Fullerene modification of WO3 electron transport layer toward high‐efficiency MA‐free perovskite solar cells with eliminated light‐soaking effect

In perovskite solar cells (PSCs), the light‐soaking effect, which means device performance changes obviously under continuous light illumination, is potentially harmful to loaded devices as well as accurately assessing their efficiency. Herein, chemically stable tungsten trioxide (WO3) with high electron mobility is used as electron transport material in methylamine (MA)‐free PSCs. However, the light‐soaking effect is observed apparently in our devices. A fullerene derivative, C60 pyrrolidine Tris‐acid (CPTA), is introduced to modify the interface between WO3 and perovskite (PVK) layers, which can bond with WO3 and PVK simultaneously, leading to the passivation of the defect and the suppression of trap‐assisted nonradiative recombination. What is more, the introduction of CPTA can enhance the built‐in electric field between WO3 and PVK layers, thereby facilitating the electron extraction and inhibiting the carrier accumulation at the interface. Consequently, the light‐soaking effect of WO3‐based PSCs has been eliminated, and the power conversion efficiency has been boosted from 17.4% for control device to 20.5% for WO3/CPTA‐based PSC with enhanced stability. This study gives guidance for the design of interfacial molecules to eliminate the light‐soaking effect.


| INTRODUCTION
During the past decade, perovskite solar cells (PSCs) have drawn great attention due to their low-cost solution processing and prominent features of the perovskite (PVK) materials, which include adjustable band gap, large absorption coefficient, high carrier mobility, large carrier diffusion length, and small exciton binding energy. [1][2][3] The certificated power conversion efficiency (PCE) of PSCs has been rapidly boosted to 25.7% in only a dozen years, [4][5][6][7][8] showing tremendous potential in a wide range of applications. However, there are still many issues to be addressed before their scalable commercialization. [9] The normal n-i-p structured PSC is the most studied cell system in PSCs. As a key building block of PSCs, an electron transport layer (ETL) located between the transparent electrode and PVK layer is applied, which can effectively transport the photogenerated electrons from the PVK layer to the cathode. [10] As an electron transport material (ETM), titanium dioxide (TiO 2 ) has been widely applied in the n-i-p structured PSCs because of their superior properties in electron extraction, transportation, well-aligned band energy, and the relatively mature manufacturing technique. [11] However, TiO 2 is sensitive to ultraviolet radiation and shows highly photocatalytic activity toward degrading the PVKs, which is detrimental to the stability of PSCs. [12,13] Zinc oxide (ZnO) is then proposed as an alternative for TiO 2 due to its high carrier mobility and nonphotocatalytic activity. [14] However, chemical instability of ZnO interface causes the decomposition of hybrid PVK film into PbI 2 under the extended heat-treatment. [15] Recently, tin dioxide (SnO 2 ) has been proverbially recognized as excellent ETMs of PSCs with impressive device performance owing to its high electron mobility and favorable band alignment with PVKs. [16,17] Nevertheless, exploring new types of ETMs is still an important issue to further push the study on PSCs. Tungsten trioxide (WO 3 ) is another promising ETM for PSCs, which has a wide bandgap (~3.4 eV), good chemical stability, and high electron mobility. [18] Mahmood et al. first explored WO 3 ETM for PSCs. [19] However, the resulting device with only WO 3 as ETL displayed a very low PCE of 3.8%, which could be attributed to rich defect density and fast charge recombination at the WO 3 /PVK interface. Introducing another n-type metal oxide (such as TiO 2 or SnO 2 ) to construct double-layer ETLs with WO 3 could suppress the charge recombination for good cell efficiency. [19][20][21] For instance, You et al. designed a TiO 2 /WO 3 bilayer as ETL to achieve an enhanced PCE of 20.14%, which was much better than the single WO 3 ETL-based device (with an efficiency of 17.04%). [22] Eze et al. used fullerene C 60 to modify WO x and obtained a PCE of 16.07%. [23] Chen et al. employed hexanolassisted low-temperature processed WO x nanocrystalline film as ETL. The increased conductivity and the reduced trap density of the resulting WO x nanocrystalline film boosted the efficiency to 20.77%, which was the highest efficiency of WO x -based PSCs up to date. [24] Nevertheless, much room still remained for the further improvement of WO x as candidate ETM for PSCs.
The light-soaking effect, which means that a continuous light illumination can gradually change the performance of PSCs, has raised concerns about the unstable power output of these devices. [25,26] Therefore, it is crucial to eliminate such light-soaking effects which could accurately assess the efficiency and stability of PSCs. It is reported that light-soaking instability of devices is mainly due to the trap-assisted recombination or charge accumulation at the ETL/PVK and/or hole transport layer/PVK interfaces. [27,28] Wang et al. introduced an n-type helical perylene diimide (PDI 2 ) as effective interfacial layer between TiO 2 and PVK, which promoted charge transport and passivated surface defects, thereby resulting in suppressed interfacial recombination, improved cell efficiency, and reduced light-soaking instability. [29] Zhang et al. used an interfacial modifier, [6,6]-phenyl-C 61 -butyric acid 2-((2-(dimethylamino)ethyl) (methyl)-amino)-ethyl ester, to modify the TiO 2 ETL of the planar PSCs. Both the electron extraction capability of ETL and the quality of PVK film deposited above it were enhanced, thus the PCE and light-soaking stability of the resultant devices were significantly improved. [30] Shao et al. employed a fullerene derivative, fulleropyrrolidine functionalized by a side chain of triethylene glycol monoethyl ether (PTEG-1) as the ETL in p-i-n structured PSCs. The PTEG-1 could effectively alleviate the trap-assisted nonradiation recombination occurred at the interface of PVK/ETL owing to the electron-donor property of the side chains, resulting in the increased efficiency and the eliminated light-soaking effect. [31] The light-soaking effect was also found in WO x -based PSCs, which could be suppressed if the WO x layer was modified by a self-assembled monolayer (C 60 -C 6 -phosphoric acid). [26] However, indepth explanation was not provided in this work.
Herein, we employed WO 3 film as ETL of PSCs, showing an obvious light-soaking effect. C 60 pyrrolidine Tris-acid (CPTA) was therefore introduced as an interface layer between WO 3 and PVK. Surprisingly, the light-soaking effect is completely removed after CPTA modification. The introduction of CPTA can simultaneously bond with W 6+ in WO 3 and Pb 2+ in PVK, which promotes a reduction in defect density and a suppression in trap-assisted nonradiation recombination. Also, both the work function of ETL and the PVK layer are changed simultaneously upon CPTA modification. The increased built-in electric field between ETL and PVK layer can facilitate the electrons extraction from PVK to ITO and restrain the carrier accumulation at the interface. Consequently, the WO 3 /CPTA-based PSCs achieve an impressive PCE of 20.5% along with the elimination of light-soaking effect. In addition, the unencapsulated WO 3 /CPTA-based PSCs exhibit an excellent stability: 90% of its initial PCE is maintained after 500 h continuous light irradiation at 1-sun, and 77% of its original PCE is obtained subjected to 1000 h thermal aging at 85°C. All of them are significantly superior to the pristine WO 3 -based cells. This work affords a viable way for eliminating the light-soaking effects of PSCs.

| RESULTS AND DISCUSSION
Considering the instability of MA + under hightemperature thermal aging, [32] we employed methylamine (MA)-free, Cs 0.15 FA 0.85 Pb(I 0.95 Br 0.03 Cl 0.02 ) 3 -based PVK (CsFA-based PVK) as the light absorber in this study. Figure 1A schematically displays an n-i-p device configuration of ITO/WO 3 ( Figure S1)/CPTA/CsFAbased PVK/Spiro-OMeTAD/Au, and the film preparation process is detailedly described ( Figure S2 and Supporting Information Experiment Section). Interestingly, it is found that the PCE of the pristine WO 3 -ETL formed device progressively increased from 14.5% to 15.8%, 16.4%, 16.8% and peaked at 17.1% with the extension of the light illumination time from 0 to 2, 4, 10, and 20 min (Table S1). The characteristic curves of current density-voltage (J-V) are displayed in Figure 1B, and the evolved process of detailed performance parameters is depicted in Figure S3. The PCE of the WO 3 -based device is no longer improved when the light-soaking time is further extended to 30 min, indicating that~20 min should be the optimal lightsoaking time for the PSC with WO 3 ETL. By contrast, modification by CPTA results in significantly higher and more stable device performance (20.1%) which is almost unchanged regardless of the light illumination time, indicating that the light-soaking effect is effectively eliminated upon CPTA modifying the WO 3 ETL.  To in-depth understand the mechanism of CPTA modification on the elimination of light-soaking effect, the photoelectric properties of ETL are first tested. We measured and calculated the direct conductivity (σ) of WO 3 films before and after CPTA modification by current-voltage (I-V) characteristic curves (shown in Figure 2A). The conductivity of the WO 3 /CPTA film is estimated to be 3.78 × 10 −3 mS·cm −1 , significantly larger than the pristine WO 3 film (2.28 × 10 −3 mS·cm −1 ). This might be due to the electron injection through Lewis coordination between CPTA and W 6+ and result in enhanced charge transport. [33] Atomic force microscopy was performed to compare the surface roughness of WO 3 and WO 3 /CPTA films. As shown in Figure 2B,C, the roughness of WO 3 film decreases from the root mean square of 5.38 to 4.26 nm after CPTA modification, which will provide a favorable interfacial contact between the ETL and PVK and is conducive to uniform growth of PVKs. [34] We further find that the CPTA layer remains on the WO 3 surface even by following the washing procedure with the mixed solvent of DMF/DMSO ( Figures S4 and S5). Moreover, to acquire the distribution of CPTA in the film, we carried out the time-of-flight secondary ion mass spectrometric characterization. [35] Since C 60 might be decomposed by the high-energy ion beams during testing affording multicarbon fragments, we attribute the signal of the multicarbon fragments (C ≥ 8) to the CPTA molecule. As illustrated in Figure S5, although most CPTA molecules are detected at the interface between WO 3 and PVK, some CPTA molecules diffuse into the PVK layer owing to the solubility of CPTA in DMF.
Both PVK films with and without CPTA modification exhibit similar morphology characteristics, thickness, UV-Vis absorption, and crystallization intensity ( Figure S6). The Urbach energy (E u ) of PVK films was used to evaluate their structural quality, which was obtained from their UV-Vis absorption spectra ( Figure S6G) according to the equation α = α 0 exp(hv/ E u ), where α is the absorption coefficient of PVK films and hv is the photon energy. The E u of WO 3 /CPTA-based PVK film is 48.78 meV, lower than the control one (64.64 meV, Figure 2D), demonstrating that WO 3 /CPTAbased PVK film has better quality and fewer defects. [36,37] Previous studies have pointed out that the carrier's accumulation at the PVK/electrode interface could result in a large capacitance effect owing to the interfacial electronic dipole polarization. [38] To assess this, frequency-dependent capacitance (C-f) measurement was performed to compare the capacitance properties of the control and target samples ( Figure S7). According to a previous report, the capacitance change of lowfrequency region (region I) is attributed to the polarization at electrode interfaces. [39] The capacitance of WO 3 / CPTA-based device is significantly lower than WO 3based device in the region I, indicating that the significant suppression of carriers accumulation at the ETL/PVK interface after CPTA modification and that is beneficial to eliminate the light-soaking effect. To further gain sight into the underly reason for the eliminated light-soaking effect, we employed the X-ray photoelectron spectroscopy (XPS) to study the interaction between WO 3 and CPTA. The W 4f spectra of the WO 3 and WO 3 /CPTA films are shown in Figure 2E. The binding energies located at 37.54 and 35.38 eV are identified to be the 4f 5/2 and 4f 7/2 characteristic peaks of W 6+ , respectively. While the two peaks shift positively to 37.74 and 35.58 eV after CPTA modified, which reveals the formation of chemical bonds between WO 3 and CPTA. [40] Moreover, we also explore the chemical interactions between CPTA and PVK. As illustrated in Figures 2F and S8, the signals of Pb 2+ move to lower binding energies upon CPTA modification, confirming again that the CPTA molecule would diffuse into the PVK layer due to the solubility of CPTA in DMF ( Figures S4 and S5) and hence the oxygen (O) atoms in CPTA donate its lone electron pair to the empty 6p orbital of Pb 2+ by coordination bond. [41,42] The strong chemical reactions between CPTA with WO 3 and PVK will passivate the defects at interface, which is conductive to suppress trap-assisted nonradiative recombination and beneficial for eliminating the light-soaking effect. [28,39] High-resolution synchrotron radiation photoelectron spectroscopy (HR-SRPES) was further carried out to evaluate the energy band levels [43,44] of the ETL and corresponding PVK films before and after CPTA treatment ( Figure 3I,II). The work function (W F ) of the WO 3 / CPTA film is −4.05 eV, higher than the WO 3 film (−4.25 eV). Moreover, the W F of PVK film is also increased from −4.00 eV to −3.73 eV upon CPTA modification, which might be attributed to the chemical interactions between CPTA and PVK. The energy-level arrangement diagram of the films is shown in Figure 3III and the corresponding parameters are listed in Table S2. Ulteriorly, the calculated built-in electrical field (Δμ u ) increases from 0.25 eV for the WO 3 -based sample to 0.32 eV for the WO 3 /CPTA-based sample, which is consistent with the results of kelvin probe force microscopy (KPFM). ( Figure S9 and Table S3. The Δμ k increases from 0.27 eV of WO 3 -based device to 0.46 eV of WO 3 /CPTA-based device.) Therefore, the CPTA modification can facilitate the electron extraction and suppress the accumulation of charge carriers at the buried interface. [45,46] Combined with the XPS, HR-SRPES, and KPFM results, the CPTA can bond with W 6+ of WO 3 and Pb 2+ of the PVK layer, resulting in effective defect passivation and alleviation of trap-assisted nonradiative recombination at the interface. Furthermore, the built-in electrical field increases significantly after CPTA modification, which can promote the interfacial carriers transport and inhibit the interfacial carrier accumulation. All these factors afford the elimination of the light-soaking effect.
Subsequently, we investigate the effect of CPTA on the electron extraction and electron-hole recombination dynamics at the interface of WO 3 /PVK through photoluminescence (PL) and time-resolved PL spectroscopies. [47] The steady-state PL peak intensity of the WO 3 / CPTA-based PVK film has been quenched more tempestuously than that of the film formed on the pristine WO 3 ETL ( Figure 4A), indicating that more efficient electron transfer from PVK layer to WO 3 /CPTA layer than to pure WO 3 layer, which is beneficial for suppressing the lightsoaking effect. [34] Moreover, the PL peak blue-shifts from 790 nm for the control sample to 785 nm for the CPTAmodified one, suggesting an effective defect passivation of PVK and ETL subject to CPTA modification. [42,44] The conclusion is reinforced by the results obtained from time-resolved PL spectra in Figure 4B and Table S4. Reduced carrier lifetime from 167 ns for the WO 3 -based PVK to 12.2 ns for the WO 3 /CPTA-based PVK demonstrates the efficient carrier transfer from PVK to CPTAmodified WO 3 .
To further understand the charge recombination mechanism, we measured the trend of short-circuit current density (J SC ) and open-circuit voltage (V OC ) of devices under different incident light intensities (I). According to the formula of J SC ∝ I α , the linear dependence between J SC and I in logarithmic form is shown in Figure 4C. The slopes of both devices are close to 1, which implies that the bimolecular recombination in both devices is negligible. [16] Besides, the dependence of V OC on Log(I) is presented in Figure 4D. The slope of WO 3 / CPTA-based device is 1.43 kT/q, which is lower than that of the WO 3 -based device (1.61 kT/q), indicating that the trap-assisted monomolecular recombination is significantly reduced in WO 3 /CPTA-based devices. [48] The reduction in trap-assisted monomolecular recombination might be attributed to trap passivation through chemical interactions between WO 3 and CPTA and/or between PVK and CPTA.
The defect density was assessed by space charge limited current method. [49,50] We fabricated electron-only devices with a configuration of ITO/WO 3 (or WO 3 /CPTA)/CsFAbased PVK/PCBM/Au, while the corresponding I-V curves are presented in Figure 4E. The trap-filled limit voltage (V TFL ) values of devices based on WO 3 and WO 3 /CPTA are 0.47 and 0.23 V, respectively. The calculated trap density (N t ) for the WO 3 /CPTA-based device is 5.4 × 10 15 cm −3 (Table S5), which is~1/2 lower than that of the WO 3 -based device (1.1 × 10 16 cm −3 ). The obvious reduction of trap density indicates the effective passivation effect of CPTA on defects.
Electrochemical impedance spectroscopy was further employed to unveil the interfacial charge transport behavior. [51] The Nyquist plots of the two devices tested in the dark condition were shown in Figure 4F, while the fitting results are summarized in Table S6. The WO 3 / CPTA-based device exhibits a larger recombination resistance (R rec ) in the low-frequency range and a smaller transfer resistance (R tr ) in the high-frequency range in comparison with the WO 3 -based cell, indicating its more efficient charge transport, which hence alleviates the accumulation of carriers at the interface and eliminates the light-soaking effect. To verify the influence of CPTA modification on the photovoltaic performance of devices, we comparatively studied the photovoltaic properties of WO 3 and WO 3 /CPTA-based PSC devices. As shown in Figure 5A, the cell with pristine WO 3 ETL exhibits a stabilized PCE of 17.4%, along with the V OC of 0.93 V, J SC of 24.7 mA·cm −2 , and fill factor (FF) of 75.9%. In contrast, the device with CPTA modification delivers a much higher PCE of 20.5%, together with the V OC of 1.03 V, J SC of 24.8 mA·cm −2 , and FF of 80.4% under identical conditions, which is among the high efficiency of PSCs with WO 3 -based ETMs (Table S7). The improvement of V OC and FF might be attributed to better contact between the ETL and the PVK layer due to the CPTA modification. The external quantum efficiency was also measured ( Figure 5B) and the integrated J SC values of WO 3 and WO 3 /CPTA-based devices are 23.50 and 23.63 mA·cm −2 , respectively. The stabilized power output was tested under continuous AM 1.5 G illumination ( Figure 5C). The stabilized PCEs are, respectively, 16.3% for the control cell and 19.6% for the CPTA-modified one, which match well with the performance deduced from the J-V curves. To confirm the reproducibility, 50 individual PSCs were fabricated, respectively, and their statistical histograms of the PV parameters are shown in Figures 5D and S10. WO 3 /CPTA-based PSCs have a narrower PCE distribution, highlighting the improved reliability of the device with CPTA modification.
Finally, we comparatively studied the stability of the unencapsulated devices based on WO 3 and WO 3 / CPTA ETLs. Figures 6A and S11 show the dark stability in the N 2 atmosphere. An~81% of its original PCE is maintained for the WO 3 /CPTA-based device after 3000 h of storage, while the PCE of the WO 3based device drops to~53% of its initial PCE. The light stability of the devices was also evaluated under continuous light irradiation (white light LED, 1 sun) for 500 h. As illustrated in Figures 6B and S12, the device based on WO 3 /CPTA ETL maintains~90% of its original PCE, while it is only~82% for the WO 3based device. The thermal stability of devices was further examined under continuous heating at 85°C for 1000 h in N 2 atmosphere. We adopted Poly[bis(4phenyl)(2,4,6-trimethylphenyl)amine] as the hole transport material [52] and the evolution of the normalized performance parameters are displayed in Figures 6C and S13. The PCE of the WO 3 /CPTA-based device remains at 77% of the initial efficiency, while the WO 3 -based device decreases to 57%. The enhanced stability of WO 3 /CPTA devices might be attributed to the reinforced interface upon the modification of CPTA and/or the reduction of defect density in bulk PVK films. All data of the PSCs were collected after illumination stabilized for 20 min. CPTA, C 60 pyrrolidine Tris-acid; EQE, external quantum efficiency; ETL, electron transport layer; FF, fill factor; PCE, power conversion efficiency; SPO, stabilized power output.

| CONCLUSION
In summary, CPTA has been introduced to modify the interface of WO 3 ETL and PVK layer. The chemical bonds can be formed between CPTA with both WO 3 and PVK, resulting in decreased trap density obviously, and thus effectively suppressed the trap-assisted nonradiative recombination. Moreover, the introduction of CPTA can boost the built-in electric field, which is conducive to facilitate the charge transfer from PVK layer to WO 3 ETL, significantly inhibiting the accumulation of charge carriers at the WO 3 /PVK interface. The above effects jointly accelerate the elimination of light-soaking effects in the present PSCs. Hence, a decent PCE of 20.5% along with excellent stability has been achieved for the WO 3 / CPTA-based MA-free PSCs. Our work paves the way for the realization of highly efficient and stable PSCs.