Optimized Carrier Extraction at Interfaces for 23.6% Efficient Tin–Lead Perovskite Solar Cells


 Carrier extraction is a key issue which limits the efficiency of perovskite solar cells. In this work, carrier extraction is improved by modifying the perovskite layers with a combination of ethylenediammonium diiodide post-treatment and glycine hydrochloride additive. Ethylenediammonium dications primarily affect the top surface of the perovskite films, while glycinium cations preferentially accumulate at the bottom region. The top and bottom interface modifications improve the crystallinity of the perovskite films and lower the density of electrical traps via surface passivation effects, resulting in long charge carrier lifetimes. The orientated aggregation of the ethylenediammonium and glycinium cations at the charge collection interfaces result in the formation of surface dipoles, which facilitate charge extraction. The performance of the treated solar cell devices also increases. The fill factor rose to 0.82, and the power conversion efficiency reaches 23.6% (23.1% certified). The open circuit voltage reaches 0.91 V, just 0.06 V below the Shockley–Queisser limit. The unencapsulated devices also show improved stability under AM 1.5G, retaining over 80% of the initial efficiency after 200 h continuous operation in inert atmosphere. Our strategy is also successfully applied to centimeter-scale devices, with efficiencies up to 21.0%.


Materials
Unless otherwise stated, all materials were used as received without further purification.

Fabrication of Perovskite Thin Films
The perovskite film preparation was conducted in an Ar-filled glove box (H2O, O2 <0. To prepare the samples with additive, 2 mol% (4.0 mg, 0.036 mmol) of GlyHCl with respect to the total amount of SnI2 and PbI2 was added to the precursor solution. The precursor solution was stirred at 45 C for 40 min and filtered through a 0.20 m PTFE filter before use. To spin coat the 3 films, 200 L of the room temperature precursor solution was applied to the substrate. A two-step spin coating program was used. The first step was at 1000 rpm for 10 s with an acceleration of 200 rpm s -1 , and the second step was 4000 rpm for 40 s with an acceleration of 1000 rpm s -1 . Room temperature chlorobenzene (400 L) was used as the antisolvent. The chlorobenzene was quickly dripped onto the surface of the spinning substrate over an interval of 1 s during the second spin coating step at 20 s before the end of the procedure. The substrate was then immediately annealed on a 100 C hot plate for 10 min, followed by annealing at 65 C for over 10 min. For the EDAI2 post-treatment, 1.0 mg EDAI2 was added to 1.0 mL IPA and 1.0 mL toluene. The mixed solvent solution was stirred at 70 C for 3 h and then filtered through a 0.20 m PTFE filter before spin coating. 120 L of the EDAI2 solution was applied to the annealed and cooled perovskite films by spin coating at 4000 rpm for 20 s with an acceleration of 1333 rpm s -1 . Following spin coating, the films were immediately annealed again at 100 C for around 5 min (see more detail at Fig. S47).
For the space-charge-limited current (SCLC) measurements, hole-only and electron-only devices were prepared. For the hole-only devices, 150 µL PTAA solution (20 mg mL -1 in chlorobenzene, doped with 1 wt% F4-TCNQ) was spin-coated onto the perovskite films fabricated on the PEDOT:PSS-coated substrates at 2000 rpm for 30 s with an acceleration of 400 rpm s -1 , and the films were immediately annealed at 65 C for around 5 min. For the electron-only devices, doublesided C60 layers (20 nm) were sequentially deposited by thermal evaporation before and after the perovskite layer preparation. For all samples, 100 nm of silver was deposited through a metallic aperture mask to form the top electrode.
For the films fabricated on PTAA/PFN-coated FTO, 120 µL PTAA solution (2 mg mL -1 in chlorobenzene) was spin-coated onto the FTO substrates at 4000 rpm for 20 s, and the films were immediately annealed at 130 C for around 10 min. Then, the PTAA-coated FTO substrates were coated with PFN-I. 120 µL PFN-I solution (0.1 mg mL -1 in DMF) was spin-coated onto the PTAAcoated FTO substrates at 4000 rpm for 20 s without annealing.

Fabrication of solar cell devices
Glass/FTO substrates (10 Ω sq -1 , AGC Inc.) were etched with zinc powder and HCl (6 M in deionized water), and consecutively cleaned with 15 min ultrasonic bath in water, acetone, detergent 4 solution (Semico Clean 56, Furuuchi chemical), water, and isopropanol, followed by drying with an air gun, and finally plasma treatment. The PEDOT:PSS hole transport layer was fabricated from an aqueous dispersion which was filtered through a 0.45 m PVDF filter and then spin coated on the FTO substrate using a spin program of 10 s at 500 rpm followed by 30 s at 4000 rpm. The films were then annealed in air at 140 C for 20 min. After transferring to an Ar-filled glove box (H2O, O2 <0.1 ppm), the substrates were degassed at 140 C for 30 min. The perovskite layer was fabricated on PEDOT:PSS following the above-mentioned procedure. The samples were moved under Ar to a vacuum deposition chamber, where 20 nm of C60 (deposition rate 0.01 nm s -1 ) and 8 nm of BCP (deposition rate 0.01 nm s -1 ) were deposited by thermal evaporation. The top electrode was prepared by depositing 100 nm of silver (Ag) through a shadow mask. The deposition rate for Ag was firstly set as 0.005 nm s -1 to reach 5 nm, then raised to 0.01 nm s -1 to reach 20 nm, and finally raised to 0.08 nm s -1 to reach the target thickness.

Characterization
For the time-resolved photoluminescence (TRPL) measurements, the samples were excited by a picosecond pulsed laser with a wavelength of 688 nm (Advanced Laser Diode System) and the excitation frequency of 10 kHz. To shorten the measurement time, the PL signal was divided with a beam splitter (transmission 50%) and detected with a pair of avalanche photodiodes (APD) (ID Quantique). The TRPL signal on each APD was recorded using time-correlated single photon counting boards (Pico Quant). Under weak excitation conditions, a defocus lens was used in order to maintain a high incident photon flux into the APDs. The initial carrier densities were calculated from the excitation photon flux and the absorption coefficient of samples at 688 nm (4.4 × 10 4 cm −1 , Fig. S29). Using a variable neutral density filter, the excitation laser power was adjusted to achieve the desired initial carrier density. The TRPL dynamics were analyzed by the rate equation model reported in our previous work 1 .
Scanning electron microscopy (SEM) was performed with a Hitachi S8010 ultra-high-resolution scanning electron microscope (Hitachi High-Tech Corporation).
UV-vis absorption measurement was performed with a Shimadzu UV-3600 plus spectrometer (Shimadzu Co., Ltd. Time-of-flight secondary ion mass spectrometry (ToF-SIMS) measurements were carried out using a ToF.SIMS 5 (IONTOF GmbH) operated in the high lateral resolution mode (burst alignment mode). A 50 kV Bi3 2+ primary ion beam with pulses width of 125 ns was used for data acquisition. Primary ion dose density (PIDD) was maintained less than approximately 5 × 10 11 ions cm -2 in each measurement cycles to prevent sample damage from the irradiation of primary ion beam. For sample etching, a 10 kV Ar gas cluster ion beam (Ar-GCIB) with a center size of approximately 1200 atoms was used as sputtering ion beam. The raster area of the primary ion beam was 30 µm × 30 µm and that of the sputtering ion beam was 500 × 500. Depth profiles and three-dimensional images of EDAI2/GlyHCl-treated ingredients (or ions) were reconstructed from 7 the raw data after data acquisition. In addition, mass spectra and depth profiles with high mass resolution were acquired from an area of 200 µm × 200 µm by using high mass resolution mode (bunching mode). A low-energy electron-flood gun was used for charge compensation.    perovskite films. Noting that, the size of Clis even smaller than that of Iwithin the perovskite composition. Thus, Clis highly mobilizable in the whole wet process of the film fabrication.
Additionally, Clis too small to be composed to the perovskite structure in this case, it is therefore expelled out of the perovskite structures, and ends up at the grain surfaces, especially the top the bottom surfaces of the resultant films.   S19a). The crystallites were collected and dried under vacuum, before re-dissolving them in dimethyl sulfoxide-d6 and measuring the 1 H NMR spectrum. Signals from GlyH + could be identified, in addition to those from FA + and MA + (Fig. S19b). By comparing the integrated peak areas, the ratio of GlyH + against the total of A site component (Cs + , FA + , MA + ) was estimated to be 0.20. Assuming all the materials bind as a full coverage monolayer to the perovskite surfaces, an analysis of the surface/volume ratios suggests this would correspond to the formation of ca. 20 nm nanocrystals. More importantly, we note that the hydrogen atoms from the ammonium head of GlyH + are shifted by 0.32 ppm compared with the pristine GlyHCl (Fig. S19c), indicating that the ammonium group of GlyH + cation has an affinity for the perovskite components.

XRD characterization
To confirm the effect of GlyHCl on the enhancement of the film crystallinity, we also compared the XRD patterns of the films fabricated with the addition of 2, 4, and 8 mol% GlyHCl without EDAI2 post-treatment. The intensity of the (100) peak increases with the amount of GlyHCl added into the precursor solution, and no new peaks or peak-shifts can be observed. We can therefore conclude that GlyHCl enhances the crystallinity of the resultant perovskite films without inducing new crystalline phases, such as the low dimension perovskite crystals.

Surface treatment
The composition of the co-solvent system (isopropanol and toluene with the volume ratio of 1:1) used for the EDAI2 post-treatment is very important. The polarity of the isopropanol helps the organic ammonium salts to anchor to the film surface, enhancing defect passivation 2 . Sncontaining perovskites are soluble in isopropanol 3 , however, so toluene was applied to reduce the overall viscosity of the solution to accelerate the dispersion of the EDAI2 solution on the films before spinning. The viscosity of isopropanol and toluene is 2.40 and 0.59, respectively.  (Table S2), confirming the polishing effect of the post-treatment with EDAI2 solution.  To further investigate the structural characters at the film surface, we conducted grazing incidence XRD (GIXRD) measurements with incident angles of 0.2, 0.4, and 1.0° (Fig. S26 and 27). As the incident angle increases, more structural information is collected from regions further away from the top surface of the perovskite layer. As incident angle increased, the interplanar space in the crystal lattice of the films with treatments ( Fig. S26a and b) expanded, indicating that the film at the surface region is more compressed than the bulk 4 . For the films fabricated with different modifications, the intensity of the (100) peaks increases especially with the incident angle of 0.2° (Fig. S26d), confirming the benefits of the treatments on the film crystallinity, particularly at the surface region.

PL dynamics
At the range of excitation intensities under consideration here, the dependences of PL decay curves, PL lifetimes, and initial PL intensities shown in Fig. 2c, d, S31 are well explained by a simple rate equation that includes single-carrier trapping and two-carrier (electron−hole) radiative recombination 1 : Here n and IPL are the photoexcited carrier density and the PL intensity, respectively. A, B, and N are the carrier trapping rate, the two-carrier radiative recombination coefficient, and the unintentional doped carrier density, respectively. Note that photoexcitation produces equal numbers of electrons and holes. The initial photoexcited carrier density n (t = 0) = n0 is given as follows: Here h, c, λ, α, Iex, and Fex are the Planck constant, the speed of light, the excitation wavelength, the absorption coefficient at the excitation wavelength, the excitation fluence, and the excitation photon flux, respectively. In our TR-PL experiments, λ = 688 nm and α = 4.4×10 4 cm −1 . The excitation fluence of Iex = 100 nJ cm −2 corresponds to n0 = 1.5×10 16 cm −3 and Fex = 3.5×10 11 photons cm −2 , comparable to the incident photon flux under AM1.5G.
By solving the rate equations, the time-dependent PL intensity can be obtained as follows: where and 0 = ( + 0 ) 0 (6) Here, I0 is the initial PL intensity, while ξ is a constant that dependent on several factors, including the detection efficiency and the charge carrier extraction efficiency. According to the global fitting using Eqs. (4) Dashed lines represent the fit of the decay signals to the rate equations reported in our previous work 1 .

Space-charge-limited currents (SCLC) characterization
Trap densities in the perovskite films were estimated from dark current−voltage (I−V) curves obtained for FTO/PEDOT:PSS/perovskite/PTAA/Ag hole-only, and FTO/C60/perovskite/C60/Ag electron-only devices 5,6 . Fig. S32 shows the I−V curves of the hole-only devices. The voltages where the currents start to sharply increase can be assigned to the trap-filled limit voltage (VTFL), from which the trap density, Ntrap, can be estimated using the following relation: VTFL = Ntrap(eL 2 )/(2εrε0), where e is the elementary charge of the electron, ε0 is the vacuum permittivity, εr is the relative dielectric constant of the perovskite (around 32) 7 , and L is the thickness of the perovskite film. The VTFL values are given in Table S2. The estimated trap densities for the control, EDAI2-treated, and EDAI2/GlyHCl-treated films were 3.0 × 10 15 , 2.9 × 10 15 , and 1.7 × 10 15 cm −3 , respectively. The EDAI2/GlyHCl-treated films have the lowest trap-state density, and therefore, the corresponding devices should have the smallest number of recombination events. Meanwhile, the EDAI2/GlyHCl-treated films also showed a low density of electron traps (2.9 × 10 14 cm -3 ), as determined from the electron only devices.          S46. The integrated ESR spectra for the data presented in Fig. 4d.

Further detail about the filming processes for better lab-to-lab reproducibility
We observed white dots forming at the bottom surface of the films processed on 1.1 mm thick ITO substrate. However, this phenomenon cannot be observed in the films deposited on 1.8 mm thick FTO-based substrates with the normal fabrication conditions. We also found that the white dots will be formed even on the films fabricated with FTO substrate, if we drip the antisolvent preheated up to 45 °C. We therefore believe that the formation of the white dots is related to the cation exchange processes during the crystal growth (see the details in the main text). The cation exchange is a dynamic process, which is closely linked to the time dependence of the wet process.
In the case of FTO, if the films were deposited with the hot antisolvent, the wet films tend to dry much faster than the case with the cold antisolvent. In the hot antisolvent dripping case, therefore, the whole cation exchange processes cannot be completed in time, resulting in white dots at the bottom surface of the films. In the ITO case, we may alleviate it by reducing the amount of GlyHCl added into the precursor solution and/or optimizing the fabrication conditions by, for example, i) changing the temperature of the antisolvent, ii) controlling the temperature inside the glovebox, or iii) dripping the antisolvent in a time window that ensures a sufficient time to complete the cation exchange within the whole crystallization process.

Impedance characterization
The complex impedance data provides an accurate and detailed picture of the equivalent circuit resistances 9,10 . Representative complex impedance scans for solar cell devices fabricated with the control, EDAI2-treated, and EDAI2/GlyHCl-treated perovskite layers are shown in Fig. S51a. The series resistance is minimal, while the parallel resistance, indicated by the diameter of the main semicircular feature, increases from 290  cm 2 for the control film, to 445  cm 2 for the EDAI2-

PLQY characterization
We fabricated perovskite films on PEDOT:PSS to evaluate the nonradiative recombination in our system via PLQY measurement. On PEDOT:PSS, the PLQY of the EDAI2/GlyHCl-treated perovskite film is 2.1%, considerably higher than that of the control film (~0.001%), verifying that the nonradiative recombination pathways in the EDAI2/GlyHCl-treated films are largely suppressed.