Influence of graphene dispersion on the photovoltaic performance of tandem solar cells with four-terminal perovskite–polycrystalline silicon configuration

Silicon’s prominence in photovoltaic technology stems from its abundance and safety. While Si-based solar cells demonstrate high energy conversion efficiency and long-term stability, they encounter challenges such as high costs, intricate fabrication processes, and suboptimal efficiency. To address these issues, researchers have developed tandem solar cells that combine silicon with perovskite cells. This research specifically investigates the use of the spin coating technique with graphene dispersion solutions to deposit graphene layers in perovskite solar cells (PSCs), providing a flexible and cost-effective alternative to conventional methods. By employing graphene as a protective sealant for the perovskite interlayer to prevent degradation, the study aims to enhance the overall performance and stability of tandem solar cells. Graphene was applied onto the hole transport layer at varying concentrations (1, 5, and 10 mg ml−1) in isopropanol. Notably, the introduction of graphene resulted in decreased power conversion efficiencies (PCEs) in PSC top cells over 60 h, with efficiency reductions of 43%, 24%, and 17% for different concentrations. Importantly, these efficiency declines were significantly lower compared to cells lacking a graphene layer, which experienced a sharp 93% decrease. This investigation underscores the critical role of graphene layers in improving the stability of PSC top cells while maintaining compatibility with the stability of poly-Si bottom cells.


Introduction
Silicon, an abundant and nontoxic material, has become the most favored material in photovoltaic (PV) technology.Si-based solar cells, such as monocrystalline, multi-crystalline, and amorphous Si solar cells, are remarkable due to their high-energy conversion efficiency that can reach up to 26.6% of the conversion efficiency of research cells [1][2][3] and long-term stability that extends until 2050 [4].However, Si-based solar cells have several drawbacks, such as demanding costing, complex fabrication, and retarded efficiency that is far below the optimum value.Facile tandem solar cells have recently been used to solve the limited efficiency of Si solar cells.Organic-inorganic lead halide perovskite solar cells (PSCs) can be used as top cells in the perovskite-Si tandem solar cells.Tandem cells are dependent on several characteristics of PSCs, including strong absorption coefficients, long diffusion lengths, band-gap tunability (1.5-1.8 eV), and steep absorption edges [5][6][7].
However, perovskite top cells face stability issues when exposed to air humidity, oxygen, ultraviolet (UV) light, thermal stress, light soaking, electric fields and many other factors [8][9][10][11], which then leads to rapid degradation.The degradation rate of perovskite top cells can affect the overall tandem module degradation and consequently reduce the module's lifetime energy and cost-effectiveness.Several methods, such as encapsulation [12,13], interfacial engineering [14][15][16][17] and chemical passivation [18,19], are used to improve the long-term stability of sensitive perovskite top cells.
Graphene plays a crucial role in silicon-perovskite tandem solar cells, serving as a versatile material with unique properties that enhance device performance.It can function as a transparent electrode, chargetransporting layer, interfacial layer, and additive, addressing various challenges in these solar cells [20].The incorporation of graphene and its derivatives in perovskite solar cells has been shown to improve their power conversion efficiency and stability [21].Furthermore, graphene can act as a protection layer, preventing metalinduced degradation and halide diffusion in flexible and transparent metal electrode-based perovskite solar cells [22].These findings highlight the significant potential of graphene in enhancing the efficiency and stability of silicon-perovskite tandem solar cells.
Typically, the chemical vapor deposition (CVD) method is used to form the graphene layer.However, this method has limitations for process reproducibility, mass production and post-synthesis treatment (transfer and deposition on the desired substrate) [23].The application of graphene dispersion solution in PSC research is an alternative to overcome the problem.Therefore, in this research, the flexible and low-cost deposition method that is the spin coating method in the formation of a graphene layer through a graphene dispersion solution makes it the best solution for the facile process.
In this study, the no-constraint orientation of the tandem mechanism, which involves four mechanically stacked terminals [24][25][26][27], was used to tandem PSCs and Si solar cells.This study also focused on graphene, which possesses high transparency, high conductivity and high carrier mobility [28][29][30][31].Graphene was utilised as a sealing mechanism to stabilise and passivate the PSC interlayer from coming into contact with oxygen and moisture, which degrade the overall performance of PSCs.The spin coating technique, which is a simple and low-cost method, was performed to coat the graphene film with graphene dispersions in isopropanol to fabricate the film layers of the PSC devices.The concentration of the graphene dispersion was varied, and Raman analysis was performed to investigate the effect of the intensity of such dispersion.The correlations between the Raman analysis results and the PV performances of the four-terminal perovskite-Si tandem solar cells were analysed.

Experimental section
The first step for building the tandem device was the fabrication of the perovskite top cells.Fluorine-doped tin oxide (FTO) glass (15 Ωsq −1 , Solaronix) was used as a substrate.Compact TiO 2 blocking layer (bl-TiO 2 ) was coated on the precleaned FTO through spin coating (3000 rpm, 30 s) by using titanium isopropoxide solution (BL, Dyesol) in ethanol solvent (1:1 weight ratio) and then annealed for 30 min at 450 °C.Mesoporous TiO 2 layer (mp-TiO 2 ) was fabricated onto the bl-TiO 2 /FTO substrate through spin coating (4000 rpm, 20 s) and annealed for 30 min at 450 °C.The spin coating solution was prepared by diluting TiO 2 paste (18 NR-T, Dyesol) with ethanol (1:9 weight ratio), and the perovskite layer (CH 3 NH 3 PbI 3 ) was created using a two-step process.Then, 1.2 M of lead dioxide (PbI 2 ; 99%, Sigma Aldrich) were dissolved in N, N dimethylformamide solvent (99.8%,R&M Chemicals) and 100 μl of 4-tertbutylpyridine (TBP; 96%, Sigma Aldrich) as the first solution.The PbI 2 precursor solution (50 μl) was spin coated on mp-TiO 2 /bl-TiO 2 /FTO substrates at 3000 rpm for 60 s and heated at 70 °C for 30 min.Subsequently, 30 mg methylammonium iodide (MAI, Sigma Aldrich) were dissolved in 1 ml of isopropanol as the second solution.The MAI precursor solution (100 μl) was then deposited at 3000 rpm for 20 s on the as-prepared PbI 2 solution and heated at 95 °C for 30 min to produce a CH 3 NH 3 PbI 3 layer.Both solutions were heated at 60 °C overnight under magnetic stirring before using.The 2,2,7,7-tetrakis-(N,N-di-p-methoxyphenylamine)-9,9-spirobifluorene (spiro-OMeTAD) solution was coated as the hole transport layer (HTL).This layer was prepared by mixing 72.3 mg of spiro-OMeTAD (99%, Sigma Aldrich) with 1 ml of chlorobenzene containing 17.5 μl of lithium-bis(tri-fluoromethanesulfonyl)-imide (Li-TFSI) solution (520 mg of Li-TFSI (99%, Sigma Aldrich) in 1 ml of acetonitrile) and 28.8 μl of TBP and deposited through spin coating (4000 rpm, 20 s).A single-layer graphene dispersion in water (1 mg ml −1 , ACS Material) was heated on a hotplate at 60 °C until it dried and became graphene flakes.The treated graphene dispersion was then dissolved in isopropanol and sonicated in ultrasonic bath for 5 h to obtain a homogenous solution.Graphene dispersions with concentrations of 1, 5, and 10 mg ml −1 were deposited using the spin coating technique (1500 rpm, 60 s).Finally, the silver (Ag) top electrode was deposited through thermal evaporation.Graphene 0, 1, 5 and 10 is denoted to the amount of graphene deposited on top of the spiro-OMETAD layer.For instance, 0 is the reference cell, and 1 means 1 mg ml −1 graphene deposited.The active cell area was 0.07 cm 2 [17,32,33].
The second step was the fabrication of the polycrystalline silicon (poly-Si) bottom cell.A commercial p-type poly-Si wafer substrate (100) crystal phase with a thickness of 200 μm and resistivity of 3-10 cm 2 ) with phosphorus oxy-chloride (POCl 3 ) was used in the fabrication process.The metallisation process produced the back and front metal contacts of the cell.The aluminium (Al) paste was screen printed onto the backside of the Si wafer, and the substrate was subsequently dried in the furnace at 150 °C for 15 min.The Ag paste was screen-printed onto the front side of the substrate and dried using the same process.Both contacts were heated using rapid thermal annealing at roughly 750 °C to form the back and front contacts.The substrate was then cut into 3 cm 2 .Finally, the PSC top cell was attached and fixed to the poly-Si bottom cell by using scotch tape [34].The measurement and characterisation of graphene intensities were conducted using a Raman spectrometer (DXR2xi, Thermo Scientific) with a laser wavelength of 532 nm, spectrograph aperture of 50 μm and slit and grating groove density of 900 lines/mm.The data were obtained at room temperature and scan range of 500-3000 cm −1 .The solar-simulated AM 1.5 G sunlight was used, and the solar simulator was calibrated using a standard Si PV cell (Daystar Solar Meters) to obtain an intensity of 100 mW cm −2 .The current density-voltage (J-V) curves were recorded using the Keithley 2400 SourceMeter at a scan rate of 0.1 V/s.The external quantum efficiency (EQE) measurements were obtained on a Newport IPCE system that was calibrated using a Si photodiode before the analyses.The cross-section images of the device and the distribution of graphene film at different concentrations were obtained by field emission scanning electron microscopy (FESEM) with a model of ZEISS MERLIN Compact at a resolution of 15 kV.The Raman spectra show two main peaks at 1342 and 1578 cm −1 in the graphene layers, which correspond to the D and G peaks, respectively.This result verifies the presence of the graphene layer.The D and 2D peaks of all graphene concentrations are more intense than the G peaks, thereby indicating the perturbation and deconvolution of the graphene structures.In addition, the full width at half maximum of the D, G, and 2D bands is extremely broad and large and displays many imperfections in the graphene structures.Several studies reported that the intense D and 2D bands denote the defect formation and amount of disordered phase in the graphene layer [35,36].The findings confirm that the deterioration of the graphene structures is caused by the transformation of graphene dispersion in water into graphene flakes by applying low heat; such a treatment process may result in the oxidation of graphene.Moreover, the oxidation process can result in the formation of defects and imperfections due to the high oxygen content, and thus increase the interplanar distance and mismatches in the lattice of the graphene.The intensity at a graphene concentration of 10 mg ml −1 is higher than at graphene concentrations of 1 and 5 mg ml −1 .In addition, the spectra of the D and G peaks were narrower in the case of the former than in the latter.This result indicates that the best coverage and most even surface layer can be achieved at a graphene concentration of 10 mg mL −1 , followed by 5 and 1 mg ml −1 .The thickness of the layer with 10 mg ml −1 of graphene (mean value = 59.5 ± 7.8 nm) is larger than those of the layers with 5 (32.6 ± 10.7 nm) and 1 mg ml −1 (47.6 ± 10.3 nm).The former also has better surface coverage with lower standard error than the latter two.The mean thickness values of the graphene layer were determined by selecting five different spots in the crosssectional field emission scanning electron microscope images of the layers with 1, 5 and 10 mg ml −1 of graphene (figures 3(a)-(c), respectively).The spectra acquired at a graphene concentration of 1 mg ml −1 graphene have excessively large broadness an extremely low intensity.This result shows the indistinct presence of graphene later with 1 mg ml −1 graphene concentration, which displays, a large of defects and mismatched layer.The presence of the defects on the graphene surface is confirmed by the high calculated I D /I G ratios of the layers with the three graphene concentrations (1.19, 1.36 and 1.29 for layers with graphene concentrations of 1, 5 and 10 mg ml −1 , respectively) due to the treatment process of the graphene dispersion.The I D peak corresponds to the disorder-induced mode in the Raman spectrum and is associated with defects or disorder in the carbonbased material.A higher intensity of the I D peak indicates a higher level of structural disorder or defects within the material.The I G peak represents the graphitic or sp 2 carbon structure in the material.It is associated with the E 2 g vibrational mode of the C-C bond stretching.The I G peak is a characteristic feature of well-defined graphitic structures.The ratio of the I D peak intensity to the I G peak intensity (I D /I G ratio) is a crucial parameter in Raman spectroscopy analysis.This ratio provides insights into the level of structural disorder and defects present in the material.A higher I D /I G ratio indicates a higher degree of disorder or defects, while a lower ratio suggests a more ordered and less defective structure.

Result and discussion
The effects of the intensities of the graphene layer on the PV performances of the perovskite-Si tandem solar cell were analysed (table 1).The PSC device without a graphene layer was set as a control sample to verify the effect of the graphene layer.The J-V curves depicted in figure 4 correspond to the parameters provided in table 1, which show the best performance of the PSC top cell with and without a graphene layer.The PSC top cell without graphene layer (denoted as Graphene 0 mg ml −1 ) has the best open-circuit voltage (V OC ; 0.897 V), short-circuit current density (J SC ; 16.2 mA cm −2 ), fill factor (FF; 0.51) and power conversion efficiency (PCE; 7.4%).The PCE values of the PSC top cells with graphene concentrations of 1, (Graphene 1 mg ml −1 ), 5 (Graphene 5 mg ml −1 ) and 10 mg ml −1 (Graphene 10 mg ml −1 ) decrease to 5.8%, 4.5% and 3.0%, respectively.Based on observations, graphene layers with high intensities and large thicknesses can increase the series resistance and deteriorate the J SC and FF values of the PSC devices.The high series resistance of PSCs demonstrates a good agreement with the high I D /I G ratio of the graphene layer that exhibits morphological defects (i.e.nonuniformity and low coverage).In conclusion, these effects are caused not only by the treatment process of the graphene dispersion, but also by the weakness of the spin coating technique applied to coat the graphene layer.The spin coating technique usually depends on the viscosity of the solution to gain a uniform and good surface coverage.Low graphene concentration signifies low viscosity that causes rapid spread of the graphene dispersion through spin coating deposition if the speed of the rotation is not precisely controlled.
A 0.7 mm 2 active area of the PSC top cell without and with a graphene layer was stacked over 2.0 cm 2 poly-Si solar cells.The two cells were attached using scotch tape to build a tandem structure, and the PCE of the perovskite-Si tandem solar cells was calculated.The PCE of the filtered poly-Si solar cells was measured under AM 1.5 sun illumination.The J-V graph indicates that the J SC and FF of these cells decrease because of the increased intensity and thickness of the graphene layer; such increments reduce the transparency of the PSC top cell.The PCE of the filtered poly-Si solar cells without graphene declined from 6.7% to 5.1%, 4.0% and 2.3% under the presence of 1, 5, and 10 mg mL −1 of graphene, respectively.
The EQE values confirm the degeneration of the photocurrents of the PSC top cell with graphene layer and poly-Si solar cell filtered using the PSC top cell as shown in figure 5.The PSC top cell without a graphene layer Table 1.Photovoltaic performances of the poly-Si solar cell, PSCs with Graphene 0, 1, 5, and 10 mg ml −1 , and tandem solar cell.

Sample devices
Voc (V) Jsc (mA/cm 2 ) FF (-) Graphene 0 mg ml exhibits high and broad spectral responses over the wavelength range of 300-800 nm, with a maximum peak that has an EQE value of 80%.Meanwhile, the EQE values of the PSC top cells with 1, 5, and 10 mg ml −1 graphene decrease to roughly 70%, 40%, and 30%, respectively.The ability of graphene to absorb light is verified by its ∼2.3% light absorption rate in most UV and visible spectra [37,38].The results suggest that the increased intensity and thickness of the graphene layer amplify its capability to absorb light.However, this parasitic absorption in the PSC leads to efficiency deterioration.This behavior is similar to that shown by the EQE of poly-Si solar cells filtered by PSC top cells.The spectral responses deteriorate, and the peaks decrease to approximately 80% when no graphene is present and to roughly 70%, 60% and 50% when the graphene concentration is 1, 5, and 10 mg ml −1 , respectively.Conversely, the EQE of the single poly-Si solar cell displays high and broad spectral responses over the wavelength range of 300-1100 nm, with a maximum peak of 90% as in figure 5.The high EQE value of this cell  indicates that the high transparency of the PSC top cell plays a crucial role in the four-terminal tandem solar cells.However, this study observes that the transparency of the PSC top cell is influenced by the intensity and thickness of every sublayer.Given that the only variable parameter is the concentration of graphene dispersion, the transparency of bulk PSCs is affected by the intensity and thickness of the graphene layer.The high opacity of the PSC top cells blocks the incident photon from penetrating the Si bottom cell and thus results in an inefficient photon-to-electron (photocurrent) conversion.When the graphene concentration is 10 mg ml −1 , the EQE drops to 50%.As shown in figure 6, the PCEs of the single PSC top cell and the poly-Si bottom cell have a larger gap than that of the four-terminal perovskite-Si tandem solar cells.This gap is the solution to improving the efficiency of the PSC top cells or Si bottom cells.The PCE value of the perovskite-Si tandem solar cells without a graphene layer is 14.1%, whereas those of the solar cells with 1, 5 and 10 mg ml −1 graphene are 10.9%, 8.5% and 5.3%, respectively.These PCEs are two times of those of all single top and bottom cells.
Stability is one of the key challenges in any PV device.This factor is critical in PSCs.To evaluate the stability of the PSC with a graphene layer, the PV performances of the PSC top cell were examined under an ambient environment (∼85% humidity) and ambient temperature (27 °C) for five days (60 h) by leaving the PSCs exposed to air.After 60 h, the PV performance of Graphene 0 mg ml −1 almost completely decays (93%).This reduction in the PCE can be ascribed to the degradation of the V OC , J SC and FF due to the major damages in the interface of the perovskite/HTL and HTL/metal electrodes exposed to moisture, oxygen and other chemicals.The Ag electrode corrodes because of the pinholes that exist in the spiro-OMeTAD layer and the diffusion of the unstable products containing iodine in the CH 3 NH 3 PbI 3 layer.The latter also causes the Ag layer to form AgI [19,39].The findings of this study reveal that the graphene layer seals the spiro-OMeTAD/CH 3 NH 3 PbI 3 interlayer and therefore improves the stability of the PSC top cell and increases the lifetime of the perovskite-Si tandem solar cell.Graphene 1, 5 and 10 mg ml −1 slowly decrease and maintain their J SC and FF values after 24 h.This phenomenon indicates that the spiro-OMeTAD/CH 3 NH 3 PbI 3 interlayer is passivated by the graphene layer, especially from the moisture and UV light under ambient conditions.This result is consistent with the findings of Kim et al, who stated that using reduced graphene oxide (rGO) as an additive in the perovskite layer inhibits recombination and improves stability.In addition, the presence of rGO in the spiro-OMeTAD layer passivates the perovskite layer and hinders it from being crystallized at high temperatures [28].Palma et al suggested that utilising rGO as hexamethylenetetramine in PSCs results in excellent endurance properties and a 36% increase in PCE after the 1987 h life test compared with spiro-OMeTAD, which obtained a 41% reduction after the same endurance time test [40].
As previously mentioned, the PV performances of the PSC top cell without the graphene layer drastically drop from the as-prepared time to 60 h, and the corresponding PCE declines to 93% as shown in table 2. The low stability of PSC top cells will cause issues in tandem solar cells because the PCE of the PSCs must sustain longterm stability, as well as compatibility with the long life of the Si bottom cell in the tandem architecture.Table 2 further indicates that the presence of a graphene layer in PSC top cells helps stabilise the PCE performances, as illustrated by the 43%, 24%, and 17% reduction in the relative variation of the PCEs of Graphene 1, 5 and 10 mg ml −1 , respectively.These values exhibit a slightly large gap compared to the PSC top cell without graphene layer.Graphene 1 mg ml −1 is selected as the most suitable PSC top cell for the tandem solar cell because of its potential to gain high PCE and long-term stability by using a better coating method to acquire the optimal surface morphology and thickness, which are both important parameters in PSC fabrication.The PV performances of Graphene 5 and 10 mg ml −1 are slightly low but still display good stability.However, using these cells to tandem with solar cells will cause difficulties.

Conclusion
The intensity and thickness of the graphene layer played crucial roles in the PV performances of the PSC top cell and poly-Si bottom cell.The increase in both factors caused parasitic absorption and high surface resistance, which deteriorated the EQE and PV performances.Using the graphene layer as a sealing mechanism in PSCs yielded substantial stability properties.Thick graphene layers blocked the photocurrent due to their high opacity and high surface resistance.The PCEs of all PSC top cells with graphene concentrations of 1, 5, and 10 mg ml −1 gradually reduced to 43%, 24%, and 17%, respectively, within 60 h.Similarly, the PCE of the PSC top cell without a graphene layer decreased by 93% under the same condition.The PSC top cell with a graphene concentration of 1 mg ml −1 can be used in tandem solar cells due to its satisfactory PV performance and stability compared with other cells.Further research should be conducted to improve the graphene surface morphology and thickness and enhance the PV performances and stability.0.815 ± 0.01 10.9 ± 1.2 0.38 ± 3.0 3.4 ± 0.5 Graphene 10 mg ml −1 As-prepared 0.812 ± 0.01 8.9 ± 1.0 0.41 ± 3.9 3.0 ± 0.5 −17 Graphene 10 mg ml −1 60 h 0.791 ± 0.02 8.1 ± 0.9 0.39 ± 3.7 2.5 ± 0.5

Figure 1
Figure1depicts the schematic of the perovskite top cell that is stacked on the poly-Si bottom cell and utilised as the mechanically stalked four-terminal tandem solar cell.The PSC top cell features an n-i-p mesoporous structure that consists of FTO/bl-TiO 2 /mp-TiO 2 /CH 3 NH 3 PbI 3 /spiro-OMeTAD/graphene/Ag contact.This study emphasises the variation in the PV performances of the perovskite-Si tandem solar cells under the presence of different graphene concentrations in the PSC devices.The presence of the graphene interlayer in the PSCs was characterised through Raman spectroscopy.The intensities of the different concentrations of the graphene layer are shown in figure2.The Raman spectra show two main peaks at 1342 and 1578 cm −1 in the graphene layers, which correspond to the D and G peaks, respectively.This result verifies the presence of the graphene layer.The D and 2D peaks of all graphene concentrations are more intense than the G peaks, thereby indicating the perturbation and deconvolution of the graphene structures.In addition, the full width at half maximum of the D, G, and 2D bands is extremely broad and large and displays many imperfections in the graphene structures.Several studies reported that the intense D and 2D bands denote the defect formation and amount of disordered phase in the graphene layer[35,36].The findings confirm that the deterioration of the graphene structures is caused by the transformation of graphene dispersion in water into graphene flakes by applying low heat; such a treatment process may result in the oxidation of graphene.Moreover, the oxidation process can result in the formation of defects and imperfections due to the high oxygen content, and thus increase the interplanar distance and mismatches in the lattice of the graphene.

Figure 6 .
Figure 6.Comparison PCE of the PSCs top cell, poly-Si bottom cell and tandem solar cells.

Table 2 .
Photovoltaic performances of PSCs top cell for the 60 h stability test with the relative PCE variation in percentage.