Zuoming
Jin
a,
Bin
Li
b,
Yutian
Xu
a,
Boya
Zhu
a,
Gaiqin
Ding
a,
Yuanqiang
Wang
a,
Jingxia
Yang
a,
Qinghong
Zhang
b and
Yichuan
Rui
*a
aCollege of Chemistry and Chemical Engineering, Shanghai University of Engineering Science, Shanghai 201620, P. R. China. E-mail: ryc713@126.com
bState Key Laboratory for Modification of Chemical Fibers and Polymer Materials, College of Materials Science and Engineering, Donghua University, Shanghai 201620, P. R. China
First published on 2nd May 2023
The rational passivation of the defects at the buried interface plays a significant role in reducing energy loss and improving the photovoltaic performance of perovskite solar cells (PSCs). Herein, we have applied a metal–organic framework (MOF)-based host–guest system consisting of zeolitic imidazolate framework-8 (ZIF-8) nanosheets confined with methylammonium chloride (MACl) to modify the SnO2/perovskite interface. The MACl@ZIF-8 not only effectively reduces the oxygen vacancies on SnO2, but also binds with the uncoordinated Pb2+ and halogen ions at the bottom of the perovskite layer. Moreover, the loaded MACl guest plays the role of both crystalline seeds involved in the formation of ZIF-8@perovskite heterojunctions and organic cation vacancy passivators. Thus, the simultaneous passivation of the buried interface and bulk phase of the perovskite is realized, and the photovoltaic performance of devices is enhanced considerably with an optimal efficiency of 22.10%. The introduction of MACl@ZIF-8 also allows the device to achieve excellent moisture stability due to better crystallinity and fewer phase defects. In particular, the UV shielding characteristic of the imidazole ligands in ZIF-8 results in a significant increase in the UV resistance capability.
In the typical organic–inorganic halide perovskite, the thermally unstable MA+ tends to partially evaporate and escape during the annealing process of the film. This could induce component stoichiometry imbalance and bulk phase defects.17,18 Thus, many cation vacancies are left and result in the unfavorable transfer of photogenerated carriers. Therefore, cation vacancy compensation for MA-based perovskite becomes an effective way to enhance device performance.
Bimolecular or host–guest passivation systems have been explored to achieve dual passivation of the buried interface and the bulk phase simultaneously. Zhu et al. developed a depth-dependent manipulation strategy to modulate both the interface defects and bulk defects by two kinds of molecules. Since different sizes exhibit different permeabilities in the perovskite layer, the defects at different locations can be passivated.19 Grätzel et al. also designed a host–guest method by employing dibenzo-21-crown-7 (DB21C7)-loaded Cs+ to form a well-defined host–guest complex, thus delivering Cs+ ions into the interior while modulating the perovskite. This provides a local nanoscale doping gradient that supports photoinduced charge separation while passivating surface and bulk defects, stabilizing the perovskite phase through synergistic effects of the host, guest and host–guest complexes.20 The defects at the buried interface and in the bulk phase are prone to inducing severe carrier recombination and both the PCE and long-term stability suffer from outside environmental factors, which result in poor photovoltaic performance. Therefore, a robust and functionalized modifier that meets multiple passivation requirements is essential.
Metal–organic frameworks (MOFs) are a class of crystalline inorganic–organic hybrid materials that have a wide range of special characteristics.21 Their multimodal porosity and open architecture enable effective mass transmission. MOFs also exhibit outstanding chemical and physical stability, leading to a wide range of applications in biology,22 sensing,23 and catalysis.24 There have been some explorations of MOFs in the field of PSCs. Nevertheless, these studies mainly focused on three-dimensional (3D) MOFs, which were used as scaffolds or modifiers.25,26 The relatively large size and insulation, which are characteristic of 3D MOFs, impede their application in PSCs. As two-dimensional (2D) materials, 2D MOFs are a series of materials that possess regular porous structures with large specific surface area,27 thus attracting much attention due to their excellent properties such as tunable structure, rich active sites, and outstanding physical and chemical stability.28,29 In particular, 2D MOFs provide more exposed active sites than 3D MOFs, which enable them to be used to improve device performance more effectively.30,31 Jen et al. reported a thiol-functionalized 2D conjugated MOF as an electron-extraction layer at the perovskite/cathode interface to capture the leaked Pb2+ from the degraded perovskite by forming water-insoluble solids; thus, excellent stability of devices was obtained and the harmful effect on the environment was suppressed.32 Therefore, 2D MOFs have sufficient capabilities and unique advantages in the field of photovoltaics.
In this study, we synthesized 2D zeolitic imidazolate framework-8 (ZIF-8) nanosheets by a unique mixed solvent method and then embedded methylammonium chloride (MACl) into the framework. The as-obtained MACl@ZIF-8 was applied as a modifier to passivate the buried interface between SnO2 and the perovskite layer, which realized excellent dual-defect passivation in both the buried interface and the bulk phase of the perovskite. With the reduced non-radiative recombination and enhanced electron transportation, the champion efficiency of the PSCs was boosted from 19.61% to 22.10%. In addition, the UV-resistant characteristics and crystallization-regulated roles of ZIF-8 empowered the devices with improved UV and humidity stability.
XRD tests were performed to characterize the crystallinity and purity of the prepared ZIF-8 and MACl@ZIF-8. Fig. 1(e) shows the main diffraction peaks located at 7.3°, 10.4°, 12.7°, and 18.1°, corresponding to the (011), (002), (112), and (222) planes of ZIF-8. Both samples exhibited excellent crystallinity and the structure of ZIF-8 did not change much after loading MACl, which also proves the excellent structure stability of MACl@ZIF-8. The FTIR results in Fig. S1 (ESI†) also confirmed the high purity of ZIF-8 and MACl@ZIF-8,36 and no significant changes in the FTIR peaks of the products were observed after loading MACl. To verify that MACl seeds successfully occupied the holes of ZIF-8, the specific surface areas of ZIF-8 and MACl@ZIF-8 were measured. As shown in Fig. 1(f), the specific surface area of MACl@ZIF-8 is 166.75 m2 g−1, which is significantly lower than 1014.95 m2 g−1 of pristine ZIF-8, indicating that most pores of ZIF-8 are occupied by MACl. The pore size distribution curves also confirmed the disappearance of the pores (Fig. 1(g)).
MACl@ZIF-8 was redispersed in DMF and spin-coated on the SnO2 ETLs. Fig. 2(a) and (b) demonstrate the surface SEM images of SnO2 films without and with MACl@ZIF-8. Both films exhibited compact and pinhole-free morphologies. Notably, the MACl@ZIF-8-modified SnO2 film is more uniform and smoother. To determine the uniformity of MACl@ZIF-8 on SnO2, we used EDS to analyze the distribution of the elements. As shown in Fig. 2(c), the EDS mappings of Zn, N, and Cl further identified the uniform distribution of MACl@ZIF-8 on SnO2. The cross-sectional SEM image of MACl@ZIF-8-modified SnO2 also exhibited a smoother surface than pristine SnO2 (Fig. 2(d) and (e)). The root mean square (RMS) of the SnO2 ETLs was reduced from 5.96 to 4.83 nm according to the AFM measurements (Fig. 2(f) and (g)). The transmittance of ZIF-8 and MACl@ZIF-8-modified SnO2 slightly increased in comparison to unmodified SnO2, as revealed by ultraviolet-visible (UV-vis) spectroscopy (Fig. S2, ESI†). We hypothesize that the increased transmittance is due to the even distribution of MACl@ZIF-8, which induces possible optical coupling in the MACl@ZIF-8/SnO2 layer.37 In typical planar PSCs, the PCE of the device is directly influenced by the transmittance, which has a close relationship to the number of photogenerated carriers produced by the active perovskite layer. Additionally, the contact angle of MACl@ZIF-8-modified SnO2 film is 39.9° (Fig. 2(i)), which is smaller than 49.4° for the pristine SnO2 (Fig. 2(h)). This change could be attributed to the smoother surface and more hydrophilic sites exposed on ZIF-8. Reasonably, the smaller contact angle would facilitate the spreading of the perovskite precursor solution and the subsequent growth of the crystals.
Fig. 3(a) schematically demonstrates the interaction between SnO2 and ZIF-8. The presence of uncoordinated N and Zn atoms at the edges of the ZIF-8 framework allows them to passivate the defective sites on the SnO2 surface38,39 To verify the speculation, the chemical states were investigated by XPS. As Fig. 3(b) shows, the Sn 3d peaks of MACl@ZIF-8-modified SnO2 shifted to lower binding energy, while the –NH characteristic peak in the N 1s XPS spectrum was reduced from 45% to 36% (Fig. 3(d)). Therefore, we can infer that the electrons were transferred from the N atoms in MACl@ZIF-8 to the Sn atoms in SnO2,40 which indicates that the N atoms in the imidazole of ZIF-8 chemically interacted with the Sn atoms in SnO2 to form Sn–N bonds.41 Thus, it is favorable to reduce the number of dangling Sn bonds on the surface of SnO2. Besides, the O 1s spectra are also divided into two peaks by deconvolution (Fig. 3(c)), where the larger binding energy peak corresponds to the vacancy oxygen and the lower one corresponds to the lattice oxygen of SnO2. The fraction of oxygen vacancy was reduced from 54% to 38% after MACl@ZIF-8 modification (Table S1, ESI†), indicating that the oxygen vacancies on SnO2 were effectively passivated.42 Thus, the MACl@ZIF-8 modification effectively eliminates trap states on SnO2 and thereby creates fewer barriers for electron transportation.
The energy levels of pristine SnO2 and MACl@ZIF-8-modified SnO2 ETLs were investigated by ultraviolet photoemission spectroscopy (UPS). The valence band (EVB), conduction band (ECB), and Fermi level (EF) were derived from the results of UV-vis and UPS spectra (Fig. S3, ESI†). The EF of the ETL decreased from 3.82 eV to 3.42 eV, which indicated that the ETL was more n-type and was conducive to electron transfer.43 Besides, the ECB of MACl@ZIF-8-modified SnO2 moved up to a higher energy level of −4.01 eV, which was closer to the −3.93 eV of the perovskite layer (Fig. 3(e)). Then, the perovskite/SnO2 interface achieved a favorable energy level alignment which led to less energy loss. Furthermore, the conductivity of MACl@ZIF-8/SnO2 was calculated from the I–V curves under a dark environment (Fig. 3(f)). The equation σ = d/AR, where d is the thickness of the ETL, R is the resistance, and A is the active area, is used to determine the conductivities (σ) of different ETLs.44 Consequently, the σ of SnO2, ZIF-8/SnO2 and MACl@ZIF-8/SnO2 are 6.67 × 10−5 S cm−1, 7.54 × 10−5 S cm−1 and 8.89 × 10−5 S cm−1, respectively. Hence, MACl@ZIF-8-modified SnO2 exhibited the best conductivity. From the results of the energy level discussed above, the greater n-type character of MACl@ZIF-8 is conducive to electron transfer and electronic conductivity.43 Therefore, the extraction and transfer of electrons from the perovskite layer to the ETL showed remarkable improvement.
Here, we adopted the triple-cation mixed perovskite ((FA0.83MA0.17)0.95Cs0.05Pb(I0.83Br0.17)3) as the light-absorbing layer. With the buried interface passivated by 2D MACl@ZIF-8, the perovskite film demonstrated optimized crystallinity and larger grain size than the control sample. As shown in Fig. 4(a)–(c), the perovskite grains on ZIF-8 or MACl@ZIF-8-modified SnO2 are larger and more compact than that on pristine SnO2. The corresponding cross-sectional SEM images (Fig. 4(d)–(f)) also clearly demonstrate that the modified perovskite films have better crystallinity, especially for the perovskite based on MACl@ZIF-8/SnO2, which shows larger grain sizes. The improved crystallinity could be attributed to the following. The surface of the substrate is smoother after 2D MACl@ZIF-8 modification, which is more conducive to the uniform growth of perovskite films with larger grain sizes.45 (ii) The pre-buried MACl guest in the ZIF-8 host could first react with PbI2/PbBr2 in the perovskite precursor solution and form perovskite seeds during the spin-coating process, which is conducive to the oriented growth of perovskite crystals.46 (iii) MACl could form the FAI–MABr–CsI–MACl–PbI2–PbBr2–DMSO intermediate phase during the solution process, expanding the time window for perovskite crystal growth and thus yielding larger grains;47 accordingly, the MACl@ZIF-8-modified perovskite films show stronger light absorption (Fig. S4, ESI†). MACl@ZIF-8 modification provides a smoother surface and slightly improves the transmittance of pristine SnO2. More importantly, MACl@ZIF-8 modification results in high-quality perovskite films with better crystallinity and a purer phase. Thus, the absorption intensity of the perovskite layer based on MACl@ZIF-8-modified SnO2 is enhanced.
Fig. 4 (a)–(c) Top-view and (d)–(f) cross-sectional SEM images of perovskite films based on SnO2, ZIF-8/SnO2 and MACl@ZIF-8/SnO2, respectively. The insets are grain size distributions. |
To further verify the crystallinity of the perovskite layer after the buried interface modification, XRD patterns of the perovskite films were obtained (Fig. S5, ESI†). The intensity of the main peak of the perovskite film based on MACl@ZIF-8/SnO2 was strengthened as compared to that of pristine SnO2. More importantly, the intensity of the PbI2 diffraction peak of the ZIF-8 and MACl@ZIF-8-modified perovskite films were greatly reduced as compared to that of the pristine SnO2. This finding shows that 2D ZIF-8 may significantly suppress the impurity phase in perovskite film by reducing the amount of residual PbI2 at the buried interface.46
Based on the above findings, we investigated the photovoltaic performances of PSCs with and without MACl@ZIF-8 modification. The device had a planar n–i–p heterostructure of FTO/SnO2/perovskite/Spiro-OMeTAD/Au. As schematically illustrated in Fig. 5(a), MACl@ZIF-8 acts as an interlayer with the effect of multi-passivation. On the one hand, the N and Zn in ZIF-8 could heal the uncoordinated Pb2+ and halogen ions at the bottom of the perovskite layer.48,49 The XPS peak shift of Pb 4f and I 3d in Fig. S6 (ESI†) indicated the interactions. On the other hand, the MA+ released from ZIF-8 could compensate for the cationic vacancies. The J–V curves of champion devices are shown in Fig. 5(b), the optimal MACl@ZIF-8-modified device achieved an impressive PCE of 22.10% with a high FF of 79.10%, VOC of 1.16 V, and JSC of 24.07 mA cm−2. As a comparison, the control device showed a relatively lower PCE of 19.61% with FF of 77.15%, VOC of 1.10 V, and JSC of 23.52 mA cm−2. Significantly, VOC was obviously enhanced after passivating by MACl@ZIF-8, because MACl@ZIF-8 at the buried interface made good contact between the perovskite layer and the SnO2 layer, which in turn accelerated electron extraction and led to suppressed carrier recombination. It should also be noted that the Cl− carried by the 2D modifier could diffuse to the bulk phase of the perovskite during the annealing process, which plays an effective role in regulating perovskite crystallization and passivating the bulk defects.50,51 To verify the light conversion capability of different devices, the external quantum efficiency (EQE) spectra were obtained. As shown in Fig. S7 (ESI†), the integrated JSC values of the control device, and devices with ZIF-8 and with MACl@ZIF-8 were 23.16, 23.31 and 23.59 mA cm−2, matching well with the JSC values obtained from the J–V curves. The EQE of the ZIF-8 device slightly increased as compared to the control, while the MACl@ZIF-8 device showed the highest EQE values, especially in the 400–600 nm range, resulting in an integrated JSC enhancement. The increased EQE agrees well with the enhanced light absorption mentioned above. Thus, more photogenerated carriers will be generated when the incident light is absorbed by the active layer, facilitating the creation of a larger current density.
The steady PCE out-puts at maximum power point (MPP) were measured as shown in Fig. 5(c). With the optimized buried interface and passivated perovskite bulk phase, the PCE of the MACl@ZIF-8 device maintained nearly 21.35%, which is higher as compared to 18.58% for the control device. The statistical distributions of the performance parameters, including VOC, JSC, FF, and PCE, are shown in Fig. 5(d). The average value of each parameter increased and showed excellent reproducibility after the passivation of MACl@ZIF-8.
With the improved device performance, we further determined the mechanism of defect passivation at the buried interface and perovskite phase. Fig. 6(a) shows the dark J–V curves of devices with and without passivation. The MACl@ZIF-8-modified device exhibited reduced dark current, which proved that the device has less energy loss during operation.52 Electrochemical impedance spectroscopy (EIS) was conducted under dark conditions to investigate the charge transfer kinetics within the device. The resultant Nyquist plots are shown in Fig. 6(b), and the series resistance (Rs) and recombination resistance (Rrec) were obtained via an equivalent circuit model.53,54 The fitting results showed that the Rrec of the MACl@ZIF-8-modified device was 8.5 kΩ, which is larger as compared to 2.5 kΩ of the control device. This demonstrates that the MACl@ZIF-8-modified devices could effectively suppress non-radiative recombination while also being more conducive to charge transfer,55 which highly matched the results of the dark J–V. It was further confirmed that the modification of the buried interface by MACl@ZIF-8 accelerated the electron transfer.4,56 To evaluate the trap density of the perovskite films on different ETLs, we performed space-charge limited current (SCLC) measurements by employing the electron-only structure of FTO/SnO2/perovskite/PCBM/Ag. Fig. 6(c) shows the current of devices under applied bias voltage in the dark. The trap-state density was calculated via the equation VTFL = eNtd2/2ε0ε.57 From the SCLC curves, it was observed that the VTFL of the device after introducing MACl@ZIF-8 decreased from 0.46 V to 0.30 V, and the calculated trap density of the MACl@ZIF-8-modified device was 5.26 × 1015 cm−3, which is a significantly reduced trap-state density as compared to the control device of 9.31 × 1015 cm−3. This could be attributed to the advantages of dual-passivation of the host–guest system. The non-radiative recombination centers in the perovskite layer and at the buried interface were significantly reduced and faster charge transportation was realized. It is well known that the photovoltage of a solar cell is directly related to its internal luminescence ability.58 Here, we qualitatively demonstrate the reduction in trap density by recording the difference in luminescence performances of the two devices using a photophysical characterization method.59 As shown in Fig. S8 (ESI†), the device without MACl@ZIF-8 showed a localized uneven luminescence under applied current and a weak maximum luminescence intensity, while the MACl@ZIF-8-based device showed a uniform and more stable electroluminescence, which suggests that the non-radiative recombination in the modified device is suppressed.
Fig. 6(d) shows the steady-state photoluminescence (PL) spectra. The PL peak intensity of the modified film is lower as compared to that of the control sample, which indicates that the introduction of MACl@ZIF-8 into the buried interface facilitates the reduced number of non-radiative recombination centers as well as higher electron mobility. In general, the uncoordinated defects are regarded as recombination centers and induce severe energy loss at the buried interface. The time-resolved photoluminescence (TRPL) decay measurement was also conducted to investigate the dynamics of photogenerated carriers. Fig. 6(e) shows the fitted results with the bi-exponential decay function. The decay curves are categorized into two decay parts, namely, a fast decay (τ1) related to the non-radiative recombination at the interface and a slow decay (τ2) related to the radiative recombination in the perovskite bulk.60 The main fitted data are shown in Table S2 (ESI†). The average carrier lifetime can be calculated by the following equation:
1/C2 = 2(Vbi − V)/A2qεε0Nd, |
The long-term stability performance was also investigated. In general, ZIF-8 could absorb the photons whose energies were lower than its bandgap due to the low-energy absorption property of the 2-MIM ligand.64,65 Therefore, the UV absorption characteristic of ZIF-8 prevented the decomposition of the perovskite layer, induced by UV radiation, as illustrated in Fig. 7(a). With the shielding of the MACl@ZIF-8 layer, the perovskite film maintained a high absorption capability after 200 hours of UV irradiation, whereas the perovskite film without MACl@ZIF-8 exhibited severe UV-induced deterioration with an obvious characteristic absorption edge of PbI2 (Fig. 7(b)). Fig. 7(c) and (e) show the SEM images of perovskite films aged for 100 h without and with the MACl@ZIF-8 layer. There was obvious phase degradation as compared to the passivated sample. To more visually monitor the decomposition process, the perovskite films were stored in air at a relative humidity of around 65%. Fig. 7(d) shows the photographs of the surface morphological changes of the samples. After storage for 800 h, the control perovskite film became completely yellow, while the perovskite film with MACl@ZIF-8 maintained a much slower rate of decomposition. The corresponding XRD patterns of perovskite films aged for 100 h maintained good crystallinity (Fig. 7(f)). In agreement with the photograph, the perovskite film without MACl@ZIF-8 showed a stronger peak of PbI2. In comparison, the perovskite film with MACl@ZIF-8 passivation exhibited an almost invariable PbI2 peak, indicating the excellent stability of passivated perovskite film crystals under the attack of moisture. The perovskite films with greater crystallinity and larger grains are more resistant to moisture and have better long-term stability.66
Finally, the photovoltaic performance evolution versus time was investigated. As Fig. 7(g) shows, the device with MACl@ZIF-8 can maintain 86% of its original efficiency after 500 hours under UV light exposure. Nevertheless, the efficiency of the control group was below 60% under the same conditions. Moreover, the stability of unencapsulated devices, when exposed to humidity, was tested and the results are shown in Fig. 7(h). The efficiency of the device with MACl@ZIF-8 maintained more than 90% of the original efficiency after 500 h of storage, while the control device only kept less than 60% of its primary efficiency. The enhanced stability can be ascribed to better perovskite crystallinity with a reduced trap state, making the overall device less prone to moisture degradation.67
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3tc00609c |
This journal is © The Royal Society of Chemistry 2023 |