Electrosprayed Polymer-Hybridized Multidoped ZnO Mesoscopic Nanocrystals Yield Highly Efficient and Stable Perovskite Solar Cells.

Solid-state perovskite solar cells have been expeditiously developed since the past few years. However, there are a number of open questions and issues related to the perovskite devices, such as their long-term ambient stability and hysteresis in current density-voltage curves. We developed highly efficient and hysteresis-less perovskite devices by changing the frequently used TiO2 mesoscopic layer with polymer-hybridized multidoped ZnO nanocrystals in a common n-i-p structure for the first time. The gradual adjustment of ZnO conduction band position using single- and multidopant atoms will likely enhance the power conversion efficiency (PCE) from 8.26 to 13.54%, with PCEmax = 15.09%. The highest PCEavg of 13.54% was demonstrated by 2 atom % boron and 6 atom % fluorine co-doped (B, F:ZnO) nanolayers (using optimized film thickness of 160 nm) owing to their highest conductivity, carrier concentration, optical transmittance, and band-gap energy compared to other doped films. We also successfully apply a fine polyethylenimine thin layer on the doped ZnO nanolayers, causing the reduction in work function and overall demonstrating the enhancement in PCE from ∼10.86% up to 16.20%. A polymer-mixed electron-transporting layer demonstrates the remarkable PCEmax of 20.74% by decreasing the trap sites in the oxide layer that probably reduces the chances of carrier interfacial recombination originated from traps and thus improves the device performance. Particularly, we produce these electron-rich multidoped ZnO nanolayers via electrospray technique, which is highly suitable for the future development of perovskite solar cells.

Introduction

The intensive interest in perovskite solar cells has led to a very rapid progress in cell performance compared to any other photovoltaic technology.[1−3] The advances in power conversion efficiency (PCE) have been achieved by optimizing cell configuration and layer fabrication processes.[4−9] PCE as high as 20.7% has been achieved using a planar structure, while a PCE of 22.1% has been obtained via a mesoscopic structure.[10] For most of the device architectures, a thin film of TiO2 has been mostly used as an electron-transporting layer (ETL) in both planar and mesoscopic perovskite cells.[11−16] New types of ETLs (e.g., WO3, Zn2SnO4, and SrTiO3) have also been explored to produce efficient perovskite cells.[17−19] However, to develop up-to-date efficient perovskite solar cells using these mesoscopic ETLs along with thin hole-blocking layers of the same materials, high-temperature sintering is required. This severely hinders the fabrication of large flexible devices.[20] Thus, in the near future, the development of perovskite devices at low temperature would be an encouraging way and movement.

Nanostructures of ZnO are highly suitable alternative materials to other mesoscopic ETLs for perovskite devices owing to their suitable energy levels, ease of fabrication at low temperatures, and relatively high electron mobility.[21−24] Only a handful of studies have evaluated the effect of ZnO-based ETLs of different morphologies.[25,26] ZnO nanorods, for example, offer efficient charge collection compared to TiO2 nanorods, owing to their more rapid charge-carrying behavior.[27] In addition, doping the ZnO by substituting the metallic atoms into its lattice not only improves its conductivity, but also facilitates the shift of the Fermi level toward conduction band direction, which will eventually enhance the performance of dye-sensitized solar cells (DSSCs), but the efficiency is not satisfactory yet.[28−30] Moreover, the alternation in the surface morphology of oxide layer will greatly influence the device performance because the interface between the perovskite and the oxide layer plays an important role in charge separation and crystal growth.[17,25] Reports are also available on the bulk modification and interfacial treatment of oxide layer to enhance the collection and injection of charges and also to reduce the carrier recombination at the same time.[31−33] However, the efficiency of these liquid-based devices is not comparable to that of perovskite solar cells, and the main drawbacks of DSSCs (instability of liquid electrolyte, costly dye and catalyst, etc.) have not been solved yet. Recently, researchers[6,22,34−36] have also presented a route of surface modification and improving the electron-transport properties to enhance the performance of perovskite devices by doping metallic elements (e.g., Al, Yt, Mg, Nb) into the nanostructures of ZnO and TiO2. However, compared to the advanced TiO2 ETL-based devices, lower device performances were achieved even in these devices.

Motivated by the desire to evaluate the synergetic influence of dopant on the nanostructures surface of oxide layer and to tune the energy levels to obtain an efficient perovskite solar cell, we introduced first the substitution of multidopant into the ZnO lattice and studied their effect on the perovskite device performance. In our previous reports, we investigated the influence of single and co-dopants (with optimized dopant amounts) on the performance of transparent electrodes (transparent conductive oxides) by enhancing the electrical conductivity, optical transmission, stability, and possibility for future commercialization.[37−39] We had found that the additional electrons could be generated using doping atoms, which will enhance the carrier density, mobility, conductivity, and surface stability because such findings of multidoped ZnO films have an excellent technological aspects and would also add to basic studies.

The use of conjugated polyelectrolytes offers the advantage of changing the work function of the oxide layer to boost the cell performance in an efficient way.[40−43] In this method, the surface of oxide layer is finely coated using a material which physically or chemically adsorbed on it; the surface transformer is selected in a sense to generate strong surface dipoles that bring a shift in vacuum level and thus alter the work function of the oxide ETLs.[44] Very recently, polyethylenimine (PEI) or polyethyleneimine-ethoxylated polymers have been investigated to modify indium tin oxide (ITO) work function by forming a dipole layer at the ITO–PEI interface and to yield efficient solar cells.[45] These polymers can easily form solutions with 2-methoxyethanol or water (known as environmentally friendly solvents) and can be simply handled in ambient atmosphere. Moreover, their ease of handling and inexpensive nature make them well suited for printed electronics with roll-to-roll large-area production.[44] The key issue with this method is that the existence of trap sites in the bulk of the ZnO ETLs still forbid the transport of electrons caused by the empty spaces between ZnO nanocrystals (NCs).

A different methodology is to form the composite of polymers and oxide layers (ZnO or TiO2) by mixing them together before deposition in the developing stage. A simple polymer-mixed composite single-layer ETL is more functional and easy to process compared to a complex bilayer film. However, only a few studies have been conducted where a composite ETL has been investigated rather than a bilayer ETL (oxide layer coated with polymer) in inverted polymer cells. For instance, only in polymer solar cells, composites of poly(ethylene oxide),[46] poly(ethylene glycol),[47] poly(vinylpyrrolidone),[44] and PEI[48] with ZnO films have been reported to construct polymer-mixed ETLs. However, such mixed polymer composite ETLs have not been investigated yet for perovskite solar cells.

In this work, we have introduced a low-temperature and easy-to-operate full-solution-processed electrospraying technique for the development of mesoporous pure ZnO and multidoped ZnO nanocrystals. We constantly increase the position of conduction of the ZnO ETLs by doping with diverse combination of metallic and nonmetallic atoms. By changing the position of oxide’s conduction band and surface morphology, we are capable of significantly improving the open-circuit voltage (VOC) to achieve the maximum values of efficiency reported so far for doped ZnO-based perovskite devices. Previously, we have studied the transparent electrodes in the form of 2 atom % boron-doped ZnO (B:ZnO), 2 atom % tantalum-doped ZnO (Ta:ZnO), 2 atom % boron and 6 atom % fluorine co-doped (B, F:ZnO), and 2 atom % tantalum and 6 atom % nitrogen co-doped (Ta, N:ZnO) thin films.[37−39] We have constructed the perovskite solar cells using these thin films and investigated their effect on the cell performance at optimized dopant concentrations reported previously. The B:ZnO films with highest conductivity, carrier concentration, optical transmittance, and band-gap energy will likely produce a PCEavg of 13.54% and a PCEmax of 15.09% among other doped films. We have substantially reduced the work functions of these films by coating them with a thin layer of PEI, which results in remarkable improvements of PCE of 16.2% in hysteresis-free mesoscopic perovskite solar cells. A remarkable PCEmax of 20.74% was achieved when a polymer-mixed ZnO ETL is used. To the best of the author’s knowledge, this is the first study that employs co-doped semiconducting nanostructured films in combination with polymer as an ETL to obtain hysteresis-less and very efficient perovskite solar cells. Furthermore, the obtained PCEmax of 20.74% exceeds the previously reported highest efficiency for the perovskite solar cells based on ZnO ETLs.[49−53]

Results and Discussion

General Scheme for Tuning Nanolayer Properties and Device Architecture

Figure a shows the schematic illustration of low-temperature, easy-to-operate, and fully solution-processed electrostatic spray deposited pure ZnO and multidoped ZnO mesoporous ETLs composed of fine nanocrystals (as shown in the inset of Figure a). Highly mesoporous ZnO ETLs could be easily achieved by providing a constant low heating (150 °C) at the bottom of the substrate during the entire deposition process. The Brunauer–Emmett–Teller (BET) surface area increased from 7.3 m2/g for pure ZnO to 27.5 m2/g for B, F:ZnO ETL. However, these ETLs have numerous traps left in the bulk of ZnO, and doped ZnO nanocrystals (NCs) lead to electron trapping due to the voids between ZnO NCs (Figure d(i)). Coating a thin layer of conjugated polyelectrolyte, such as polymer (PEI), on top of the ETL (as illustrated in Figure b) can reduce the work function of ETL and will also reduce the surface traps, which enhances electronic pairing of ETL/perovskite layer and thus boost the device performance (as explained in Figure d(ii)). The functionality of these films was further enhanced by mixing the ZnO solution with PEI and the blended solution was electrosprayed directly onto the heated substrates, as illustrated in Figure c. The PEI is finely dispersed in the entire film (inset of Figure c), which decreased the surface roughness and trap states for ZnO film, thereby possibly improving the physical contact and stimulating the strong dipoles movements between the perovskite absorber layer and ITO conducting substrate, generating the cells with improved performance (Figure d(iii)).

Figure 1

Schematic illustration of low-temperature and fully solution-processed electrosspray-deposited (a) pure ZnO and multidoped ZnO nanocrystals (the inset is the magnified view of mesoporous nanocrystals); (b) PEI-coated ZnO and multidoped ZnO nanolayers; and (c) PEI-mixed pure ZnO and multidoped ZnO nanocrystals (the inset is the magnified view of PEI-blended nanocrystals). (d (i)–(iii)) Schematic illustration of device architecture based on the above oxide nanolayers showing the electron trapping due to the voids between ZnO nanocrystals.

Figure S1 displays the surface nanostructures of pure ZnO and multidoped ZnO (such as B:ZnO, Ta:ZnO, B, F:ZnO, and Ta, N:ZnO) mesoscopic ETLs formed by electrospraying at the optimized dopant concentrations and the process parameters (flow rate, distance between the nozzle and the substrate, solution concentration, substrate temperature, and applied voltage as summarized in Table S1) as reported elsewhere.[37−39] As the film morphology is critically influenced by the dopant concentrations and process parameters. The grain size of the doped ETLs was gradually reduced as the dopant amount was increased in the precursor solution, which will eventually affect their optical and electrical characteristics and hence the device performance.[37−39] It was also examined that the additions of dopant atoms have strongly influence the surface roughness of ETLs. It was also examined that when dopant atoms were substituted into pure ZnO films, a substantial reduction in surface roughness was observed.[37−39] The surface modification and morphology refinement of the oxide ETLs is a better route to refine the morphology of light harvester and transfer of electrons in the mesostructured perovskite solar cells. The electrical and optical properties of the pure ZnO and doped ZnO films at the optimized dopant concentrations are plotted in Figure and summarized in Table S2. The B, F:ZnO nanolayer demonstrates the lowest resistivity (9.70 × 10–5 Ω cm), highest optical transmittance (99.8%), maximum carrier concentration (3.41 × 1021 cm–3), and a band-gap energy of 3.42 eV compared to other pure and doped ZnO films. A side-view scanning electron microscopy (SEM) image of full cell based on B, F:ZnO nanolayer as an electron-transporting material is shown in Figure S1f.

Figure 2

Plots of (a) carrier concentration and electrical resistivity and (b) band-gap energy and optical transmittance (at 520 nm) for five different pure ZnO and doped ZnO nanolayers.

Device Structure and Energy-Level Diagram

The schematic demonstration of the perovskite device architecture (glass/ITO/ZnO or doped ZnO nanolayers/MAPbI(3–Cl/spiro-OMeTAD/Ag) and the energy-level description of the corresponding materials (measured by Kelvin probe) are shown in Figure a,b, respectively. The incorporation of dopant atoms into ZnO lattice produced dopant-substituted crystal lattice in ZnO, allowing ZnO ETLs to have higher values of conduction band, as illustrated in Figure b. It is important to note that the electronic structure of ZnO can be tuned using doping, which results in increase of the band gap. The single-doped (B:ZnO and Ta:ZnO) ZnO nanolayers do not contribute significantly to the rise of conduction band edge. Therefore, co-doped ZnO nanolayers, such as Ta, N:ZnO and B, F:ZnO, are also employed to raise the conduction band prominently, and a significant shift of 0.22 eV is caused by B, F:ZnO nanolayer compared to pure ZnO. The continuous elevation of conduction band edge will increase the VOC for the perovskite cell and suppressed the recombination at the doped ZnO/perovskite interface.[26]

Figure 3

(a) Illustration of device architecture with PEI-coated and PEI-mixed ZnO and doped ZnO nanolayers and (b) energy-level diagram of the corresponding devices with five various types of pure and doped ZnO nanolayers. Continuous raising of the conduction band edge using various combinations of dopants causes a 0.22 eV shift in the conduction band compared to the pure ZnO films, and also the conduction band edge of B, F:ZnO nanolayer decreased to 0.38 eV owing to dipole interaction by the PEI layer.

Device Performance of Pure and Doped ZnO Nanolayers with Optimized Dopant Concentrations

In Figures a and S2, we show the J–V plots and external quantum efficiency (EQE) of the perovskite devices with pure ZnO and doped ZnO nanolayers with film thickness of 160 nm. The photovoltaic parameters, such as open-circuit voltage (VOC), short-circuit current density (JSC), fill factor (FF), and power conversion efficiency (PCE) of the related cells, are outlined in Table . The cells with B, F:ZnO nanolayer show the best JSC of 20.75 mA/cm2, FF of 0.68, VOC of 960 mV, average PCE (PCEavg) of 13.54%, and maximum PCE (PCEmax) of 15.09%, which is much improved compared to the devices based on pure ZnO nanolayer with similar film thickness (PCEavg of 8.26% and PCEmax of 9.16%) and also higher than other doped films as well. Actually, doping of ZnO enhances the electron density, which results in an increase in JSC and PCE as well.[37−39] Moreover, the rise in VOC (from 860 to 960 mV) after doping originated from an increase in Fermi energy levels and electron density, and reduced recombination losses, which decreases the restriction to the transfer of electrons.[37−39] Furthermore, continuous raising of the conduction band edge using various combinations of dopants causes a 0.22 eV shift in the conduction band compared to pure ZnO films, which will eventually improve the VOC and enhance the faster transfer of electrons.[49] The devices fabricated with thicker pure ZnO and doped ZnO nanolayers (240 and 300 nm rather than 160 nm) showed lower VOC and JSC (in the reverse direction only) (Figure S3), showing that thinner nanolayers are better for highly efficient devices. The enhanced performance of the nanolayers with small thickness is attributed to the complete infiltration of the perovskite absorber into the oxide layer and to the improvement of charge-transfer characteristics.[22]Figure S2 shows the EQE spectra of the perovskite devices with pure and doped ZnO nanolayers. A plateau of over 85% EQE across the visible region is clearly seen, proposing the full conversion due to the improved absorbance of perovskite absorber for the devices based on B, F:ZnO nanolayers. The solar cells exhibit a considerable hysteresis-less behavior when altering the scan direction, as shown in Figure a, and the related photovoltaic parameters are listed in Table S3. As exhibited in Figure b,c, device performance in terms of JSC, VOC, FF, and PCEavg was enhanced upon the addition of dopants into ZnO. The devices based on B, F:ZnO nanolayers exhibit the best performance compared to other nanolayers. The variation in the performances of 50 different perovskite devices based on pure ZnO and B, F:ZnO nanolayers is presented in Figure d, demonstrating a PCEavg of 8.26 and 13.54%, respectively.

Figure 4

(a) J–V plots of five various types of pure and doped ZnO nanolayers in different scanning directions at a rate of 0.01 V/s. Plots of (b) JSC and FF and (c) VOC and PCEavg for ZnO and four different dopant schemes. (d) Bar graphs demonstrating the changes in PCE for 50 separate devices with pure ZnO and B, F:ZnO nanolayers.

Table 1

Device Parameters (in the Reverse Scan Direction Only) with Pure and Doped ZnO Nanolayers as ETLs under 1 sun Illumination (AM 1.5G, 100 mW/cm2)

ETL type	JSC (mA/cm2)	VOC (mV)	FF (%)	PCEavg (%)	PCEmax (%)
pure ZnO	16.0	860	61	8.39 ± 0.15	9.16
B:ZnO	18.0	925	63	10.48 ± 0.13	11.49
Ta:ZnO	18.5	930	65	11.18 ± 0.12	12.67
Ta, N:ZnO	20.0	945	66	12.47 ± 0.14	13.99
B, F:ZnO	21.0	960	68	13.70 ± 0.13	15.09

Device Performance of Doped ZnO Films with Unoptimized Dopant Concentrations

We have also constructed perovskite devices using highly doped ZnO nanolayer to evaluate their performance with the devices based on optimized dopant concentration. It was observed that the devices based on highly doped ZnO films have low device performance (see Figure S4 and Table S4) compared to the optimized ones (as shown in Figure a). Actually, at higher dopant amounts, film morphology gets destroyed, which increases the surface roughness.[37−39] Furthermore, the substitution of heavily doped atoms into the ZnO sites will result in lower carrier concentration and less conductive and transparent films,[37−39] which will eventually affect the device performance in terms of low-efficient ones, as can be seen in Figure S4.

Device Performance for PEI-Coated Pure and Doped ZnO Films

We have now spin-coated the best-performing pure ZnO and B, F:ZnO nanocrystals using a thin layer of PEI, keeping in mind the strong interaction between nonpolar surface of ZnO and PEI,[45] to achieve a fine coating that reduces the work function of ZnO further.[26] The schematic representation of the energy-level diagram based on the PEI-coated B, F:ZfunO nanolayer is shown in Figure b. We observe a 0.38 eV shift (measured by Kelvin probe) in work function between PEI-uncoated B, F:ZnO and PEI-coated B, F:ZnO nanolayer, caused by PEI coating. These findings also demonstrate that PEI fine coating reduces the work function of B, F:ZnO ETL from 4.16 to 3.78 eV.

We have also confirmed whether the PEI thin layer was actually coated onto the ZnO surface or not since it has a thickness of only 2–3 nm. To prove the successful coverage of ZnO surface with PEI layer, X-ray photoelectron spectroscopy (XPS) was performed, as exhibited in Figure S5. As presented in Figure S5a, the representative peak (N 1s), which is collected from the nitrogen element in the PEI polymer (see Figure b), was clearly seen for PEI-coated B, F:ZnO nanolayer and PEI-coated ITO substrate samples around 400 eV. The presence of boron (B) and fluorine (F) peaks in Figure S5a confirms the successful substitution of B and F atoms into the ZnO lattice. These findings confirm that the PEI layer was essentially coated on the B, F:ZnO nanolayer surface. Especially, we observe that the positions of O1s and Zn2p peaks in the B, F:ZnO nanolayer were moved in the presence of PEI layer (Supporting Information, Figure S5c,d), which shows that the PEI layer definitely disturbs the ZnO electronic structure. The PEI coating covering the B, F:ZnO nanolayer surface results in the increase of JSC rise from 20.75 to 21.85 mA/cm2, FF from 68 to 70%, and VOC from 960 to 990 mV, as listed in Table and exhibited in Figure S6, demonstrating a PCEavg of 15.14% (PCEmax = 16.20%). Thus, PEI-coated B, F:ZnO nanolayer produced considerably better devices than uncoated nanolayers (PCEavg = 13.54%; PCEmax = 15.09%) and this enhancement is attributed to the reduced series resistance and rapid electron extraction.[45] Furthermore, coating the surface of pure ZnO nanolayer using PEI polymer will also improve the device performance, but not much as observed for doped samples. The enhanced performance of doped films features the outstanding performance realized by the combined implementation of doping, surface modification of semiconducting surface, and PEI coating.[23] The J–V plots for the best-performing perovskite devices after PEI coating also showed certainly improved efficiency, producing PCEmax values of 16.20 and 10.86% for the devices with B, F:ZnO and pure ZnO nanolayers, respectively (as plotted in Figure a). We have also prepared 50 cells (using PEI-coated B, F:ZnO and pure ZnO nanolayers) independently under the same experimental conditions. The histogram of the PCEavg (Figure b) exhibits that the cells (based on PEI-coated B, F:ZnO nanolayer) had an average efficiency of 15.14% and a PCEmax of 16.20%.

Table 2

Device Parameters (in Both Forward and Reverse Scans) with PEI-Coated Pure ZnO and B, F:ZnO Nanolayers Having Optimal ETL Thickness of 160 nm

film type	scan mode	JSC (mA/cm2)	VOC (mV)	FF (%)	PCEavg (%)	PCEmax (%)
pure ZnO	reverse	17.0	895	63	9.58 ± 0.12	10.95
	forward	16.8	895	63	9.47 ± 0.14	10.77
	average	16.9	895	63	9.52 ± 0.13	10.86
B, F:ZnO	reverse	22.0	990	70	15.24 ± 0.11	16.28
	forward	21.70	990	70	15.03 ± 0.10	16.13
	average	21.85	990	70	15.14 ± 0.10	16.20

Figure 5

J–V plots for the champion devices with (a) PEI-coated and (b) PEI-mixed ZnO and B, F:ZnO nanolayers using different sweep directions and bar graphs exhibiting the alteration in PCEavg for 50 individual devices with (c) PEI-coated and (d) PEI-mixed pure ZnO and B, F:ZnO nanolayers.

Device Performance for PEI-Mixed Pure and Doped ZnO Films

Earlier, we demonstrated that PEI coating covering the surface of pure and doped ZnO nanocrystals generated enhanced device efficiency compared to the only pure ZnO and doped ZnO ETLs. Next, we have electrosprayed PEI-mixed pure and doped ZnO nanocrystals to improve the device performance further thanks to the reduced traps found in the oxide nanocrystals filled by PEI and decreased surface roughness and reduced work function. Figure S7 displays the surface atomic force microscopy (AFM) monographs of the pure ZnO, PEI-coated ZnO, and ZnO/PEI composite layers. It is worth noting that for PEI-coated ETLs, surface roughness reduced from 6.78 nm for ZnO film to 5.21 nm and further decreased to 3.46 nm for polymer-mixed ZnO nanocrystals, possibly enhancing the physical contact and prompting robust molecular dipoles between the absorber layer and ITO substrate, yielding the enhanced PCEs.[48] The PEI-mixed B, F:ZnO nanocrystals cause the JSC to increase from 21.85 to 22.75 mA/cm2, the FF to rise from 70 to 75%, and the VOC to increase from 990 to 1000 mV, as outlined in Table and presented in Figures S8 and 5c, yielding a PCEavg of 17.06% (PCEmax = 20.74%). Thus, the PEI-mixed ETLs yielded substantially better devices than PEI-coated ETL-based devices (PCEavg = 15.14%; PCEmax = 16.2%) and is believed to benefit the devices by reducing the traps inside the oxide layer and having better physical contact between the polymer and oxide layer.[48] Furthermore, a single layer existed between the perovskite layer and the PEI-mixed ZnO composite nanolayer interface; on the other hand, two interfaces are found between the PEI-coated ZnO bilayer ETL and perovskite layer, which will reduce the ability of the carrier to transport vertically due to surface traps. The best-performing cell revealed a PCEmax of 20.74%, the best efficiency ever stated for perovskite devices with ZnO ETLs. The PEI-mixed B, F:ZnO ETLs commit the improved absorption of perovskite absorber compared to the ETLs discussed above. This hypothesis is showed by the ultraviolet–visible absorption spectra presented in Figure S9 (Supporting Information). The perovskite layer based on PEI-mixed B, F:ZnO ETLs exhibits outstanding light absorption, committing the performance improvements of the perovskite devices. This is also proved by the EQE results exhibited in Figure S10 (Supporting Information), which are essentially in accordance with the trend of JSC in the J–V curves, as displayed in Figure b. The integrated JSC is 23.1 mA/cm2 (Figure S10, Supporting Information), which is satisfactory since the measured JSC (23.7 mA/cm2) from J–V curves is achieved in the reverse voltage scanning. In Figure d, the bar chart of the averaged PCE in the reverse and forward scanning directions for 50 separate cells based on PEI-mixed ZnO and B, F:ZnO nanolayers demonstrates good reproducibility with >85% of cells producing PCE > 10.76 and >17.06%, respectively.

Table 3

Device Parameters (in Different Scanning Directions) with PEI-Mixed Pure ZnO and B, F:ZnO Nanolayers Having Optimal ETL Thickness of 160 nm

film type	scan mode	JSC (mA/cm2)	VOC (mV)	FF (%)	PCEavg (%)	PCEmax (%)
pure ZnO	reverse	17.5	923	67	10.82 ± 0.10	13.04
	forward	17.3	923	67	10.69 ± 0.14	12.78
	average	17.4	923	67	10.76 ± 0.12	12.91
B, F:ZnO	reverse	23.0	1000	75	17.25 ± 0.12	20.83
	forward	22.50	1000	75	16.80 ± 0.14	20.65
	average	22.75	1000	75	17.06 ± 0.13	20.74

Potential Key Summary Plots for Cells with PEI-Coated and PEI-Mixed ETLs

The champion devices with PEI-coated B, F:ZnO and PEI-mixed B, F:ZnO nanolayers also reveal the hysteresis-free current–voltage curves (Figure a,c), and the PCEmax (averaged) values calculated from those curves are in good agreement with that achieved from maximum power point tracking measurements (i.e., 0.835 V) at 1 sun illumination, as plotted in Figure a. It is also examined that the devices based on PEI-mixed B, F:ZnO nanolayers have a constant PCEmax of 20.74%, whereas the PCE of PEI-coated devices (16.2%) deteriorates a little with light soaking. The statistics of maximum and average PCEs obtained for the perovskite devices based on various ETLs (discussed in this study) are summarized in Figure b,c. We constantly demonstrate that multidoped nanolayer with highest conductivity and optical transmittance, high carrier concentration, and larger band-gap energy caused a maximum shift in the conduction band, surpassing pure ZnO nanolayers in all schemes and eventually producing devices with better PCEavg (2–3 points higher in each case). This highlights the potential benefits of multidopants on the electrical, optical, and morphological characteristics of electrosprayed deposited ZnO at low temperature. Improvements in device performance are further extended by applying conformal coating of PEI on the surface of ETLs (will reduce the work function further), resulting in highly efficient cells. The state-of-the-art devices (with PCEmax of 20.74%) are produced using a polymer-mixed oxide nanolayer thanks to the decrease in the trap sites of the ETLs, which would surely minimize the chances of trap-assisted carrier interfacial recombination[48] and eventually produce hysteresis-free, stable, and most efficient ZnO-based perovskite solar cells reported to date. Although the devices based on PEI-mixed nanolayers showed reduced hysteresis with almost no difference in the forward and backward scans, cells with lower PCEs presented more hysteresis for other ETL schemes. Interestingly, hysteresis index[26] (used to evaluate the hysteresis response) varied significantly for the different ETL schemes (discuss in this study), as plotted in Figure d, and can be calculated from the following equationwhere Jscan(VOC/2) is the current density at VOC/2 for the forward scan and Jscan(VOC/2) is the current density at VOC/2 for the backward scan. Hysteric devices demonstrate higher numbers of hysteresis index, whereas the lowest numbers represent the hysteresis-free devices. Particularly, in our case, the trend of hysteresis revealed that it is consistently lower for multidoped ZnO nanolayers with and without PEI coating on their surface and is further reduced to a minimum number when a polymer-mixed oxide nanolayer is used. It is important to mention that the decrease of hysteresis depends mainly on interfacial engineering of the perovskite devices by surface modification of ETLs surface.[26] The lower hysteresis for devices based on multidoped nanolayers is mainly due to their better electrical, optical, and surface characteristics and the proven lower work function facilitating electron extraction and transfer through the ETL.[26] A dipole layer (created on the ETLs surface by coating a thin layer of PEI) will further reduce the work function, resulting in the elimination of hysteresis significantly. Remarkably, the perovskite devices are also almost hysteresis-free by using a polymer-mixed nanolayer by reducing the traps in the oxide layer, which will definitely minimize trap-assisted interfacial recombination of carriers. Additionally, the hysteresis strongly relies on sweep directions of applied bias, voltage sweep rates, and light soaking.[26] The devices based on PEI-mixed B, F:ZnO nanolayer demonstrate nonhysteric behavior compared to PEI-coated B, F:ZnO nanolayer (Figures S6 and S8), with negligible variation detected in the JSC when changing the voltage sweep rate or direction, as presented in Figure S11. The perovskite devices with PEI-coated and PEI-mixed B, F:ZnO nanolayers further undergo long-term stability test by light soaking in ambient environment (relative humidity = 40–45%) and storing the devices at 45 °C for 15 days continuously. In general, the perovskite solar cells are extremely susceptible to moisture and eventually rapidly deteriorate in humid environment because of their ionic nature.[49] The devices based on PEI-mixed nanolayers yielded marginally more stable data in terms of photovoltaic parameters (VOC, JSC, FF, and PCEavg) compared to devices based on PEI-coated nanolayers (decreased by only ∼6–8%) over a 15 day period (Figures S12 and S13, Supporting Information). Thus, the newly proposed polymer-mixed ETLs serve as a key addition and a scalable technique toward industrialization of perovskite solar cells.

Figure 6

(a) Steady-state PCEmax as a function of time for PEI-coated and PEI-mixed B, F:ZnO nanolayers, obtained by applying 835 mV bias voltage, which is identical to the maximum power output potential. Bar graphs of (b) PCEavg and (c) PCEmax summarizing the performance data for various types of ZnO and B, F:ZnO nanolayers and (d) hysteresis index for different types of ZnO and B, F:ZnO nanolayers.

Conclusions

In summary, an atmospheric pressure-based electrospray method is utilized for the formation of electron-rich multidoped ZnO nanolayers as ETLs for high-efficiency perovskite solar cells. The position of conduction band edge is continuously raised using various combinations of dopants, which causes a 0.22 eV shift in the conduction band compared to pure ZnO films, which will eventually improve the VOC, suppress the recombination, and enhance the faster transfer of electrons. The B, F:ZnO nanolayer with highest conductivity and optical transmittance, high carrier density, and larger band-gap energy and also benefited by the maximum shift in the conduction band demonstrates the highest PCEavg of 13.54% and PCEmax of 15.09%, which are much higher than those of pure ZnO film-based devices. Further improvements in the performance by reducing the work function of the oxide layer is achieved by coating the nanolayer surface with a thin layer of PEI, which results in remarkable highly efficient and hysteresis-free devices with a PCEmax of 16.20%. Eventually, a polymer-mixed ETL demonstrates a remarkable PCEmax of 20.74% by decreasing the bulk traps inside the oxide layer, which probably reduces the chances of trap-assisted surface recombination of charges and, subsequently, improves the device performance. The current work has notably confirmed the optimization of voltage by adjusting the energetic structure of the charge-removing material together with the perovskite absorber, which will be one of the best reasonable approaches to boost the device efficiency.

Experimental Section

Electrosprayed Deposition of Pure and Multidoped ZnO Mesoscopic Nanolayers

Pure and doped ZnO (such as B:ZnO, Ta:ZnO, B, F:ZnO, and Ta, N:ZnO) mesoscopic nanolayers were coated onto the surface of ZnO-blocking layer (deposited on indium tin oxide (ITO) substrates) using atmospheric pressure-based electrospraying method, as reported previously.[37−39] Precursors such as zinc acetate dihydrate Zn(O2CCH3)2(H2O)2, boric acid (B(OH)3), tantalum (v) chloride (TaCl5), ammonium fluoride (NH4F), and ammonium acetate (C2H3O2NH4) were used as a source of zinc, boron, tantalum, fluorine, and nitrogen, respectively. The applied high voltages, precursor solution flow rates, and nozzle-to-substrate distances were maintained at 5.6 kV, 0.003 mL/min, and 4.0 cm, respectively. The precursor solution was sprayed for 1 h, while the hot plate temperature was retained at 150 °C (Table S1), and finally, the nanolayers were annealed at 410 °C in air atmosphere for 1 h. To synthesize the composite ETLs, the ZnO precursor was mixed using a fixed amount of PEI (7 wt %) and continuously stirred for 12 h, and the resultant solution was electrosprayed using the same conditions mentioned above.

Fabrication of Cells Using Pure and Doped ZnO Mesoscopic Nanolayers

First, a dense ZnO-blocking layer was spin-coated onto well-cleaned ITO substrates, followed by sintering at 500 °C for 30 min.[22] After that, mesoporous pure and multidoped ZnO nanolayers were formed by electrospraying method, as explained above. A thin layer of PEI was formed over the surface of ZnO ETLs by spin-coating a PEI solution (prepared in deionized water with a concentration of 0.01 wt %) at 4500 rpm for 2 min and subsequent annealing at 150 °C for 5 min. After deposition, the perovskite absorber layer was formed over the surface of ZnO ETLs via the one-step spin-coating route, as explained below. The CH3NH3PbI(3–x)Cl solution (CH3NH3I/PbCl2 = 3:1, in N,N-dimethylformamide) was deposited by spin-coating at 2500 rpm for 30 s and subsequent drying at 95 °C for 1 h. Afterward, 30 μL of a hole-transport material was dropped on the substrate and spin-coated at 3000 rpm for 30 s. Finally, 80 nm thick silver as a top electrode layer was formed by thermal evaporation method.

Characterization

The plane-view surface images of ZnO ETLs and cross-sectional view of complete devices were obtained using a scanning electron microscope (JSM-7600F, JEOL). The surface roughness was obtained by atomic force microscopy (AFM, SPA 400). X-ray photoelectron spectroscopy (XPS) was conducted to confirm the elemental composition of ETLs. The optical transmittance of the ETLs was measured using a UV–visible spectrometer (UV-3101PC). Electrical resistivity data were collected using four-point probe equipment (CMT-SR1000N). Hall effect measurement instrument (HMS-3000) was used to obtain the Hall mobility and carrier concentration data for various ETLs. The Brunauer–Emmett–Teller (BET) gas adsorption measurement route was used to measure the surface area of the powders using Quantochrome NOVA 1000 (Boynton Beach, FL). The current density vs voltage (J–V) plots were collected using a Keithley digital source meter by adjusting the intensity to 1 sun (100 mW/cm2). Both the reverse and forward scans were obtained at a rate of 10 mV/s with a delay time of 5 s. The active area of 0.25 cm2 was set using a metal mask aperture.