Efficient Bifacial Semitransparent Perovskite Solar Cells Using Down‐Conversion 2D Perovskite Nanoplatelets–Poly(Methyl Methacrylate) Composite Film

Semitransparent perovskite solar cells (ST‐PSCs) hold significant appeal for various applications in smart windows, multijunction tandem devices, bifacial and chargeable devices, etc. Unfortunately, to possess high transparency, the perovskite layer in the ST‐PSCs must be kept relatively thin (<400 nm), which in turn causes insufficient light absorption and thus inferior device performance. Herein, a 2D perovskite nanoplatelets (NPLs)/poly(methyl methacrylate) (PMMA) composite thin layer is applied in the ST‐PSCs to solve these problems. Thanks to its dual function of down‐conversion (DC) effect, converting high‐energy UV photons into low‐energy visible photons to enhance the photocurrent, and interfacial passivation, reducing the nonradiative recombination at the interface, the 2D NPLs–PMMA‐based devices with the different average visible transmittance (AVT) values of perovskite film demonstrate significantly improved power‐conversion efficiency (PCE) compared to the pristine devices, and remarkable UV stability, retaining over 77% of initial PCE after aging under continuous UV illumination for 280 h. More importantly, the full bifacial ST‐PSCs using a transparent MoO3/Au/MoO3 rear electrode exhibits a record PCE of 14.26% and 10.65% with a whole device AVT of 19.4% and 26.9%, respectively, which are among the highest performing ST‐PSCs of the kind reported to date.


Introduction
Semitransparent perovskite solar cells (ST-PSCs) have shown great potential as an emerging technology for the use in building integrated photovoltaics (BIPVs), tandem devices, as well as portable and wearable electronics. [1]Due to their high transparency, ST-PSCs can be applied in windows, skylights, and other building surfaces to generate solar electricity without significantly impacting the aesthetics and functionality of the buildings. [2]dditionally, ST-PSCs with great nearinfrared (NIR) transmittance can work as the top cell in tandem solar cells, where the ST-PSC absorbs the light with short wavelength, while the bottom cell absorbs the light with long wavelengths transmitted through the top ST cell, achieving higher efficiency than individual cells. [3]Being fabricated on flexible substrates, ST-PSCs also can conform to curved or irregular surfaces, making them suitable for applications in flexible or wearable devices. [4]urthermore, ST-PSCs exhibit the capacity for bifacial operation enabling them to capture sunlight from both the front and rear sides of the cell.Compared to the conventional monofacial PSCs, this results in a higher power output per unit area with the minimal additional manufacturing costs through the utilization of the diffuse irradiation from albedo in the bifacial PSCs. [5]Nevertheless, compared with the opaque or nontransparent PSCs, ST-PSCs often exhibit lower power-conversion efficiency (PCE) because of the loss of photocurrent.Although some work reported higher performance with ST perovskite films, the whole devices were not truly transparent due to the use of traditional opaque rear electrodes.Consequently, these devices are not suitable for the applications in BIPVs requiring full transparency.Up to date, ST-PSCs with average visible transmittance (AVT) ranging from 20% to 30% can only achieve the PCEs of 8%-14%. [6]To obtain high transparency, the most common strategy in ST-PSCs is to reduce the thickness of perovskite films.However, the thin perovskite films (<400 nm) are unable to absorb light efficiently thereby causing low photocurrent of the solar cells.
It is well known that the metal halide perovskite light absorbers exhibit the strong absorbance across a broad range of spectral wavelengths (400-800 nm), including both visible and NIR regions. [7]This wide light absorption allows them to capture sunlight and convert it into electrical energy effectively.Nevertheless, the light from the UV region almost does not contribute to the photocurrent in the PSCs.Instead, it leads to degradation of the devices. [8]It has been widely reported that UV light can decompose perovskite layers and induce oxygen defects and vacancies thereby causing poor device performance and stability. [9]One of the effective approaches to enhance the photocurrent is using down conversion (DC) materials.With DC materials, the high-energy UV photons can be converted into low-energy photons which are then absorbed by the perovskite material to generate extra photocurrent.The process of converting photon energy by DC materials not only extends the spectral response range of PSCs to the UV-blue region, but also protects the perovskite materials in the devices from decomposition caused by UV light.
This strategy has been effectively used to enhance performance of the opaque PSCs previously.For instance, Eu 3þ dopant, the most effective lanthanide luminescence material, was adopted to convert damaging UV light to additional visible light for PSCs.Peng et al. [10] synthesized the luminescent Eu-TiO 2 thin films as electron-transport layers (ETL) via chemical-bath deposition.Benefiting from this DC effect in the ETL, the corresponding planar PSCs yielded an improved power conversion efficiency (PCE) of 21.40% with remarkable stability under UV illumination.Similarly, ZnGa 2 O 4 :Eu 3þ (ZGO) nanophosphor was synthesized and incorporated with mesoporous TiO 2 layer for PSCs by Xian et al. [11] A much higher photocurrent of 25.02 mA cm À2 was obtained for the ZGO-based devices compared to the ones without ZGO (20.2 mA cm À2 ).In addition, a samarium-based DC nanomaterial (Sr 2 CeO 4 :Sm 3þ ) was employed in fabricating PSCs. [12]Thanks to the harvesting and reemission of UV light to the visible light range through the DC layer, the photocurrent and average PCE of the devices were increased by 11.4% and 16.2%, respectively, compared to the control devices.Despite these, the commonly used lanthanide-based DC materials often request high synthesis temperature to be incorporated into the ETL.Moreover, these materials are scarce on Earth, posing significant obstacles to their widespread applications in practice.In addition, the metal halide perovskite materials themselves are also promising to be used as DC materials, in particular 2D perovskite, due to their unique properties such as large Stokes shift and high photoluminescence (PL) quantum yield (PLQY). [13] The colloidal of layered 2D perovskite even offer controllable Stokes shift and enhanced PLQY by controlling organic cations and particles sizes, making them excellent candidates as absorbers/emitters for DC application.Despite limited studies showing the potential of using DC materials to boost PCE for opaque PSCs, there is a notable lack of research on utilizing DC materials, particularly low-dimensional perovskite DC materials in ST-PSCs to increase their photocurrent, UV stability, and thus the overall device performance.
In this work, we presented a simple yet effective method to improve the device performance and stability of ST-PSCs by using a PEA 2 MA 3 Pb 4 Br 13 -based 2D nanoplatelets (NPLs)-poly (methyl methacrylate) (PMMA) composite thin layer.The synthesized 2D perovskite NPLs demonstrated a strong DC ability, efficiently converting UV-blue light into visible light.This led to improved photon harvesting for the perovskite film and thus enhanced short-circuit current density ( J sc ) for the devices.The devices also exhibited higher open-circuit voltage (V oc ) and fill factor (FF), ascribed to the passivation effect of 2D NPLs-PMMA thin layer, which effectively reduced the interfacial nonradiative recombination.Therefore, compared with the control devices, the 2D NPLs-PMMA-based devices with opaque Au electrode delivered the significantly higher PCEs of 18.47%, 17.06%, 11.07%, and 5.26% when the AVT values of the perovskite layers were 17%, 27%, 37%, and 40%, respectively.The devices also demonstrated impressive long-term stability and UV tolerance.After aging at an ambient condition (room temperature, 30% relative humidity [RH]) for approximately 60 d, the unencapsulated 2D NPLs-PMMA devices retained an average of over 95% of their initial efficiency.Moreover, more than 77% of the average performance retention was achieved after being exposed to UV illumination for 280 h.Most importantly, when coupled with a transparent electrode of MoO 3 /Au/MoO 3 to assemble the full ST devices, the 2D NPLs-PMMA-based ST-PSCs produced the champion PCEs of 14.26% and 10.65% for the devices with a total AVT of 19.4% and 26.9%, respectively.

Optical Property and Morphology of 2D NPLs-PMMA Layer
The nanocrystals of 2D perovskite often exhibit a large Stokes shift, thus promising in term of DC application. [13]In this work, we employed PEA 2 MA 3 Pb 4 Br 13 -based 2D NPLs as the DC materials.The NPLs were synthesized using a reprecipitation method [14] (Supporting Information).The transmission electron microscopy (TEM) and selected area electron diffraction (SAED) images (Figure 1a,b) demonstrate the highly crystalline nature of the synthesized PEA 2 MA 3 Pb 4 Br 13 -based 2D NPLs.Obviously, the 2D NPLs possess a rectangular shape with a length up to 600 nm and a width ranging from 40 to 80 nm.The clear lattice fringes of 0.604 nm corresponding to (100) plane of the perovskite NPLs could be found in Figure 1b.As exhibited in atomic force microscopy (AFM) image (Figure 1c), the thickness of the layer-structured 2D NPLs is around 80 nm.The X-Ray diffraction (XRD) pattern (Figure 1d) shows the cubic phase of 2D NPLs with diffraction peaks at 14.9°, 21.2°, 30.1°, and 33.7°, which correspond to the crystal planes of (100), (110), (200), and (210), respectively. [15]Particularly, there is a strong peak at 5.3°, which indicates the formation of 2D perovskite nanostructure. [16]It is worth noting that there is a small diffraction peak at 3.8°, which can be assigned to PEA 2 MA 2 Pb 3 Br 10 (PEA 2 MA n-1 Pb n Br 3nþ1 , n = 3)-based 2D NPLs (Figure S1, Supporting Information), suggesting that the synthesized PEA 2 MA 3 Pb 4 Br 13 (n = 4)-based 2D NPLs contain a small amount of PEA 2 MA 2 Pb 3 Br 10 component.The optical properties of the synthesized 2D NPLs were investigated by UV-visible absorption and photoluminescence (PL) measurement.As shown in Figure S2a, Supporting Information, under the UV excitation at 365 nm, the 2D NPLs dispersion presents a strong emission peak at around 474 nm which is the typical PL peak for PEA 2 MA 3 Pb 4 Br 13 (n = 4)-based 2D NPLs, and a weak emission peak at 456 nm that belongs to PEA 2 MA 2 Pb 3 Br 10 (n = 3)-based 2D NPLs.This reveals that the synthesized 2D NPLs comprise the dominant PEA 2 MA 3 Pb 4 Br 13 (n = 4) NPLs and a small amount of PEA 2 MA 2 Pb 3 Br 10 (n = 3) NPLs.These 2D NPLs mixture generate the bright cyan light when exposed to UV light.Similarly, it is found that two absorption onsets are located at approximately 456 and 474 nm, respectively, in the UV-vis spectra, suggesting the presence of two different components (n = 3 and n = 4) of the synthesized 2D NPLs.The Stokes shift of this 2D NPLs mixture is around 27 nm, which is very important to avoid self-absorption problem as DC materials.The calculated energy bandgap of these two phases of 2D NPLs from the Tauc plot (Figure S2b, Supporting Information) are 2.68 eV for n = 3 and 2.59 eV for n = 4, respectively.
Various 2D NPLs-PMMA composite films were prepared by tuning the volume ratio between the 2D NPLs colloidal and the PMMA precursor.Since the 2D perovskite NPLs tend to aggregate to reduce the high surface energy, the PMMA with organic long chain was employed to work as an encapsulation matrix to protect the 2D NPLs from aggregation, thereby forming a uniform composite layer of well-dispersed 2D NPLs.In the meantime, it also has a role in surface passivation of perovskite bulk layer. [17]Even though PMMA itself can absorb the UV light, as demonstrated in Figure S3, Supporting Information, the ultrathin layer of PMMA exhibits negligible effects on blocking UV radiation.Therefore, it barely affects the DC capability of 2D NPLs in the composite films.For simplicity, the perovskite film with different 2D NPLs and PMMA ratios are defined as follows: 0:1 (namely PMMA), 1:9 (0.1-2D NPLs-PMMA), 1:3 (0.25-2D NPLs-PMMA), and 1:1 (0.5-2D NPLs-PMMA).As illustrated in Figure S4, Supporting Information, all 2D NPLs-PMMA composite layers present a strong cyan light emission under the UV light, an indication of their suitability as a DC layer in the ST-PSCs.The scanning electron microscope (SEM) images (Figure S5, Supporting Information) show the uniform morphology of ST perovskite films with the grain size ranging between 200 and 800 nm.After depositing a 2D NPLs-PMMA composite layer, there are no significant changes in terms of the morphology of bulk 3D perovskite film, but bright nanoparticles can be found on the surface.The shape and size of the nanoparticles are similar to that of the synthesized 2D NPLs shown in TEM and AFM (Figure 1a,c), indicting the presence of 2D NPLs on the surface of bulk perovskite film.Meanwhile, by increasing the concentration of 2D NPLs in the composite film from 10% to 50%, more quantity of 2D NPLs can be observed on the perovskite surface.

ST Perovskite Layer with Opaque Electrode
For comparison, the conventional opaque PSCs were fabricated first by using a thick perovskite layer from a 1.5 M precursor solution.As shown in Figure S6, Supporting Information, the device demonstrated an impressive PCE of 22.41% with a high J sc of 25.35 mA cm À2 (24.4 mA cm À2 from external quantum efficiency [EQE] measurement).However, the perovskite layer made from the high-concentration precursor solution is not suitable as the light absorber in ST-PSCs, due to the low transmittance of the thick perovskite layer with an AVT of 4.42%.Therefore, we reduced the precursor concentration (<0.5 M) to produce thinner perovskite layer with a higher AVT for ST-PSC application.
To investigate the effect of the 2D NPLs-PMMA composite layer on the device performance, solar cells with the structure of fluorine doped tin oxide (FTO) conducive glass/SnO 2 / FA 0.9 MA 0.1 PbI 2.7 Br 0.3 /2D NPLs-PMMA composite layer/2,2',7,7'-Tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9'-spirobifluorene (Spiro-OMeTAD)/opaque Au were assembled.The crosssectional SEM image (Figure 2a) demonstrates that the thickness of SnO 2 , Spiro-OMeTAD, and opaque Au layers are around 30, 200, and 100 nm, respectively.The transparency of the perovskite films was tuned by adjusting the concentration of the perovskite precursor solution.As shown in Figure 2a and S7, Supporting Information, the thickness of these ST perovskite films are approximately 250, 200, 150, and 120 nm.Correspondingly, the fabricated ST perovskite films exhibit the AVT values of 16.0%, 26.4%, 36.7%, and 39.0%, respectively, within the wavelength of 400-800 nm (Figure 2b).In addition, compared to the XRD patterns of the control perovskite film and perovskite film/ PMMA, there is an additional peak of 2D NPLs located at 5°( Figure S8a, Supporting Information) for the perovskite film with 2D NPLs-PMMA thin layers, which is consistent with the XRD pattern of the synthesized 2D NPLs in Figure 1d.All perovskite layers with different AVT values display the same primary diffraction peaks of the photoactive perovskite phase at 13.9°, 28.2°, and 31.6°(FigureS8b, Supporting Information), corresponding to crystal planes of ( 101), (202), and (211), respectively. [18]The UV-vis absorption spectra (Figure S9a, Supporting Information) exhibits that the light absorbance of the perovskite film is slightly reduced after being coated with a 2D NPLs-PMMA thin layer.With the increase of amount of 2D NPLs, the absorbance of the film gradually decreases in the wavelengths between 480 and 800 nm accompanied by the slight increase in transmittance from 26.4% to 28.6% (Figure S9b, Supporting Information).The decrease in absorbance of the perovskite film is attributed to the increase of amount of 2D NPLs, which are unable to effectively harvest the light in the spectral range of 480-800 nm, leading to the slightly lower light trapping.However, this 2DNPLs-PMMA thin layer may serve as an antireflective layer, allowing more light to pass through the whole sample, thereby increasing the total transmittance.
The current density-voltage ( J-V ) curves of the opaque devices with various 2D NPLs-PMMA composite layers were measured to examine the device performance and DC effect.When using the ST perovskite film with an AVT of ≈17% as the light absorber, as shown in Figure 2c and Table 1, the control device without a 2D NPLs-PMMA composite layer only delivered a PCE of 16.48% with a J sc of 20.16 mA cm À2 , a V oc of 1.12 V, and an FF of 72.72% from the reverse scan (Rev, from V oc to J sc ).In contrast, the devices with 2D NPLs-PMMA composite film were found to produce superior PCEs.Particularly, the 0.25-2D NPLs-PMMA device yielded a champion PCE of 18.47% with a J sc of 20.95 mA cm À2 , a V oc of 1.12 V, and an FF of 78.72% (Figure S10, Supporting Information).The enhanced efficiency is mainly ascribed to the improvement toward J sc and FF.In the meantime, with the increase of amount of 2D NPLs in the composite film, the PCE increased from 16.71% (PMMA) to 18.47% (0.25-2D NPLs-PMMA), and then decreased to 18.02% (0.5-2D NPLs-PMMA) (Figure S11 and Table S1, Supporting Information).To verify the photocurrent increment induced by the DC layer of 2D NPLs-PMMA, EQE measurement was carried out.As depicted in Figure 2d, the integrated J sc for the control and 0.25-2D NPLs-PMMA devices are 19.87 and 20.46 mA cm À2 , respectively, which are in alignment with the J sc extracted from the J-V curves.It is noticed that an obvious EQE boost for the 0.25-2D NPLs-PMMA device is observed when comparing with the control device in the UV-blue region (<440 nm), suggesting that the luminescent down-shifting of the 2D NPLs-PMMA composite film contributes to the enhanced photocurrent.More interestingly, this DC effect is more noticeable in the devices with a higher AVT/thinner perovskite film, as shown in Figure S12, Supporting Information.The reason for this behavior is probably because more UV light could reach the 2D NPLs-PMMA layer through the thinner perovskite and be converted to visible light, leading to the higher EQE at UV-blue region.To quantify the contribution of DC of 2D NPLs-PMMA composite layer to the integrated J sc , we calculated the integrated J sc from 340 to 440 nm ( J sc-440 ) and the contribution index.As listed in Table S2, Supporting Information, the contribution index of the device with a lower AVT/thicker perovskite layer (17% of AVT) is around 11.9%.However, when the perovskite layer is thinner with a higher AVT of 27%, the DC effect is more remarkable, contributing 34.3% to the improvement of integrated J sc .In addition, due to the enhancement of efficient hole extraction and suppressed defects at the interface of perovskite/Spiro-OMeTAD by inserting a 2D NPLs-PMMA layer, there is also an EQE increment in the longer wavelength region of 600-800 nm.Both DC and interface passivation effects of the 2D NPLs-PMMA layer led to the improved photocurrent.To further confirm the influence of 2D NPLs-PMMA-based DC layer on photocurrent, we measured the J-V curves of the control and 2D NPLs-PMMA-based devices under an UV illumination (365 nm, 4 W).As demonstrated in Figure S13 and Table S3, Supporting Information, the devices with 2D NPLs-PMMA layer produced the higher J sc , 1.14 mA cm À2 for 0.1-2D NPLs-PMMA, 1.16 mA cm À2 for 0.25-2D NPLs-PMMA, and 1.25 mA cm À2 for 0.5-2D NPLs-PMMA, than the control device, which only delivered a J sc of 0.98 mA cm À2 under the UV-365 nm illumination.This result further reveals the photocurrent improvement achieved by the luminescence conversion from UV light to visible light via the 2D NPLs-PMMA DC layer in the devices.
The AVT increase of the perovskite layer is accompanied by the lowering of PCE and J sc of both control and 0.25-2D NPLs-PMMA devices.Figure 2e presents the correlation of the thickness of perovskite film with AVT, and the thickness of perovskite film with the J sc of the device.With the increase of AVT from 17% to 40% (decrease of perovskite film thickness from 250 to 120 nm), the PCE and J sc of the control device deceases from 16.48% to 4.62% and from 20.16 to 8.41 mA cm À2 , respectively, while the PCE and J sc of the 0.25-2D NPLs-PMMA device reduces from 18.47% to 5.26% and from 20.95 to 8.90 mA cm À2 (Table 1).This is attributed to the relative decrease in light absorbance of the perovskite layer.Significantly, the devices with a 2D NPLs-PMMA composite layer always exhibit a higher J sc compared with the control device at a similar AVT, indicating the contribution of DC effect to the device performance.Here, we propose the down-shifting process of 2D NPLs-PMMA layer in the devices as illustrated in Figure 2f.The 2D NPLs in the composite layer are excited by absorbing the UV-blue photons with the energy higher than their bandgap (>2.59 eV), which are not effectively absorbed Hysteresis index: HI = (PCE Rev -PCE FW )/PCE Rev .
by the bulk 3D perovskite film.This excitation could happen either through the direct absorption of incident UV light or through energy transfer from the adjacent perovskite materials. [19]Upon absorption of UV photons, the excitons consisting of electron-hole pairs are created in the NPLs.Owing to the layered structure of 2D NPLs, the excitons are spatially confined within the 2D layers, resulting in quantum confinement effects.This confinement leads to a quantization of energy levels and enables the excitons to possess discrete energy state. [20]The excitons in 2D NPLs undergo the radiative recombination at lower energy band, emitting lower-energy photons, namely visible photons.The energy of the emitted photons matches well with the bandgap of the bulk 3D perovskite film, making them become more effectively absorbed by the bulk perovskite light harvester and generate higher photocurrent.Therefore, by introducing a thin DC layer of 2D NPLs-PMMA, the UV-blue light can be converted into the visible light absorbed by the bulk 3D perovskite film, contributing to the photocurrent enhancement of the devices.The statistic PCE distributions of the control and 0.25-2D NPLs-PMMA devices (10 cells for each type) with two different AVT values of perovskite layer (top ≈17%, bottom ≈27%) are shown in Figure 3a, representing a good reproducibility and reliability of the devices.The statistics of the photovoltaic parameters including PCE, V oc , J sc , and FF are plotted in Figure S14 and S15, Supporting Information, which display the similar trend to the champion devices for each type.Apart from the evident J sc enhancement, all the devices with 2D NPLs-PMMA layer show a higher V oc and FF, which are attributed to the passivation effect of the 2D NPLs-PMMA thin layer at the interface of perovskite and Spiro-OMeTAD.The 2D NPLs-PMMA layer is able to passivate the surface defects of the perovskite bulk layer to decrease the interfacial nonradiative recombination, since the donor electrons from oxygen atoms in the carbonyl (C = O) groups of PMMA can reduce the charge state of Pb 2þ defect ions at the perovskite/HTL interface. [17,21]To analyze this surface passivation effect, the steady-state PL was performed.As illustrated in Figure 3b, compared with the control perovskite film without 2D NPLs-PMMA composite layer, the 0.25-2D NPLs-PMMAbased sample presents a much stronger PL peak at approximately 783 nm, which indicates more radiative recombination and less defects in the bulk 3D perovskite film.The better radiative recombination and suppressed nonradiative recombination process in 0.25-2D NPLs-PMMA are beneficial to reducing energy loss and thus enhancing device efficiency.This is also confirmed by the time-resolved PL (TR-PL) spectra (Figure S16, Supporting Information), where the average charge carrier lifetime can be estimated by fitting the TR-PL decay spectra with a triexponential function of time for the perovskite films with/without 2D NPLs-PMMA layer on glass substrates.Table S4, Supporting Information, lists the detailed fitting parameters.The perovskite film alone presented an average charge carrier lifetime (τ ave ) of 225 ns, exhibiting a typical exciton lifetime characteristic of the ST perovskite layer.In contrast, when coated with the 2D NPLs-PMMA thin layer, the perovskite films demonstrated longer τ ave .In particular, the highest τ ave of 394 ns was obtained in 0.25-2D NPLs-PMMA sample, indicating the significant suppression of nonradiative recombination in the bulk perovskite film, in accordance with the steady-state PL results.
Moreover, the evolution of V oc as a function of light illumination intensity (Figure 3c) was also studied to evaluate the trapinduced nonradiative recombination in the devices.The ideality factor (m Ф ) is calculated according to the following equation [22] V where k B T, q, and Ф ph are thermal energy, elementary charge, and absorbed light flux, respectively.The fitted m Ф for the control device is 1.81, which is much higher than the value of 1.40 for the 0.25-2D NPLs-PMMA.Generally, the higher m Ф suggest more severe trap-assisted recombination and higher carrier recombination rate at the interfaces.In other word, the carrier recombination at the interface between perovskite and Spiro-OMeTAD films can be effectively suppressed by introducing a thin layer of 2D NPLs-PMMA composite in the devices.When the amount of 2D NPLs in the composite film further increased, the device performance of 0.5-2D NPLs-PMMA dropped slightly.It is probably due to the increased grain boundaries and ununiform interfacial contact caused by the large amount of 2D NPLs on the surface of bulk 3D perovskite film, which lead to decreased radiative recombination but increased nonradiative recombination of perovskite absorber (Figure S17, Supporting Information).In addition, the PEA 2 MA 3 Pb 4 Br 13 -based 2D perovskite NPLs are essentially a periodic assembly of quantum-confined 2D hybrid perovskite sheets dielectrically separated by a spacer layer of the phenylethyl ammonium (PEA) cations. [23]Owning to the insulating nature of the long-chain PEA cations in 2D NPLs, their conductivity is much lower than the 3D perovskite materials. [24]The higher amount of 2D NPLs are able to increase the resistance for charge transfer from perovskite to HTL, leading to the device performance deterioration.
In addition, the impedance spectrum (IS) measurements of the devices in dark condition were recorded to investigate the dynamic process of charge carrier transfer and recombination in the devices.There are two arcs in the Nyquist plot which can be well fitted by using an equivalent circuit as shown in Figure 3d and S18, Supporting Information.The series resistance (R s ) represents the resistance of conductive glass substrate and metal electrode, the resistance (R 3 ) extracted from the first semicircle at high-frequency zone is associated with the interfaces of selective contacts and perovskite layer, while the resistance (R 1 ) of the extracted second semicircle at low-frequency zone reflects the recombination in the bulk perovskite layer.In general, the sum of R 3 and R 1 is regarded as the recombination resistance (R rec ) of charge carrier at interfaces. [22,25]Table S5, Supporting Information, lists all the detailed parameters extracted from the Nyquist plot.Since the electron and holetransport layers are identical in this study, the difference of R rec comes from the 2D NPLs-PMMA thin layer.Compared with the control device that presents an R 3 of 2.30 kΩ and an R rec of 108.20 kΩ, the remarkably lower R 3 (1.79kΩ) and higher R rec (337.39 kΩ) of 0.25-2D NPLs-PMMA device indicates a more efficient hole carrier transfer and suppressed nonradiative recombination process at the interfaces in the device.
To understand the hole extraction ability, we plotted the energy band alignment diagram (Figure 3e) of the perovskite films with/without different 2D NPLs-PMMA composite layers in the devices by using ultraviolet photoelectron spectroscopy (UPS).Combined with the energy bandgap value of perovskite film derived from the Tauc plot (Figure S19, Supporting Information), the perovskite film demonstrated an optical bandgap of 1.57 eV with a valence band (EVB) of À5.82 eV and a conduction band of À4.25 eV.When an ultrathin layer of PMMA is coated on the top of perovskite film, the VB of the perovskite/PMMA film is down-shifted to À6.25 eV, which is attributed to the insulating nature of PMMA. [26]As depicted in Figure S20, Supporting Information, with the increase of the amount of 2D NPLs in the composite layer, both cutoff and VB maximum binding energy of the film gradually decrease, thereby leading to a slightly increased VB.Compared with the control perovskite film, the 2D NPLs-PMMA thin layer can modify the VB of perovskite film to possess a more favorable smaller energy band offset between the perovskite and Spiro-OMeTAD.Therefore, such a better energy band alignment enables more efficient hole transfer and reduced nonradiative recombination at the interface of perovskite/Spiro-OMeTAD.

Long-Term and UV Stability
The long-term stability of the unencapsulated devices with/ without 2D NPLs-PMMA composite layer was compared via monitoring the performance of six devices (unencapsulated) for each type for around 60 d in the ambient environment in dark condition with an relarive humidity (RH) of 30% at room temperature.As shown in Figure 3f, the PCE for both control and 0.25-2D NPLs-PMMA devices increased gradually in the first 10 d to the maximum value.There are two main reasons for this increment at the beginning test period.Due to the effects of ion migration, perovskite absorber requires time to reach a steady state. [27]Additionally, the lithium salt (LiTFSI) dopant needs sufficient time to promote the oxidation process of Spiro-OMeTAD by oxygen for improving the hole mobility and conductivity of the HTL. [28]After aging for around 60 d, the 0.25-2D NPLs-PMMA devices retained an average value of over 95% of their initial efficiency, but the PCE for the control ones significantly decreased to only 86% of the original value.The performance drop is attributed to the decease of J sc and FF, as Voc for both devices almost remained the same level during the test (Figure S21, Supporting Information).To uncover the reason for this difference, the surface contact angles (CAs) of the perovskite films with/without 2D NPLs-PMMA layer were measured to check their vulnerability to water.Owing to the hydrophobicity of PMMA and 2D perovskite NPLs, [29] the CA of the water droplet on the surface of perovskite/ composite film keeps at around 80°, as depicted in Figure S22, Supporting Information, whereas the perovskite film without 2D NPLs-PMMA composite layer only presents a CA of 21.3°.Clearly, the 2D NPLs-PMMA composite thin layer enables the repulsion of perovskite film against water penetration thereby leading to the better long-term stability of the unencapsulated devices in ambient environment.
It has been reported that the UV illumination can decompose perovskite materials and deteriorate the device performance by inducing oxygen vacancies and defects. [30]To investigate the UV tolerance of the devices, six samples for each type without encapsulation were placed under the continuous UV-light irradiation (365 nm, 4 W) for over 280 h and their photovoltaic performances were monitored with aging time.As shown in Figure 3g, after aging for 280 h, the control devices only retained an average value of 67% of their initial PCE.In contrast, more than 77% of the average performance retention was obtained for the 0.25-2D NPLs-PMMA devices.By analyzing the photovoltaic parameters of the devices as a function of aging time under UV light (Figure S23, Supporting Information), we noticed that the V oc for both devices were quite stable against UV illumination, while the degradation rates of J sc and FF were relatively more apparent in the control devices than that in 0.25-2D NPLs-PMMA devices.The rapid degradation of J sc and FF in the control devices is attributed to the accelerated deterioration of polycrystalline perovskite film and its interfaces by the UV light irradiation. [9,12]By contrast, the 2D NPLs-PMMA thin layer is able to convert UV photons into low-energy photons, which protects the perovskite materials from UV-light-induced degradation, thereby improving the photostability under UV light irradiation and obtaining higher retention of the original efficiency.

Bifacial ST PSCs
To fabricate the full ST PSCs, a dielectric/metal/dielectric (DMD) transparent counter electrode of MoO 3 /Au/MoO 3 instead of opaque Au electrode was used as rear electrode by using e-beam evaporation deposition technique.As shown in Figure 4a, the transparent MoO 3 /Au/MoO 3 electrode is composed of three layers with the thickness of approximately 20, 8, and 20 nm, respectively.The bottom MoO 3 layer works as the seed layer to grow up continuous thin Au film.In the meantime, the top MoO 3 layer plays a role of optical spacer.It not only can control the reflectance of the metal defining the transmission window of DMD structure, but also can protect the underneath thin layers. [31]The fabricated MoO 3 /Au/MoO 3 -based counter electrode demonstrated the excellent transparency and decent conductivity.Over 77.5% transmittance in the visible range between 400 and 800 nm with the sheet resistance of 6.5 Ω À2 was achieved (Figure S24, Supporting Information).
Figure 4b presents the transmittance spectra of the 0.25-2D NPLs-PMMA ST-PSCs using the DMD electrode with two different perovskite absorber thicknesses, where the AVT values for the full devices were also included.The J-V characteristics of the devices with the light irradiated from both FTO side (bottom side, BS) and DMD side (top side, TS) were recorded, respectively, as illustrated in Figure 4c and S25, Supporting Information.It is notable that under the BS illumination, the 0.25-2D NPLs-PMMA ST-PSCs delivered a decent PCE of 14.26% with an AVT of 19.4% and a PCE of 10.65% with an AVT of 26.9%, leading to a light utilization efficiency (LUE = AVT Â PCE) of 2.77% and 2.86%, respectively (Table 2).Compared with the opaque counterparts (Table 1), the ST-PSCs presented a lower device performance, which is attributed to the reduced light-trap and charge collecting capability of the rear electrode.Due to the high transparency and relatively low conductivity of the MoO 3 /Au/MoO 3 -based rear electrode compared to the thick Au electrode, the lower J sc , V oc , and FF were obtained with the ST-PSCs.However, the transparent DMD electrode enables the potential of bifacial operation, where the light can be absorbed from both the BS and the TS of the ST-PSCs.The bifaciality of the 0.25-2D NPLs-PMMA ST-PSCs, which is defined as the PCE ratio of the device under front and bottom illumination, is 0.77 (0.79) for an AVT of 19.4% (26.9%), indicating the efficient utilization of photon incident from both sides of the ST-PSCs.Thanks to the DC and passivation effects of 2D NPLs-PMMA thin layer, the 0.25-2D NPLs-PMMA ST-PSCs demonstrated better device performance with negligible hysteresis compared to the control ST-PSCs without 2D NPLs-PMMA layer, which only produced a PCE of 12.51% (AVT ≈20%) and 9.61% (AVT ≈27%), respectively, as shown in Figure S25, Supporting Information.
Furthermore, the ST-PSCs illuminated from the FTO side often exhibits a higher efficiency than that from the DMD side, which is mainly ascribed to the higher J sc .To uncover the underlying reason for this phenomenon, the EQE spectrum (Figure 4d and S26, Supporting Information) was measured by illuminating through either FTO or DMD side for the ST-PSCs.It was observed that from the FTO side over 80% of EQE value was achieved at the low wavelengths ranging from 400 to 550 nm.While irradiating from the DMD side, the EQE curves exhibited a rapid drop at the wavelength below 430 nm and slight decrease between 430 and 600 nm.The reason for this remarkable EQE reduction at the lower wavelength range is because of the strong absorption of Spiro-OMeTAD in the UV region. [32]As shown in Figure S27, Supporting Information, Spiro-OMeTAD layer presents the significant absorption starting from 430 nm toward the low-wavelength region.The calculated J sc extracted from EQE spectrum of the bifacial ST-PSCs measured from different illumination directions are listed in Table 2, which demonstrates the great agreement with the J sc obtained from the J-V plots.In addition, the time-dependent power output of the ST-PSCs illuminated from different directions by monitoring the J sc and PCE under a constant voltage bias at the maximum power point was further studied (Figure 4e and S28, Supporting Information).Both control and the 0.25-2D NPLs-PMMA ST-PSCs showed steady output and rapid transient photocurrent response within 120 s, resulting in a slightly reduced PCE (by an average of ≈0.24%) compared to the value calculated from the J-V curves measured under Rev scan.The stability test of 0.25-2D NPLs-PMMA ST-PSC under continuous 1 sun illumination was carried out as presented in Figure S29, Supporting Information.Around 70% of performance retention was achieved for the device without encapsulation after 4 h operation at a bias voltage of maximum power point in ambient air (≈35% RH).It is noticed that the temperature of the device during the test went up to over 65 °C, due to the released heat from the Xe lamp of the solar simulator, which is probably the main reason for the PCE drop during the test.Figure 4f presents the champion PCE as a function of AVT of the 0.25-2D NPLs-PMMA ST-PSCs in our work and the comparison with the state-of-the-art ST-PSCs using similar structure of DMD-based transparent counter electrode in literature.It is worth pointing out that, to the best of our knowledge, the efficiencies of our ST-PSCs (14.26% for an AVT of 19.4% and 10.65% for an AVT of 26.9%) are the highest among all the reported DMD-based ST-PSCs at a similar AVT value in the previous published work.Even compared to the ST-PSCs with other type of transparent electrode, our results are in the top level (Table S6,

Supporting Information).[6d]
Although the required transmittance for the ST-PSCs depends on the specific applications, it is generally recommended that a threshold transmittance for the window applications is an AVT of 20%-30% for the whole devices. [33]Our ST-PSCs present excellent PCEs within this range of AVT, making them highly promising in BIPV applications.

Conclusion
In summary, we have demonstrated an effective approach to improve the device performance of ST-PSCs by using a composite thin layer of 2D NPLs-PMMA.Due to the DC and passivation effects of this thin layer, the devices with opaque Au electrode presented enhanced photocurrent and suppressed interfacial nonradiative recombination.Consequently, by tuning the thickness of perovskite film and concentration of 2D NPLs in the composite layer, the opaque 0.25-2D NPLs-PMMA devices yielded significantly improved PCEs with various AVT values.In addition, these unencapsulated devices exhibited impressive UV stability, retaining over 77% of the original PCE after aging at the continuous UV illumination in ambient condition (room temperature, ≈30% RH) for 280 h.More encouragingly, when equipped with a transparent MoO 3 /Au/MoO 3 -based counter electrode, the bifacial 0.25-2D NPLs-PMMA ST-PSCs achieved an outstanding PCE of 14.26% (10.65%) with a total AVT of 19.4% (26.9%), which represented the highest PCEs reported for DMD-based ST-PSCs with the similar AVT values.Our findings pave a promising way for enhancing device efficiency and stability of ST-PSCs for BIPV applications.

Figure 2 .
Figure 2. a) The cross-sectional SEM image of the opaque device with an AVT value of ≈17% (perovskite layer).b) Transmittance spectra of the perovskite films on glass slide with tuned AVT.c) The J-V curves of the control and 0.25-2D NPLs-PMMA opaque devices with different AVT values of perovskite films.d) EQE spectrum and the integrated current density of the control and 0.25-2D NPLs-PMMA opaque devices with an AVT value of ≈17% (perovskite layer).e) The plot of the AVT and the current density of the device as a function of the thickness of perovskite film.f ) Schematic illustration explaining the mechanism of enhanced photocurrent of the device by employing a 2D-NPLs-PMMA composite layer.

Figure 3 .
Figure 3. a) Statistic PCE distributions of the control and 0.25-2D NPLs-PMMA opaque PSCs with an AVT of ≈17% (top) and ≈27% (bottom) for perovskite layer.b) The steady-state PL spectrum of perovskite film and perovskite/0.25-2DNPLs-PMMA film on glass substrates.The excitation wavelength is 532 nm.c) The plot of open-circuit voltage as a function of light illumination intensity in the control and 0.25-2D NPLs-PMMA opaque PSCs.d) Nyquist plots and fitted curves of the control and 0.25-2D NPLs-PMMA devices under dark condition.The inset is the equivalent circuit.e) Energy band alignment diagram of different layers in devices.f ) Evaluation of normalized PCE for the unencapsulated control and 0.25-2D NPLs-PMMA opaque PSCs based on six devices for each type for around 60 d (storage conditions: ≈30% RH, room temperature, dark condition).g) Stability test for the unencapsulated control and 0.25-2D NPLs-PMMA opaque PSCs based on six devices for each type under continuous UV illumination in ambient conditions (≈30% RH, room temperature).Inset: the digital image of the devices under UV light illumination.

Figure 4 .
Figure 4. a) The cross-sectional SEM image of the ST-PSC with a MoO 3 /Au/MoO 3 transparent electrode.b) Transmittance spectra of the 0.25-2D NPLs-PMMA ST-PSCs with different AVT values.c) The J-V curves and d) EQE spectrum and the integrated current density of the 0.25-2D NPLs-PMMA ST-PSCs with a total AVT of 19.4% (top) or 26.9% (bottom).e) Stabilized photocurrent measurement for the 0.25-2D NPLs-PMMA ST-PSCs with different AVT values at a bias voltage of maximum power point under 1 sun illumination.f ) The plot of PCE as a function of AVT for comparison with other ST-PSCs with a DMD transparent electrode.

Table 1 .
Characteristic photovoltaic parameters of the champion opaque PSCs with/without 2D NLPs-PMMA thin layer under reverse (Rev) and forward (FW) scans.

Table 2 .
Characteristic photovoltaic parameters of the champion ST-PSCs with/without 2D NLPs-PMMA thin layer measured from both bottom and top sides under Rev and FW scans.Hysteresis index: HI = (PCE Rev -PCE FW )/PCE Rev .b) Bifaciality = PCE Top /PCE Bottom .