Photoexcitation-induced passivation of SnO2 thin film for efficient perovskite solar cells

ABSTRACT A high-quality tin oxide electron transport layer (ETL) is a key common factor to achieve high-performance perovskite solar cells (PSCs). However, the conventional annealing technique to prepare high-quality ETLs by continuous heating under near-equilibrium conditions requires high temperatures and a long fabrication time. Alternatively, we present a non-equilibrium, photoexcitation-induced passivation technique that uses multiple ultrashort laser pulses. The ultrafast photoexcitation and following electron–electron and electron–phonon scattering processes induce ultrafast annealing to efficiently passivate surface and bulk defects, and improve the crystallinity of SnO2, resulting in suppressing the carrier recombination and facilitating the charge transport between the ETL and perovskite interface. By rapidly scanning the laser beam, the annealing time is reduced to several minutes, which is much more efficient compared with conventional thermal annealing. To demonstrate the university and scalability of this technique, typical antisolvent and antisolvent-free processed hybrid organic–inorganic metal halide PSCs have been fabricated and achieved the power conversion efficiency (PCE) of 24.14% and 22.75% respectively, and a 12-square-centimeter module antisolvent-free processed perovskite solar module achieves a PCE of 20.26%, with significantly enhanced performance both in PCE and stability. This study establishes a new approach towards the commercialization of efficient low-temperature manufacturing of PSCs.


INTRODUCTION
Inorganic metal oxide semiconductors are widely used as electron transport layers (ETLs) for perovskite solar cells (PSCs), which efficiently extract and transport electrons from the perovskite layer into the electrode [ 1 -5 ].Tin (IV) oxide (SnO 2 )based ETLs offer desirable band alignment and electron mobility, while being processable at much lower temperatures [ 6 , 7 ].Among various deposition methods [ 8 -10 ], SnO 2 nanoparticle-based PSCs prepared by using chemical bath deposition (CBD), which provides conformal coverage between the perovskite layer and the electrode, have demonstrated the best performance so far [ 11 -13 ].However, the intrinsic defects (e.g.oxygen vacancies, tin interstitials) and surface defects (e.g.hydroxyl groups and Sn dangling bonds) that are unavoidably formed in SnO 2 generate massive shallow trap states near the conduction band and result in carrier recombination at the SnO 2 /perovskite interface [ 14 , 15 ].To reduce the carrier loss, the ETL film prepared by using CBD generally requires an annealing condition with a temperature of ≥170°C for ≥1 hour [ 11 -13 , 16 ].This is in part because the ETLs normally show poor heat transfer and large thermal inertia for passivation under near thermal equilibrium, which makes it challenging to reduce the annealing temperature and time, bringing additional energy costs.Currently, except for thermal annealing at high temperatures for at least 1 h, there is no better way to reduce the annealing temperature of tin oxide films prepared by using the CBD method [ 17 , 18 ].At the same time, thermal annealing is general ly fol lowed by buried interface passivation to further reduce surface defects, which requires additional thermal annealing and complicates the preparation process [ 19 , 20 ].
Herein, we report a novel non-equilibrium, photoexcitation-induced passivation (PiP) technique that uses ultrashort laser pulses.The ultrafast photoexcitation and following electron-electron and electron-phonon scattering processes induce an ultrafast electron and phonon heating process, thus ensuring efficient low-temperature annealing of SnO 2 nanoparticle-based ETLs prepared by using CBD.By rapidly scanning a laser beam, the annealing process of a 5 cm × 5 cm sample can be finished within 30 s at room temperature, which facilitates the scale-up and energy saving of fabricating PSCs.The two representative perovskite-based PSCs fabricated via this method achieve a great enhancement in comparison with the control devices, and a power conversion efficiency (PCE) of 20.26% for a 12-square-centimeter perovskite solar module (PSM) is achieved as well.

PiP of SnO 2 thin film
The schematic of a home-built PiP set-up and the mechanism of PiP are shown in Fig. 1 a.A femtosecond laser beam is reflected by a high-speed rotating polygon mirror and then focused on the sample.The set-up has a laser scanning speed of > 36 m s −1 in a working range of 300 mm, with a stage movement speed of > 1.5 mm s −1 , which ensures annealing a 5 cm × 5 cm sample in 30 s and a 10 cm × 10 cm sample in 60 s, as shown in Fig. 1 a (i).The detailed home-built PiP experimental set-up can be found in the Methods and Supplementary material .The demonstration and actual processing video can be found in the Supplementary movie .
The PiP process is achieved by using ultrafast intensive excitation after laser pulse energy deposition in tens of femtoseconds, followed by the electronelectron and electron-phonon scattering processes (in hundreds of femtoseconds), as shown in Fig. 1 a (ii).The interaction between the ultrafast laser and the SnO 2 film starts from direct photon absorption by excitation of intrinsic free carriers and shallow trap states.The absorption coefficient is 1650 cm −1 at 1030 nm as measured.The carrier population wi l l rapidly rise up to the anti-bonding states (conduction band) by impact ionization via the electronelectron scattering process in tens of femtoseconds.The population of anti-bonding states raises the free energy of the excited system, generating a stretching force on each bond, which efficiently cleaves weak or metastable bonds, thus passivating the surface defects and oxygen vacancies of the SnO 2 film [ 21 -23 ].
The following electron-phonon scattering process to reach an equilibrium temperature of the electron and phonon system induces an ultrafast phonon heating process (for 28 mJ cm −2 laser fluence, the maximum temperature rise is 22.6 K) with a heating rate of > 10 13 K s −1 in hundreds of femtoseconds (see the Supplementary material for theoretical analysis of temperature rise by PiP) and an ultrafast thermal gradient with the Gaussian distribution is generated in the focus zone ( Supplementary Fig. S1a ) [ 24 ].On the way to cooling the excited system, atoms obtain excessive kinetic energy to align into an ordered arrangement, thus passivating the bulk defects of the SnO 2 and improving the crystallinity of the SnO 2 [ 21 , 25 ].This temperature gradient decays rapidly via thermal diffusion in 25 ns with a quenching rate of ∼10 9 K s −1 .The accumulation temperature after 50 consecutive laser pulses is only 6.5 K ( Supplementary Fig. S1b ), showing that the temperature rise after each single pulse drops to near room temperature before the next pulse arrives, demonstrating that the accumulation effect is negligible in the PiP process of the SnO 2 film.This shows that the whole passivation process is completed very fast with much more energy and time efficiency compared with the conventional annealing technique.
The microscopic characterization of the SnO 2 thin film cross sections using high-resolution transmission electron microscope (HRTEM) and the corresponding energy-disperse X-ray spectroscopy of Sn, O, F and Pt ( Supplementary Fig. S2 ) present a complete coverage over the fluorine-doped tin oxide (FTO) surface.By comparing the HRTEM images before and after the PiP, it is found that a large number of amorphous particles existed in the original SnO 2 (Fig. 1 b, part i-1).This is more clearly identified from the random distribution of Sn atoms in the atomic-resolution image and the corresponding fast Fourier transform (FFT) pattern (Fig. 1

Morphology, optoelectronic properties and surface defect characterization of SnO 2 films
To investigate the reasons why the PiP-SnO 2 thin film efficiently improves the cell performance, the surface morphology of SnO 2 thin films was studied by using scanning electron microscopy (SEM) images, as shown in Fig. 2 a.Compared with the smooth, bare FTO surface, the SnO 2 thin film deposited on the FTO is composed of compact nanoparticles.All SnO 2 thin films are uniform and full-coverage except for the much higher laser fluence.This is because at high laser fluence, SnO 2 nanoparticles tend to coalesce into large crystallites, resulting in pinholes in the SnO 2 thin films.These large crystallites and pinholes act as carrier trap sites in the SnO 2 /perovskite interface where the energy is lost through non-radiative recombination pathways, resulting in a reduction in carrier lifetimes.The contact angle first increases with increasing laser fluence (Fig. 2 b), which indicates a reduction in surface hydrophilic groups.However, when the laser fluence is further increased, the non-uniformity of the surface increases the hydrophilicity.The transmittance spectra of the original-SnO 2 , 170°C thermal-annealed SnO 2 for 1 hour (170°C-SnO 2 ) and PiP-SnO 2 thin films on FTO substrates are shown in Fig. 2 c and Supplementary Fig. S5a .There is a slight increase in the transmittance of the substrate with the PiP-SnO 2 thin films, which could contribute to the high J SC of the PSCs.Since the SnO 2 thin film is very thin relative to the FTO layer, there is no clear difference in transmittance.The conductivity of the SnO 2 thin films on the FTO shows an increase of > 20% by the PiP (Fig. 2 d and Supplementary Fig. S5b ).Moreover, the electron mobility of various ETLs measured by using a space charge-limited current method is shown in Supplementary Fig. S6 [ 26 ].It can be found that the electron mobility of the SnO 2 is higher than that of the original-SnO 2 and 170°C-SnO 2 , which is beneficial to SnO 2 thin films to transport electrons generated from the perovskite layer, thereby improving the cell performance.
Since the film deposition takes place in a waterbased solution, OH groups (surface defects) inevitably appear in the SnO 2 thin film, which is proved by the X-ray photoelectron spectrum (XPS) in Fig. 2 e-g and Supplementary Fig. S7a .After the calibration of the spectra by using the C1s peak, the deconvolution of the O1s peak was executed in detail to detect the four oxygen species for each sample.Generally, there are two kinds of OH groups commonly existing on the SnO 2 thin film surface [ 27 ].One is OH (bridge) integrated with two metal sites at 531.5 ( ±0.1) eV and the other is OH (terminal) , which binds to one metal site at 532.2 ( ±0.1) eV [ 27 -29 ].The relative proportions of OH (bridge) and OH (terminal) species for each sample are listed in Supplementary Table S1 .The relative proportions of OH (bridge) and OH (terminal) decrease both for 170°C-SnO 2 and PiP-SnO 2 thin films, which indicates that the PiP achieves a similar effect to conventional hightemperature thermal annealing.Compared with the 170°C annealing for 1 hour, the PiP is much more time-efficient.From the XPS results, we can confirm that the PiP process can not only drive the SnO 2 undergoing a phase transition from the amorphous state to the crystalline state, but also reduce the metastable surface functional groups on SnO 2 thin films, as shown in Supplementary Fig. S7b .
The metastable OH (bridge) and OH (terminal) could be cleaving after photoexcitation, which is also indicated by the enhanced Raman peak at 632 cm −1 (corresponding to the fundamental active Raman vibration modes A 1g ) in Supplementary Fig. S8 [ 30 , 31 ].

Carrier transport behaviors between SnO 2 films and perovskite films
An ideal band alignment is critical to efficiently extracting electrons from perovskite to ETL and preventing hole quenching at the ETL/perovskite interface [ 15 , 32 , 33 ].As shown in Fig. 3 a, the first principle density functional theory (DTF) calculations reveal that the removal of metastable surface OH groups reduces the conduction band minimum, facilitating charge extraction.At the same time, it increases the valence band maximum, so as to efficiently transfer electrons and block holes ( Supplementary Fig. S9 ) [ 7 ].Ultraviolet photoelectron spectroscopy (UPS) supports that the deeper valence band is achieved (Fig. 3 b) [ 34 ].Steady-state photoluminescence (PL) and time-resolved photoluminescence (TRPL) measurements were also carried out to probe the photocarrier dynamics between the SnO 2 layers and the perovskite layer (Fig. 3 c and Supplementary Fig. S10 ).The crystallinity of the deposited perovskite film on different SnO 2 films is consistent, which is confirmed by the results of X-ray diffraction (XRD) analysis ( Supplementary Fig. S11 ).The PiP-SnO 2 /perovskite film has a lower PL intensity and shorter carrier lifetime than the other films.Timeresolved confocal PL lifetime mapping of selectively annealed SnO 2 by PiP-based perovskite film and the TRPL spectra of the corresponding region demonstrate the more efficient electron extraction after the PiP (Fig. 3 d).The dotted box in the middle is the selectively PiP-SnO 2 /perovskite region, which exhibits a shorter PL lifetime than the original part.These results further support the reduction in charge recombination in the PiP-SnO 2 /perovskite interface.To further investigate the photocarrier dynamics at the ETL/perovskite interface, femtosecond transient absorption spectroscopy was employed, as shown in Supplementary Fig. S12 , to show the time-resolved difference absorption spectra in the 10 ps-1 ns timescale of different SnO 2 /perovskite films.The amplitude of the ground-state bleaching (GSB) signal is proportional to the number of photocarriers in the excited state.(a) J-V curves and photovoltaic parameters of a representative antisolvent-processed PSC that used original-SnO 2 /FA 0.95 MA 0.05 , 170°C-SnO 2 /FA 0.95 MA 0.05 and PiP-SnO 2 /FA 0.95 MA 0.05 .(b) J-V curves and photovoltaic parameters of a representative antisolvent-free-processed PSC that used original-SnO 2 /FA 0.83 Cs 0.17 , 170°C-SnO 2 /FA 0.83 Cs 0.17 and PiP-SnO 2 /FA 0.83 Cs 0.17 .(c) J-V curves of the PSM recorded in the lab using a mask with an aperture area of 12 cm 2 .Average PCE evolution of the unencapsulated (d) FA 0.95 MA 0.05 and (e) FA 0.83 Cs 0.17 devices measured over a 180-h stability test under light soaking with AM1.5G-simulated illumination (RH ∼ 20 ± 5%).The shaded regions represent the variation range of the PCE obtained from six cells.(f) Cross-sectional SEM images of original-SnO 2 /FA 0.83 Cs 0.17 /i-BABr/spiro/Au and PiP-SnO 2 /FA 0.83 Cs 0.17 /i-BABr/spiro/Au aged under continuous AM1.5G-simulated illumination for 180 hours.The degraded area of the perovskite film at the SnO 2 /perovskite interface is marked by the dashed line.
This signal decays when the perovskite undergoes electron-hole recombination or charge transfer to an accepting layer [ 35 ].It can be clearly seen that the GSB signal of the perovskite based on the PiP-SnO 2 decays faster compared with the other two samples in the same timescale, which supports the more rapid charge transfer to the SnO 2 ETL [ 36 ].The decay traces of SnO 2 /perovskite have been well fitted with a three-exponential function (Fig. 3 e) and the relevant decay time constants are collected in the inserted table.As shown in Fig. 3 f, the fastest time constant τ 1 more likely arises from the excitation of electrons [ 37 ].The faster time constant τ 2 is plausibly attributed to the electron transfer from the photoexcited perovskite to the SnO 2 conduction band, while the slowest time constant τ 3 is all within 40 0-60 0 ps, which can be ascribed to the charge recombination dynamics of the perovskite as reported previously [ 38 , 39 ].The τ 2 for the PiP-SnO 2 /perovskite film is much shorter, which indicates the significant enhancement of the electron-transfer rate (1/ τ 2 ).There is a slight increase in τ 3 , indicating that the electrons can survive for a longer time in the PiP-SnO 2 -based perovskite.stability of the devices with PiP-SnO 2 under light soaking with AM1.5G-simulated i l lumination at a temperature of 20 ± 5°C is greatly enhanced compared with other annealing conditions when stored under ambient air conditions with 20% ± 5% relative humidity (RH) without encapsulation over 180 hours.Upon introducing the PiP strategy to passivate the SnO 2 layer, the bulk morphology of the SnO 2 /perovskite bilayer featuring well-defined interfaces is markedly stabilized under continuous AM1.5G-simulated i l lumination for 180 hours (Fig. 4 f and Supplementary Fig. S22 ).Taken together, our results suggest that the PiP of SnO 2 is an effective and universal approach to significantly enhance PSC performance and facilitates the scale-up of PSCs.

CONCLUSION
In summary, in this study, we developed a novel nonequilibrium PiP technique for the low-temperature fabrication of PSCs.The ultrafast and efficient electron-electron and electron-phonon scattering ensure room-temperature annealing of SnO 2 nanoparticle-based ETLs prepared by using CBD with an ultrafast heating rate, which efficiently passivates the surface and bulk defects of SnO 2 .By rapidly scanning a laser beam, the annealing process at room temperature can be finished in a few minutes, which normally takes hours by continuous thermal annealing at 170°C, reducing the energy cost and fabrication time.Our PiP technique has been demonstrated to be a universal and scalable approach in different types of PSCs and PSMs, and shows significantly enhanced performance in both PCE and stability ( Supplementary Table S2 ).This study establishes a new approach towards the commercialization of efficient low-temperature manufacturing of PSCs.

METHODS SnO 2 ETL fabrication
The FTO substrate was first etched by using a femtosecond laser scanning system and then cleaned by sonicating in detergent, deionized water and ethyl alcohol for 30 min each.The CBD solution was prepared by dissolving 5 g of urea, 100 μL of mercaptoacetic acid and 5 mL of HCl (37 wt%) into 400 mL of deionized water and then adding 1.096 g of SnCl 2 2H 2 O to the solution (0.012 M).The cleaned FTO substrate was immersed in the diluted CBD solution (0.002 M) and reacted at 90°C.After 3 hours, the SnO 2 deposited FTO substrate was washed by sonicating in deionized water for 5 min.The thus prepared SnO 2 thin film was then processed under different annealing conditions.

The home-built PiP set-up
A femtosecond laser source (TANGOR-100 W, Amplitude) with a center wavelength of 1030/515 nm, pulse duration of ∼425 fs and osci l lator frequency close to 40 MHz was integrated into a polygon laser scanning system (LSE3 00 STD, Nex t scan technology) that was equipped with a high-precision linear stage (PRO165LM, Aerotech).An attenuator (2-EWP-R-0515-M, Altechna) was placed in the optical path to adjust the laser pulse energy fluence.The polygon laser scanning speed could be up to 100 m s −1 .To measure the exact pulse energy, a photodiode power meter was used (S120VC, Thorlabs).The in-scan resolution and the cross-scan resolution were set to 0.9 and 20 μm, respectively, so that the laser spot (45 μm) could fully cover the sample.The laser scanning speed was 36 m s −1 and the stage moving speed was 1.66 mm s −1 .

PSM laser etching
P1, P2 and P3 were all scribed by using a femtosecond laser machine (Pharos-10 W, light conversion).The FTO glass was first etched to form the module substrate with six strips (P1) with a laser power of 0.15 W, pulse duration of 10 ps, repetition rate of 10 KHz and laser scanning speed of 250 mm s −1 .After the deposition of the spiro-OMeTAD film, the sample was re-etched to form P2 lines with a laser power of 0.3 W, pulse duration of 2 ps, repetition rate of 200 KHz and laser scanning speed of 500 mm s −1 .Finally, an effective series of connected modules was formed by etching the Au to form P3 lines with a laser power of 5 mW, pulse duration of 260 fs, repetition rate of 1 KHz and laser scanning speed of 20 mm s −1 .With precise control of the P1-P2-P3 process, we achieved a high geometric FF ( ∼97.7%).

Natl 2 Defect Amorphous SnO 2 LFigure 1 .
Figure 1.Diagram of the photoexcitation-induced ultrafast passivation (PiP) process and phase transition of SnO 2 thin films.(a) Schematic of the PiP procedure during the annealing of SnO 2 thin films and the specific mechanisms of the annealing process including photoexcitation, electron-electron and electron-phonon scattering, rearrangement of atoms and defects passivation.(b) High-resolution TEM images of (i) original-SnO 2 and (ii) PiP-SnO 2 (the insets show fast Fourier transform patterns of the TEM images).
b, part i-2).More representative HRTEM images of the original SnO 2 are shown in Supplementary Fig. S3 .As shown in Fig. 1 b, part ii-1, the PiP process leads to the SnO 2 undergoing a transformation from the amorphous phase to the crystalline phase, thus improv-ing the crystallinity of the SnO 2 , which is confirmed by the massive appearance of lattice fringes and regularly arranged Sn atoms.The FFT pattern of the SnO 2 atomic-resolution image (Fig. 1 b, part ii-2) after PiP corresponds to the [100] zone axes.More representative HRTEM images of the SnO 2 after the PiP are shown in Supplementary Fig. S4 .

Figure 2 .
Figure 2. Characterization of SnO 2 films.(a) Schematic illustration of morphologic change of original-SnO 2 , 170°C-SnO 2 and PiP-SnO 2 thin films on FTO substrate and corresponding top-view SEM.Scale bars are 200 nm.Comparison of (b) contact angle, (c) transmittance spectra and (d) conductivity of the original-SnO 2 , 170°C-SnO 2 and PiP-SnO 2 thin films on the FTO substrate.(e-g) The XPS spectrum of O1s of original-SnO 2 , PiP-SnO 2 and high-laser-fluence-treated SnO 2 thin films.

Figure 3 .
Figure 3. Influence of PiP of SnO 2 on its band alignment with perovskite and the carrier transfer processes at the interface of the SnO 2 and the perovskite.(a) Band structure and partial density of states of original-SnO 2 and PiP-SnO 2 .(b) The energy onset and cut-off of UPS spectra of original-SnO 2 and PiP-SnO 2 thin films, respectively, where their work function and valence band maximum can be derived.(c) Steady-state photoluminescence (PL) spectral photon flux for the perovskite film with different ETLs.(d) Time-resolved confocal PL lifetime map and corresponding time-resolved PL traces for selective PiP SnO 2based perovskite film.(e) Transient absorption decay kinetic traces of SnO 2 /perovskite at 750 nm.The kinetic analyses with three-exponential fit are also shown by solid lines.(f) Schematic diagram of the overall photocarrier dynamics between the SnO 2 and the perovskite.

Figure 4 .
Figure 4. Universality and scalability of PiP for PSCs and performance comparison of PSCs.(a)J-V curves and photovoltaic parameters of a representative antisolvent-processed PSC that used original-SnO 2 /FA 0.95 MA 0.05 , 170°C-SnO 2 /FA 0.95 MA 0.05 and PiP-SnO 2 /FA 0.95 MA 0.05 .(b) J-V curves and photovoltaic parameters of a representative antisolvent-free-processed PSC that used original-SnO 2 /FA 0.83 Cs 0.17 , 170°C-SnO 2 /FA 0.83 Cs 0.17 and PiP-SnO 2 /FA 0.83 Cs 0.17 .(c) J-V curves of the PSM recorded in the lab using a mask with an aperture area of 12 cm 2 .Average PCE evolution of the unencapsulated (d) FA 0.95 MA 0.05 and (e) FA 0.83 Cs 0.17 devices measured over a 180-h stability test under light soaking with AM1.5G-simulated illumination (RH ∼ 20 ± 5%).The shaded regions represent the variation range of the PCE obtained from six cells.(f) Cross-sectional SEM images of original-SnO 2 /FA 0.83 Cs 0.17 /i-BABr/spiro/Au and PiP-SnO 2 /FA 0.83 Cs 0.17 /i-BABr/spiro/Au aged under continuous AM1.5G-simulated illumination for 180 hours.The degraded area of the perovskite film at the SnO 2 /perovskite interface is marked by the dashed line.