Optical Absorption, Charge Separation and Recombination Dynamics in Pb and Sn/Pb Cocktail Perovskite Solar Cells and Their Relationships to the Photovoltaic Properties

Due to the unique characteristics such as simple low-temperature preparation method and high efficiency with a record of over 20%, organometal trihalide perovskite (CH 3 NH 3 PbI 3 )-based solid-state hybrid solar cells have attracted an increasing interest since 2012 when it was reported. During the last several years, some of the fundamental photophysical properties of perovskite related to the high photovoltaic performance have been investigated. Optical absorption, charge separation and recombination are very im‐ portant factors influencing the perovskite solar cell performance. In this chapter, our re‐ cent results of optical absorption, charge separation (electron and hole injection) and charge recombination dynamics at each interface in perovskite solar cells, and their rela‐ tionships to photovoltaic properties will be introduced. Our results suggest that charge recombination is a key factor in improving the performance of the perovskite solar cells. tively. The faster decay process could be attributed to nonradiative recombination of electrons and holes through defects or trap states in MAPbI 3 or at the MAPbI 3 /Y 2 O 3 interface. The relative contribution of the faster decay is less than 10%, which suggests that the defect or trap state density in MAPbI 3 or at the interface is very small, which is consistent with a thermally stimulated current measurement. [39, 40] The slower decay process could be attributed to the recombination of free electrons and holes in MAPbI 3 . This result suggests that the lifetime of the photoexcited charge carriers was very long, which was confirmed to be as much as the order of microseconds by the ns-TA measurement results shown in Figure 3. The TA decay was fitted with eq. (2), and two decay processes were found with lifetimes of 3.7±0.1 µs (70%) and 60±1 µs (30%), respectively. Combining the fs-TA and ns-TA results for MAPbI 3 /Y 2 O 3 , the lifetime of the photoexcited charge carriers in MAPbI 3 can be mostly considered to be as long as the order of microseconds. that the great differences in the IPCE and J sc of the two kinds of cell result from the great difference in the charge recombination between the electrons in TiO 2 and holes in spiro -OMeTAD. These results indicate that the key for improving the IPCE and J sc of the perovskite-based solar cells is charge collection efficiency, instead of charge separation efficiency. Thus suppressing the recombination through proper interfacial engineering for perovskite-based solid hybrid solar cells is crucial for achieving a high efficiency. For the Sn/Pb cocktail perovskite solar cells, it is determined that the bandgap is 1.21 eV and light harvesting can be extended to a wider wavelength over 1000 nm. The Urbach energy is calculated to be 34 meV, which is more than twice that of the MAPbI 3 perovskite. Three recombination processes for the photoexcited carriers were found with lifetimes being 4 ps (55%), 650 ps (25%) and one much larger than 3 ns (20%). The larger Urbach energy and the faster recombination suggest that there are defects in the prepared Sn/Pb cocktail perovskite. These results are significantly different from those of the MAPbI 3 perovskite, in which case the photocarrier lifetimes are larger than 100 ns and almost no defects were observed. Moreover, the charge separation and charge recombination dynamics were explored. We find that both the electron injection into TiO 2 and the charge recombination lifetime became shorter, i.e., 190 µs, when P3HT was employed as a HTM. This implies that the charges can be more effectively collected without the HTM. On the basis of the above results, we find that the loss in the charge separation efficiency originates from fast photoexcited carrier recombination with a lifetime of 4 ps, and the loss in charge collection efficiency was due to the charge recombination occurring at the TiO 2 /P3HT interface, thus leading to lower IPCE values. Also, the low V oc and FF were found to result from the large Urbach energy and the recombination occurring at the TiO 2 /P3HT interface. These findings indicate that the photo‐ voltaic performance of Sn/Pb cocktail perovskite solar cells can be further improved by reducing the defects in the material and the recombination occurring at the TiO 2 /HTM interface through proper interfacial engineering.


Introduction
Organolead halide perovskites in the format of AMX3 (A=organic molecule, e.g., CH 3 NH 3 (MA), B=Pb, X=Cl, Br, and I) can be easily crystallized from solution at relatively low temperature (i.e., ≤100 °C), which makes it possible to use them as light absorbing materials in different kinds of solar cells. Following the recently reported certified high power conversion efficiency (PCE) of over 20%, [1][2][3][4][5][6][7][8][9][10][11][12] the interest in organolead halide perovskite-based organic-inorganic hybrid solid-state solar cells has increased dramatically over the past several years. The higher PCEs of organolead halide perovskites (especially MAPbI 3 ) result from their unique properties that are key for achieving high photovoltaic performance, which are (1) a direct band gap and a high optical absorption coefficient; [13,14] (2) large dielectric coefficient leading to smaller exciton binding energy; [15] (3) long photoexcited carrier lifetimes (>100 ns) and long diffusion lengths (100 -1000 nm or even longer); [16,17] (4) no deep state defects and very small Urbach energy. [18] It is reasonable to expect that further improvements in photovoltaic performance can be achieved by increasing the light harvesting up to NIR wavelengths of 1000 nm, since MAP-bI 3 perovskite only absorbs light at wavelengths below 800 nm constrained by its optical band gap of 1.5 eV. In addition, practically, Pb-free organometal halide perovskites are preferred due to the potential toxicology issue of Pb. Replacing Pb with Sn or mixing Pb and Sn in organometal halide perovskites can result in increased light harvesting in the NIR region up to 1000 nm [19,20] and, at the same time, reduce the toxicity issue related to Pb. Several research groups have reported, very recently, Sn-based or Pb/Sn cocktail MASn x Pb 1-x I 3 (0≦x≦1) perovskite solar cells. [21][22][23][24] However, it is found that the PCE of Sn/Pb cocktail MASn x Pb 1x I 3 perovskite solar cells is far inferior to that of MAPbI 3 perovskite solar cells. In order to improve the photovoltaic performance of Pb and Sn/Pb cocktail perovskite solar cells, it is critical to gain a thorough understanding of the optical absorption properties, the photoexcited carrier lifetimes, as well as the charge separation and recombination dynamics at each interface. In this chapter, we will focus on recent studies of the optical absorption, photoexcited carrier lifetime, the charge separation and charge recombination dynamics at each interface in perovskite solar cells, including Pb-based and Sn/Pb cocktail perovskite solar cells. The relationships of each of these physical properties to the photovoltaic performance of the solar cells will be discussed and the methodologies for improving the photovoltaic performance of perovskite solar cells will be proposed.

Sample preparation
Samples of Pb-based perovskite hybrid solar cells were prepared by the following method. [25] F-doped SnO 2 layered glass (FTO glass, Nippon Sheet Glass Co. Ltd) was patterned using Zn powder and 6 N HCl aqueous solution. Titanium diisopropoxide bis(acetylacetonate) solution in ethanol was sprayed onto this patterned FTO glass at 300 °C to prepare compact TiO 2 layers. Porous TiO 2 layers were fabricated by spin-coating TiO 2 pastes of different nanoparticle sizes (18 nm: PST-18NR or 30 nm: PST-30NRD, JGC Catalysts and Chemicals Ltd.) in ethanol (TiO 2 paste : ethanol = 1:2.5 weight ratio for PST-18NR or TiO 2 paste : ethanol = 2:7 weight ratio for PST-30NRD), followed by heating the substrates at 550 °C for 30 min. For some TA measurements, glass, instead of FTO was used as the substrate and a porous Y 2 O 3 layer was fabricated on the glass substrates. Next, CH 3 NH 3 I and PbCl 2 were mixed with a 3:1 molar ratio for preparing a 40 % solution of perovskite in N,N-dimethylformamide and the mixture was spin-coated on the TiO 2 and Y 2 O 3 porous substrates. After heating at 100 °C for 45 minutes, the substrates were spin-coated with a mixture of 55 mM of tert-butylpyridine, 9 mM of lithium bis(trifluoromethylsyfonyl)imide salt, and 68 mM of spiro-OMeTAD. Finally, Ag and Au electrodes were fabricated by vacuum deposition for the photovoltaic measurements. The photovoltaic performance was evaluated using an AM1.5G 100 mW/cm 2 irradiance solar simulator (CEP-2000, Bunkoukeiki Inc) with a 0.4 cm x 0.4 cm mask.
The Sn/Pb cocktail perovskite samples were prepared using the following method. [23] PbI 2 (Purity: 99.999 %), SnI 2 , and regioregular poly(3-hexylthiophene-2,5-diyl) (P3HT) were purchased from Sigma Aldrich and used as received. F-doped SnO 2 coated glass ((FTO glass), Nippon Sheet Glass Co. Ltd) was patterned using Zn and a 6 N HCl aqueous solution. A compact TiO 2 layer was produced by spraying titanium diisopropoxide bis(acetylacetonate) solution in ethanol onto this patterned glass at 300 °C. The substrate was then dipped in a 40 mM solution of TiCl 4 in water for 30 min. Then the substrate was baked at 500 °C for 20 min. Next, a porous TiO 2 layer was produced by spin-coating with a TiO 2 paste (PST-18NR, JGC Catalysts and Chemicals Ltd.) in ethanol (TiO 2 paste : ethanol = 2:7 weight ratio). The substrate was then heated at 550 °C for 30 min. For certain TA measurements, glass substrates other than FTO ones, with a porous Al 2 O 3 layer deposited on them, were used. The substrates were then spin-coated with a mixture of SnI 2 , PbI 2 , and CH 3 NH 3 I (0.5 : 0.5 : 1.0 (molar ratio)) in dimethylformamide (DMF) (40 wt %), and they were baked at 70 °C for 30 min. Next, P3HT in chlorobenzene solution (15 mg/ml) was spin-coated on the prepared perovskite layer and the substrate was put in nitrogen at ambient temperature for 1 h. For conducting TA measurements, samples of Sn/Pb cocktail perovskite deposited on both Al 2 O 3 and TiO 2 substrates with and without P3HT as a hole transport material (HTM) were employed. Ag and Au electrodes were finally deposited by vacuum deposition for characterizing the photovoltaic performance, which was evaluated using an AM 1.5G 100 mW/cm 2 irradiance solar simulator (CEP-2000SRR, Bunkoukeiki Inc) with a 0.4 cm x 0.4 cm mask.

Optical absorption measurements
A gas-microphone photoacoustic (PA) technique [26] was used to study the optical absorption properties of the samples. The light source was a 300 W xenon arc lamp. By passing the light through a monochromator, a monochromatic light beam was obtained. This beam was modulated with a mechanical chopper and focused onto the surface of the sample placed in a sealed PA cell filled with N 2 gas. The measurements of PA spectrum were carried out in the wavelength range of 500-1200 nm with a modulation frequency of 33 Hz at room temperature. The PA signal was measured by first passing the microphone output through a preamplifier and then to a lock-in amplifier. The PA spectra were normalized with the PA spectrum obtained from a carbon black sheet.

Transient Absorption (TA) Measurements
Two different TA setups were used: the charge separation (electron injection and hole injection) dynamics was characterized by a femtosecond TA technique (fs-TA) [27,28] and charge recombination dynamics was characterized by a nanosecond TA technique (ns-TA) [28][29][30][31]. In the fs-TA setup for characterizing the charge separation, [27,28,31] the laser source used was a titanium/sapphire laser (CPA-2010, Clark-MXR Inc.) with a wavelength of 775 nm, a repetition rate of 1 kHz, and a pulse width of 150 fs. The light was separated into two parts. One part was used as a probe pulse. The other part was used to pump an optical parametric amplifier (OPA) (a TOAPS from Quantronix) to generate light pulses with a wavelength tunable from 290 nm to 3 µm. This was used as a pump light to excite the sample. In this study, a pump light wavelength of 470 nm and a probe beam wavelength of 775 nm were used. Deschler and co-workers [32] reported very recently that the TA response of ground state bleaching (GSB) for CH 3 NH 3 PbClI 2 on glass peaked at 1.65 eV and was spectrally unchanged up to times beyond 200 ns. Therefore, the probe light wavelength of 775 nm (i.e., the photon energy of 1.64 eV) used in this study is appropriate to monitor the GSB of CH 3 NH 3 PbClI 2 on Y 2 O 3 and TiO 2 substrates with or without spiro-OMeTAD, and thus electron transfer and hole transfer processes at each interface can be investigated systematically. On the other hand, in the case of Sn/Pb cocktail perovskite, the probe light of 775 nm was used to detect the photoexcited carrier absorption.
In the ns-TA setup for characterizing the charge recombination [28][29][30][31], the pump light source was an OPO (Surelite II -10FP) output excited by a Nd:YAG nanosecond pulse laser (Panther, Continuum, Electro-Optics Inc.), with a pulse width of 5 ns and a repetition rate of 0.5 Hz. A pulse light with a wavelength of 470 nm was used as the pump light to excite the sample. The probe light was produced from a fiber coupled CW semiconductor. Three different probe wavelengths, of 785 nm, 658 nm and 1310 nm, were used. Specifically, the probe beam of 785 nm was employed to measure the TA responses of GSB for MAPbClI 2 on Y 2 O 3 substrates, i.e., the recombination of electrons and holes in MAPbClI 2 . [ 32] The probe beam of 658 nm was used to measure the trapped electrons in TiO 2 [33] based on the research of Yoshihara and coworkers, which was used to investigate charge recombination between the electrons in TiO 2 and the holes in the perovskite. The probe beam of 1310 nm was used to monitor the holes in spiro-OMeTAD [34] and thus measure the charge recombination between holes in spiro-OMeTAD and electrons in TiO 2 and/or in perovskite. For all measurements, the pump and probe lights were irradiated from the glass side and the TA measurements were carried out in a nitrogen atmosphere. Figure 1 shows the optical absorption spectra of the Pb (MAPbI 2 ) and Sn/Pb cocktail (MASn 0.5 Pb 0.5 I 3 ) perovskite samples on a porous TiO 2 substrate measured using the PA technique (we refer to this as the PA spectrum in the following) at room temperature. From the position of the shoulder in each PA spectrum, [35] the bandgap energies of MAPbI 2 and MASn 0.5 Pb 0.5 I 3 are determined to be 1.52 and 1.21 eV, respectively, which are almost the same as those given in our earlier reports. [23,25] We find that, below the shoulder, the trend of the absorption coefficient is exponential. The slope of this exponential tail, known as the Urbach tail, corresponds to the absorption tail states, and is usually quantitatively expressed by the Urbach Energy E U . [18,31,36] Investigation of these exponential tails can offer important information on the band structure, disorder, defects, impurities, and electron-phonon interactions in semiconductor materials. The numerical relationship between the optical absorption coefficient α in the exponential tail and the photon energy (hυ) is shown in eq. (1), through which the values of E U of the samples can be calculated. [37,38] 4 determined to be 1.52 and 1.21 eV, respectively, which are almost the same as those given in our earlier re below the shoulder, the trend of the absorption coefficient is exponential. The slope of this exponential tai tail, corresponds to the absorption tail states, and is usually quantitatively expressed by the Urb Investigation of these exponential tails can offer important information on the band structure, disorder, electron-phonon interactions in semiconductor materials. The numerical relationship between the optical a in the exponential tail and the photon energy (hυ) is shown in eq. (1), through which the values of E U calculated. 37,38

Optical Absorption Study: Bandgap and Urbach Energy of Pb and Sn/Pb cocktail perovskite
where h is Planck's constant, and α 0 , υ 0 , E u are fitting parameters. The values of E u for Pb and Sn/Pb co determined to be 22 meV and 34 meV, respectively. The value of E u for Sn/Pb cocktail perovskite is lar perovskite. It can be assumed that the value of E u is a reflection of the disorder and/or defects in the semic where h is Planck's constant, and α 0 , υ 0 , E u are fitting parameters. The values of E u for Pb and Sn/Pb cocktail perovskites were determined to be 22 meV and 34 meV, respectively. The value of E u for Sn/Pb cocktail perovskite is larger than that for the Pb perovskite. It can be assumed that the value of E u is a reflection of the disorder and/or defects in the semiconductor crystal. [18,[36][37][38] Thus, the width of the exponential tail increases with increasing density of defects. Therefore, the larger E u determined for Sn/Pb cocktail perovskite (i.e., 34 meV) implies that there could be higher defect states in the sample compared to MAPbI 3 (i.e., 22 meV). This will lead to faster recombination of photoexcited carriers in the Sn/Pb cocktail MASn 0.5 Pb 0.5 I 3 perovskite and lower values of V oc and FF, which we will discuss in detail below together with the TA measurement results.

Dynamics of Photoexcited Carrier Recombination and Charge Transfer in Pb-based Perovskite Solar Cells and their Relationships to the Photovoltaic Properties [25, 39, 40]
To perform systematic investigations on the charge separation and recombination dynamics in perovskite solar cells, TA measurements were conducted for MAPbI 3 on either Y 2 O 3 or TiO 2 substrates, with and without a spiro-OMeTAD layer. For perovskite deposited on Y 2 O 3 substrate, the lifetime of photoexcited charge carriers in MAPbI 3 and the charge separation dynamics at the MAPbI 3 /spiro-OMeTAD interface were studied by the fs-TA technique. The pump light wavelength used to excite MAPbI 3 only was 470 nm, and the probe light wavelength used was 775 nm, which is just at the optical absorption edge of MAPbI 3 . Figure 2 shows the normalized TA response of MAPbI 3 /Y 2 O 3 for a time scale of up to 3 ns with a lower pump light intensity (0.9 µJ/cm 2 ) with and without the spiro-OMeTAD layer. We can observe a bleaching signal with a very slow decay in the figure. Because electron injection from the perovskite to Y 2 O 3 cannot occur due to the large band gap of Y 2 O 3 , it is reasonable to attribute the slow decay to the slow recombination process of photoexcited charge carriers in MAPbI 3 . Two processes were found in the TA decay process and the TA signal obtained can be fitted to the biexponential function shown below very well: To perform systematic investigations on the charge separation and recombination dynam measurements were conducted for MAPbI 3 on either Y 2 O 3 or TiO 2 substrates, with and with perovskite deposited on Y 2 O 3 substrate, the lifetime of photoexcited charge carriers in M dynamics at the MAPbI 3 /spiro-OMeTAD interface were studied by the fs-TA technique. Th excite MAPbI 3 only was 470 nm, and the probe light wavelength used was 775 nm, which is ju Figure 2 shows the normalized TA response of MAPbI 3 /Y 2 O 3 for a time scale of up intensity (0.9 µJ/cm 2 ) with and without the spiro-OMeTAD layer. We can observe a bleaching the figure. Because electron injection from the perovskite to Y 2 O 3 cannot occur due to th reasonable to attribute the slow decay to the slow recombination process of photoexcited processes were found in the TA decay process and the TA signal obtained can be fitted to below very well: (2) where t 1 and t 2 are two time constants, A 1 and A 2 are the contributions from the corresponding compo where t 1 and t 2 are two time constants, A 1 and A 2 are the contributions from the corresponding components. The time constants t 1 and t 2 of the two charge recombination processes were calculated to be 34 ps (A 1 /(A 1 +A 2 ): 9%) and much larger than 10 ns (A 2 /(A 1 +A 2 ): 91%), respec-tively. The faster decay process could be attributed to nonradiative recombination of electrons and holes through defects or trap states in MAPbI 3 or at the MAPbI 3 /Y 2 O 3 interface. The relative contribution of the faster decay is less than 10%, which suggests that the defect or trap state density in MAPbI 3 or at the interface is very small, which is consistent with a thermally stimulated current measurement. [39,40] The slower decay process could be attributed to the recombination of free electrons and holes in MAPbI 3 . This result suggests that the lifetime of the photoexcited charge carriers was very long, which was confirmed to be as much as the order of microseconds by the ns-TA measurement results shown in Figure 3. The TA decay was fitted with eq. (2), and two decay processes were found with lifetimes of 3.7±0.1 µs (70%) and 60±1 µs (30%), respectively. Combining the fs-TA and ns-TA results for MAPbI 3 /Y 2 O 3 , the lifetime of the photoexcited charge carriers in MAPbI 3 can be mostly considered to be as long as the order of microseconds.
suggested that the lifetime of photoexcited charge carriers was quite long, which was confirmed to be as long as µs by the ns-TA measurement result as shown in Figure 3. By fitting the TA decay with eq. (1), it was found that two decay processes existed with lifetimes of 3.7±0.1 µs (70%) and 60±1 µs (30%), respectively. Combing the fs-TA and ns-TA results for MAPbI 3 /Y 2 O 3 , the lifetime of photoexcited charge carriers in MAPbI 3 can be mainly considered to be as long as µs. The TA response of MAPbI 3 /Y 2 O 3 with spiro-OMeTAD was fitted to eq. (2), and only one exponential decay with a time constant of 16±2 ns was found. Compared with the long lifetime of the TA decay in Y 2 O 3 /MAPbI 3 , the fast decay process in Y 2 O 3 /MAPbI 3 /spiro-OMeTAD can be considered to originate from the photoexcited hole injection from MAPbI 3 to spiro-OMe-TAD. This result indicates that charge separation occurred at the interface between MAPbI 3 and spiro-OMeTAD.
[25] Figure 4 shows the TA response of MAPbI 3 /Y 2 O 3 with a layer of spiro-OMeTAD measured at a probe wavelength of 1310 nm. This result clearly indicates that charge separation at the MAPbI 3 /spiro-OMeTAD interface truly occurred, which is entirely consistent with the fs-TA results shown in Figure 2. The decrease of the TA signal in the sample of MAPbI 3 /Y 2 O 3 with a layer of spiro-OMeTAD originated from the recombination process between the electrons in MAPbI 3 and the holes in spiro-OMeTAD, which can be fitted very well with eq. (2). Only one exponential decay process was observed, with a time constant determined to be 0.37±0.07 µs, which corresponds to the recombination time [ 25]. Figure 5 shows the photoexcited charge carrier (electrons and holes) dynamics in MAPbI 3 deposited on a Y 2 O 3 substrate with spiro-OMeTAD as a HTM. 6 result clearly indicates that charge separation at the MAPbI 3 /spiro-OMeTAD interface truly occurred, fs-TA results shown in Figure 2. The decrease of the TA signal in the sample of MAPbI 3 /Y 2 O 3 with a from the recombination process between the electrons in MAPbI 3 and the holes in spiro-OMeTAD, w (2). Only one exponential decay process was observed, with a time constant determined to be 0.37±0. recombination time 25 . Figure 5 shows the photoexcited charge carrier (electrons and holes) dynamics substrate with spiro-OMeTAD as a HTM.
1.0x10 -6 1.5x10 -6 2.0x10  The photoexcited electron injection and the recombination dynamics were measured using the TA techniques for MAPbI 3 deposited on TiO 2 substrates. Figure 6(a) shows the normalized TA responses of MAPbI 3 /TiO 2 for 400 ps measured with different pump light intensities. The decay processes in the normalized TA responses overlapped very well with each other when the pump light intensity was lower. However, when the pump light intensity became larger than 3.75 µJ/cm 2 , a faster decay process, resulting from Auger recombination, appeared in the TA response. [25] To determine the electron injection dynamics from MAPbI 3 to TiO 2 , the TA response of MAPbI 3 /TiO 2 measured under a lower pump light intensity (0.9 µJ/cm 2 ) was used to make a comparison with that of MAPbI 3 /Y 2 O 3 as shown in Figure 6(b). The TA response of MAPbI 3 /TiO 2 decayed much faster than that of MAPbI 3 /Y 2 O 3 and can be fitted to a single exponential decay function very well, with a time constant determined to be 1.8±0.1 ns, which is approximately 2 -3 orders smaller than the photoexcited charge carrier lifetimes (∼µs) of MAPbI 3 as shown above. Then the electron injection time t ET and electron injection rate k ET were calculated to be about 1.8 ns and 5.5×10 8 s -1 , respectively. [25] Thus, the electron injection efficiency η Einj was calculated to be about 100% for all electrons with lifetimes longer than 100 ns. It is supposed here that the other 9% of charge carriers that recombined with a lifetime of 34 ps were not present in the MAPbI 3 prepared on TiO 2 , because no faster decay component was observed in the TA responses of MAPbI 3 /TiO 2 . [25] Next, the recombination dynamics in MAPbI 3 /TiO 2 without the layer of spiro-OMeTAD, i.e., the recombination process between the electrons in TiO 2 and the holes in MAPbI 3 was measured using ns-TA. Figure 7 shows the TA response of MAPbI 3 /TiO 2 measured with a pump light wavelength of 470 nm and a probe light wavelength of 658 nm. [25] According to the TA signal in nanocrystalline TiO 2 films reported by Yoshikawa [33] and our earlier studies, [18][19][20][21] it is reasonable to think that the TA response in MAPbI 3 /TiO 2 observed at 658 nm corresponds to the electrons in TiO 2 injected from MAPbI 3 . [25] Then the recombination time between the electrons in TiO 2 and the holes in perovskite was determined to be 0.14 µs by fitting Figure 7 to eq. (2).
Then, the recombination dynamics in MAPbI 3 /TiO 2 with spiro-OMeTAD as a hole transport material, i.e., the recombination between the electrons in TiO 2 and the holes in spiro-OMe-TAD, were measured using ns-TA. Figure 8 shows the TA responses of MAPbI 3 /TiO 2 without and with spiro-OMeTAD, which were measured with a pump light wavelength of 470 nm and a probe light wavelength of 1310 nm, which was used to monitor the relaxation dynamics of holes in spiro-OMeTAD. From fig. 8 we can see that no TA signal was observed for the The decay processes in the normalized TA responses overlapped very well with each other when t However, when the pump light intensity became larger than 3.75 µJ/cm 2 , a faster decay process, re appeared in the TA response. 25 To determine the electron injection dynamics from MAPbI 3 to TiO 2 measured under a lower pump light intensity (0.9 µJ/cm 2 ) was used to make a comparison with that o 6(b). The TA response of MAPbI 3 /TiO 2 decayed much faster than that of MAPbI 3 /Y 2 O 3 and can be function very well, with a time constant determined to be 1.8±0.1 ns, which is approximately 2 -3 o    Then another kind of cell with 30 nm sized TiO 2 nanoparticles were also studied, which we call cell B. The energy conversion efficiency of cell B was typically 8-10%. It was found that the recombination dynamics in the Perovskite solar cells has a great influence on the IPCE, Jsc and the energy conversion efficiency. [25] Two kinds of TiO 2 /MAPbI 3 /spiro-OMeTAD solar cells showing different IPCE spectra and photovoltaic performance were studied, and the relationships between the charge separation and recombination dynamics and the photovoltaic properties were investigated.
[25] Figure 9 shows To investigate the correlation between the photoexcited charge carrier dynamics and the photovoltaic properties, especially the IPCE, the charge separation and recombination dynamics of cell B were also characterized [25]. Figure 10 compares the charge carrier dynamics of the two cells A and B, it can be seen that the recombination time of electrons in TiO 2 and holes in the perovskite was almost identical (i.e., 0.13-0.14 µs), and the electron injection time was a little faster (0.7 ns) in cell B. However, the recombination time of electrons in TiO 2 and holes in spiro-OMeTAD in cell B became as long as 600 µs, which is ten times longer than that in cell A (60 µs). It is important to see how these changed separation and recombination dynamics influence the IPCE and J sc of the two kinds of cell. It is well known that IPCE is proportional to the charge separation efficiency η Csep and the charge collection efficiency η Ccol, where η Csep is the product of the electron injection efficiency η Einj and the hole injection efficiency η Hinj . η Einj , η Hinj and η Ccol can be calculated using the photoexcited carrier lifetimes, electron and hole injection times, and recombination times as shown in Figures 5 and 10. As discussed in detail in reference 25, the charge separation efficiency of the two kinds of cell were found to be almost the same, about 90%. However, the charge collection efficiency of cell B was found to be almost 100%, which is much larger than that of cell A, only 70%.

Dynamics of Photoexcited Carrier Recombination and Charge Transfer in Sn/Pb Cocktail Perovskite Solar Ce
Relationships to the Photovoltaic Properties 31 To find the mechanism responsible for the low PCE, especially the low Voc and FF in Sn/Pb cocktail perovskit which has been reported 23 , both the fs-TA and ns-TA techniques were used to investigate the charge sep recombination dynamics in these solar cells systematically. 31 Sn/Pb cocktail perovskites deposited on either Al2O3 or with and without P3HT, were measured by the TA techniques using probe beams with different wavelengths. S Pb-based perovskite solar cells discussed above, firstly, photoexcited carrier lifetimes, hole injection to P3HT and re at the Sn/Pb cocktail perovskite /P3HT interface were clarified as shown in Figure 11. 31 The photoexcited carrier 4.2±0.4 ps (A1/(A1+A2+y0): 55%), 650±57 ps (A2/(A1+A2+y0): 25%), and much longer than 3 ns (y0/(A1+A2 respectively. Therefore, the recombination dynamics and lifetimes of photoexcited carriers in Sn/Pb cocktail perovsk faster and shorter than those in Pb-based perovskite, for which case they are much greater than 100 ns for over photoexcited carriers as mentioned above. It is considered that the faster decay processes are due to some

Dynamics of Photoexcited Carrier Recombination and Charge Transfer in Sn/Pb Cocktail Perovskite Solar Cells and their Relationships to the Photovoltaic Properties [31]
To find the mechanism responsible for the low PCE, especially the low V oc and FF in Sn/Pb cocktail perovskite solar cells, which has been reported [23], both the fs-TA and ns-TA techniques were used to investigate the charge separation and recombination dynamics in these solar cells systematically. [31] Sn/Pb cocktail perovskites deposited on either Al 2 O 3 or TiO 2 films, with and without P3HT, were measured by the TA techniques using probe beams with different wavelengths. Similar to the Pb-based perovskite solar cells discussed above, firstly, photoexcited carrier lifetimes, hole injection to P3HT and recombination at the Sn/Pb cocktail perovskite /P3HT interface were clarified as shown in Figure 11. [31] The photoexcited carrier lifetimes are 4.2±0.4 ps (A 1 /(A 1 +A 2 +y 0 ): 55%), 650±57 ps (A 2 /(A 1 +A 2 +y 0 ): 25%), and much longer than 3 ns (y 0 /(A 1 +A 2 +y 0 ): 20%), respectively. Therefore, the recombination dynamics and lifetimes of photoexcited carriers in Sn/Pb cocktail perovskite are much faster and shorter than those in Pb-based perovskite, for which case they are much greater than 100 ns for over 90% of the photoexcited carriers as mentioned above. It is considered that the faster decay processes are due to some nonradiative recombination of electrons and holes through defects or trap states in the Sn/Pb cocktail perovskite, which is consistent with the measured larger Urbach energy in the Sn/Pb cocktail perovskite (i.e., 34 meV). It is larger than that in Pb-based perovskite (i.e., 22 mev). On the other hand, the charge separation at the interface between Sn/ Pb cocktail perovskite and P3HT occurred in approximately 1 ps and interfacial charge recombination between the Sn/Pb cocktail perovskite (on an Al 2 O 3 substrate) and P3HT occurred in approximately 16 ps. [31] Secondly, the photoexcited charge separation and recombination dynamics of Sn/Pb cocktail perovskite deposited on a TiO 2 substrate with P3HT as a hole transport material were evaluated, which is shown in Figure 12. The charge separation at the perovskite/TiO 2 interface occurs in as fast as 1 ps. It is worth noting that the recombination time at the Sn/Pb cocktail perovskite/ TiO 2 interface is 880 µs, which is about two to three orders of magnitude slower compared to that occurring at the MAPbI 3 /TiO 2 interface as shown in Figure 10. On the other hand, the recombination time at the TiO 2 /P3HT interface is 190 µs, which is much faster compared to that without P3HT. This indicates that the recombination of electrons in the TiO 2 becomes faster when P3HT is used, which originates from the recombination at the TiO 2 /P3HT interface. This recombination is one of the main reasons for the lower V oc and FF of the Sn/Pb cocktail solar cells. [23] Pinhole-free Sn/Pb cocktail perovskite needs to be prepared in order to reduce direct interaction between the TiO 2 and P3HT, thus suppressing potential recombination. [31] Figure 12 Schematic illustration of photoexcited charge separation and recombination dyn perovskite solar cells with TiO 2 as an electron transport material (ETM) and P3HT as a HTM It is important to understand how the optical absorption, charge separation and recombination dynamics relate to the incident photon to current conversion efficiency (IPCE) spectrum and the current-voltage (I-V) characteristics of the solar cell. Figure 13 shows a typical IPCE spectrum of a Sn/Pb cocktail perovskite solar cell. Based on the photoexcited carrier relaxation dynamics of Sn/Pb perovskite on an Al 2 O 3 substrate, and the charge transfer dynamics at TiO 2 /perovskite/P3HT interfaces, which are summarized in Table 1, it is possible to calculate the charge separation efficiency. [31] First, it is determined that the hole injection efficiency η Hinj is almost 100% and the electron injection efficiency η Einj is about 86%. Thus, the charge separation efficiency η Csep (η Csep =η Hinj η Einj ) is 86%. The 14% loss in η Csep results from the fast recombination component, with a lifetime of 4 ps, of photoexcited carriers in the Sn/Pb cocktail perovskite, which may originate from the nonradiative recombination owing to the defects in the sample. Therefore, in order to achieve 100% charge separation efficiency, the crystalline quality of the Sn/Pb cocktail perovskite has to be improved. The optical absorption is estimated to be about 95% in this case, due to the fact that the perovskite material is much more absorbing over a broader range up to 1000 nm, which could result in complete absorption (such as at 470 nm) in thin films such as 500 nm. [23] As shown in Figure 13, the IPCE value at 470 nm, corresponding to the excitation wavelength used for the TA measurements, is 68%. So the charge collection efficiency η Ccol is estimated to be 83%. [31] The 17% loss in η Ccol mainly originates from the charge recombination occurring at the TiO 2 /P3HT interface. In order to improve IPCE and Jsc for the solar cell, direct contact between P3HT and TiO 2 should be avoided or at least suppressed by preparing a pinhole-free Sn/Pb cocktail perovskite layer on TiO 2 or by inserting a barrier/blocking layer at the interface. [31]  As shown in Figure 13(b), J sc , V oc , FF and PCE of the Sn/Pb cocktail perovskite solar cells are 22.61 mA/cm 2 , 0.21 V, 0.37 and 1.77%, respectively. It is obvious that the low PCE is mainly due to the low V oc and FF. J sc could be improved to as high as 30 mA/cm 2 if the charge separation efficiency can be increased to 100% by reducing the fast nonradiative recombination rate and if the charge collection efficiency can be increased to more than 95% through suppression of the recombination occurring at the TiO 2 /P3HT interface. For V oc , the bandgap-voltage offset E G -qV oc is calculated to be as large as 1 eV, where E G is the bandgap of 1.21 eV and q is the elementary charge. This bandgap-voltage offset is quite similar to that of amorphous silicon, and is thought to be closely related to the larger Urbach energy as shown in Figure 14. The recombination resistance from the I-V curve was estimated to be 27 Ω cm 2 , which could also result in a lower V oc . The smaller recombination resistance estimated is mostly due to the faster recombination of electrons and holes at the TiO 2 /P3HT interface. Thus, the lower V oc in the Sn/ Pb cocktail perovskite solar cells is due to two reasons: (1) the larger Urbach energy; (2) the smaller recombination resistance. The lower FF is also due to reason (2). Therefore, these findings imply that the photovoltaic performance of the Sn/Pb cocktail perovskite solar cells can be further improved by decreasing the Urbach energy, i.e., reducing the defects in the perovskite, and by reducing the recombination occurring at the TiO 2 /P3HT interface through appropriate interface engineering such as passivation or inserting a barrier/blocking layer. [31] recombination occurring at the TiO 2 /P3HT interface through appropriate interface engineering such as passiv barrier/blocking layer. 31 Figure 14 The bandgap-voltage offset (E G -qV oc ) versus Urbach energy for typical photovoltaic absorber materials (fro for the Pb and Sn/Pb cocktail perovskites used in our study at room temperature.

Conclusions
In summary, by conducting a systematic study on the optical absorption, photoexcited charge carrier lif Figure 14. The bandgap-voltage offset (E G -qV oc ) versus Urbach energy for typical photovoltaic absorber materials (from reference 18) and for the Pb and Sn/Pb cocktail perovskites used in our study at room temperature.

Conclusions
In summary, by conducting a systematic study on the optical absorption, photoexcited charge carrier lifetimes, and charge separation and charge recombination dynamics, we have explored the ways through which the photovoltaic performance of Pb and Sn/Pb cocktail perovskite solar cells can be improved.
For MAPbI 3 solar cells, we find that the great differences in the IPCE and J sc of the two kinds of cell result from the great difference in the charge recombination between the electrons in TiO 2 and holes in spiro-OMeTAD. These results indicate that the key for improving the IPCE and J sc of the perovskite-based solar cells is charge collection efficiency, instead of charge separation efficiency. Thus suppressing the recombination through proper interfacial engineering for perovskite-based solid hybrid solar cells is crucial for achieving a high efficiency.
For the Sn/Pb cocktail perovskite solar cells, it is determined that the bandgap is 1.21 eV and light harvesting can be extended to a wider wavelength over 1000 nm. The Urbach energy is calculated to be 34 meV, which is more than twice that of the MAPbI 3 perovskite. Three recombination processes for the photoexcited carriers were found with lifetimes being 4 ps (55%), 650 ps (25%) and one much larger than 3 ns (20%). The larger Urbach energy and the faster recombination suggest that there are defects in the prepared Sn/Pb cocktail perovskite. These results are significantly different from those of the MAPbI 3 perovskite, in which case the photocarrier lifetimes are larger than 100 ns and almost no defects were observed. Moreover, the charge separation and charge recombination dynamics were explored. We find that both the electron injection into TiO 2 and hole transfer to the P3HT layer occurred on a timescale of 1 ps. It was surprising to find that the charge recombination lifetime at the Sn/Pb cocktail perovskite/TiO 2 interface was as long as 880 µs, which is 2-3 orders of magnitude greater than that occurring at the MAPbI 3 /TiO 2 interface. However, the charge recombination lifetime became shorter, i.e., 190 µs, when P3HT was employed as a HTM. This implies that the charges can be more effectively collected without the HTM. On the basis of the above results, we find that the loss in the charge separation efficiency originates from fast photoexcited carrier recombination with a lifetime of 4 ps, and the loss in charge collection efficiency was due to the charge recombination occurring at the TiO 2 /P3HT interface, thus leading to lower IPCE values. Also, the low V oc and FF were found to result from the large Urbach energy and the recombination occurring at the TiO 2 /P3HT interface. These findings indicate that the photovoltaic performance of Sn/Pb cocktail perovskite solar cells can be further improved by reducing the defects in the material and the recombination occurring at the TiO 2 /HTM interface through proper interfacial engineering.