Contactless Determination of Optimal Chloride Concentration for Power Conversion Efficiency in CH3NH3Pb(Cl,I)3 Using Photoluminescence Spectroscopy

We applied room-temperature photoluminescence (PL) spectroscopy for the compositional engineering of a CH3NH3Pb(Cl,I)3 light harvester in an alloy-based perovskite solar cell. This spectroscopic characterization determines the optimal Cl concentration where the power conversion efficiency shows its maximum in a contactless and non-destructive manner. The PL quenching ratio evaluated from the comparative PL studies between the films grown on glass/ZrO2 and SnO2:F/TiO2 substrates exhibited its maximum at a Cl concentration of 10 mol%, which agrees with the Cl concentration determined from the current–voltage measurement-based device performance. We also discuss the possible reasons for the coincidence mentioned above regarding the charge extraction effect induced by Cl incorporation.


Introduction
Much attention has been recently paid to an inorganic-organic lead halide perovskite CH 3 NH 3 PbI 3 (MAPbI 3 ) as a new class of light harvester in a solar cell [1,2]. These perovskite materials demonstrate rapid progress in their performance as a solar cell, the central part of which is owing to their material properties such as their direct nature in the interband transition, high absorption coefficient, and high charge carrier mobility [3][4][5]. Furthermore, since their first discovery in 2009 [6], the power conversion efficiency (PCE) is now around 25%, exceeding that of other emerging photovoltaic technologies such as dye-sensitized solar cells and organic photovoltaics [7][8][9]. These properties and the possibility of flexible synthesis have made halide perovskites promising for next-generation photovoltaics [10][11][12][13].
To enhance the PCEs of the device, many groups have proposed the mixing of halide anions (X = Br, Cl) in the MAPbI 3 structure [1,14,15]. Chloride ions have positive roles in manufacturing high-quality perovskite films. For example, both diffusion lengths for electrons and holes of mixed halides are larger than those of iodide [16,17].
Despite the enormous profits implemented by chloride addition, the question related to an optimal ratio is still a topic that is under debate. Throughout this paper, one defines the term 'optimal ratio' as the Cl concentration showing the highest PCE in the perovskitebased MAPb(Cl,I) 3 . Although some research groups proposed that the optimal ratio should be around 0.33 [18], a look through the literature has revealed that the optimal ratio depends on the device preparation methods and growth conditions [11,19]. In other words, the determination of the optimal ratio requires the preparation and characterization of many devices with different Cl concentrations. This could indeed be time-consuming because electrical contact processes are needed to evaluate the PCE. Among well-established optical techniques, photoluminescence spectroscopy (PLS) is a non-contact characterization tool. This technique is suitable for materials intended for optoelectronic applications. So far, there have been several reports on the PL properties of MAPb(Cl,I) 3 films with charge extractors [17,19,20], attributing their observed PL quenching to the charge extraction from the light harvester. Because it is well known that charge extraction plays an essential role in PCE determination, we reached a conjecture that this technique can be used as a contactless and non-destructive determination tool for the optimal ratio. Furthermore, the PL measurement does not require electrical contact on the samples, which expectedly enables rapid characterization by applying, e.g., a composition-spread technique [21]. For the verification, we characterized and compared PL properties and current-voltage (J-V) characteristics by preparing several MAPb(Cl,I) 3 films whose Cl concentration ranges from 0 to 25 mol%.
After the introduction, the paper begins with a description of the methods for film preparation, spectroscopic characterizations, electrical characterizations, and theoretical calculations. It will then go on to the results and discussions related to the optical properties of our prepared films concerning device performance from the viewpoint of compositional engineering. Finally, we draw our conclusions.

Experimental and Calculation Procedures
We first deposited porous titanium oxide (mesoporous TiO 2 ) on the blocking layer (compact TiO 2 ). The preparation methods consisted of spin coating (dip-casting) and screen printing.

Results and Discussion
We evaluated the Cl concentration dependence of the PL spectra in MAPb(Cl,I) 3 films grown on (1) glass/ZrO 2 and on (2) FTO/TiO 2 [19,30,31]. The study of the first structures was intended to quantify the number of photocarriers absorbed in the light harvesting layer [32]. In this structure, ZrO 2 , an insulator, was used instead of TiO 2 (a charge extractor) as a scaffold [33,34]. As will be explained later in detail, by evaluating the PL's intensity ratio between the first and the second structures, hereafter, we aim to evaluate a quantity related to the charge extraction efficiency and hence the PCE. Figure 1a,b show the spectra and integrated intensities of the PL in the film grown on glass/ZrO 2 , respectively, as a function of the Cl concentrations. The central photon energy of the PL spectrum was about 1.61 to 1.59 eV for each Cl concentration [35]. The observed energy is in good agreement with the reported bandgap energy of MAPbI 3, ranging from 1.5 to 1.6 eV [36,37]. Ernzerhof (PBE) functional, which belongs to generalized gradient approximation (GGA) functionals.

Results and Discussion
We evaluated the Cl concentration dependence of the PL spectra in MAPb(Cl,I)3 films grown on (1) glass/ZrO2 and on (2) FTO/TiO2 [19,30,31]. The study of the first structures was intended to quantify the number of photocarriers absorbed in the light harvesting layer [32]. In this structure, ZrO2, an insulator, was used instead of TiO2 (a charge extractor) as a scaffold [33,34]. As will be explained later in detail, by evaluating the PL's intensity ratio between the first and the second structures, hereafter, we aim to evaluate a quantity related to the charge extraction efficiency and hence the PCE. Figure 1a,b show the spectra and integrated intensities of the PL in the film grown on glass/ZrO2, respectively, as a function of the Cl concentrations. The central photon energy of the PL spectrum was about 1.61 to 1.59 eV for each Cl concentration [35]. The observed energy is in good agreement with the reported bandgap energy of MAPbI3, ranging from 1.5 to 1.6 eV [36,37]. The central photon energy did not shift even with chloride addition [13]. Chloride addition tends to blueshift the bandgap energy because the bandgap energy of MAPbCl3 is significantly larger than that of MAPbI3 [37]. This observation is probably due to the cancellation of the bandgap energy shift with the carrier localization effect. It is well known that the bandgap fluctuation introduced by Cl incorporation induces the red-shifting carrier localization [38]. As can be understood from Figure 1b, the integrated PL intensity decreases as the Cl concentration increases, which is in good agreement with the 'theoretical' concentration dependence shown in Figure 2 [26]. Here, we plotted the density of states spectrally integrated between the gap energy and 3.5 eV, which is considered to be proportional to the number of photocarriers absorbed in the light harvesting layer. The central photon energy did not shift even with chloride addition [13]. Chloride addition tends to blueshift the bandgap energy because the bandgap energy of MAPbCl 3 is significantly larger than that of MAPbI 3 [37]. This observation is probably due to the cancellation of the bandgap energy shift with the carrier localization effect. It is well known that the bandgap fluctuation introduced by Cl incorporation induces the redshifting carrier localization [38]. As can be understood from Figure 1b, the integrated PL intensity decreases as the Cl concentration increases, which is in good agreement with the 'theoretical' concentration dependence shown in Figure 2 [26]. Here, we plotted the density of states spectrally integrated between the gap energy and 3.5 eV, which is considered to be proportional to the number of photocarriers absorbed in the light harvesting layer. Figure 3a,b show PL's spectra and intensities, respectively, for the film grown on FTO/TiO 2 [39,40]. The PL is quenched, compared to the first samples' cases shown in Figure 1b. This result is probably due to photocarriers' charge extraction from the light harvesting layer into the electron transport layer [41]. Letting the integrated PL intensity of the first and second structures be I 1 and I 2 , we define an 'efficiency' term hereafter as [(I 1 − I 2 )/I 1 ] to evaluate the effect induced by replacing ZrO 2 with TiO 2 . We plot this quantity as a function of the Cl concentration in Figure 3c. We notice an interesting concentration dependence. The maximum efficiency found itself at 10 mol% of chlorine, which is in good agreement with the PCE shown in Table 1 [42,43]. We measured the current-voltage (J-V) curves in MAPb(Cl,I) 3 devices and summarized their parameters, such as PCE in Table 1. We can safely conclude that the 'efficiency' is intimately related to the charge extraction efficiency. Our results indicate that PL spectroscopy could determine the optimal ratio by evaluating the 'efficiency' because the charge extraction efficiency is known to be one of the principal components determining the PCE of solar cells. However, it is not good practice to draw a conclusion by using only one characterization tool. For further justification, we conducted a similar study with time-resolved differential absorption spectroscopy, which showed an identical concentration dependence to the case of PL spectroscopy [44,45].  Figure 3a,b show PL's spectra and intensities, respectively, for the film grown on FTO/TiO2 [39,40]. The PL is quenched, compared to the first samples' cases shown in Figure 1b. This result is probably due to photocarriers' charge extraction from the light harvesting layer into the electron transport layer [41]. Letting the integrated PL intensity of the first and second structures be I1 and I2, we define an 'efficiency' term hereafter as [(I1 − I2)/I1] to evaluate the effect induced by replacing ZrO2 with TiO2. We plot this quantity as a function of the Cl concentration in Figure 3c. We notice an interesting concentration dependence. The maximum efficiency found itself at 10 mol% of chlorine, which is in good agreement with the PCE shown in Table 1 [42,43]. We measured the current-voltage (J-V) curves in MAPb(Cl,I)3 devices and summarized their parameters, such as PCE in Table 1. We can safely conclude that the 'efficiency' is intimately related to the charge extraction efficiency. Our results indicate that PL spectroscopy could determine the optimal ratio by evaluating the 'efficiency' because the charge extraction efficiency is known to be one of the principal components determining the PCE of solar cells. However, it is not good practice to draw a conclusion by using only one characterization tool. For further justification, we conducted a similar study with time-resolved differential absorption spectroscopy, which showed an identical concentration dependence to the case of PL spectroscopy [44,45].    Figure 3a,b show PL's spectra and intensities, respectively, for the film grown on FTO/TiO2 [39,40]. The PL is quenched, compared to the first samples' cases shown in Figure 1b. This result is probably due to photocarriers' charge extraction from the light harvesting layer into the electron transport layer [41]. Letting the integrated PL intensity of the first and second structures be I1 and I2, we define an 'efficiency' term hereafter as [(I1 − I2)/I1] to evaluate the effect induced by replacing ZrO2 with TiO2. We plot this quantity as a function of the Cl concentration in Figure 3c. We notice an interesting concentration dependence. The maximum efficiency found itself at 10 mol% of chlorine, which is in good agreement with the PCE shown in Table 1 [42,43]. We measured the current-voltage (J-V) curves in MAPb(Cl,I)3 devices and summarized their parameters, such as PCE in Table 1. We can safely conclude that the 'efficiency' is intimately related to the charge extraction efficiency. Our results indicate that PL spectroscopy could determine the optimal ratio by evaluating the 'efficiency' because the charge extraction efficiency is known to be one of the principal components determining the PCE of solar cells. However, it is not good practice to draw a conclusion by using only one characterization tool. For further justification, we conducted a similar study with time-resolved differential absorption spectroscopy, which showed an identical concentration dependence to the case of PL spectroscopy [44,45].   Table 1. Summary of device performance of CH 3 NH 3 Pb(Cl,I) 3 solar cells with different Cl concentrations obtained from photocurrent density-voltage (J-V) curves. The device parameters were from the data obtained in a forward scanning direction. It is noted that the hysteresis in the J-V curve is negligible.  Figure 4a,b are SEM images of CH 3 NH 3 PbI 3 and CH 3 NH 3 Pb(Cl,I) 3 with a Cl concentration of 25 mol% [14,46]. The particle size of CH 3 NH 3 PbI 3 of ca. 150 nm is significantly different from that of CH 3 NH 3 Pb(Cl,I) 3 (ca. 400 nm). A previous study insisted that photocarriers' localization length scale is comparable to the length scale of the particles' sizes. Because photocarriers' localization length likely influences the PCE of the device, our observed change in the particles' size may have some impact on the PCE. Further studies are necessary to clarify this issue. Figure 4a,b are SEM images of CH3NH3PbI3 and CH3NH3Pb(Cl,I)3 with a Cl concentration of 25mol% [14,46]. The particle size of CH3NH3PbI3 of ca. 150 nm is significantly different from that of CH3NH3Pb(Cl,I)3 (ca. 400 nm). A previous study insisted that photocarriers' localization length scale is comparable to the length scale of the particles' sizes. Because photocarriers' localization length likely influences the PCE of the device, our observed change in the particles' size may have some impact on the PCE. Further studies are necessary to clarify this issue.

Conclusions
The present research aimed to examine the usefulness of PL spectroscopy for the compositional engineering of a MAPb(Cl,I)3 light harvester. The comparative investigation of the Cl concentration dependence of PL intensities for the films on glass/ZrO2 and FTO/TiO2 has shown its maximum at a Cl concentration of 10 mol%, which is in good agreement with that determined from J-V measurements. This spectroscopy can determine the optimal Cl ratio in a contactless manner, which will expectedly speed up the material development cycle.

Conclusions
The present research aimed to examine the usefulness of PL spectroscopy for the compositional engineering of a MAPb(Cl,I) 3 light harvester. The comparative investigation of the Cl concentration dependence of PL intensities for the films on glass/ZrO 2 and FTO/TiO 2 has shown its maximum at a Cl concentration of 10 mol%, which is in good agreement with that determined from J-V measurements. This spectroscopy can determine the optimal Cl ratio in a contactless manner, which will expectedly speed up the material development cycle.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy.