The Effect of Hole Transporting Layer in Charge Accumulation Properties of p-i-n Perovskite Solar Cells

The charge accumulation properties of p-i-n perovskite solar cells were investigated using three representative organic and inorganic hole transporting layer (HTLs): a) Poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate) (PEDOT:PSS, Al 4083), b) copper-doped nickel oxide (Cu:NiOx) and c) Copper oxide (CuO). Through impedance spectroscopy analysis and modelling it is shown that charge accumulation is decreased in the HTL/Perovskite interface, between PEDOT:PSS to Cu:NiOx and CuO respectively. This was indicative from the decrease in double layer capacitance (Cdl) and interfacial charge accumulation capacitance (Cel), resulting in an increase to recombination resistance (Rrec), thus decreased charge recombination events between the three HTLs. Through AFM measurements it is also shown that the reduced recombination events (followed by the increase in Rrec) is also a result of increased grain size between the three HTLs, thus reduction in the grain boundaries area. These charge accumulation properties of the three HTLs have resulted in an increase to the power conversion efficiency between the PEDOT:PSS (8.44%), Cu:NiOx (11.45%) and CuO (15.3%)-based devices.

events between the three HTLs. Through AFM measurements it is also shown that the reduced recombination events (followed by the increase in Rrec) is also a result of increased grain size between the three HTLs, thus reduction in the grain boundaries area. These charge accumulation properties of the three HTLs have resulted in an increase to the power conversion efficiency between the PEDOT:PSS (8.44%), Cu:NiOx (11.45%) and CuO (15.3%)-based devices. 1 stelios.choulis@cut.ac.cy 2 Over the last couple of years perovskite-based solar cells have shown advancements, with an incredibly fast power conversion efficiency (PCE) improvement from 3.8% to 21%. 1,2 The high PCE values are a result of several attractive features that hybrid perovskite semiconductors offer, even if casted from solution at low processing temperatures. 3 Their main solar cell related attractive features are the low exciton binding energy, 4,5 tunable bandgap, 6,7 high absorption coefficient, 8 long carrier diffusion length 9 and high carrier mobility. 10 The p-i-n structured perovskite solar cell is distinguished compared to other perovskite solar cell device architectures due to its facile processing. In this structure, the perovskite active layer is placed between two carrier selective electrodes and is in direct contact with the HTL of the transparent bottom electrode and the electron-transport layer (ETL) of the opaque back electrode. Each electrode is responsible for allowing only a specific type of carrier to be collected by blocking the opposite carrier type. 8 Efficient carrier generation, transport and collection are crucial for proper photovoltaic operation.
One of the most important aspects of the solar cell device performance is the crystallinity of the perovskite active layer. A critical factor affecting the crystallinity of the perovskite sensitizer is the layer on which the perovskite photoactive layer is grown from solution 11 . It has been shown that the under-layer has a major influence on the perovskite crystal orientation, grain size and morphological defects. These factors are crucial for highly efficient perovskite solar cells. 11,12 . Large perovskite grain size is generally believed to result in increased carrier diffusion length and mobility of carriers. 12 Furthermore, a crystal structure with large grains can provide good surface coverage and small grain boundaries area, which are also desirable since they can minimize the presence of pinholes and charge traps, effectively reducing carrier recombination. 3 Incorporating HTLs in perovskite solar cells has been a crucial part towards the achievement of high performance devices. The main features of an efficient HTL are: a) efficient electron blocking, b) efficient hole extraction. These features can have a major impact to device parameters, e.g. Voc via quasi Fermi level splitting. 13 Different types of materials have been studied and used as HTLs thus far, including inorganic, polymeric 14 and small organic molecules. 14,15 Considerable effort has been put in the process of HTL optimization in terms of stability and efficient semiconducting properties. An example is the use of cobalt based dopants to enhance the charge carrier mobility of the widely used spiro-OMeTAD 15 .
The focus of this study is to analyze the effect of the HTL in the charge accumulation of a p-i-n structure perovskite solar cell. The exact transport and charge collection mechanisms in perovskite solar cells are not yet completely understood. Impedance spectroscopy has been used to provide valuable information about the physical processes manifesting inside the perovskitebased solar cells. 8 By studying these processes, both from various experimental methods and by fitting the data in an equivalent circuit model (ECM) we were able to extract valuable information regarding the charge accumulation at the p-i-n structure perovskite solar cell device.
In the present study, three p-i-n CH3NH3PbI3-based devices with different HTLs were fabricated and extensively analyzed using a series of measurements. For all the devices the ETL respectively. As previously mentioned the increased grain size, and therefore reduced grain boundaries area of the perovskite photoactive layers, influences the charge transport properties and subsequently the performance of corresponding devices. 12 The exact effect of the grain size and grain boundaries towards charge accumulation is discussed further down this manuscript.
Representative devices from each structure were chosen and characterized using J-V and impedance spectroscopy. The devices with the highest PCE values were chosen for study from each architecture. The active area for all the devices was at 0.09 cm 2 . No hysteresis was observed, between the forward and reverse scan of the current density-voltage (J/V) characteristics, for all the devices under study. The preparation of PEDOT: PSS and Cu:NiOxbased devices was based on the methodology of Jung et. al. 17 resulting in a thickness of ~40 nm 5 for each layer. The CuO-based synthesis and fabrication details can be found in our recent study. 18 The thickness of the CuO interfacial layer was measured at 15 nm. J/V measurements were performed using a solar simulator from Newport equipped with a Xenon lamp at 100 mW/cm 2 providing an AM 1.5 G spectrum and a scan rate of 0.1V/s. Impedance measurements were performed using a Metrohm Autolab PGSTAT 302N. A red LED (at 625 nm) was used as the light source. A small AC perturbation of 20 mV was applied to the devices and the different current output was measured throughout a frequency range of 1MHz-1Hz. The steady state DC bias was kept at 0 V throughout the impedance spectroscopy experiments.   Table I, the best performing device in terms of PCE was the CuO-based device followed by Cu:NiOx-based device and then the PEDOT:PSS-based device. The origin of the PCE improvement between the three p-i-n solar cells could be attributed in two main factors: a) The lower work function of CuO (5.6 eV) 19 and Cu:NiOx (5.3 6 eV) 20 compared to PEDOT:PSS (4.9 eV) 21 , offering a more favorable energy level alignment with the perovskite (Ec= 3.9 eV, Ev=5.4 eV) 22 and thus more efficient hole extraction, leading to increased Voc and Jsc for the corresponding solar cells, b) the increased grain size of the CH3NH3PbI3 when fabricated on Cu:NiOx and CuO-based devices compared to PEDOT:PSSbased devices (see Fig.1). The increased grain size (and thus the reduced grain boundaries area) potentially leads to decreased carrier recombination due to decreased charge trap densities as reported elsewhere. 12,23 The photovoltaic parameters as well as fitted parameters from impedance spectroscopy are shown in Table I. To further understand the effect of the HTL in charge accumulation of p-i-n perovskite solar cells, impedance spectroscopy has been used. Fig. 2b shows three representative Nyquist plots of the devices as well as their simulated fits from the ECMs. It can be clearly seen that the plots follow the general shape of two frequency responses for perovskite based solar cells. 24,25 The feature at high frequencies has been previously attributed to the charge transport resistance  27 This inductive loop is a feature previously observed in battery systems attributed to the constant formation of the solid electrolyte interphase products, which involves ionic movement 28 and has also been previously observed in n-i-p perovskite solar cell at intermediate frequencies 26 . The absence of this feature was previously believed to be a sign of charge accumulation at the device interfaces, 26  Since the devices with Cu:NiOx and CuO exhibit reduced grain boundaries area, a possible explanation can be that there are less pathways for ion migration in the device. It is beyond the scope of this paper to study the exact origin of the inductive loop. However, the absence of this feature from our best performing Cu:NiOx and CuO devices leads us to believe that a potential explanation could lie in the ionic nature of perovskite solar cells using inorganic HTLs.
Because of the different shape of the Nquist plots we were unable to use a universal ECM for the three devices under investigation. Instead, two different ECMs have been used, which are 8 shown in Fig. 2c and 2d. The blueprint for the ECM used was based on the circuit model described from Bag et.al. 21 Fig. 2c presents the circuit that was used to model PEDOT:PSSbased devices, whereas Fig. 2d shows the circuit that was used to model Cu:NiOx and CuObased devices. Both models share a variety of similar elements, however they also have distinct differences. Rs denotes the series resistance from external circuitry (from wires and equipment).  Table I. From  Fig.2b as well as the value extracted from the ECM, an increase to recombination resistance (Rrec) for Cu:NiOxbased (11000 Ω) and CuO-based (25000 Ω) devices compared with PEDOT:PSS-based device (9000 Ω) was observed. The increase in Rrec is directly related to our previous observations. The progressively increased grain size and reduced grain boundaries area between the three different under layers indicate that less charges are accumulating in the grain boundaries of the devices, resulting in decreased carrier recombination. Furthermore, due to the more favorable energy level alignments between the three HTLs, previously discussed, the extraction of holes is energetically favored and thus decreased interfacial charge accumulation occurs, resulting in a decrease in carrier recombination in the HTL/Pvsk interface. This is in accordance with the increase in Rrec between the three devices, which denotes less frequent recombination events.
Furthermore, we observe that the double layer capacitance (Cdl) is decreased for CuObased device (3.5x10 -7 F/cm 2 ) compared with Cu:NiOx-based device (1.5 x10 -5 F/cm 2 ) and PEDOT:PSS-based device (2.8x10 -5 F/cm 2 ). Cdl arises from charge accumulation at the interfaces from both ionic and electronic accumulation, although under light, Cdl is predominantly affected by the accumulation of photo-generated carriers. 26 Since Cdl is tied to charge accumulation at the interfaces, having a small value is favorable for increased device performance. The reduction in the value of Cdl that we observe between the three devices is directly correlated to our previous observations regarding the decrease in charge accumulation at the HTL/Pvsk interface. Both observations regarding the decreased charge accumulation at the grain boundaries area as well as the HTL/Pvsk interface have a direct impact to the increased Voc and Jsc of the corresponding solar cell devices.
Apart from the above observations, to be able to obtain good fits for the CuO based device the Cdl had to be changed from an ideal capacitor to a constant phase element (CPE).
CPEs are usually used to model the behavior of a non-ideal capacitor. This could mean that the capacitance of a layer or interface is not constant and varies between regions of the layer. This To better understand the charge accumulation mechanisms in p-i-n perovskite solar cells capacitance-frequency (C-F) measurements are shown in Fig. 3.

FIG.3. Capacitance versus frequency (C-F) plot
The C-F plot shows the data representation in terms of capacitance instead of impedance. Fig. 3 shows the C-F plot of the three representative devices. From the C-F plot it is observed that there are two distinct capacitive plateaus for both devices, one at the 0.1 kHz range and another at 100 kHz range. The plateau at 100 kHz range is labeled Cbulk whereas the plateau at 0.1 kHz is labelled Cel. Cbulk is generally attributed to the capacitance arising from the various polarization effects of the perovskite photoactive layer (e.g. octahedra reorientation and ionic defects). 26 The second capacitive plateau, Cel was attributed to electronic and ionic accumulation to the interfaces of devices by previous reports. 26 As it was previously stated, Cel is predominantly affected by the photogenerated charge accumulation at the device interfaces.
From the above figure it can be seen that Cel plateau is much lower for Cu:NiOx and even lower for CuO-based device compared to PEDOT:PSS-based device. This indicates decreased charge accumulation at the HTL/Pvsk interface and subsequently related to the increase in Voc and Jsc.
The above observations are also in accordance with the increase in Rrec observed before.