Fabrication of Flexible Quasi-Interdigitated Back-Contact Perovskite Solar Cells

: Perovskites are a promising class of semiconductor materials, which are being studied intensively for their applications in emerging new ﬂexible optoelectronic devices. In this paper, device manufacturing and characterization of quasi-interdigitated back-contact perovskite solar cells fabricated on ﬂexible substrates are studied. The photovoltaic parameters of the prepared ﬂexible quasi-interdigitated back-contact perovskite solar cells (FQIBC PSCs) are obtained for the front- and rear-side illumination options. The dependences of the device’s open-circuit potential and short-circuit current on the illumination intensity are investigated to determine the main recombination pathways in the devices. Spectral response analysis of the devices demonstrates that the optical transmission losses can be minimized when FQIBC PSCs are illuminated from the front-side. Optoelectronic simulations are used to rationalize the experimental results. It is determined that the obtained FQIBC PSCs have high surface recombination losses, which hinder the device performance. The ﬁndings demonstrate a process for the fabrication of ﬂexible back-contact PSCs and provide some directions for device performance improvements.


Introduction
The use of photovoltaic (PV) technologies in producing clean and renewable energy is increasing at a fast pace [1]. In recent years, more and more countries have been trying to increase the percentage of energy production from the renewable sources [2]. At this moment, the most commonly used PV devices are silicon solar cells. However, the progression in improving the power conversion efficiencies (PCEs) of silicon solar cells is slow and somewhat stagnant [3,4].
Recently, new types of organic and organic-inorganic semiconductor materials have been actively researched [5]. One type of such materials is perovskites. Perovskites have found a wide range of applications in solar cells [6][7][8] and other optoelectronic devices [9][10][11]. This is due to their optimal optoelectronic properties such as tunable and direct bandgap [12,13] long charge carrier diffusion length [14] and high absorption coefficient [15]. In addition, the methods for producing thin films with perovskites are rather cheap and simple [16,17]. Nowadays, the PCEs of perovskite solar cells (PSCs) reached values as high as 25.5% and promise to increase further [18,19].
Currently, there is a fast-growing interest in developing flexible PV and other optoelectronic devices due to their lucrative application potential in wearable/portable electronics [20][21][22] and IoT technologies [23,24]. Due to this, perovskites, owing to their solution processability, low cost [25,26], and abundance, have been actively investigated for developing flexible PSCs [27,28]. In addition, the low-temperature processability coupled with the roll-to-roll manufacturing possibilities make it ideal for fabricating PSCs on plastic substrates such as polyethylene terephthalate (PET) [29,30] or polyethylene naphthalate (PEN) [31,32], which are low-cost, lightweight materials [31]. Currently, lab-scale flexible PSCs on PET substrates achieved PCEs as high as 21.1% and this is anticipated to increase further [33].
While the quest for new materials and the development of already known materials for optoelectronic applications continues, there is also a quest for new designs of solar cells. Quasi-interdigitated back-contact (QIBC) PSCs are devices with a novel and advanced device architecture [34][35][36]. The manufacturing principle of QIBC PSCs relies on performing the device electrodes on substrates, followed by deposition of the perovskite absorbing layer on top of the electrodes to complete the device fabrication process [34,37]. The idea of the proposed back-contact architecture is to illuminate QIBC PSCs directly from the perovskite side (front-side). In this case, the incident light directly enters the exposed perovskite photo-active layer [34]. This is in contrast to the typical planar PSCs with a sandwich architecture, in which some transmission losses occur as the incident light passes through the supporting substrate and device functional layers (transparent metallic layer and either electron-or hole-transporting layer (ETL or HTL)) before reaching the perovskite-absorbing layer [34,38]. Thus, the back-contact architecture for PSCs promises to improve the PCEs of devices even further by addressing the transmission loss issues [39].
Currently, a great deal of research efforts have been invested in developing flexible PSCs. However, very little work has been completed on developing flexible PSCs with back-contact architecture [40]. Therefore, in this work we demonstrate a fabrication method for flexible QIBC PSCs (hereon referred to as FQIBC PSCs) on PET substrates and conduct device characterization studies on the manufactured devices. The opto-electrical behavior of the FQIBC PSCs are studied at various light intensities using a current-voltage (J-V) measurement system; the spectral response of the devices are analyzed using an external quantum efficiency (EQE) measurement system. Theoretical modeling studies are used to rationalize the experimental findings and to provide an insight into the device physics of FQIBC PSCs.

Device Fabrication
To fabricate back-contact electrodes, first, the PET/ITO substrates are cleaned in a detergent water, deionized water, acetone, and ethanol for 15 min each while sonicating in an ultrasonic bath. Then, a 30 nm ZnO layer (serves as the ETL) is deposited on the freshly cleaned PET/ITO substrates using a radio frequency magnetron sputtering system (Model: Spectros, Kurt J. Lesker, Jefferson Hills, PA, USA). Afterwards, a polymer mask with the desired features for the second electrode is formed using photolithography tools. Here, a positive photoresist is deposited using a spin-coating method at 3000 rpm spinning speed and a spinning time of 30 s. After that, the substrates are placed on a hotplate for 2 min at 110 • C for baking. The final thickness of the photoresist after the baking step is~1.5 µm. Then, the substrates with the photoresist layer are exposed to UV light (dose-370 mJ/cm 2 ) through a photo-mask using the SUSS MicroTec MA/BA Gen4 mask aligner system. The UV-exposed areas of the photoresist are removed by washing the substrates in a developer solution (3:1 mix of AZ 726 MIF with water) for 90 s and subsequent rinsing with water.
The second electrode is formed on the substrates by sequential deposition of an Al 2 O 3 insulating layer (thickness~100 nm) and a gold layer (thickness~70 nm) by reactive magnetron sputtering and thermal evaporation (Model: Mini SPECTROS, Kurt J. Lesker, Jefferson Hills, PA, USA), respectively. Here, the Al 2 O 3 serves as a spacer between the cathode (ITO/ZnO) and the anode (gold) layers. A thin layer of chromium (~10 nm) is deposited between the gold and the Al 2 O 3 layers via thermal evaporation to improve the adhesion of the gold layer to the Al 2 O 3 insulating layer. Afterwards, the remains of the polymer mask are removed by washing the substrates in acetone to obtain the back-contact electrodes for FQIBC PSCs.
The last stage of the FQIBC PSC fabrication process is deposition of the perovskite photo-absorber layer. The precursor solution for the perovskite layer contains 0.461 g PbI 2 and 0.159 g MAI in a 1:1 molar ratio and is dissolved in l mL of anhydrous DMF [41]. The perovskite layer is deposited onto the freshly prepared back-contact electrodes using a gas-assisted spin-coating method [42,43]. For this, 100 µL of the perovskite precursor solution is dropped onto the back-contact electrodes and spun for 30 s at 6500 rpm. After 2 s of spinning, the substrates are blown with a stream of nitrogen (pressure: 150 kPa) for 28 s using a nitrogen gun. The distance between the substrate and the nitrogen gun is 10 cm. Once the spin-coating process is complete, the substrates are annealed at 100 • C on a hotplate for 15 min. This affords the formation of a crystalline CH 3 NH 3 PbI 3 perovskite photo-absorber layer. The schematic representation of the device fabrication process and schematic diagram of the final device are shown in Figure 1.

Characterization
The cross-section images of the electrodes are obtained using a scanning electron microscope (Model: Crossbeam 540, Zeiss, Oberkochen, Germany). The optical images of the electrodes are obtained using an optical microscope (Model: Leica DM4000 B, Leica Microsystems, Wetzlar, Germany). The J-V curves of the FQIBC PSCs are measured using a solar simulator (Model: ORIEL Sol3A, Newport, Irvine, CA, USA) with AM1.5G solar irradiation at 100 mW/cm 2 and a parameter analyzer (Model: B1500A, Keysight, Santa Rosa, CA, USA). The J-V curves of the devices are recorded with and without using neutral density optical filters (Model: NEK01S, Thorlabs, Newton, MA, USA). The spectral responses of the FQIBC PSCs are measured using an EQE measurement system (Model: ORIEL IQE 200, Newport, Irvine, CA, USA).

Experimental Part
In order to evaluate the dimensions and structural quality of the back-contact electrodes, glass substrates are placed together with the PET substrates during the electrode functional layers (ZnO, Al 2 O 3 , and gold) deposition stages. Then, the cross-sections of these glass substrates with the deposited layers are analyzed using SEM imaging. A cross-section SEM micrograph of a sample shown in Figure 3a indicates that the electrode functional layers are homogeneous and conformal. An optical micrograph (see Figure 3b) of a prepared flexible back-contact electrode indicates that the electrodes are defect-free and have the desired structure. Figure 3c shows a photographic image of a prepared flexible back-contact electrode without the perovskite layer. Current-voltage characteristics of the obtained FQIBC PSCs (shown in Figure 4a) are measured under AM1.5G irradiation. The measurements are performed for both front-(perovskite-side) and rear-side (PET-side) illumination options. The obtained FQIBC PSCs exhibit a small hysteresis effect as the J-V curves for the forward and reverse scans more or less coincide with each other (see Figure 4a). The photovoltaic parameters of the FQIBC PSC extracted from the J-V curves under illumination are displayed in Table 1. We note that the effective device area (0.04 cm 2 ) is taken to be the same for both illumination options.  Comparison of the photovoltaic parameters for the front-and rear-side illumination indicate that except for the fill factor (FF), the short-circuit current (J sc ), the open-circuit voltage (V oc ), and the PCE of the device under testing are slightly lower for the front-side illumination. This is a rather surprising result as it is expected that the photovoltaic parameters for the front-side illumination should be higher than the ones for the rear-side illumination. Indeed, the performance of FQIBC PSCs illuminated from the rear side should be lower due to light transmission losses and the shadowing effect of the metallic contacts of the anode layer (Al 2 O 3 /gold). However, the experimental results show the contrary. It is possible that this effect can be associated with the surface recombination of the photogenerated charge carriers. The surface of the typical solution-processed perovskite films is not ideal, and there are various macroscopic and microscopic defects and dangling chemical bonds on the surface of the perovskite films [46]. If the perovskite surface is exposed to the ambient air, as in the case of the current work, this can lead to further degradation of the perovskite layer surface quality due to chemical or/and physical adsorption of various atoms and molecules present in the environment. All this leads to appearance of the localized surface states on the perovskite surface, which can act as the recombination sites for charge carriers [47,48].
The dependence of J-V characteristics on light intensity is an effective and simple method for analyzing the dominant recombination mechanisms in PSCs [47,49]. The measurements of the J-V curves under different light intensities are also carried out when the devices are illuminated from the rear and the front side. Here, the measurements are performed under AM1.5G irradiation using neutral density filters with different optical densities. Figure 4b shows the J-V curves of a FQIBC PSC measured under different light intensities for the rear-side illumination. The J-V curves measured under different light intensities for the front-side illumination are shown in Supplementary Materials Figure S1. The dependence of the J sc of the FQIBC PSC on the light intensity is linear for both illumination options (inset graph in Figure 4c). This indicates small charge carrier recombination losses under the short circuit conditions [50]. A relationship that describes the dependence of the open-circuit voltage on light intensity is given by the following equation [47,51]: Here, k B is the Boltzmann's constant, n idL is the light ideality factor, T is the absolute temperature, J ph is the photocurrent, q is the unit charge, and J s is the dark saturation current. Since the photocurrent J ph is much higher than the dark saturation current and linearly depends on the light intensity, Equation (1) can be rewritten as the following: where L is the light intensity. Thus, it is possible to determine the light ideality factor n idL from the slope of the dependence of V oc on ln (L). The value of n idL is obtained from the dependence of V oc of the FQIBC PSC on the light intensity (see Figure 4c) is 1.61, which indicates the presence of trap-assisted Shockley-Reed-Hall (SRH) recombination pathway in the device [50][51][52]. A high value of n idL (values close to 2) indicates recombination via deeper traps [52]. When the FQIBC PSC is illuminated from the front side, the value of n idL becomes more than 2. This indicates a greater influence of traps on carrier recombination, which causes lower efficiency values. The physical origin of the traps can be caused by the polycrystalline nature of the perovskite film and the presence of defects at the grain boundaries [46]. In addition, since flexible substrates are used for the manufacturing of FQIBC PSCs, mechanical bending of the substrates is inevitable during the perovskite layer deposition and the device characterization steps. This leads to the formation of microcracks and deterioration of the structural integrity of devices. Figure 4d shows the EQE spectra of a FQIBC PSC obtained for both illumination options. The EQE spectrum of the device under testing for the rear-side illumination exhibits an abrupt decrease below~370 nm. The EQE spectrum of the device under test for the front-side illumination, on the other hand, remains high and even increases slowly as the wavelength of the incident light goes below~370 nm. Such a disparity between the EQE spectra of the FQIBC PSC for different illumination options can be rationalized by the light transmission losses (absorption and reflection) occurring at the PET/ITO/ZnO layers when back-contact PSCs are illuminated from the rear side. This result nicely demonstrates the advantage of the back-contact configuration for solar cell device architecture in terms of reducing the light transmission losses and manufacturing PV and other optoelectronic devices (e.g., photodetectors) with wider spectral responses. A slightly lower EQE value for the front-side illumination is consistent with the J-V characteristics shown in Figure 4a.

Theoretical Part
To elucidate the underlying physical reason for the higher photovoltaic performance of the obtained FQIBC PSCs for the rear-side illumination and also to test our proposition on high values of the surface recombination rates in the devices, we examined the effect of the surface recombination rate on the performance of FQIBC PSCs using numerical simulation methods [53,54]. We used the standard drift-diffusion model [35,55] to calculate various electrical parameters of devices. For this, a set of equations of the standard driftdiffusion model is solved for the steady-state conditions (i.e., ∂n ∂t = 0 and ∂p ∂t = 0) using the COMSOL Multiphysics software for the presented device architecture. More details about the standard drift-diffusion model and the physical parameters of the materials used in the simulation experiments are given in Supplementary Materials Table S1.
A sketch of the cross-section of FQIBC PSCs used in the simulation experiments is shown in Supplementary Materials Figure S2. First, the simulation experiments are performed for the ideal case, in which the parameters of the device functional layers are the most optimum (see Supplementary Materials Table S1). The simulated J-V curves for the front-and rear-side illumination options and the most optimum device parameters are shown in Figure S3. The calculated maximum theoretical PCEs for the idealized device are 7.6% and 6.1% for the front-and rear-side illumination options, respectively.
Next, we adjusted the input parameters for the simulation experiments to approach the real experimental values (see Supplementary Materials Table S1) and simulated the J-V curves for FQIBC PSCs by varying the surface recombination velocity for electrons and holes (S n,p ) from 1 cm/s to 10 5 cm/s. As shown in Figure 5a, the J-V curves of FQIBC PSCs exhibit a somewhat moderate behavior when the material parameters of the device functional layer are adjusted to approach the real experimental values and the shunt and sheet resistance values obtained from the experiments are taken into account. The device currents and PCE values for the front-side illumination are higher than the ones for the rear-side illumination. Here, the S n,p value is taken to be equal to 1 cm/s. Increasing the S n,p value to 10 cm/s does not seem to affect the J-V characteristics of devices, as shown in Figure 5b. However, starting from S n,p = 10 3 cm/s, the J-V curves of FQIBC PSCs start degrading, as depicted in Figure 5c,d. When the S n,p value becomes 10 4 cm/s, not only a further degradation of the J-V curves is observed (see Figure 5e), but the device currents and PCE values for the front-side illumination become lower than the ones for the rear-side illumination. This is similar to the J-V curves of the experimentally obtained FQIBC PSCs. When the S n,p value is as high as 10 5 cm/s, the simulated J-V curves (see Figure 5f) most closely approach the experimental J-V curves shown in Figure 4a. The photovoltaic parameters of FQIBC PSC extracted from the J-V curves shown in Figure 5 are displayed in the Supplementary Materials, Table S2.
Contrasting the experimentally and numerically obtained results suggest that the factor that is responsible for higher values of the device currents and the PCEs for the rearside illumination could indeed be the high surface recombination rates in the perovskite layer of obtained FQIBC PSCs. Reducing the surface recombination rates (lower values of S n,p ≤ 10 3 cm/s) in the perovskite layer of FQIBC PSCs promise to result in higher device currents and PCEs values for the front-side illumination. One way to achieve this is to optimize the perovskite layer deposition process to obtain the perovskite photo-absorber layers with an improved crystalline and electronic nature [35]. In this regard, different deposition methods such as anti-solvent crystallization, hot-casting, thermal evaporation, and many other exciting new methods can be tested to achieve a perovskite photo-absorber layer with desired properties. Additionally, various passivation methods [56,57] can be employed to suppress the perovskite surface trap states and also to protect the perovskite layer from the degradation effects of the environment. Once these tasks are fulfilled, FQIBC PSCs may have a significant application potential in the emerging new technologies such as wearable/portable electronics, IoT technologies, and beyond.

Conclusions
In summary, a fabrication method for functional, flexible back-contact PSCs with quasi-interdigitated electrodes manufactured on PET/ITO substrates is demonstrated. Investigation of the photovoltaic properties of the obtained FQIBC PSCs shows that the devices have a higher performance when illuminated from the rear side. The analysis of the dependence of the open-circuit voltage on the light intensity indicates that the dominant recombination pathway in the devices is the Shockley-Read-Hall recombination. The measurements of the EQE spectra of the devices demonstrate that FQIBC PSCs have wider spectral response when illuminated from the front side. The computer simulation experiments on studying the effect of the surface recombination rate on the J-V characteristics of FQIBC PSCs have been carried out to rationalize the experimentally obtained results. Contrasting the experimental results with the simulation results reveals that high surface recombination rates in the perovskite layer might be the cause for the higher device performance values for the rear-side illumination. The simulation results also demonstrate that high surface recombination rates may also be one of the main factors that is limiting the device performance of the obtained FQIBC PSCs. Although the photovoltaic performance of the devices shown in this work are rather moderate, the computer simulation experiments demonstrate that with further optimization and improvements in device fabri-cation processes, FQIBC PSCs with the desired performances can be obtained. The findings of this work can be used as an example and guideline for the design and fabrication of high-performance FQIBC PSCs.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/en15093056/s1, Figure S1: J-V curves of a FQIBC PSC for the front-side illumination and different light intensities; Figure S2: Sketch of the cross-section view (shown in xz-plane) of the sunlight fall on FQIBC PSCs from the front-(perovskite-side) (a) and the rear-side (PET-side) (b). The areas enclosed with dashed white rectangles indicate the base unit areas used in simulation experiments; Figure S3: Comparison of the simulated J-V curves of FQIBC PSC for the front-side (dash red line) and the rear-side (solid black line) illumination; Table S1: Physical parameters used for simulation experiments; Table S2: The calculated photovoltaic parameters of FQIBC PSCs under AM1.5G solar irradiation [58][59][60][61][62][63][64][65][66][67][68].

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.