Electrochemically Deposited CZTSSe Thin Films for Monolithic Perovskite Tandem Solar Cells with Efficiencies Over 17%

In spite of the high potential economic feasibility of the tandem solar cells consisting of the halide perovskite and the kesterite Cu2ZnSn(S,Se)4 (CZTSSe), they have rarely been demonstrated due to the difficulty in implementing solution‐processed perovskite top cell on the rough surface of the bottom cells. Here, we firstly demonstrate an efficient monolithic two‐terminal perovskite/CZTSSe tandem solar cell by significantly reducing the surface roughness of the electrochemically deposited CZTSSe bottom cell. The surface roughness (Rrms) of the CZTSSe thin film could be reduced from 424 to 86 nm by using the potentiostatic mode rather than using the conventional galvanostatic mode, which can be further reduced to 22 nm after the subsequent ion‐milling process. The perovskite top cell with a bandgap of 1.65 eV could be prepared using a solution process on the flattened CZTSSe bottom cell, resulting in the efficient perovskite/CZTSSe tandem solar cells. After the current matching between two subcells involving the thickness control of the perovskite layer, the best performing tandem device exhibited a high conversion efficiency of 17.5% without the hysteresis effect.


Introduction
[3][4][5][6] They also have ideal electrical and optical properties for solar cell applications, such as intrinsic p-type conductivity, tunable bandgap energy from S/Se ratio, and high absorption coefficient due to the direct transition bandgap. [5,6][11][12][13] One feasible way to drastically increase the efficiency of the CZTSSe solar cell is to make tandem solar cells with wider bandgap materials like halide perovskites.
[16][17][18][19][20][21] The perovskite top cells have so far been utilized mainly for the Si tandems, and the highest certified conversion efficiency of the perovskite/Si tandem solar cells has rapidly increased to 29.8%. [9,22,23]][26] Among them, perovskite/CIGS tandem solar cells exhibit high conversion efficiency of 24.2%, [9,27] but expensive rare elements like In and Ga in the CIGS solar cells might be a hurdle for the successful commercialization.Therefore, the CZTSSe solar cell consisting of earth-abundant, cheap, and eco-friendly elements must be a good candidate for bottom cell of the perovskite tandems. [15]In addition, the economic feasibility of the CZTSSe solar cells can be further enhanced by preparing them via cheap and commercial solution processes like the electrochemical deposition.However, only a few studies have been reported on the perovskite/ CZTSSe tandem solar cells, and the highest reported conversion efficiency is only 4.6%. [26]The biggest issue of fabricating the efficient perovskite/CZTSSe tandem is the rough surface of the CZTSSe bottom cell.Given that the thickness of the thin layers in the perovskite solar cell (e.g., charge-transporting layers) is approximately tens of nanometers, it is very important to reduce the surface roughness of the bottom cell to avoid the short failure in the top cell observed from our preliminary experiments (Figure S1, Supporting Information).
In this study, we demonstrated a low-cost tandem solar cell consisting of a solution-processed perovskite top cell and an electrochemically In spite of the high potential economic feasibility of the tandem solar cells consisting of the halide perovskite and the kesterite Cu 2 ZnSn(S,Se) 4 (CZTSSe), they have rarely been demonstrated due to the difficulty in implementing solution-processed perovskite top cell on the rough surface of the bottom cells.Here, we firstly demonstrate an efficient monolithic twoterminal perovskite/CZTSSe tandem solar cell by significantly reducing the surface roughness of the electrochemically deposited CZTSSe bottom cell.The surface roughness (R rms ) of the CZTSSe thin film could be reduced from 424 to 86 nm by using the potentiostatic mode rather than using the conventional galvanostatic mode, which can be further reduced to 22 nm after the subsequent ion-milling process.The perovskite top cell with a bandgap of 1.65 eV could be prepared using a solution process on the flattened CZTSSe bottom cell, resulting in the efficient perovskite/CZTSSe tandem solar cells.After the current matching between two subcells involving the thickness control of the perovskite layer, the best performing tandem device exhibited a high conversion efficiency of 17.5% without the hysteresis effect.
deposited CZTSSe bottom cell by effectively reducing the surface roughness of the bottom cell.We proposed two strategies for flattening the surface of the CZTSSe bottom cell: 1) a bottom-up approach using the potentiostatic electrodeposition; and 2) a top-down approach utilizing the ion-milling process.The root-mean-square roughness (R rms ) values of the metal alloy precursor and corresponding CZTSSe films prepared by the potentiostatic mode are significantly decreased from 198 to 42 nm and from 424 to 86 nm, respectively, compared with the conventional galvanostatic mode.The surface roughness (R rms ) of the bottom cell prepared via the potentiostatic mode could be further reduced to 22 nm by the subsequent ion-milling of the indium tin oxide (ITO) layer, which was uniform enough to monolithically prepare functioning perovskite top cells.The subcell current densities of the perovskite/ CZTSSe tandem solar cell could be matched to a maximum value by controlling the thickness of the perovskite layer, resulting in the recordhigh conversion efficiency of 17.5%.

Results and Discussion
Figure 1a-d show field-emission scanning electron microscopy (FE-SEM) and atomic force microscopy (AFM) images of the CZT alloy films grown via galvanostatic or potentiostatic modes and those of the corresponding CZTSSe thin films after annealing under the S/Se atmosphere.As shown in Figure 1a, the CZT precursor film prepared via the galvanostatic mode features a lot of abnormally grown spherical bumps with a diameter of approximately 1 lm, which can be ascribed to the Sn out-growth. [4]Such a rough surface of the precursor film is likely to lead to a highly rough CZTSSe thin film after the annealing process (Figure 1b).In the previous study, we found out the origin of the irregular growth during the conventional galvanostatic coelectrodeposition, and have demonstrated that the surface of the precursor and CZTSSe thin films can become smoother and more uniform by introducing an initial nucleation stage involving a higher deposition current prior to the steady-state deposition. [4]However, the root-mean-square roughness (R rms ) of 150 nm that has been reduced by the two-step electrodeposition is still too high for applying to a tandem solar cell.On the other hand, the CZT precursor film deposited via the potentiostatic mode is consisting of uniform and smaller spherical features with diameters of several hundred nanometers (Figure 1c).The significantly reduced surface roughness of the precursor film would lead to the smooth surface of the CZTSSe thin film after annealing, as can be seen in Figure 1d.The AFM analysis confirmed that both precursor and CZTSSe films grown via the potentiostatic mode exhibit significantly reduced R rms values compared with the galvanostatic mode (from 198 to 42 nm for precursors, from 424 to 86 nm for CZTSSe films).The different surface roughness depending on the deposition mode is attributed to the different deposition mechanisms in the early stage.The standard reduction potentials of metal cations are very different (Zn 2+ : À0.7626 V, Sn 2+ : À0.1375 V, and Cu 2+ : +0.340 V vs normal hydrogen electrode (NHE)). [28]During the galvanostatic deposition, Sn 2+ and Cu 2+ which have higher (or less negative) reduction potentials than Zn 2+ are prone to be preferentially reduced in the early stage of the deposition. [4]The deposition potential shows distinct steps at relatively low potentials of around À0.75 V (vs Ag/AgCl) before reaching the steady-state deposition potential of À1.2 V (vs Ag/ AgCl) due to the different reduction potentials of metal cations (Figure S2A, Supporting Information).The sparse nucleation under relatively small overpotentials in the early state of the deposition is known to induce rough surface morphology of the CZT precursor film prepared via the galvanostatic mode. [2,4]On the other hand, Zn 2+ , Sn 2+ , and Cu 2+ are simultaneously reduced during the potentiostatic deposition due to the fixed overpotentials for each metal cation determined by the applied voltage of À1.24 V (Figure S2B, Supporting Information).The applied voltage in the early stage of the deposition is higher than the galvanostatic mode, providing large enough overpotentials for all cations.The large overpotantials for all cations also lead to the uniform nucleation and the formation of smooth CZT precursor films.The Xray diffraction (XRD) revealed that the asdeposited precursor films are consisting of Cu 5 Zn 8 (I4̅ 3m), b-Sn (I4 1 /amd), and Cu 6 Sn 5 (C2/c) phases regardless of the deposition modes, but the thin film from the potentiostatic mode shows smaller b-Sn peaks and larger Cu 6 Sn 5 peaks (Figure 1e). Figure 1f shows the current density-voltage (J-V) curves of the single-junction CZTSSe thin film solar cells prepared by both methods, and their solar cell parameters are summarized in Table S1, Supporting Information.The solar cell with the potentiostatically deposited CZTSSe thin film exhibited slightly increased J SC (by 1.3 mA cm À2 ) and V OC (by 6 mV), but the main parameter which led to a higher PCE was the increased fill factor (by 8.3%).The lower fill factor (FF) of the device prepared via the galvanostatic method is resulting from the increased shunt conductance and series resistance which can be apparently seen from the J-V curve.The rough surface of the CZTSSe thin film can induce poor and nonuniform formation of CdS layers, which leads to the higher series resistance in thick regions and the higher shunt conductance in thin regions.In addition, the macroscopic rough surface may cause the agglomeration of the CdS clusters on the CZTSSe thin film, which deteriorates the p-n junction quality. [4,6]On the other hand, the potentiostatically deposited CZTSSe thin film with a smooth surface is favorable to form uniform layers during the subsequent deposition of CdS and TCO layers, leading to the reduced shunt paths and superior diode property.The J SC values calculated from the external quantum efficiency (EQE) curves (Figure 1g) are qualitatively consistent with the J-V measurement, and the optical bandgaps calculated from EQE (Figure S3, Supporting Information) were 1.14 eV for the galvanostatic mode and 1.12 eV for the potentiostatic mode.The small difference in the bandgap is due to the slightly different S-to-Se ratio in CZTSSe films (Table S2, Supporting Information).
The reliable fabrication of the perovskite top cell onto the CZTSSe bottom cell via a solution process requires an extremely flat surface of bottom cells.Although one can significantly reduce the surface roughness of the CZTSSe bottom cells by using the potentiostatic deposition technique, the bottom cells were still not flat enough to form uniform perovskite top cells.Therefore, we adopted the ion-milling process for further flattening the surface of the bottom cell (Figure 2a).Given the multilayer structure of the CZTSSe bottom cell consisting of glass/Mo/ CZTSSe/CdS/i-ZnO/ITO, we intentionally deposited a thicker ITO recombination layer and then flattened its top surface by ion-milling.The etched thickness of the ITO film was controlled by the ion-milling time.The thickness of the etched ITO layer was linearly dependent on the etching time with an average etching rate of 4.64 nm min À1 as shown in Figure S4, Supporting Information.After ion-milling, the ITO films exhibited slightly increased average transmittance than the unetched ITO film regardless of the etching time without showing deterioration of the electrical conductivity (Table S3, Supporting Information).Thinner ITO films are beneficial to light harvesting by the bottom cell, but excess etching time (i.e., 40 and 60 min) significantly decreased the efficiency of tandem devices by damaging the p-n junction area underneath the ITO recombination layer (Figure S5, Supporting Information).Figure 2b-d display top-view and cross-sectional FE-SEM images of the samples with milling times of 0, 10, and 20 min, respectively.The top surface of the ITO layer without ion-milling is quite rough because of the rough surface of the CZTSSe layer as well as the aggregation of the CdS clusters during the chemical bath deposition process, but the top surface of the ITO layer becomes much smoother with the increasing milling time.The AFM analysis shown in Figure 2e-g and Figure S6, Supporting Information was consistent with SEM results, and the R rms of the ITO surface decreases to 22.39 nm after etching for 20 min (cf.107.67 nm for the unetched sample).As a result, the optimum etching time of 20 min was chosen for the fabrication of tandem device.We investigated the effect of ion-milling on photovoltaic properties of the CZTSSe solar cell.The statistical solar cell parameters with and without ion-milling for 20 min shown in Figure S7, Supporting Information reveal that the effect of the etching process is insignificant within a range of the normal experimental error.
Figure 3a depicts the structure of the transparent perovskite top cell and the CZTSSe bottom cell.Prior to integrating the perovskite cell onto the bottom cell, we investigated the optimum thickness of the perovskite layer to maximize the matching current density between two subcells.The perovskite top cell is consisting of glass/ITO/PTAA/perovskite/ C 60 /ITO/Ag, where the composition of the perovskite layer is (FA 0.7 MA 0.15 Cs 0.15 )Pb(I 0.85 Br 0.15 ) 3 .The film thickness was controlled by varying the concentration of the precursor solutions, and the crosssectional FE-SEM images show that the thickness of the perovskite layer is 165, 360, and 480 nm for the precursor concentration of 0.625, 0.94, and 1.25 M, respectively (Figure S8, Supporting Information).The absorbance of the perovskite thin films increased with the increasing thickness, whereas the same optical bandgap of 1.65 eV was observed regardless of the thickness (Figure S9, Supporting Information).Given that the bandgap of the CZTSSe bottom cell is 1.12 eV, the bandgap of the perovskite top cell is suitable for the current matching between the subcells.The J-V curves of the perovskite single junction solar cells with various thicknesses are presented in Figure 3b and the corresponding solar cell parameters are summarized in Table S4, Supporting Information.The J SC increases from 18.99 to 21.56 mA cm À2 as the film thickness increases from 165 to 480 nm due to the increased light absorption.The statistical information shown in Figure S10, Supporting Information also exhibit the same trend as the J-V curves.The EQE spectra in Figure 3c show that the photoresponses particularly at the wavelengths longer than 550 nm decrease with decreasing film thickness, whereas the EQE at shorter wavelengths barely changes with the thickness.The thickness dependency of the photoresponses at longer wavelengths is attributed to the low absorption coefficient of the perovskite absorber near the band edge. [29]The J SC values calculated from the EQE curves are 21.13, 20.55, and 18.24 mA cm À2 for the thickness of 480, 360, and 165 nm, respectively, which are consistent with those from the J-V measurements.The J SC of the CZTSSe bottom cells can be estimated from the EQE curves of the bottom cells filtered by the transparent top cells with different thicknesses, as can be seen in Figure 3d.The EQE curves of the filtered CZTSSe bottom cells are very similar to the transmittance spectra of the transparent top cells (Figure S11, Supporting Information), indicating that the photons transmitted by the top cell directly influence the J SC of the CZTSSe bottom cells.The J SC of the bottom cell calculated from the filtered EQE shown in Figure 3d increases from 14.78 to 18.34 mA cm À2 with decreasing thickness of the perovskite layer, which can be mainly attributed to the photoresponses in the wavelength range of 450-750 nm.Given that the J SC of the unfiltered CZTSSe solar cell estimated from the EQE curve is approximately 34 mA cm À2 , we assumed that the optimum matching current density would be around 17 mA cm À2 .Then, we compared the J SC values of the top cells and the filtered bottom cells depending on the thickness of the perovskite layer as can be seen from Figure 3e, where a current drop of 20% was assumed from opaque to transparent cells.As a result, the optimum thickness of the perovskite layer was determined to be 400 nm.The transparent perovskite top cell with the optimum thickness of 400 nm exhibited a J SC of 16.45 mA cm À2 , V OC of 1.102 V, and FF of 0.76, leading a PCE of 13.79% (Figure 3f).
Figure 4a shows a cross-sectional SEM image of the two-terminal monolithic perovskite/CZTSSe tandem solar cell, which is prepared by fabricating the top cell with the 400-nm-thick perovskite layer on top of the CZTSSe bottom cell flattened by ionmilling for 20 min.To reduce the reflection of the incident light, a LiF layer was deposited on the top electrode as an anti-reflection (AR) layer.Our preliminary experiments revealed that the additional ion-milling process is crucial for the fabrication of functioning tandem devices even though the bottom cells were prepared via the potentiostatic deposition (Figure S12 and Table S5, Supporting Information).The tandem cells with ion-milling for 10 min or less exhibited resistor-like J-V curves with very low FF, which is mainly ascribed to the undesirable electrical contacts between the recombination layer and the top ITO electrode like the case of Figure S1, Supporting Information.The tandem cell with ion-milling for 20 min exhibited the highest PCE of 17.5% with J SC , V OC , and FF of 17.67 mA cm À2 , 1.460 V, and 0.68, respectively.These results support our claim that the roughness of the bottom cell is the most critical factor to achieve a highly efficient device.Figure 4b compares the J-V curves of the single-junction transparent perovskite, CZTSSe, filtered CZTSSe, and the perovskite/CZTSSe tandem solar cells with the optimized conditions, and their solar cell parameters are summarized in Table 1.The J SC of filtered CZTSSe and transparent perovskite solar cells were well matched, leading to the tandem J SC of 17.67 mA cm À2 .It should be noted that the slightly higher tandem J SC compared with the matched subcell J SC is obtained thanks to the AR  coating only on the tandem device.No difference in the tandem J-V curves depending on the direction of the voltage scan was observed, indicating that the hysteresis effect is negligible.The tandem cell also shows good operational stability by maintaining the stable steady-state power output (SPO) with a constant voltage close to the maximum power voltage (Figure 4c).To the best of our knowledge, the efficiency of 17.5% achieved in this study is the highest among monolithic twoterminal perovskite/CZTSSe tandem solar cells reported till date.Figure 4d presents the EQE spectra of the subcells and the 1-R spectrum of the two-terminal perovskite/CZTSSe tandem cells.The integrated J SC of the top and bottom cells are 17.82 and 17.54 mA cm À2 , respectively, indicating that the tandem photocurrent is slightly limited by the bottom cell.The expected tandem J SC of 17.54 mA cm À2 is very close to the J SC from the J-V curve (17.67 mA cm À2 ) with a difference <1%.The region between the 1-R and EQE spectra corresponds to the parasitic absorption loss or the charge recombination loss.The absorption loss in the top cell is mainly due to the absorption by the C 60 electrontransporting layer, whereas the absorption loss in the bottom cell is mainly due to the relatively thick ITO recombination layer.The poor collection efficiency of the charge carriers generated by the low energy photons near the absorption band edge (i.e., 1000-1200 nm) is another origin of the low EQE of the bottom cell.Therefore, the performances of the perovskite/CZTSSe tandem solar cells can be further improved by optimizing the interfacial layer thickness and the defect passivation.

Conclusion
We firstly demonstrated efficient two-terminal tandem solar cells consisting of low-cost solution-processed perovskite top cell and electrochemically deposited CZTSSe thin film bottom cell.The surface roughness of the CZTSSe bottom cell has been found to be a mostly critical factor for the successful preparation of the monolithic tandem solar cell.The surface roughness of the CZTSSe thin film could be significantly reduced by using the potentiostatic method rather than the conventional galvanostatic method during the electrochemical deposition, which was attributed to the constant and large overpotential applied to each metal cation particularly in the early stage of the deposition.The surface roughness could be further decreased to a level suitable for the successful fabrication of the tandem devices (e.g., R rms of 22 nm after ion-milling for 20 min) by etching the top surface of the ITO layer with the aid of ion-milling.The matching current density between the subcells could be maximized after optimizing the thickness of the perovskite top cell with a bandgap energy of 1.65 eV, eventually leading to the record efficiency of 17.5%.The champion efficiency is much higher than the highest reported efficiency of the monolithic perovskite/CZTSSe tandem (4.6%) and the CZTSSe single-junction solar cell (13.0%).

Experimental Section
Full details of experimental procedures are provided in the Supporting Information.

Figure 2 .
Figure 2. Effects of ion-milling on the CZTSSe bottom cells.a) A schematic illustration of the ionmilling process and photos of thin films before and after ion-milling.b-d) Top-view and crosssectional-view (insets) SEM images of the CZTSSe/CdS/ITO layers with the ion-milling time of 0, 10, and 20 min.e-g) AFM mapping images and corresponding roughness histogram of the samples shown in b-d.

Figure 3 .
Figure 3. Thickness control of perovskite solar cells.a) A schematic illustration of the perovskite top cell and CZTSSe thin film bottom cell.b, c) J-V curves and EQE spectra of opaque perovskite solar cells with different thickness of the perovskite layer.d) EQE spectra of the single-junction CZTSSe cells with and without filtering by perovskite films with different thickness.e) Expected J SC values of the subcells with different thickness of the perovskite layer.f) A J-V curve of the transparent perovskite solar cell with the optimum thickness of 400 nm.

Figure 4 .
Figure 4. Performance of the monolithic perovskite/CZTSSe tandem cells.a) A cross-sectional SEM image of the two-terminal monolithic perovskite/ CZTSSe tandem solar cell.b) J-V curves of the unfiltered and filtered CZTSSe, transparent perovskite, and tandem solar cells, where the AR-layer was coated only on the tandem device.c) Steady-state power output (SPO) of the tandem cells.d) EQE spectra of the subcells and the 1-R curve of the tandem cell.