{TFL} = \frac{\varepsilon n\_{trap} L^2}{2\varepsilon\_0 \varepsilon} \tag{1}$$

where *L* is the perovskite film thickness, (≈5.7565) is the relative dielectric constant of the CH3NH3PbI3 perovskite film [38], 0 is the vacuum permittivity, and *e* is the elementary charge. As a result, the *ntrap* values of the undoped and 5 mol%-Cs-doped devices are 5.8 × 10<sup>16</sup> cm<sup>−</sup><sup>3</sup> and 3.6 × 10<sup>16</sup> cm<sup>−</sup>3, respectively. The reduced trap density in the cesiumdoped sample can be explained by the reduced pinholes and improved crystallinity of the perovskite layers (shown in Figure 2). In Figure 4d, it is shown that adding 5 mol% cesium enhanced the average PCE.

XPS spectra for the 5 mol% cesium-doped perovskite film (Figure 5a) show the Cs 3d5/2 peak at 724.41 eV, confirming the presence of cesium in the sample. In Figure 5b, the cesium doping slightly increases the binding energy of Pb 4f5/2 from 137.24 to 137.86 eV. For I, the 3d5/2 peak is also blue-shifted by the cesium doping from 618.16 to 618.8 eV, as shown in Figure 5c. We speculate that the doped cesium atoms cause local distortion in the lattice, which may affect the binding energies of the Pb and I ions.

**Figure 5.** X-ray photoelectron spectroscopy (XPS) results for perovskite films without and with 5% cesium doping: (**a**) Cs 3d5/2, (**b**) Pb 4f5/2, and (**c**) I 3d5/2.

Our results demonstrate that Cs doping is effective for improving the crystallinity and morphology of Pb(Ac)2-based perovskite layers, suppressing the formation of secondary phases such as PbI2. Thus, Cs doping is promising for enhancing the PCEs of Pb(Ac)2-based PSCs by improving the quality of perovskite films.
