Carbon-based perovskite solar cells with electron and hole-transporting/-blocking layers

Towards commercialization of perovskite solar cells (PSCs), further reducing the cost and increasing the stability of PSCs have been the most important tasks of researchers, as the efficiency of single-junction PSCs has reached a competitive level among all kinds of single-junction solar cells. Carbon-electrode-based PSCs (CPSCs), as one of the most promising constructions for achieving stable economical PSCs, now attract enormous attention for their cost-effectiveness and stability. Here, we briefly review the development of CPSCs and reveal the importance of n-i-p architecture for state-of-the-art CPSCs. However, despite their promising potential, challenges still exist in CPSCs in the n-i-p architecture, which mainly stem from the incompact contact of the hole-transporting layer (HTL)/carbon electrode. Thus, new carbon materials and/or novel manufacturing methods should be proposed. In addition, HTL is yet to be appropriate for state-of-the-art CPSCs because the fabrication of carbon electrode could result in the destruction of the underlayer. To further enhance the performance of CPSCs, both the HTL and electron transport layer as well as their interfaces with perovskite active layer need to be improved. We recommend that the perovskite active layer, with its long carrier lifetime, strong carrier transport capability, and long-term stability, is necessary as well for improved performance of CPSCs. We also highlight current researches on CPSCs and provide a systematic review of various types of regulation tools.


Introduction
Perovskite solar cell (PSC) has been one of the most promising photovoltaic technologies because of its low-cost largescale manufacturing process and inspiring photovoltaic performance, with power conversion efficiency (PCE) over 25% [1][2][3][4]. Towards commercialization of PSCs, further reducing the cost and increasing the stability are necessary. Due to the high chemical reactiveness of the metal halide perovskite, even the inert noble metal electrodes, such as Au and Ag, cannot survive [5,6]. Carbon electrodes, one of the most promising substitutions, now attract more and more attention (figure 1(a)) for their low cost and potential stability. They also fit in well with the manufacturing of large-scale and flexible PSCs, which commercialization asks for. Rapid progress of carbon-electrodebased PSC (CPSC) has been achieved in recent years, as shown in figure 1(b).
In 1996, Kay and Grätzel first reported a dye-sensitized solar cell with a promising PCE of 6.7%, using graphite/ carbon black composite pastes [7]. As for the PSC applications, in 2013, Han et al first reported a full printable processed PSC with carbon electrode (figure 1(c) (i)) [8]. Since then, CPSCs became one of the focus tasks in PSC research field. Before 2016, all researchers focused on holetransporting-layer (HTL)-free devices (figure 1(c) (ii)), as the solvent of carbon pastes can dissolve the commonly used HTL 2,2 ′ ,7,7 ′ -tetrakis[N, N-di(4-methoxyphenyl)amino]-9,9 ′ -spirobifluorene (Spiro-OMeTAD), and perovskite can transport holes well itself [9]. However, the efficiencies of CPSCs based on this HTL-free architecture were restricted for inefficient charge collecting by interfacial energy barriers, and thus open-circuit voltage V OC loss. Besides, the commercialization of CPSCs with this architecture suffered from the high-temperature process of mesoporous TiO 2 fabrication.
In 2016, researchers published complete n-i-p architectures by using alternative HTL (figure 1(c) (iii)), such as copper(II) phthalocyanine (CuPc) [10]. Since the mismatch of energy level and the poor contact at 6,6-phenyl C 61 -butyric acid methyl ester (PCBM)/carbon-electrode interface, it had long been a challenge for achieving CPSC with p-i-n architectures (figure 1(c) (iv)). In 2017, Jeon et al achieved the p-i-n architecture by mechanically transferring the prefabricated carbon nanotubes as electrodes [11]. We propose the n-i-p architecture for state-of-the-art CPSCs to achieve appropriate energy level of carbon electrode for hole extraction and transport.
In recent years, CPSCs have made great progress, with PCE as high as 22.07% [12], approaching that of metal electrode PSCs and possessing excellent operational stability. In addition to the most commonly used commercial carbon pastes, a variety of carbon-electrode materials have been developed over the past few years for CPSCs, including carbon nanotubes [12], carbon black [13,14], and graphene [15]. The emergence of these new carbon electrodes has also liberated the use of HTL commonly used in high-efficiency metal PSCs, such as Spiro-OMeTAD, which has led to a major advancement in the PCE of CPSCs. However, the high cost and complicated process of special carbon-electrode materials and the poor stability of the traditional HTL Spiro-OMeTAD are not fully compatible with the original intention of developing CPSCs that are low cost, stable, and suitable for large-scale commercial manufacturing, while there are still gaps in the PCE of CPSCs fabricated by traditional inexpensive carbon paste. Challenges come from the defects of each functional layer, as well as the delays and obstacles of charge transport and transfer kinetics due to the intrinsic poor conductivity of carbon electrodes and the interfacial defects. Therefore, optimization at the perovskite layer, transporting layer/blocking layer, electrode, and interfaces is essential for the improvement of the overall device performance.
Herein, we highlight the current researches on CPSCs and provide a systematic review of various types of regulation tools. As a multipart device, the properties of each part of CPSCs have their own impact on the overall device performance and are different from those of PSCs with conventional metal electrodes. Therefore, in combining the unique structural features of CPSCs, we review research progress on the following aspects of CPSCs: modulation of electron transport layers (ETLs) and HTLs, interfacial modulation, modulation of perovskite layers, and modulation of carbon-electrode materials. This review aims to systematically identify the optimization strategies for each part of CPSC devices and to promote new work to enhance the PCE of CPSCs and make them more advantageous for large-scale commercial manufacturing.

Transporting layer design
The charge transporting layer is a vital component of CPSCs with n-i-p architecture, and the performance of the ETL/hole blocking layer (HBL) and HTL/electron blocking layer (EBL) determines the efficiency of selective charge extraction and transport (tables 1 and 2).

ETL/HBL for CPSCs
In the CPSCs of the n-i-p architecture, due to excellent energylevel alignment and electron mobility, TiO 2 and SnO 2 are often considered as the suitable ETLs [9,[16][17][18][19]. Many researchers focused on optimizing the performance of the ETL to improve charge separation and transport to enhance the relatively low PCE of CPSCs. Liu et al first demonstrated CPSCs using Nidoped rutile TiO 2 as ETL [20]. They found that Ni doping can shift up the Fermi level of the ETL and increase the charge mobility of the TiO 2 , therefore enhancing the charge transport and extraction. Ultraviolet photoelectron spectroscopy (UPS), displayed in figure 2(a), reveals the impact of Ni doping on the TiO 2 electronic structure. The distances between the valence band (VB) and the Fermi level are about 3.40 eV and 3.38 eV for the pristine TiO 2 and Ni:TiO 2 , respectively. The upward shift of the Fermi level promotes extraction of the charge. As a result, they achieved a high PCE of 17.46% for the CPSCs. Liu et al reported a low-temperature (70 • C) solution-processed Mg-doped rutile TiO 2 as ETL in CPSCs (figure 2(b)) [21]. The efficient charge extraction, better electrical conductivity, and suppressed charge recombination of Mg-doped TiO 2 resulted in a higher PCE of 15.73% compared to the PCE of pristine TiO 2 -based device.
However, the results of studies conducted by several authors have shown that SnO 2 could be a better candidate for ETL because of its high mobility, good stability, and low photoactivity [22][23][24]. Replacing TiO 2 with SnO 2 helps to improve the PCE of CPSCs. Besides, several reports have demonstrated that rational doping of SnO 2 can improve the performance of CPSCs. Ye et al first reported lowtemperature, processed, Zn-doped SnO 2 as an effective ETL in CPSCs with CuPc HTL [25]. They found that Zn doping contributes to a more suitable energy-level alignment (figure 2(f)) and an improved conductivity of SnO 2 films. More importantly, Zn doping elevated the Fermi level of the ETL, contributing to a more powerful built-in electric field in the device and a wider depletion region. For this reason, the electron transfer and extraction are improved, and the charge recombination is suppressed. Finally, CPSCs with Zn:SnO 2 ETL achieve a champion PCE of 17.78%, V OC of 1.098 V, short-circuit current density J SC of 23.4 mA cm −2 , and fill factor (FF) of 0.692. It is worth noting that the PCEs of CPSCs using SnO 2 ETL are mostly higher than TiO 2 -based CPSCs; thus, there is a trend to gradually replace TiO 2 by SnO 2 in CPSCs. Besides, Khambunkoed et al achieved fully covered and compact ZnO ETL via a homemade slot-die coater setup on a three-axis CNC platform. With the optimized ZnO ETL, CPSCs offered the highest PCE of 10.81% [26].
Another effective way to build a high-performance ETL is to use a two-layer structure. In the all-inorganic system, Liu et al achieved the highest PCE for CsPbBr 3 CPSCs by  [20]. Reprinted from [20], Copyright (2018), with permission from Elsevier. (b) Mg-doped rutile TiO 2 as ETL in CPSCs [21]. Reprinted from [21], Copyright (2018), with permission from Elsevier. (c) SEM image of the device with TiO 2 /SnO 2 ETL [27]. Reprinted from [27], Copyright (2019), with permission from Elsevier. (d) The energy-level arrangement of CPSCs with TiO 2 /SnO 2 bilayer [28]. Reproduced from [28] with permission from the Royal Society of Chemistry. (e) SEM image of the PCBM-coated CdS [29]. Reprinted from [29], Copyright (2019), with permission from Elsevier. (f) Energy-level alignment of CPSCs with Zn-doped SnO 2 ETL [25]. Reprinted from [25], Copyright (2019), with permission from Elsevier. applying SnO 2 to passivate the compact TiO 2 (c-TiO 2 ) ETL [27]. As shown in the scanning electron microscope (SEM) image in figure 2(c), the TiO 2 /SnO 2 bilayered ETL possesses an excellent electron extraction capability and contributes to the charge transport and suppression of the interfacial trapassisted recombination. The SnO 2 layer can not only improve the surface morphology of the ETL but also reduce the current shunting pathways in the c-TiO 2 . They found fewer trap states in the CsPbBr 3 film deposited on the SnO 2 -passivated TiO 2 film via photoluminescence (PL) spectroscopy measurements. As a result, the champion PCE of the as-prepared CsPbBr 3 CPSCs is promoted from 7.05% to 8.79%. Liu et al reported a TiO 2 /SnO 2 bilayer as ETL for CPSCs, yielding a high PCE of 15.39% and excellent stability over 1200 h [28]. The TiO 2 /SnO 2 bilayer is beneficial for electron transport, thus resulting in a suppressed charge recombination and appropriate energy-level arrangement (figure 2(d)). In addition, devices with TiO 2 /SnO 2 bilayer had a wider depletion region in favor of enhanced photovoltaic performance.  [40]. Reprinted from [40], Copyright (2020), with permission from Elsevier. (e) Illustration of CPSCs with P3HT HTL and P3HT/graphene composited HTL, respectively [38]. Reprinted from [38], Copyright (2019), with permission from Elsevier.  [41] accumulation and recombination, which results in a lower PCE compared to PSCs with metal electrodes. The appropriate HTL could efficiently improve hole extraction to carbon electrodes [30]. Most of the HTLs commonly used in CPSCs are organic materials, such as poly(3-hexylthiophene-2,5-diyl) (P3HT) and CuPc, for low-cost solution processing and commercial production. Jin et al reported a typical p-type P3HT as HTL for CPSCs with ZnO ETL, and the energy-level alignment of P3HT-based PSCs is shown in figure 3(a) [31]. The highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) levels of P3HT are −5.1 eV and −3.0 eV, respectively. The LUMO level of P3HT is higher than the CB of CH 3 NH 3 PbI 3 (−3.93 eV), and the HOMO level is lower than the Fermi level of carbon (−5.0 eV) and higher than the VB of CH 3 NH 3 PbI 3 (−5.43 eV). The appropriate arrangement of energy levels could block the electron in perovskite and promote the hole extraction to carbon. Therefore, the appropriate P3HT HTL suppressed the carrier recombination and achieved PCE as high as 16.05%. Zhao et al reported a novel strategy that uses P3HT as HTL for lead-free CPSCs [32]. They found that P3HT has a suitable energy level and high hole-transport rate with low carrier non-radiative recombination. As a result, the Cs 3 Sb 2 I 9−x Cl x solar cell achieved a higher PCE of 1.67%. Liu et al applied CuPc as HTL for CPSCs (figure 3(b)) [33]. The CB offset between the CsPbBr 3 layer and the CuPc layer provides an energy barrier that prevents electrons from flowing to the CuPc layer, whereas the VB offset provides an additional driving force for the flow of holes to the CuPc layer. Moreover, depositing thin CuPc can also passivate defects on the surface of CsPbBr 3 as well as induce a large interfacial area of CuPc, correspondingly favoring the hole transported from CuPc to carbon. Finally, the optimal device acquired a decent PCE of 6.21%, over 60% higher than those of the HTL-free devices. Zhang et al applied CuPc nanorods as HTL for printable low-temperature processed CPSCs [10]. The strong π-π stacking between layered CuPc molecules favors the formation of high carrier mobility [34]. In addition, depositing thin CuPc contributes to a good contact with counter electrode, which favors the hole transported from CuPc to carbon. They achieved an impressive PCE of 16.1%, which is the highest PCE for the CPSCs at that time.
In addition, quantum dots and inorganic materials can also be used as efficient HTL. Duan et al demonstrated alkylchain-regulated quantum dots as HTL to suppress charge recombination [35]. They precisely controlled alkyl-chain length of ligands to maximize charge extraction by balancing the surface dipole-induced charge coulomb repulsive force and quantum tunneling distance. Finally, the inorganic CsPbBr 3 CPSC achieves a champion PCE up to 10.85%. Duan et al reported p-type Cu(Cr, M)O 2 (where M = Ba 2+ , Ca 2+ , or Ni 2+ ) nanocrystals with improved hole-transport properties by increasing interstitial oxygen to effectively extract holes from perovskite [36]. The larger ions induced lattice expansion allows more interstitial oxygen atoms to be inserted into the host CuCrO 2 lattice, contributing to improved hole conductivity. Therefore, inorganic Cu(Cr, M)O 2 HTL conduces to enhancing hole extraction from perovskite and reducing carrier recombination. Besides, the energy-level distribution is compatible with that of CsPbBr 3 perovskite layer and carbon electrode, as shown in figure 3(c). The optimized CsPbBr 3 solar cell with Cu(Cr, Ba)O 2 HTL achieves a champion PCE as high as 10.79%.
Constructing a double-layer HTL is also an effective method to improve the PCE of CPSCs. Yang et al reported that the PCE of CPSCs was increased up to 18.1% by employing a double layer of P3HT/ tantalum-doped tungsten oxide (Ta-WO X ) as the HTL [37]. They prepared P3HT/Ta-WO X using a blade-coating process, which solves the problem of the shallow HOMO level of P3HT resulting in a lower V OC and strong electronic coupling between the flat P3HT molecules and perovskites. By incorporating P3HT into CPSCs, the V OC of CPSCs was greatly improved to 1.12 V. Chu et al developed a low-temperature solutionprocessed P3HT/graphene composite HTL, which exhibits outstanding charge mobility and thermal tolerance, with high hole mobility, as shown in figure 3(e) [38]. By SEM, HTL films appear homogeneous, indicating that the graphene is beneficial for promoting hole transport to the carbon electrode. The PL decay of the P3HT/graphene samples (43 ns) is slightly faster than the pure P3HT samples (55 ns), indicating improved hole extraction at the perovskite/HTL interface. As a result, the P3HT/graphene composite HTL device yielded an enhanced PCE of 18.2% compared to the pure P3HT device with a PCE of 11.1%. Liu et al reported a P3HT/zinc phthalocyanine (ZnPc) composite with a tunable energy level, which was employed as the HTL in CsPbBr 3 CPSCs [39]. P3HT/ZnPc HTL is in favor of reducing the energy-level difference and passivating the surface defects of perovskite, promoting charge separation and reducing charge recombination at CsPbBr 3 /carbon interface. As a result, the authors achieved a champion PCE of 10.03%, which suggests that P3HT/ZnPc composite is an efficient HTL for high-efficiency CsPbBr 3 CPSCs. Zong et al reported an MoO 2 /N-doped carbon nanosphere (NC) composite with high hole mobility and matched energy level, which is prepared by a facile one-step pyrolysis process and introduced into all-inorganic CsPbBr 3 CPSCs as the HTL [40]. The MoO 2 /NC HTL not only enhances energylevel alignment, interface contact, and charge extraction, but it also passivates CsPbBr 3 surface defects, thus leading to the reduction of energy loss and charge recombination. As shown in figure 3(d), all the work functions (WFs) of MoO 2 /NC-x composites are between the VB of CsPbBr 3 and the WF of carbon electrode, indicating that the introduction of MoO 2 /NC composited HTL can reduce the interface energy offset and therefore energy loss. Finally, the authors achieved a maximum PCE of 9.40%, which is much higher than those of HTLfree devices. Mashhoun et al achieved a PCE of 12.31% for CPSCs utilizing doped CuSCN as HTL [41]. They coated the nanoparticular Ta-WO X layer on top of CuSCN, contributing to well-matched energy levels between the VBs of CuSCN and CH 3 NH 3 PbI 3 and the high hole mobility, as well as improving the hole extraction property of HTL. As a result, the PCE was improved from 8.59% to 12.31%. In addition, the Ta-WO X layer can protect the HTL interface from the negative impact of the carbon ink's solvent.
In conclusion, the HTLs employed in CPSCs are mainly p-type wide-bandgap semiconductors. From previous reports, P3HT and CuPc are the most commonly used HTLs with good stability, high conductivity, and energy-level matching.

Interface engineering for perovskite active layer
The interfaces (buried and upper interfaces) in CPSCs of the n-i-p architecture largely influence the charge transfer and recombination behavior. In general, carbon electrodes obtained from common commercial carbon paste do not have the same transport capability as metal electrodes. In addition, due to the special semiconductor nature of all-inorganic perovskite, the photogenerated carrier separation generally occurs in the interface of ETL and the perovskite layer, which, together with the certain hole-transport capability of carbon electrodes, makes all-inorganic CPSCs usually omit the HTL. As a result, the ETL/perovskite interface of the CPSC suffers from severe carrier non-radiative recombination, and the charge separation and transfer are hindered. In the case of organic-inorganic hybrid perovskite, charge extraction and transport at both the upper and buried interfaces are critical; otherwise, it will lead to carrier transport imbalance, and thus carrier accumulation and quenching at the interface. In addition, the introduction of interfacial modification layers not only helps to enhance carrier transport, reduce non-radiative recombination, and improve the overall device performance, but it also helps to attenuate ion migration and damage to the lower layer caused by carbon paste or HTL.
Zhu et al modified the TiO 2 as well as perovskite films simultaneously by introducing ammonium chloride (NH 4 Cl) into the interface, which not only improved the contact and energy-level matching at the TiO 2 /perovskite interface but also provided high-quality perovskite films with larger grain size by modulating the crystallization kinetics of CsPbBr 3 through ion binding and passivation, resulting in a significant enhancement of charge extraction and achieving up to 10.12% of PCE. In addition, the optimized device, without any encapsulation,  [47]. Reproduced from [47] with permission from the Royal Society of Chemistry. (b) Time-resolved photoluminescence spectra of the perovskite deposited on the different ETLs [14]. Reprinted from [14], with the permission of AIP Publishing. (c) Energy band diagram of the device and the organic silane SAM between the TiO 2 surface and the perovskite [51]. Reprinted with permission from [51]. Copyright In addition, the long-term stability of the unencapsulated PSCs exceeded 1400 h at 80% RH [53].
Similar to the buried interface, the main problems to be solved for the upper interfaces of CPSCs are poor contact due to the rough surface of the perovskite and carbon electrodes, poor hole extraction and transport ability, severe surface defects resulting in non-radiative recombination at the interface, and energy-level mismatch between the layers. As a result, researchers have used organic polymers [54][55][56][57], organic salts [58][59][60][61][62][63][64][65][66], inorganic materials [67,68], and carbon nanotubes [12,69,70] for interface engineering to improve the quality of the interface, enhance the film crystallinity quality of perovskite, passivate interfacial defects, improve the interfacial energy band alignment, enhance the interfacial contact, build a hole transfer channel and act as an EBL to reduce the carrier non-radiative recombination, and accelerate the hole extraction and transfer at the interface.
By introducing a functional polymethyl methacrylate (PMMA) layer on the surface of the all-inorganic leadfree Cs 2 AgBiBr 6 double perovskite, Li et al avoided direct contact between the carbon and the underlying chargetransfer layer and passivated the surface defects, inhibiting the non-radiative carrier recombination from poor film quality and accelerating the interfacial charge extraction. In addition, the unencapsulated devices maintain nearly 100% of their initial PCE after 80 d of storage at 25 • C and 5% RH or 60 d of storage at 85 • C and 0% RH due to the PMMA layer's resistance to corrosion by water and oxygen [55]. Ding et al inserted a carbonyl-modified polyvinyl acetate (PVAc) polymer as an interfacial modification at the CsPbBr 3 /carbon interface, which passivated the surface defects of perovskite and improved the energy-level alignment between the VB of CsPbBr 3 and the WF of carbon, inhibiting carrier recombination and accelerating charge separation. A graphene oxide (GO) layer is also introduced to further promote hole extraction and reduce the energy-level mismatch. The FTO/c-TiO 2 /m-TiO 2 /CsPbBr 3 /PVAc/GO/Carbonbased PSC achieved a champion PCE of up to 9.53%, with a 44.0% improvement compared to the control device, and significant long-term stability under high humidity, high temperature, and continuous illumination in air environment [56]. Wu et al introduced a thin layer of poly(ethylene oxide) (PEO) to modify the perovskite/carbon interface to achieve an improved interfacial energy-level alignment for efficient charge transfer, and they obtained an increase in PCE from 12.2% to 14.9% with good stability under double-85 aging conditions without encapsulation (figure 4(d)) [57]. Liu et al used MABr to in situ grow a methylammonium lead mixed halide (MAPbI x Br3x ) perovskite layer on the top of an MAPbI 3 perovskite layer to form an MAPbI 3 /MAPbI x Br3x perovskite stacking structure, which formed a favorable interfacial energy-level alignment and reduced the carrier recombination at the perovskite/carbonelectrode interface. An optimized device with a PCE of 16.2% was obtained, and good storage stability was also exhibited under continuous light and dark conditions (figure 4(e)) [60].
Lee et al post-treated a carbon electrode with phenylethylammonium iodide (PEAI) to grow an ultrathin 2D perovskite PEA 2 PbI 4 layer at the interface between the 3D perovskite and carbon layers, which improved the poor perovskite/carbon contact and greatly inhibited the interfacial charge recombination, increasing the average PCE of the device from 11.5% to 14.5%, reducing the hysteresis loss, achieving a maximum PCE of 15.6%, and showing good environmental stability and thermal stability [62]. Similarly, Zouhair et al introduced a 2D perovskite passivation layer (OA) 2 PbI 4 as an EBL at the perovskite/carbon interface to significantly reduce the interfacial recombination loss, thereby increasing the FF and V OC , obtaining a high efficiency of 18.5%, and greatly improving device stability [63]. Wang et al constructed an in situ perovskite transition layer (CsFA) PbI3x Br x film with an intermediate energy level between CsPbI 2.2 Br 0.8 and carbon electrode by ion exchange through a simple formamidinium iodide (FAI) post-treatment strategy to extend the light absorption edge of CsPbI 2.2 Br 0.8 from 657 nm to 680 nm. In addition, the transition layer can also serve as a hole transfer channel between CsPbI 2.2 Br 0.8 and the carbon electrode due to its suitable intermediate energy level and effective defect passivation. As a result, the optimized CPSC achieves a champion PCE of 15.03% and an ultrahigh FF of 0.81, and the stability of the device is also improved (figure 4(f)) [64].
Wang et al used CsCl for surface treatment, and the CsCl residue tended to react with CsPbI 3 to generate 2D Cs 2 PbI 2 Cl 2 nanosheets as an EBL. The synergistic effect of the suitable energy-level gradient and the EBL well improved the hole selectivity at the CsPbI 3 /carbon interface to reduce the carrier recombination loss, achieving an efficiency of 15.23% [67]. Similarly, Han et al used a 2D perovskite Cs 2 PtI 6 to modify the surface of CsPbI 2 Br to adjust the energy-level matching and suppress carrier recombination, and the excellent hydrophobic property of Cs 2 PtI 6 retarded the degradation of the perovskite, and the optimized devices obtained higher PCE (13.69 %) and stability [68].
Wang et al used defective multi-walled carbon nanotube (D-MWCNT) to tune the charge transfer kinetics at the interface between the HTL and the carbon electrodes. Electrostatic dipole moment interaction between the terminal oxygen-containing group of D-MWCNT and the Spiro-OMeTAD allowed the establishment of interfacial coupling at the molecular level through edge  [14]. Reprinted from [14], with the permission of AIP Publishing. (b) Schematic illustration of the MAAc additive strategy with perovskite crystallization processes for the preparation of CsPbI 2 Br inorganic perovskite films through two-step deposition [84]. Reprinted with permission from [84]. Copyright (2021) American Chemical Society. (c) Energy-level diagram of device with Cs 3 Sb 2 ClxI9−x/Rb 0.15 Cs 2.85 Sb 2 ClxI 9−x as absorber layer and TiO 2 /Nb 2 O 5 as ETL [89]. Reprinted from [89], Copyright (2021), with permission from Elsevier. (d) High-resolution cross-sectional SEM images of the whole devices with a configuration of FTO/c-TiO 2 /Li-CsPbIBr 2 /CuPc/carbon and device energy-level diagram [90]. Reprinted from [90], Copyright (2020), with permission from Elsevier. effect-induced electron redistribution and 1D hyper-channels to achieve fast charge transfer. Meanwhile, the seamless connection between the HTL and the carbon electrodes was promoted due to the induced interfacial coupling between D-MWCNT and graphene at the nanometer scale. Based on this strategy, a high PCE of 22.07% (certified PCE of 21.9%) and excellent operational stability were achieved (figure 4(g)) [12].

Design of perovskite active layer
Towards the optimization of the perovskite layer in CPSCs, we need to take the special structure and properties of the CPSCs into account, in addition to the similar approach to obtain high-quality films for PCE enhancement as in metal PSCs. In systems where the carrier transport capability of the electrode and transporting layer has drawbacks, we need to work on obtaining high-quality thin films of the perovskite layer to obtain good interfacial contact, long carrier lifetime, and strong carrier transport capability. In addition to the incompatibility caused by the mutual destruction of the conventional HTL Spiro-OMeTAD and commercial carbon pastes [71], the instability of the devices caused by the commonly used HTL dopants, and the lack of transport capability of undoped HTL, it is necessary for us to precisely tune the energy band structure of the perovskite surface to obtain a suitable energylevel match for CPSCs, combing with the development of new efficient transport materials. CPSCs are very promising for future commercialization due to their low cost and simple fabrication process, which requires us to explore production processes that can be manufactured in air environment on a large scale, at low cost, and with low energy consumption. Therefore, researchers usually use composition engineering [72][73][74], solvent engineering [75][76][77][78], additive engineering [14,[79][80][81][82], intermediate-phase engineering [50,83,84], and optimization of film deposition methods [27,[85][86][87][88] to obtain high-quality perovskite films (table 4).
Zou et al used 2,3,5,6-tetrafluoro-7,7,8,8-tetracyanodi methane (F4TCNQ) as an additive in the perovskite precursor solution, which not only acts as passivation of defects and assists in crystallization of the film, but it also modulates to obtain more favorable interfacial energy band bending and energy-level matching while forming charge transfer complexes with perovskite due to its strong electronwithdrawing ability. Having all these features simultaneously allows the CPSCs added with F4TCNQ to obtain high-quality low-defect-density films with suppressed nonradiative recombination within the layers and at the interfaces, as well as extremely fast carrier separation and extraction capabilities, and the optimized CPSCs obtain an enhanced PCE of 15.1% and high V OC of 1.07 V with long-term  [91] stability ( figure 5(a)) [14]. Using additive engineering, Ullah et al fabricated polyvinylpyrrolidone (PVP) polymer-coated CsPbI 2 Br stable phases at low temperatures (120 • C) to reduce trap states in perovskite films, thereby accelerating charge carrier separation and inhibiting carrier recombination. The optimized devices achieved a champion PCE of 10.47% [82].
Li et al innovatively proposed a strategy for two-step sequential deposition of methylamine acetate (MAAc) additive for air made of high-quality CsPbI 2 Br films. By introducing MAAc into the CsX precursor solution, PbX 2 in the films was first converted to the intermediate-phase MA 2 Pb 3 X 8 , which could then be completely converted to δ-CsPbI 2 Br. Ethylene glycol (EG) was added to the CsX precursor solution to improve the homogeneity of the CsPbI 2 Br films. The carbon-based CsPbI 2 Br planar PSCs fabricated under an ambient atmosphere of 15% RH can achieve a champion PCE of 11.2% and maintain an initial efficiency value of 79.3% after aging under 15% RH at room temperature for 360 h ( figure 5(b)) [84].
Guo et al manipulated the crystallization and improved the film quality of Cs 3 Sb 2 Cl x I 9-x by partially exchanging the Asite inorganic cation with Rb or K, inhibiting the 0D phase generation and reducing the pinholes in the film to obtain 2.46% PCE of CPSCs (figure 5(c)) [89]. Similarly, Tan et al obtained highly crystalline and well-oriented CsPbIBr 2 crystals with higher film coverage on the substrate, larger grains, and fewer grain boundaries through Li doping. The trap state density in the CsPbIBr 2 films was also effectively mitigated, leading to longer carrier lifetime and reduced energy losses. A PCE of 9.25% was achieved, along with excellent air and thermal stability (figure 5(d)) [90].
Liu et al developed a facile multistep spin-coating strategy to deposit CsPbBr 3 films with higher phase purity and larger average grain sizes (1 µm) by tuning the number of CsBr spin-coating cycles. The enhanced film crystallinity and light-harvesting ability, as well as the reduction of trap states, enabled the planar CsPbBr 3 CPSC to achieve a champion PCE of 8.79% with good humidity and thermal stability [27]. Liu et al developed a novel vapor-assisted deposition strategy. PbBr 2 thin films were fabricated by an anti-solventwashing technique, followed by vacuum evaporation of CsI onto the PbBr 2 layer. By precisely tuning the thickness of the CsI films, highly phase-pure and crystallized CsPbIBr 2 films were successfully obtained, exhibiting uniform morphology and comprehensive coverage of the substrate with grain sizes as large as micrometer scale and ultrahigh light absorption capability. The corresponding CPSCs achieved a champion PCE of 8.76% and an excellent V OC of 1.289 V. The unencapsulated devices exhibited good moisture and thermal stability [91].

Carbon-electrode exploration
Many carbon materials have been applied in electronics for decades since fullerene was first fabricated in 1985 [92]. Among these options, graphite and carbon black are two alternatives of electrodes in commercial solar cells for their excellent conductivity and low cost [7]. Because of the suitable WF of 5.0 eV, solution-processible property, and material stability of these electrodes, they possess compatibility for commercial large-scale PSC manufacturing.
Because the key efficiency and stability problems, as well as their solutions in CPSCs, have been discussed in the previous sections, only the modification and innovation of carbon electrodes are discussed here. As for the PSC applications with n-i-p architectures, carbon pastes should fit in the under HTL, including energy level and solution compatibility. In consideration of large-scale manufacturing and commercialization, low-temperature-process compatible carbon-electrode materials are preferred.
As for the n-i-p device architecture, with the commonly used Spiro-OMeTAD as HTL, the challenge is on how to avoid the destruction of Spiro-OMeTAD by the solvent of carbon pastes. Zhang et al developed a sort of self-adhesive macroporous carbon electrode by a room-temperature solvent-exchange method (figures 6(a) and (b)) [13]. They bladed carbon pastes on a glass substrate and then soaked it into ethanol to remove residual solvent. After drying, the carbon electrode was peeled off from the substrate and directly adhered to the surface of Spiro-OMeTAD. Finally, they got a PCE up to 19.2%, with J SC of 23.33 mA cm −2 , V OC of 1.08 V, and FF of 0.76. Because the commercial carbon pastes commonly contain detrimental solvents, such as chlorobenzene and xylene, replacement of HTL is necessary, as discussed previously. Recently, using similar methods, they achieved a remarkable PCE of up to 20.04%, which is competitive with the corresponding  [95]. Reprinted with permission from [95]. Copyright (2018) American Chemical Society. (f and g) PCE and stability enhancement of Ti/rGO-based CPSCs [15]. Reproduced from [15], with permission from Springer Nature.
metal-based PSCs [93]. Furthermore, the as-prepared devices exhibited excellent long-term stability, which retained 94% of its initial performance after storage for 1000 h at room temperature and 25% RH without encapsulation.
The properties of carbon pastes determine the performance of carbon electrodes in solar cells. Behrouznejad et al investigated how the weight ratio of carbon black to graphite and type of binder affect sheet resistance and resistivity of carbon composite layer, based on CuIn 0.75 Ga 0.25 S 2 HTL (figures 6(c) and (d)) [94]. The conductivity of the carbon composite layer with PMMA (4 wt.%) as a binder is significantly increased compared with that of PS (4 wt.%). They also found that the resistivity of deposited carbon layer increases by increasing the ratio of carbon black to carbon black plus graphite from 30% to 40% and 50%, but the FF of the prepared PSC is increased. They achieved the best efficiency of 15.9% in the champion cell, with J SC of 23.86 mA cm −2 , V OC of 1.08 V, and FF of 0.62. This efficiency is close to the best efficiency of the reference cell with conventional Spiro-OMeTAD/gold hole collector (16.3%) in this research.
In 2021, Zhang et al reported the application of reduced GO (rGO) as the back electrode in C-PSCs (figures 6(f) and (g)) [15]. They further tuned the electronic properties of rGO by anchoring single titanium (Ti) adatoms on it in a well-defined Ti 1 O 4 -OH configuration. As a result, the series resistance of the carbon-based electrode was notably minimized. They achieved a steady-state PCE of up to 20.6% for CPSCs by combining with an advanced modular cell architecture, by stacking a semicell A (FTO/ETL/perovskite/ spiro-OMeTAD/graphene) and charge collector B (FTO/graphene). Furthermore, the devices without encapsulation retained 98% and 95% of their initial PCEs for 1300 h under 1 sun continuous illumination at 25 • C and 60 • C, respectively.
As for the p-i-n device architecture, PCBM is commonly used as ETM. Because of its specific characteristics, PCBM cannot be replaced easily. Therefore, doping strategies and/or modification of PCBM have been carried out to improve the stability [96,97]. However, challenge still exists when it comes to the fabrication of carbon electrodes on PCBM. Besides solvent destruction, energy-level alignment and interface compatibility are further two crucial problems. Zhou et al modified cross-stacked super-aligned carbon nanotube film on PCBM by employing polyethylenimine (PEI) (figure 6(e)) [95]. With a certain concentration of PEI (0.5 wt.%) doping, they achieved suitable energy-level alignment and promoted interfacial charge transfer, leading to a significant enhancement in the photovoltaic performance.
Finally, a maximum PCE of ∼11% was obtained, with J SC of 18.7 mA cm −2 , V OC of 0.95 V, and FF of 0.61. Moreover, the devices showed superior stability compared with Agelectrode-based devices.
Up to now, few researchers focus on the development of new carbon-electrode materials, which are efficient, cheap, and environment friendly, while a lot of effort goes to underlayers. More work should be done in this field.

Summary and perspectives
CPSC based on the n-i-p structure has good commercial potential due to its high PCE, good stability, simple fabrication process, and low cost. In this paper, research progress on n-i-p CPSCs has been reviewed from the aspects of transporting/blocking layer design, interface engineering, perovskite layer regulation, and carbon-electrode modification.
The carrier recombination problem caused by the incompatibility between the traditional HTL material Spiro-OMeTAD and the commercial carbon paste in CPSCs makes the improvement of electron and hole extraction and transport become the basis of improving the PCE of CPSCs. Specifically, the following points can be considered in n-i-p structure CPSC.
(i) As the condition that the traditional HTL side in CPSCs possesses poor transport capacity is hard to improve, optimization of the ETL side becomes essential. We can obtain the appropriate Fermi energy level of the transporting layer and the interface energy-level match, obtain higher carrier mobility, and better defect passivation effect through the selection and optimization of the ETL materials. Thus, we can get an excellent electron extraction and reduce the electron-hole recombination at the perovskite layer and interface. (ii) It is necessary to further explore HTL with great stability, high hole mobility, appropriate energy levels, and compatibility with the upper carbon electrode. Therefore, more innovative methods should be proposed, such as the development of a new HTL, the use of appropriate doping, the combination of double-layer HTL, and the introduction of interface protective layers. (iii) In addition to considering the use of ETL/HTL as the carrier transport medium, we can also consider introducing HBL/EBL at the corresponding position to block the unwanted carrier transport by the energy barrier to ensure the effective transport of another kind of carrier we want and reduce the carrier recombination inside the device. (iv) The interface between the upper and buried interfaces of perovskite and the transporting/blocking layer also has an important impact on carrier transport. Inserting an appropriate interface layer can play a very positive role, such as improving the crystallization quality of perovskite film, passivating interface defects, adjusting the arrangement of interfacial energy levels, introducing interfacial dipole layer, enhancing interfacial contact to inhibit the occurrence of non-radiative recombination, and establishing carrier transport channels to accelerate the extraction and transfer of interfacial charges. Up to now, additives, such as organic salts, inorganic compounds, metal oxides, ionic liquids, fullerenes, SAMs, and organic polymers, are widely used in the interface layer.
When optimizing the perovskite layer in CPSCs, we should also take into account the special structure and properties of CPSCs. Because the carrier transport capacity of the electrode and transporting layer has intrinsic drawbacks, we should focus on achieving high-quality perovskite film to obtain good interfacial contact, long carrier lifetime, and strong carrier transport capacity, making up for the drawbacks of the overall device. We must precisely adjust the energy band structure of the perovskite surface to obtain a special energy-level matching suitable for CPSCs. We need to focus on exploring the production process with large area, low cost, high stability, and low energy consumption, and can be fabricated in the air to match the characteristics and application intention of the lowcost and simple manufacturing process of CPSCs to contribute to future commercial application.
The PCE of CPSC has exceeded 22%; yet, a huge gap still exists compared with PSCs based on metal electrodes. The main causes of its low PCE is the intrinsic poor conductivity of the carbon electrode and the poor contact between the carbon electrode and the transporting layer. These drawbacks result in high contact resistance, which affects the transport and collection of carriers. We would like to highlight the following solutions to further enhance the performance of CPSCs: (a) develop new commercial carbon-electrode materials with high conductivity; (b) use strategies, such as interface modification, which may improve contact between the carbon electrode and the transporting layer; (c) achieve nondestructive contact between the carbon electrode and the transporting layer by hotpressing or spraying method; (d) pay attention to the orientation of the spatial stacking of each layer of materials to obtain close stacking and efficient interfacial charge transfer; and (e) further develop large-area carbon-electrode deposition methods, such as blading or screen printing, to fabricate commercial modules of CPSCs.
In addition, it is necessary to further study the longterm stability of CPSCs and its influencing factors, while the matching encapsulation technology also awaits further improvement.