Introduction of cadmium chloride additive to improve the performance and stability of perovskite solar cells

With the increase in the importance of using green energy sources to meet the world's energy demands, attempts have been made to push perovskite solar cell technology toward industrialization all around the world. Improving the properties of perovskite materials as the heart of PSCs is one of the methods to fabricate favorable photovoltaic (PV) solar cells based on perovskites. Here, cadmium chloride (CdCl2) was used as an additive source for the perovskite precursor to improve its PV properties. Results indicated CdCl2 improves the perovskite growth and tailors its crystalline properties, suggesting boosted charge transport processes in the bulk and interfaces of the perovskite layer with electron–hole transport layers. Overall, by incorporation of 1.0% into the MAPbI3 layer, a maximum power conversion efficiency of 15.28% was recorded for perovskite-based solar cells, higher than the 12.17% for the control devices. The developed method not only improved the PV performance of devices but also boosted the stability behavior of solar cells due to the passivated domain boundaries and enhanced hydrophobicity in the CdCl2-based devices.


Introduction
Due to their exceptional photoelectric performance and easy fabrication processes, hybrid perovskite solar cells (HPSCs) have attracted considerable attention as a promising candidate for new-generation photovoltaic (PV) devices. [1][2][3] This could be related to the high absorption coefficient, high charge transfer phenomenon, and adjustable bandgap along with high charge diffusion length, which are all important requirements. [4][5][6] The efficiency of HPSCs has substantially increased from a low of 3.8% (in 2009) 7 to a veried 25.2% (in 2021), 8 which has encouraged researchers in the HPSC's future development. Typically, according to the direction of the photocurrent, the fundamental architecture of HPSCs can be categorized into conventional (n-i-p) and inverted (p-i-n) architectures. 9,10 It's worth mentioning that the perovskite material is the heart of an HPSC device, and its overall quality determines how well it performs. The power conversion efficiency (PCE), hysteresis, and long-term stability of HPSCs are all inuenced by the morphology, crystalline nature, and surface covering of the perovskite lm. [11][12][13] A high content of pinholes and grain boundaries (GBs) in the perovskite lm, particularly, produces severe shorting and trap centers for non-radiative recombination, which is one of the primary sources of PCE degradation in HPSCs. 14,15 Charge recombination aided by pinholes and GBs is also associated with abnormal hysteresis and instability concerns. As a result, enhancing the perovskite properties via passivating defects and optimizing perovskite quality has become a critical technique for decreasing recombination rates, removing unfavorable hysteresis, prolonging HPSC longevity, and increasing PCE. 16,17 To date, several ways of controlling perovskite morphology have been explored, involving manipulating the deposition method, additive engineering, interfacial engineering, and compositional engineering. [18][19][20][21][22] Additive engineering has been shown to be effective in improving the quality of the perovskite layer since it is a more simple and reproducible approach. Some agents, including copper chloride, N,1-diiodoformamidine, potassium hexauorophosphate, formamidine acetate salt, etc., can form an intermediate phase with perovskite precursors, which could regulate the dynamic process of crystal formation and thereby minimize GBs. [23][24][25][26][27] The incorporation of cadmium (Cd 2+ ) additive by Suneth et al. was a fundamental advance in controlling methylammonium lead iodide (MAPbI 3 ) crystallinity using a two-step deposition method. With this treatment, the efficiency of MAPbI 3 -based HPSCs was signicantly boosted from 7.1 to 13.8%. 28 Yong et al. reported the addition of Cd 2+ into organic-inorganic cation-based Cs x FA 1Àx PbI 3 lattice to enhance the crystallinity and the charge carrier lifetime of perovskite lm. An efficiency up to 20.59% was achieved for HPSC with a structure of ITO glass/SnO 2 /Cs x FA 1Àx PbI 3 /spiro-OMeTAD/Au. 29 In another study, a suppression of atomic vacancies via strain relaxation of CsMAFA triple-cation perovskite lm was reported by Makhsud et al. This suppression was achieved via incorporation of Cd isovalent small ions, which let to boost the stability of HPSCs in ambient air. 30 Recently, Cd 2+ introduced into all-inorganic perovskite for carbon-based HPSCs with a framework of FTO/SnO 2 /CsPbIBr 2 /carbon. They demonstrated that Cd ions improves the crystalline nature and morphological merits of perovskite lm, resulting in better optical absorption and carrier transfer in perovskite with less recombination rates. Also, in compared to the HPSCs without Cd 2+ doping (4.36%), the improved HPSCs attained the highest PCE of 6.79%. 31 Yi et al. introduced ammonium benzenesulfonate as a passivator into the MAPbI 3 solution to assist the lm crystallization and mitigate trap density. A champion PCE of 20.6% is obtained for the modied HPSCs. 32 Hua et al. reported the incorporation of N,1-diiodoformamidine modier into FAMAPbI 3 to improve the quality of perovskite with bigger grain size and fewer defects, hence improving the PCE from 19.07% to 21.22% as well as their light and thermal stabilities. 24 Recently, a dual additive approach for enhancing the performance of HPSCs was reported by Yang et al. 33 The highest PCE of 23.2% is yielded via a mixed modier of MACl/CsCl with a 2 : 1 ratio, which is better than that for a single modier of MACl or CsCl. Despite their high performance, the majority of HPSCs fabricated with organic hole transport layers (HTLs) such as spiro-OMeTAD have low stability and are excessively expensive, making them unsuitable for commercialization. Therefore, further work is needed to nd a cheaper alternative to spiro-OMeTAD while simultaneously utilizing the additive engineering approach.
Herein, we take advantage of using an efficient additive of cadmium chloride (CdCl 2 ) alongside with using a copper phthalocyanine (CuPc) semiconductor as a model HTL for regular (n-i-p) HPSCs. CuPc is known to possess high hole mobility, excellent interfacial bonding features, and superb stability while also being relatively inexpensive, which is why it has been used as a p-type semiconductor in different optoelectronic devices. [34][35][36] We investigate the impact of employing low CdCl 2 concentrations and high MAI contents during MAPbI 3 development in the anti-solvent deposition procedure. In addition to better crystalline phase and increased grain size, steady-state photoluminescence (PL) tests revealed a signicant reduction in charge recombination process. The bestperforming CdCl 2 -based HPSCs using a CuPc HTL yielded a PCE of 15.28% with a V OC of 1.0 V and a FF of 73.2%, recorded under reverse scanning. The modied devices demonstrated enhanced long-term stability as compared to unmodied devices fabricated with mesoscopic structures.

Solution preparation
922 mg of PbI 2 were dissolved in 1.27 mL of N,N-dimethylformamide (DMF, 99.8%) and 0.142 mL dimethyl sulfoxide (DMSO, 99.8%). The resultant product was mixed at 80 C for 45 min. Aer cooling down the PbI 2 solution, 318 mg of MAI was inserted into it and stirred for 15 min to prepare MAPbI 3 . For incorporation of CdCl 2 into the perovskite layer, different amounts of CdCl 2 in molar ratios to MAI are added into the prepared perovskite solutions, followed by further stirring at RT for 60 min. 500 mg of TiO 2 paste is dissolved in 3.0 g ethanol (EtOH, 99.8%) and is stirred at RT for 24 h to attain mesoporous TiO 2 (mp-TiO 2 ) solution. 700 mg of TTIP is added into 10 mL of 2-propanol (IPA, 99.9%) containing 70 mL of 2 M HCl, followed by stirring in an ice bath for 30 min to prepare compact TiO 2 (c-TiO 2 ) precursor. The spiro-OMeTAD precursor was prepared as reported in the literature of (ref. 37) by dissolving 36.2 mg of spiro-OMeTAD (99%) in 500 mL chlorobenzene (CB, 99.9%) doped with LiTFSI, and 4-tert-butylpyridine (98%).

HPSC fabrication
Transparent conductive F-doped SnO 2 (FTO, Pilkington) glasses with a sheet resistance of 15 U sq À1 are ultrasonically cleaned with distilled water, soap, distilled water, acetone, and IPA successively for 10 min. 45 mL of the c-TiO 2 solution is spread in each FTO piece and spin-coated at 4000 rpm, followed by annealing at 500 C for 30 min to fabricate a c-TiO 2 layer. 60 mL of mp-TiO 2 precursor is spin-coated on each c-TiO 2 layer at 4000 rpm for 30 s, followed by baking at 450 C for 60 min. To fabricate perovskite layer, 60 mL of perovskite precursors with or without CdCl 2 additive are spread on each substrate and spin coated at 1000 rpm for 8 s and then accelerate up to 6000 for 30 s. During the faster step, 200 mL of CB was pipetted on perovskite layer to assist crystal growth of perovskite layer. Perovskite layers were annealed for 30 min at 100 C. The CuPc hole transport layer (HTL) with thickness of 30 nm was thermally evaporated on the pre-formed perovskite lms in a vacuum of 10 À3 Pa with deposition rate of 1 A s À1 . To fabricate spiro-OMeTAD HTL, its precursor was spin-coated at 5000 rpm for 30 s on top the perovskites. Finally, a 100 nm gold thin-layer was thermally deposited the HTLs.

Characterizations
A Mira3, TESCAN eld emission scanning electron microscopy (FESEM) was used to monitor the effect of the CdCl 2 additive on the micro-morphology of perovskite layers. A Bruker, D8 Advance X-ray diffractometer was used to collect XRD pattern of perovskite lms. A LAMBDA 1050 spectrophotometer was used to study the light-harvesting ability of perovskite layers. A HORIBA Fluorolog-III spectrometer with an exciting wavelength at 405 nm was used to record steady-state photoluminescence (PL) responses of perovskite layers. Photovoltaic performance of MAPbI 3 -based solar cells with an active area of 2 mm Â 3 mm was characterized using a Keithley Model 2401 under a calibrated AM 1.5 light simulator to record current density-voltage curves. A calibrated Newport IPCE equipment was employed to record incident photon-to-electron conversion efficiency (IPCE) of solar cells. Fig. 1 presents the FESEM micrographs of the perovskites prepared with and without the CdCl 2 additives in the MAI precursors that were utilized to complete the MAPbI 3 conversion. The inuence of the CdCl 2 incorporation is clear in that good uniformity and large grains are observed in the CdCl 2added perovskites immediately aer heat treatment ( Fig. 1b and  c), while lm deposited without solvent additive shows low uniformity and small grains with obvious pinholes at the grain edges (Fig. 1a). As illustrated in Fig. 1e-h, the addition of small amounts of CdCl 2 enlarged the average grain size vefold, from 280 nm to 590 nm. This is due to the inclusion of CdCl 2 into the MAI solution, which produces Cl ions, resulting in the formation of an intermediate phase that acts as a seed and enhances the development of bigger MAPbI 3 crystals. 38 With a further increase of CdCl 2 amount to 1.5%, the surface of the perovskite layer was identical to the 1% CdCl 2 -added lm but revealed small particles with obvious grain boundaries (GBs). From the FESEM results, we can argue that the grain size of the CdCl 2 -modied lms was not greatly enlarged. Nevertheless, smaller grains have combined into bigger ones, the gaps between grains have been passivated, and the lm has become denser and more compact. This could help improve HPSC performance by reducing nonradiative charge recombination. 39 The performance of HPSCs is directly dependent on the crystal phase and purity of the MAPbI 3 layers. 40 Fig. 2a demonstrates the X-ray patterns of MAPbI 3 perovskites without and with 1% CdCl 2 inclusion into the MAI solution. As demonstrated, multiple main XRD peaks at 13.82 , 20.10 , 28.18 , and 31.59 were indexed as (110), (112), (220), and (310) planes of the tetragonal MAPbI 3 phase, respectively. 21 The sharp and intensive peaks in the XRD plots of the MAPbI 3 perovskites implied that the lms were well crystallized. Furthermore, the intensities of the prominent peaks all rose when the CdCl 2 solvent was incorporated, implying that a superior crystalline nature and larger crystallite size for perovskite lm were developed as compared to a lm prepared without CdCl 2 . 41 In MAPbI 3 samples prepared without and with CdCl 2 , the full width at half maximum intensity of the (110) peak was 0.108 and 0.097, respectively. According to the Scherrer equation, 42 the average crystallite sizes of the corresponding samples were 73.46 and 81.21 nm, respectively. These ndings suggest that the moderate amount of CdCl 2 in perovskite could increase both crystallinity and grain size. This result is also in line with the ndings of the FESEM images.

Results and discussion
The low XRD signal at 12.4 is indexed as the cubic PbI 2 , because of the slight degradation of the MAPbI 3 to PbI 2 and/or incomplete crystallization of perovskite. 23 The presence of PbI 2 in the perovskite would be detrimental to photo-induced carrier transfer, lowering lm stability and increasing the hysteresis behavior of HPSCs. Interestingly, it was apparently seen that the typical peak of PbI 2 was completely removed with CdCl 2 treatment. Fig. 2b shows the XRD patterns of perovskite samples zoomed to the major (110) peaks. The (110) peak was altered to higher values of 2q with the addition of CdCl 2 . This is most likely owing to a variation in the lattice parameters caused by the replacement of smaller Cd 2+ species for Pb 2+ sites, 28 therefore leading to the shrinkage of the MAPbI 3 lattice. The reduced lead amounts in perovskite could be benecial for increasing the tolerance factor, improves the stability of the MAPbI 3 structure. 31 We used UV-vis absorption and PL spectroscopy to explore the inuence of the CdCl 2 solvent on light absorption and the mechanism of photoexcited carriers transport in perovskite lm deposited under the same conditions unless the variation in CdCl 2 concentrations in MAI solution. The UV-vis spectra of MAPbI 3 perovskite layers with various concentrations of CdCl 2 additive are shown in Fig. 3a. The absorbance plots measured for MAPbI 3 layers with and without CdCl 2 treatments exhibited remarkably identical absorption proles (edge-absorption position at 777 nm) thanks to the unique energy bandgap (E g ) absorption of MAPbI 3 as a sunlight harvester. 43 As we can see, all samples fabricated by the spin-coating clearly show strong absorbance over the full light absorption below 700 nm. When the concentration of CdCl 2 solution is raised to 1%, the absorption performance of the associated MAPbI 3 layer slightly increases within the visible range. This could be due to the increased grain size and improved crystallinity, while the lm thickness remains similar with the insertion of CdCl 2 . 44 However, when more CdCl 2 (1%) was introduced, the perovskite's light absorption capacity dropped because the presence of unwanted defects and GBs disrupted the lm's compact structure. Fig. 3b describes the Tauc plots for all samples prepared with and without CdCl 2 additives. The value of E g can be evaluated by extrapolating the variation between photo energy (hn) and (ahn) 2 . 45 The values of E g were found to be 1.59 eV, which is similar to the values reported in the literature. 13,40 Fig . 3c shows the steady-state PL measurements of perovskite lms spin-coated on glass substrates without HTL and ETL lms in order to reduce charge extraction at the interfaces and enable the inherent features of the MAPbI 3 lm to govern the recombination dynamics. All the steady-state PL plots of perovskites reveal a typical and strong luminescence peak at locations identical to the absorbance edges detected in UV-vis measurements. The increased PL intensity and reduced linewidth of CdCl 2 -treated lms in comparison to the control layer indicate that the incorporation of solvent additive can signicantly inhibit trap-mediated, non-radiative recombination in those perovskites. 46 This could be due to the CdCl 2 additive treatments improving the crystallinity of the developed perovskites. The enhanced crystalline structure decreased defects in the entire MAPbI 3 , resulting in fewer recombination taps in the lm. This inhibited the non-radiation recombination pathway, making it more suitable for optoelectronics manufacturing. Conversely, as the CdCl 2 concentration was increased to 1.5%, the PL intensity of the perovskite decreased, indicating an increase in the density of microstructural defects.
We fabricated regular (n-i-p) HPSC devices using the architecture of glass/FTO/c-TiO 2 /mp-TiO 2 /MAPbI 3 /CuPc/Au, as illustrated in Fig. 4a. The ETL lms are fabricated based on our previously prepared mp-TiO 2 via the spin-coating method, while HTLs utilize a p-type inorganic lm named CuPc. The mixed solutions of MAI : PbI 2 were used to form perovskite lm with an anti-solvent assistance approach. An additive engineering method was employed in our perovskite lms to improve the HPSC performance. The CdCl 2 additive was selected as a modier to assist perovskite crystallization during the deposition process. We have investigated the PV parameters of our designed HPSCs with reverse J-V sweep being performed under the AM 1.5 sunlight simulator. Fig. 4b demonstrates the bestperforming J-V plots of the HPSCs fabricated with and without additive treatments, and the summarized cell parameters are displayed in Table 1. The performance of HPSCs with and without CdCl 2 additives varies signicantly. Without CdCl 2 , the HPSCs show low PV performances with a J SC of 19.78 mA cm À2 , V OC of 0.962 V, FF of 65.70% and, a champion PCE of 12.17%. The HPSC has an enhanced PCE of 15.28% with a high J SC of 20.87 mA cm À2 , a V OC of 1.00 V, and a good FF of 73% aer adding 1% CdCl 2 . When the amount of CdCl 2 is increased to 1.5%, the PCE of the HPSC has a notable reduction to 14.10% with reduced J SC and practically unchanged V OC . As previously mentioned, an adequate dose of CdCl 2 modier can assist in removing PbI 2 residual without forming a non-perovskite phase, resulting in a compact high-quality layer with large crystals, which can greatly reduce interior recombination and enhance HPSC performance. The integrated J SC values of the unmodied and 1% CdCl 2 -based HPSCs are 19.08 mA cm À2 and 20.14 mA cm À2 , respectively, as shown in Fig. 4c, which match with the J SC values determined from the detailed J-V scan  characteristics. The IPCE measurements conrm the accuracy of our HPSC efficiency characterizations.
To compare the performance of CuPc-based PSCs with conventional spiro-OMeTAD PSCs, we fabricated PSCs with spiro-OMeTAD HTL. A champion PCE of 18.65% with a J SC of 21.51 mA cm À2 , V OC of 1.117 mV, and FF of 77.60% was obtained (Fig. 4d). It notes that CuPc can be considered a low-cost HTL in perovskite-based solar cells. The CuPc HTL recorded a J SC near to spiro-OMeTAD devices, with lower V OC and FF than spiro-OMeTAD HTL one. Finding refers that recording higher PCE with CuPc HTL is reachable by more engineering at the interface of perovskite/CuPc, which will be targeted in our future experiments. Fig. 5 depicts the distribution of PV parameters depending on the concentration of MAI to CdCl 2 , where the ratio of MAI : CdCl 2 is changed from 0% to 1.5%. As shown in Fig. 5a, adjusting the ratio has a notable effect on the mean J SC of nearly 20 mA cm À2 . V OC improves slightly with increasing the CdCl 2 concentration (Fig. 5b). The average V OC value of 0.946 V for the control device is increased to 0.964, 0.994, and 0.985 V as the ratio varies from 0% to 1.5%. Mean FF is also enhanced as the concentration of CdCl 2 increases, reaching 72.29% at 1% and  then decreasing to 69.68% at 1.5% (Fig. 5c). As a consequence, PCE is boosted from 11.44% for the control device to 14.61% for the 1% CdCl 2 -modied device (Fig. 5d). Also, the statics show the narrow distribution of PV parameters of HPSCs based on the optimized 1% CdCl 2 , indicating excellent reproducibility of solar cells.
To further study the role of CdCl 2 material on the PV properties of space-charge-limited current (SCLC) measurements for electron-only devices with a structure of FTO/c-TiO 2 /mp-TiO 2 / MAPbI 3 with or without CdCl 2 /PCBM/Au were investigated. The electron trap density (N t ) of the perovskites layers were calculated as follows: 47 where q is the electric charge, L indicates the thickness of the perovskite lm, 3 r is the relative dielectric constant of the MAPbI 3 perovskite (3 r ¼ 32), 3 0 refers to the free space permittivity, and V TFL indicates for the trap-lled limit voltage. The V TFL for the electron-only devices for the control and 1%-CdCl 2 the 2D/3D perovskite layer were found to be 0.63 V and 0.40 V respectively. The hole trap density values for the control and 1%-CdCl 2 PSCs are 13.19 Â 10 15 cm À3 and 8.38 Â 10 15 cm À3 respectively. It indicates that the CdCl 2 additive facilitates the electron transfer process and reduces the charge recombination by suppressing the carrier traps within the bulk of perovskite layer. In addition, the hysteresis index (HI) of PSCs was recorded by measuring J-V curves in reverse and forward scan directions (Fig. 6b). HI was obtained using a formula of PCE reverse À PCE forward PCE reverse Â 100%: 40 Compared with the control PSCs with a HI value of 11.99%, the 1% CdCl 2 -modied PSCs showed a HI of 2.87%. The weaken HI in the CdCl 2 -based PSCs can be attributed to the passivated domain boundaries in the modied perovskite layer. In addition, as SCLC results showed, CdCl 2 incorporation into the perovskite layer accelerates electron transfer within the bulk of perovskite layer and suppresses HI in the PSCs. 48,49 One of the most critical performance factors for future commercial uses of HPSCs is long-term stability. Therefore, the HPSC longevity of related MAPbI 3 -based regular HPSCs (unmodied and with 1% CdCl 2 ) was assessed by tracing the PCE progression under ambient conditions with a RH of 25-40% in the dark (see Fig. 7). Aer 2040 h of age, the PCEs of the HPSC with 1% CdCl 2 additive demonstrate outstanding stability, maintaining 83% of the original value. In contrast, the performances of the unmodied HPSC dropped to less than 59% of their initial values aer 2040 h, mainly attributable to perovskite degradation. Hence, the incorporation of CdCl 2 in MAPbI 3 -based HPSCs, results in outstanding PV parameters and environmental stability.

Conclusion
In conclusion, we have proven a simple method based on additive engineering to improve the PCE and stability of MAPbI 3 -based HPSCs by adding a certain concentration of CdCl 2 additive. The impacts of varying concentrations of CdCl 2 additives on perovskite layer characteristics and PV properties were studied. High crystallinity, large grain size, improved optical absorption ability, and fewer pinholes and GBs were obtained for perovskite lm fabricated with the optimized CdCl 2 concentration, resulting in excellent photovoltaic performance. The synergistic inuence of crystal growth control and pinhole passivation by CdCl 2 effectively decreases the charge carrier recombination. With the incorporation of 1% CdCl 2 into the MAI solution, the HPSC yielded a champion PCE of over 15.28% under the reverse scan conditions. In comparison with unmodied HPSC, the HPSC based on 1% CdCl 2 also shows excellent ambient stability, which maintained 83% of its initial PCE aer being stored for 2040 h without encapsulation.
Our ndings indicate that CdCl 2 is a good additive for making inexpensive HPSCs with excellent stability.

Conflicts of interest
The authors declare no conict of interest.