Efficient Monolithic Perovskite/Silicon Tandem Photovoltaics

Tunable bandgaps make halide perovskites promising candidates for developing tandem solar cells (TSCs), a strategy to break the radiative limit of 33.7% for single‐junction solar cells. Combining perovskites with market‐dominant crystalline silicon (c‐Si) is particularly attractive; simple estimates based on the bandgap matching indicate that the efficiency limit in such tandem device is as high as 46%. However, state‐of‐the‐art perovskite/c‐Si TSCs only achieve an efficiency of ~32.5%, implying significant challenges and also rich opportunities. In this review, we start with the operating mechanism and efficiency limit of TSCs, followed by systematical discussions on wide‐bandgap perovskite front cells, interface selective contacts, and electrical interconnection layer, as well as photon management for highly efficient perovskite/c‐Si TSCs. We highlight the challenges in this field and provide our understanding of future research directions toward highly efficient and stable large‐scale wide‐bandgap perovskite front cells for the commercialization of perovskite/c‐Si TSCs.


Introduction
[3] While the efficiency of single-junction PSCs may improve further, increasing difficulties are expected when approaching a radiative limit. [4,5]Along with the efforts to increase the PCE of single-junction PSCs, strong efforts are dedicated to developing tandem solar cells (TSCs), where different parts of the solar spectrum can be better utilized and the radiative limit is no longer applicable. [6] unique advantage of PSCs is their bandgap tunability, making them promising candidates for high-efficiency TSCs.[16] This may be the most direct route to commercialize PSCs and one of the fastest routes to achieve efficiency improvements in commercial c-Si modules.
In 2015, Jonathan et al. [8] reported the first monolithic perovskite/c-Si TSC based on MAPbI 3 perovskite front cell.The TSC delivered an efficiency of 13.7%, limited by poor matching of MAPbI 3 (1.56 eV) and c-Si (1.12 eV) cell bandgaps, as well as excessive optical losses.[19][20][21] However, stateof-the-art perovskite/c-Si TSCs ( ∼ 32.5%)only reach ∼ 70% of their theoretical limit ( ∼ 46%) because of the unideal bandgap of perovskite front cells, large open circuit voltage (V oc ) loss, excessive optical loss in ICLs and window layers, etc. [14,15,22] Here, we provide the operating mechanism and efficiency limit of TSCs, followed by systematical discussions on promising wide-bandgap perovskite front cells.The strategies for reducing V oc losses of perovskite/c-Si TSCs are then introduced.In addition, the optoelectronic configuration for highly efficient perovskite/c-Si TSCs is discussed from the perspective of electrical ICL and texture manipulation of the substrate.Finally, we highlight the challenges in this field and provide our understanding of future research directions toward largescale, efficient, and stable perovskite/c-Si TSCs.

Operating Mechanisms and Efficiency Limits of Tandem Cells
Figure 1a shows the solar irradiance spectrum, in which visible light in the blue region and near-infrared light in the red region can be absorbed by wide-bandgap and narrow-bandgap semiconductors, respectively. [23]In a single-junction solar cell, only part of the solar spectrum can be used to generate electricity, and spectrum loss will reduce the efficiency of the solar cell. [24]The semiconductor with a bandgap of Eg can absorb photons with energy hv ≥ Eg (h is Planck's constant, ν is frequency), allowing electrons to cross the bandgap from the valence band (VB) into the higher energetically conduction band (CB) (Figure 1b). [12]The created electron-hole pair will contribute to the current of the solar cell.For the photons with energy higher than E g , the excess energy is transferred into the lattice and cannot be extracted as electrical energy, that is, thermalization losses.Photons with energy below E g cannot excite electrons and are transmitted, and hence all the photon's energy below E g is lost as transmission losses.The radiative limit is used to predict the best potential efficiency of single-junction solar cells with different bandgaps, in which the maximum efficiency of 33.7% can be reached based on a ∼ 1.35 eV bandgap semiconductor. [25,26]ne promising way to break the radiative efficiency limit of singlejunction solar cells is to couple multiple active absorber materials to develop TSCs, which can fully utilize the solar spectrum and reduce thermalization losses of the high-energy photon. [27,28]In multijunction TSCs, photoactive absorber layers with different bandgaps harvest different wavelengths of solar spectrum.Taking a two-cell configuration as an example, the high-energy region of the spectrum is absorbed by the front cell, while the transmitted low-energy photon is absorbed by the rear cell (Figure 1c). [29]ased on different connection modes, the two-cell configuration TSCs have two types: mechanically stacked four-terminal (4T) and 2terminal monolithic terminal (2T) configurations, as shown in Figure 1d. [23]In the 4T TSCs, the two subcells are two independent complete solar cells and are optomechanically stacked together.Because 4T tandem devices are not necessarily connected in series, they are not limited by current matching, making them less sensitive to the exact bandgap combination adaptation.The power density of the 4T TSCs (P 4T ) is given as: [30][31][32] where V and J are the voltage and current density of the subcells.
The subscripts 1 and 2 denote the front and rear subcells.Unlike 4T TSCs, two subcells in the 2T TSCs are optically and electrically connected, where the front and rear subcells share the middle electrode.According to the current continuity in 2T TSCs, the current density through the front and rear subcells (J 1 and J 2 ) must be identical, that is, "current matching." [33]The power density of a 2T TSC is: The requirement of "current matching" in 2T TSCs brings many challenges for realizing highly efficient 2T TSCs, including bandgap matching of the two subcells, highly efficient electrical ICLs, photon management, etc.
][36] The 4T perovskite/c-Sibased TSCs are easier to fabricate but have very high requirements for electrodes.Three of the four electrodes are required to be transparent electrodes.The light incident surface electrode needs to have high transmittance over a wide spectral range, while the two middle electrodes need to have high transmittance over the infrared spectral range.Because of high requirements of electrodes, 4T TSCs always suffer from  [23] Copyright 2017, Nature Publishing Group.e) The maximum theoretical efficiency of 2T TSCs as a function of the bandgap of front and rear subcells, assuming no absorption losses.Reproduced with permission. [14]Copyright 2018, Nature Publishing Group.
additional reflection and parasitic absorption, resulting in relatively low efficiencies.Compared with 4T TSCs, the 2T perovskite/c-Si TSCs, which are the focus of this review, require only one broad-spectrum transparent electrode and have the potential to achieve higher efficiency.
In a 2T perovskite/c-Si TSC, the wide-bandgap perovskite absorbs the blue part spectrum in Figure 1a, while the 1.12 eV silicon absorbs the red part spectrum.The solar spectrum can be better utilized and the thermalization loss of high-energy photons is significantly decreased.As such, 2T perovskite/c-Si TSCs can overcome the efficiency limit of single-junction silicon cells and PSCs. Figure 1e depicts theoretical PCEs versus bandgaps of front and rear subcells in the monolithic 2T configuration. [14]The black dotted line in Figure 1e indicated that combining ∼ 1.12 eV c-Si with 1.65-1.70eV perovskites can potentially yield ∼ 46% efficiency.Such a result suggests that the perovskite front cell with a suitable bandgap is crucial for realizing high-efficiency perovskite/c-Si TSCs.

Composition Engineering for Efficient Wide-Bandgap Perovskite Front Cells
Figure 2a shows the schematic of ABX 3 halide perovskite crystal structure, where A is a monovalent cation (MA + , FA + , Cs + ), B is a divalent cation (Pb 2+ , Sn 2+ ), and X is the halide anion (I À , Br À , Cl À ). [37]In the halide perovskites, primary energetic band structures are created by the metal-halide octahedral network, that is, the orbital interaction between lead and halides, as shown in Figure 2b. [38,39]Compositional changes in A, B, and X sites can affect the metal-halide octahedral configuration and also the band structure.[42] Cation substitution and (or) anion alloying engineering have created many perovskites with different bandgaps.The resulting wide-bandgap perovskites can be classified into two types, namely mixed halide wide-bandgap perovskites and pure iodine wide-bandgap perovskites.
In 2015, MAPbI 3 perovskite with a bandgap of 1.56 eV was firstly employed as a front cell to pair with a silicon rear cell.However, the efficiency of the TSCs was as low as 13.7% because of the poorly matched bandgap of MAPbI 3 and Si layers. [8]Figure 1e indicates that the optimal bandgap range of perovskite front cell is 1.65 ∼ 1.70 eV in 2T perovskite/c-Si TSCs, which is crucial to balance the light absorption for obtaining high photocurrents and high-efficiency tandem cells.A series of wide-bandgap perovskites were then developed by incorporating mixed halides (I, Br) into Cs/MA/FA-based perovskites to finely tune the bandgap (Figure 2c,d). [43]The new 1.63 eV Cs 0.17 FA 0.83 Pb(Br 0.17 I 0.83 ) 3 perovskite was employed in tandem with the silicon rear cell. [44]Optimized bandgap increased solar spectrum absorption and enabled a better match between the photocurrent of front and rear cells, resulting in an improved efficiency of 23.6%.This efficiency exceeded both subcells and surpassed the record single-junction PSC at that time.In addition, McGehee et al. [45] used triple-halide alloys (Cl, Br, I) to tailor the bandgap and develop an efficient 1.67 eV wide-bandgap perovskite front cells.By integrating this front cell with silicon rear cell, a PCE of 27% is achieved in 1 cm 2 2T perovskite/c-Si TSC.
While most perovskite front cells in the reported 2T perovskite/c-Si TSCs were based on mixed halide perovskites, [15,48] they are intrinsically easy to segregate into I-rich and Br-rich domains under the illumination or thermal pressure conditions, which will decrease the efficiency and stability of the tandem devices. [49]Pure iodide widebandgap perovskites without mixed halides provide an alternative as front cells for highly efficient perovskite/c-Si TSCs. [46,50,51]Recently, a pure iodide Cs 0.3 DMA 0.2 MA 0.5 PbI 3 perovskite with a bandgap of 1.69 eV was developed and the corresponding perovskite front cell exhibited excellent photo-stability (1% degradation after 1000 h continuous operation), which almost exceeded all of the mixed halide perovskites with similar bandgaps. [52]ompared with MA-based pure iodide wide-bandgap perovskites, pure iodide FA-Cs perovskite, that is, FA 1-x Cs x PbI 3 , can eliminate the concerns of volatile MA + cation and phase segregation induced by mixed halides. [50,51]The mildly deformed octahedral framework in Cs 1-x FA x PbI 3 perovskites make it much more stable than those mixed halide perovskites with similar wide bandgaps (Figure 2e). [46]However, depositing high-quality FA 1-x Cs x PbI 3 films has been challenging because of the different precipitation kinetics of FA + and Cs + cations.Recently, some new methods, such as slow solid phase cation, have been developed to fabricate high-quality FA 1-x Cs x PbI 3 films. [46]hermodynamically stable CsPbI 3 inorganic perovskite with suitable bandgap ( ∼ 1.7 eV) is considered as another promising candidate for front cells of perovskite/c-Si TSCs (Figure 2f). [53]However, the phase instability of Cs-based inorganic perovskites caused by unideal tolerance factors seriously hinders their tandem application with c-Si cells. [54]][57][58][59][60] Currently, the phase stability of CsPbI 3 perovskite has been significantly improved, and the efficiency of CsPbI 3 PSCs has approached ∼ 21%.
Overall, the challenges in pure iodide wide-bandgap perovskites, including deposition methods, structural stability, etc., hinder the development of pure iodide wide bandgap of perovskite/silicon tandem cells.With the breakthroughs of pure iodide wide-bandgap perovskites in terms of chemical deposition methods, stability, etc. in the past few years, highly efficient and stable pure iodide perovskite/c-Si TSCs are highly expected to be realized in the near future.

Strategies to Reduce the Voltage Loss in Perovskite/c-Si TSCs
The ideal bandgap perovskite front cells are beneficial to balance photon absorption for current matching and high photocurrent, while high V oc is an important factor to achieve highly efficient perovskite/c-Si tandem devices.In a 2T perovskite/c-Si TSC, the voltage of the whole device is equal to the sum of the voltages of the two subcells. [29,35,61]Boccard et al. [62] compared two different rear-cell designs and highlight the importance of high V oc perovskite front cell to achieve high-efficiency perovskite/c-Si TSCs.69] Many strategies, for example, crystallization and growth manipulation, defects passivation, and interface contact regulation, have been developed to reduce the V oc loss and improve the performance of the perovskite/c-Si tandem device.

Crystallization and Growth Manipulation of Wide-Bandgap Perovskites
In perovskite/c-Si TSCs, depositing dense and pinhole-free mixed halide wide-bandgap perovskite films is more challenging as the textured silicon substrate and the rapid crystallization of these perovskites can easily lead to poor contact and a large number of holes at the interface between perovskites and silicon (Figure 3a). [70]Regulating crystallization and growth kinetics is a common approach to improve the film quality of wide-bandgap perovskites and reduce V oc loss in PV devices. [71]Multiadditive synergistic approach has been successfully developed to manipulate the crystallization dynamics and improve the quality of wide-bandgap perovskite films.For example, when MACl and MAH 2 PO 2 were simultaneously incorporated into the perovskite precursor to slow down crystal grain growth and obtain high-quality perovskite film. [72]MACl enlarged the grain size by slowing the perovskite crystal formation process, while MAH 2 PO 2 further slowed down the rate of MACl leaving the film and amplified the effect of MACl enlarging the grain size.As a result, large grains and smooth morphology were achieved in the wide-bandgap perovskite films (Figure 3b).Large grains were beneficial to reduce grain boundary (GB) defects, while smooth morphology facilitated the formation of a more uniform electron transport layer (ETL) to create a continuous, shunt-free interconnection layer.Finally, the V oc deficit of the wide-bandgap perovskites PSCs was reduced to 0.49-0.51V.The TSCs based on a 1.64 eV perovskite, and c-Si yields a high V oc of 1.80 V. Additionally, complementary additives of PEAI and Pb(SCN) 2 have also been demonstrated to improve the quality of (FA 0.65 MA 0.20 Cs 0.15 )Pb(I 0.8 Br 0.2 ) 3 films (Figure 3c). [73]The coupling of PEA + and SCN À provided a synergistic effect to overcome the individual challenges associated with each additive, leading to markedly improved structural and optoelectronic properties of perovskite films: It increased the perovskite film quality with improved crystalline properties, reduced excess PbI 2 formation associated with the use of Pb(SCN) 2 , decreased defect density, and improved charge mobility and lifetime.In addition to regulating crystal growth, stable carbazole was also introduced into Cs 0.05 FA 0.8 MA 0.15 Pb (I 0.75 Br 0.25 ) 3 to simultaneously regulate growth dynamics and manipulate GBs to reduce nonradiative recombination losses (Figure 3d,e). [74]ith carbazole, the wide-bandgap PSCs exhibited an improved V oc of 1.22 V and a high PCE up to 20.2%, further enabling a stabilized PCE of 28.6% in a monolithic perovskite/c-Si TSCs.
Compared with these mixed halide perovskite films, depositing high-quality pure iodide wide-bandgap perovskites is more difficult because of the phase separation/nonphotoactive phase formation induced by large size mismatch of A-site cations, seriously hindering the development of pure iodide wide-bandgap perovskite/c-Si TSCs.Until recently, a new method, that is, slow solid phase cation exchange, has been successfully developed to effectively resolve the phase separation induced by the Cs + and FA + cation mismatch in Cs-rich perovskite and fabricate high-quality pure Cs 1-x FA x PbI 3 (0 < x < 0.5) perovskites, which potentially facilitate the development of Cs 1-x FA x PbI 3 /c-Si tandem TSCs. [46]Subsequently, two-dimensional (2D) Ruddlesden-Popper (RP) phase (Cs 2 PbI 2 Cl 2 ) was developed to suppress the secondary phases of δ-CsPbI 3 and facilitate the formation of the pure-iodide widebandgap Cs 0.3 DMA 0.2 MA 0.5 PbI 3 perovskite films. [52]The corresponding 2T TSCs achieve a high V oc of 1.90 V, leading to a high efficiency of 29.51%.These new approaches to fabricate pure iodide wide-bandgap perovskite will further facilitate the development of pure iodide widebandgap perovskite/c-Si TSCs.

Passivating Wide-Bandgap Perovskites Front Cells
Defects passivation engineering is another effective approach to reduce the V oc loss in the wide-bandgap PSCs. [75]In general, passivating GBs can not only reduce defects but also mitigate ion migration and eases phase separation induced by mixed halides.For example, a self-limiting passivation molecule (SLP) (1-butanethiol) has successfully been anchored on GBs of Cs 0.05 MA 0.15 FA 0.8 PbI 2.25 Br 0.75 films to reduce the defects, enhance the diffusion length, and mitigate phase separation of the I-Br (Figure 4a). [76]The perovskite/c-Si TSCs yield a certified efficiency of 25.7% with a V oc of 1.70 V. Incorporating 2D perovskites into 3D perovskite is another strategy to passivate the GBs in wide-bandgap perovskite films.Kim et al. [77] successfully incorporated 2D perovskites into the GBs of wide-bandgap perovskites through anion engineering, which significantly reduces the GBs defects and improves the charge transport properties (Figure 4b).The resulting wide-bandgap PSCs exhibited a PCE of 20.7% and corresponding 2T TSCs yield an efficiency of 26.7% with V oc of 1.756 V.In addition to GBs passivation, multifunctional molecules, such as phenformin hydrochloride molecule PhenHCl with electron-rich and electron-poor domains, have been used to concurrently passivate cationic and anionic defects in mixed halide perovskites (Figure 4c). [78]Based on a 1.68 eV wide-bandgap perovskite with PhenHCl treatment, the V oc in p-i-n PSCs is improved from 1.08 to 1.18 V, resulting in PCEs up to 20.5%.Moreover, the concurrent cationic and anionic defect passivation strategy suppressed the light-induced halide segregation in the wide-bandgap perovskites.Finally, the corresponding perovskite/c-Si TSCs exhibited an increased V oc from 1.76 to 1.84 V and PCE from 25.4% to 27.4%, as well as improved stability.

Improving the Interface Selective Contact
Effective interface contact in perovskite/c-Si TSCs is crucial to improve charge carriers' extraction, reduce interface nonrecombination and contribute to high V oc .Some functional molecules/inorganic compounds have been used to modulate the charge transport layer, which in turn regulates the contact between the charge transport layer and the  [70] Copyright 2020, Elsevier.b) The wide-bandgap perovskite films fabricated by MACl and/or MAH 2 PO 2 additive methods.Reproduced with permission. [72]Copyright 2021, Elsevier.c) Schematic illustration of PEAI and Pb(SCN) 2 additives improving the quality of wide-bandgap perovskite films.Reproduced with permission. [73]Copyright 2019, Elsevier.d) Carbazole manipulates GBs of the wide-bandgap perovskites (scale bar: 200 nm), and e) evolution of steady-state PL spectra for the films with and without carbazole addition.Reproduced with permission. [74]Copyright 2021, Elsevier.
Energy Environ.Mater.2024, 7, e12639 perovskite layer, thereby inhibiting nonradiative recombination and reducing V oc loss. [79,80]ncorporating a D-sorbitol gradient dopant into PEDOT:PSS has been employed to manipulate the orientation of PEDOT across the film and improve the contact between perovskite and charge transport layers. [81]he 2T TSCs show a reduced V oc loss and deliver a steady-state efficiency of 21.0% with a FF of 80.4%.Besides, C 60 with self-assembled molecular monolayer (C60-SAM) is successfully anchored on the a-NbO x surface as an efficient electron-selective contact (Figure 5a), [79] in which the pyrrolidine group from the C60-SAM molecule interacts with the metal oxide electron-selective layers and the C60 interacts with (negatively charged) iodide ions originating from noncoordinated Pb 2+ or Pb-I antisite defects.C60-anchoring strategy improves electric field and provides effective passivation at the interface of perovskite and electron-selective layer, finally decreasing voltage losses at the perovskite/holes selective interface.These advancements enable 27% efficiency perovskite/c-Si TSCs with a V oc of 1.83 V.In addition, SAM with methyl group substitution has been designed as a hole-selective layer, which shows a fast hole extraction (Figure 5b). [82]The SAM provided both fast extraction and efficient passivation at the hole-selective interface, thereby leading to wide-bandgap PSCs with high V OC of >1.23 V and mitigated light-induced halide segregation.The improvements of perovskite front cell transferred into TSCs delivers a PCE of 29.15% with a high V oc of 1.90 V.
In addition to organic molecule, stable inorganic compounds (such as MgF x , LiF, etc.) are successfully applied to improve the contact between perovskite and charge transport layers.Recently, a MgF x interlayer with a thickness of ∼ 1 nm is successfully incorporated into the perovskite/C 60 interface, which favorably adjusts the surface energy of the perovskite layer and facilitates efficient electron extraction to mitigate nonradiative recombination (Figure 5c).Finally, the champion TSCs exhibit a V oc of 1.92 V and enable a certified stabilized PCE of 29.3%.Additionally, a thermally evaporated CsBr thin layer is also introduced between the perovskite layer and hole transport layer (HTL) to construct a gradient perovskite absorber, which not only eliminates the adverse effect of the residual PbI 2 at the bottom of the perovskite films but also optimizes energy level alignment (Figure 5d).[87] In order to simultaneously address the film decomposition and redox chemistry at the perovskite/NiO x interface, Cl rich NiO x /perovskite interface is developed.Because Cl is more electronegative than I, the enrichment of Cl at the perovskite/NiO x interface can effectively inhibit perovskite decomposition at the interface.Moreover, Cl can enhance the intensity of the p orbital contribution to the VB of I  [76] Copyright 2020, American Association for the Advancement of Science (AAAS).b) The TEM image of 2D phase at the GBs of 3D perovskite.Reproduced with permission. [77]Copyright 2021, American Association for the Advancement of Science (AAAS).c) Schematic diagram of multifunctional molecular passivation.Reproduced with permission. [78]Copyright 2021, Elsevier.
Energy Environ.Mater.2024, 7, e12639 that enhances FA/I interactions and stabilizes the NiO x interface (Figure 5e). [84]Finally, the efficiency of wide-bandgap PSCs increased from 17.82% to 19.76%, leading to the perovskite/c-Si TSCs with PCEs of up to 27.26% with a V oc of 1.86 V.

Efficient Electrical Interconnection between Perovskite and Silicon Subcells
Efficient electrical interconnection between perovskite and silicon subcells is another key factor determining the performance of perovskite/ c-Si 2T TSCs.For p-i-n or n-i-p single-junction solar cells, the n-type layer extracts and transfers electrons, while the p-type layer extracts and transfers holes generated in the active layers.In the 2T TSCs, if the n-type layer of one subcell and the p-type layer of the other subcell are simply connected in series, a reverse n-p diode will form to limit the current flow. [88]91] Highly efficient electrical ICL in 2T perovskite/c-Si TSCs needs to fulfill two functions: 1 Good optical properties, balancing light absorption of double light absorbing layer.In a perovskite/c-Si 2T TSC, sunlight firstly passes through the wide-bandgap (E g 1) perovskite absorber, and the photons with energy above E g 1 are absorbed by the wide-bandgap perovskite layer.The remaining sunlight will transmit through the ICL (part of the light is absorbed by the ICL) and be absorbed by the Si absorber.A good ICL requires reduced light reflection and parasitic absorption, all of which can significantly affect the performances of silicon subcells and hence the whole TSCs.Moreover, ICL also requires high optical transmittance performance to reduce self-parasitic absorption.In a highly efficient perovskite/c-Si TSC, both ICL composition and thickness should be optimized for increasing optical transmittance and reducing parasitic absorption. 2 Excellent electrical properties, ensuring the extraction and transport of charge carriers and providing an effective carrier recombination area.For the ICL with a recombination layer, electrons and holes extracted from individual subcells will recombine within the ICL, which will facilitate current flow within the tandem device.For low-conductivity ICL, electrons and holes will Reproduced with permission. [79]Copyright 2021, Royal Society of Chemistry.b) The FF of PSCs correlated to nonradiative interface loss.Reproduced with permission. [82]Copyright 2020, American Association for the Advancement of Science (AAAS).c) Quasi-Fermi-level splitting (QFLS) mapping for the perovskite, and perovskite/MgF x /C 60 samples on a 2PACz coated Si cell. [80]Copyright 2022, American Association for the Advancement of Science (AAAS).d) Schematic illustration of a fully-textured perovskite/c-Si TSC with the addition of CsBr layer. [83]Copyright 2021, Wiley-VCH.e) The addition of excess CsCl regulated the perovskite crystal, preventing the formation of PbI 2 À x Br x at the interface.Reproduced with permission. [84]Copyright 2022, Wiley-VCH.
accumulate on both sides of the recombination layer.In this scenario, the recombination layer ICL acts as a capacitor whose accumulated carriers introduce an electric field in the opposite direction to the photovoltage and compensate for the output voltage, which can reduce the V oc of the tandem device. [20]We also note that the requirements for the ICL conductivity are not so high as those for electrodes, since the recombination layer electrically connects the subcells in the vertical direction and high lateral electrical conductivity is not required. [92]For example, transparent ultrathin metal was employed in polymer tandem devices serving as the recombination layer.The gold film with a thickness of only 0.5 nm can support efficient electron and hole recombination.The lateral electrical conductivity of 0.5-nm-thick gold is low as the film may not be continuous.

Tunnel Junction-Based ICL
There are two types of tunnel junctions: 1) formed only by silicon, and 2) silicon and other interfacial materials, such as SnO 2 or TiO 2 .
An n ++ /p ++ tunnel junction by depositing heavily doped n ++ hydrogenated amorphous silicon (a-Si:H) was firstly developed for facilitating majority-carrier charge recombination between perovskite front subcell and silicon rear subcell (Figure 6a). [8]In the tunnel junction layer, holes from the n-type Si layer pass through the p-type emitter and electrons from the perovskite layer pass through TiO 2 ETL, respectively (Figure 6b).The corresponding 2T TSCs demonstrated a PCE of 13.7%.In order to optimize refractive index matching between the tunnel junction layer and underlying crystalline silicon, a nanocrystalline silicon tunnel junction ICL was further developed, with silicon heterojunction (SHJ) solar cells as the rear subcells (Figure 6c). [93]The p + and n + layers were fabricated by plasmaenhanced chemical vapor deposition (PECVD) with trimethyl boron or phosphine as the dopant.The electrons generated in perovskite subcell were extracted by the C 60 ETL and transported to the nanocrystalline n + layer, while the holes generated in the silicon subcell transported to the nanocrystalline p + layer.Finally, the electrons and holes recombined in the tunnel junction.Because of a better refractive index matching nanocrystalline silicon and underlying crystalline silicon, the light reflectance was further suppressed in the tunnel junction.The TSC exhibits high PCEs of 22.7% at 0.25 cm 2 and 18% at 12.96 cm 2 .Based on the nanocrystalline silicon tunnel junction, double-textured SHJ solar cells are further employed as rear cells to reduce optical loss and improve the efficiency of perovskite/c-Si TSCs to 25.2% at 1.42 cm 2 (Figure 6d).The combination of silicon and other interfacial materials, such as SnO 2 /p ++ silicon, is another tunnel junction, where interfacial materials SnO 2 provide dual functions of ETL for the perovskite front subcell and recombination contact with p ++ silicon (Figure 6e). [94,95]The SnO 2 /p ++ junction plays an important role in the electrical connection of these two subcells without increasing additional optical loss.Although there is a barrier from the native SiO 2 layer of silicon, the barrier is thin enough for the current transport through tunneling and the J-V curves of SnO 2 /p ++ junction follow a diode behavior.As a result, PCEs of 21.8% on 16 cm 2 perovskite/c-Si TSCs are consecutively achieved.
In summary, tunnel junction-based ICLs have the following advantages: 1) low parasitic absorption which would increase the performance of the rear subcell; 2) low light reflection loss from better refractive index matching; 3) low lateral electrical conductivity that is compatible with large-area TSCs that only provide recombination area.However, preparing high-quality, large-scale tunnel junction-based ICLs has always been challenging through PECVD/Atomic Layer Deposition techniques, thus limiting the application of tunnel junction in the commercialization of perovskite/c-Si TSCs.

Recombination Layer-Based ICL
Common transparent conductive oxides (TCOs) mainly include Indium Zinc Oxide (IZO), Zinc Tin Oxide (ZTO), and Indium tin oxide (ITO).[98] Based on the optimized IZO ICL, the TSCs yield a PCE of up to 20.5%.However, the current of rear cell shows IZO thickness dependence, that is, with increasing thickness of IZO, the current in the rear subcell decreases because of optical interferences.Moreover, IZO is neither electrically nor optically stable when annealed in an oxygen-containing environment.ZTO is then employed as the recombination layer in the ICL of mesoporous perovskite/c-Si TSCs (Figure 7a). [99]N-type ZTO shows high electron mobility and high transparency, as well as good temperature stability.Similar to IZO, the thickness of this recombination layer has a strong effect on the optical absorption of the rear cell (Figure 7b).Subsequently, ITO ICL is introduced into perovskite/c-Si TSCs. [100]Below 100 nm, the thickness of ITO has a negligible effect on the optical absorption of the rear cell.Based on n-i-p structure PSC front subcell and SHJ silicon rear subcell, the TSCs with ITO ICL achieved high PCE of up to 18.1%.McGehee et al. [44] also reported a perovskite-SHJ tandem device using ITO as the recombination layer (Figure 7c).In their ICL structure, a sputtered 20nm-thick ITO serves as the recombination layer, where n-type amorphous Si:H extracts electrons from the SHJ rear subcell, and NiO x servers as HTL of the perovskite front subcell.Finally, they demonstrated a PCE of 23.6% on an area of 1 cm 2 TSCs.Later, they further enhanced the PCE to 27% at 1 cm 2 devices by optimizing the optical design to minimize the reflection loss. [45,61]ransparent conductive oxides with good electrical and optical properties are promising candidates for the ICL recombination layer in perovskite/c-Si TSCs. [101,102]Since crystal silicon is generally robust enough to withstand damage from energetic ions during the sputtering process, it is not necessary to consider sputtering damage during the ICL fabrication process.However, parasitic absorption and reflection losses are unavoidable in the TCO layer.Considering the absence of chemical protection and high lateral conductivity requirements, TCO layer can be fabricated thinner to decrease the parasitic absorption and reflection loss compared with their application as the transparent electrode.

Photon Management in Perovskite/c-Si Tandem Cells
The maximum total number of absorbed photons available to a TSC, which can be calculated from the sum of subcell external quantum efficiencies weighted with the incident spectrum, is limited by the subcell with the lowest bandgap. [103]In the perovskite/c-Si TSC, the theoretical limit of subcell current matching is supposed to be approximately half Energy Environ.Mater.2024, 7, e12639 of a single-junction silicon cell ( ∼ 43.3 mA cm À2 ). [104,105]However, at present, the J sc in perovskite/c-Si TSCs is still limited to the range of 15-19 mA cm À2 . [35,106]In addition to the undesirable bandgap of the perovskite front cell, large optical loss can account for the low current density.For 2T perovskite/c-Si configurations, the optical loss mainly comes from reflection and parasitic absorption loss. [19,107,108]revious investigations have indicated that the parasitic absorption can be significantly reduced through optimizing the perovskite front cell and ICLs.The optical loss from reflection can be reduced by using antireflective coatings and graded index structures.For example, Sahli and workers successfully fabricated PSCs onto the textured front surface of silicon cells by developing a conformal deposition process for a perovskite thin film that combines co-evaporation and spin coating, providing necessary antireflection and light trapping (Figure 8a). [21]he TSCs overcome the output current limit in the case of the frontplanar Si solar cell.An impressive certified J sc of 19.5 mA cm À2 is achieved, which is an 8% improvement with respect to that of the previously reported certified perovskite/Si TSCs.Additionally, perovskite/ silicon TSCs with gentle periodic nanotextures are also developed, which feature various advantages without compromising the material quality of solution-processed perovskite layers (Figure 8b). [109]Reflection losses are reduced in comparison to planar tandems and the devices are less sensitive upon deviations from optimum layer thicknesses.The nanotextures also enable excellent perovskite film formation and greatly improve fabrication yields.The optically advanced rear reflector with a dielectric buffer layer reduces the parasitic absorption at near-infrared wavelengths.Finally, the improvements enable a certified PCE of 29.80%.
The above results suggested that, in principle, although limited, there is room to further improve the device's current density by further reducing reflection and parasitic absorption losses.Grant et al. used an optical model to identify the optical loss mechanisms in a 2T perovskite/c-Si TSCs (Figure 8c,d). [110]The results highlight the importance of the low absorption in window layers, and the correct choice of the refractive index and thickness of charge transport layers, in order to minimize reflection at the interfaces formed by these layers.Through adopting suitable charge transport materials, optimizing perovskite absorber thickness, and introducing light trapping in the silicon cell, a matched current of over 20 mA cm À2 can be achieved, enabling efficiencies >30% and even more.For example, a TSCs based on a double-side polished c-Si wafer suffers from large reflection losses and has a relatively low matched device current density of 17.28 mA cm À2 (Figure 8e). [111]Through a rear side texture, a front side texture with burial layer, and an MgF 2 antireflective coating, reflection losses could be reduced systematically, which can increase the current density to 19.57mA cm À2 .The TSCs based on 1.66 eV perovskite bandgap and both-side textured device have the potential to achieve an efficiency of 32.5%. [19] [8]Copyright 2015, American Institute of Physics.c) Nanocrystalline silicon tunnel junction-based tandem cells.Reproduced with permission. [93]Copyright 2018, Wiley-VCH.d) The cross-section of perovskite front cell deposited on the SHJ rear cell.Reproduced with permission. [18]Copyright 2018, Nature Publishing Group.e) Schematic device design of interface-layer-free perovskite/ silicon homojunction cells.Reproduced with permission. [94]Copyright 2018, Royal Society of Chemistry.

Challenges and Outlook
The increase in perovskite/c-Si TSC efficiency has been driven by the in-depth understanding of the wide-bandgap perovskite materials properties, new passivation approaches, the fast development of electrical ICL, photon management strategies, etc.The PCEs of perovskite/c-Si TSCs have achieved a certified 32.5% and exceeded all single-junction cells by up to 31%, but are still well below its theoretical limit.Meanwhile, these high-efficiency devices are almost based on small-scale TSCs and also suffer from long-term stability challenges.Therefore, further improving the efficiency and stability as well as reducing the cellto-module gap are important aspects for future research aiming at accelerating the commercialization of perovskite/c-Si TSCs.
The intrinsic and extrinsic instability of perovskite front cells limits the lifetime of perovskite/c-Si TSCs.For example, the wide-bandgap perovskites currently used in perovskite/c-Si TSCs are mostly fabricated by mixing A-site cation and X-site halide ions, which suffer from serious phase segregation challenges, that is, forming I-rich and Br-rich domains under the illumination or thermal conditions.A complete understanding of phase segregation in mixed halide perovskites is still lacking.[114] Therefore, it is necessary to employ advanced techniques to closely monitor the phase segregation of mixed halide perovskites and reveal the phase segregation mechanism.More targeted strategies are needed to improve the stability of wide-bandgap perovskites and finally increase the lifetime of perovskite/c-Si TSCs.Thermodynamically stable pure iodide wide-bandgap perovskites, such as Cs x FA 1-x PbI 3 and CsPbI 3 , should be explored for tandem with silicon rear cells, which may be a fundamental approach to achieving highly stable perovskite/c-Si TSCs.
The-state-of-art small-scale perovskite/c-Si TSCs have efficiencies approaching 31%, but corresponding module performance lags significantly.This is because finding a suitable deposition method for ICLs and large-area perovskite films with submicron thicknesses on textured silicon substrates remains challenging, which require improvements in fabrication methods such as blade coating or vacuum deposition.Enhancing the diffusion length of the perovskite layer by increasing the carrier lifetime can achieve a thicker perovskite film (>1 μm) that could fully cover the roughness of the texture c-Si rear cell.In addition, previous studies have suggested the importance of defect passivation for high-efficiency and stable TSCs.As such, more fundamental  [99] Copyright 2016, American Institute of Physics.c) Perovskite/c-Si device with ITO recombination layer.Reproduced with permission. [44]Copyright 2017, Nature Publishing Group.
Energy Environ.Mater.2024, 7, e12639 understanding and advanced passivation strategies should be developed for passivating and controlling defects in large-area wide-bandgap perovskite films.
Efficient photon management is necessary to reduce parasitic absorption and reflection loss, and increase the harvest of photon in the TSCs.Firstly, it is important to search for new HTLs with low parasitic absorption and excellent hole transport properties that are compatible with low-temperature preparation.Meanwhile, it is essential to develop some other methods to avoid sputtering damage or search for novel buffer layers with better transmittance.Additionally, further development and optimization of the tunneling junction should also be implemented to decrease optical loss.Secondly, efficient light trapping techniques are urgently required to reduce reflection loss and achieve higher J sc .The pyramid features are a promising structure; although nearly 20 mA cm À2 J sc has been obtained with random pyramids, further improvement is still important to achieve ultra-high efficiency.It is important to develop regular pyramids because of their possible benefits for better perovskite film quality, thus influencing the electrical properties of devices.Further efforts in light management need to be devoted to thorough optical simulation and careful design of devices to achieve  [21] Copyright 2021, Nature Publishing Group.b) Cross-section through a meshed unit cell of a tandem device with nanotextures between perovskite and silicon subcells, as used for simulations with the finite element method.Reproduced with permission. [109]Copyright 2022, Nature Publishing Group.c) Schematics of device structures for simulation in the p-i-n orientations (the arrows indicate the direction of incident light) and d) light distribution for the p-i-n perovskite/c-Si TSCs.Reproduced with permission. [110]Copyright 2016, Optica Publishing Group.e) Perovskite/c-Si TSCs with and without texture. [111]ergy Environ.Mater.2024, 7, e12639 the best performance.Theoretical understanding can provide reliable guidelines for the exploration of tandem device techniques.
In addition, employing suitable silicon solar cells is also important to develop perovskite/c-Si 2T TSCs.Now, among silicon solar cell technologies, the aluminum back surface field (Al-BSF) solar cells, passivated emitter and rear cells (PERC), and TOPCon solar cells, predominate in large-scale industrial PV device production.Besides these homojunction solar cells, another promising route is silicon heterojunction solar cells (SHJ), which employ layers of different materials to improve efficiency.In the case of perovskite/c-Si TSCs, the compatibility of structure matching and manufacturing process between c-Si solar cells and perovskite front cells needs to be further explored.
We believe that most of the above-mentioned issues can be addressed in the near future through combined efforts from academia and industry, and also hope that Perovskite/c-Si TSCs can play an important role in the PV technology.

Figure 1 .
Figure 1.Operating mechanisms and efficiency limit for TSCs.a) Solar irradiance spectrum.Schematic diagram of light absorption in the b) single-junction and c) multijunction solar cells.d) Architecture diagram of 4T and 2T TSCs.Reproduced with permission.[23]Copyright 2017, Nature Publishing Group.e) The maximum theoretical efficiency of 2T TSCs as a function of the bandgap of front and rear subcells, assuming no absorption losses.Reproduced with permission.[14]Copyright 2018, Nature Publishing Group.

Figure 3 .
Figure 3. Crystal manipulation of wide-bandgap perovskites.a) Simple solution-processed perovskite films on textured silicon wafers.Reproduced with permission.[70]Copyright 2020, Elsevier.b) The wide-bandgap perovskite films fabricated by MACl and/or MAH 2 PO 2 additive methods.Reproduced with permission.[72]Copyright 2021, Elsevier.c) Schematic illustration of PEAI and Pb(SCN) 2 additives improving the quality of wide-bandgap perovskite films.Reproduced with permission.[73]Copyright 2019, Elsevier.d) Carbazole manipulates GBs of the wide-bandgap perovskites (scale bar: 200 nm), and e) evolution of steady-state PL spectra for the films with and without carbazole addition.Reproduced with permission.[74]Copyright 2021, Elsevier.

Figure 4 .
Figure 4. Defects passivation strategies in wide-bandgap perovskite front cells.a) Effects of SLP on charge diffusion length and phase segregation.Reproduced with permission.[76]Copyright 2020, American Association for the Advancement of Science (AAAS).b) The TEM image of 2D phase at the GBs of 3D perovskite.Reproduced with permission.[77]Copyright 2021, American Association for the Advancement of Science (AAAS).c) Schematic diagram of multifunctional molecular passivation.Reproduced with permission.[78]Copyright 2021, Elsevier.

Figure 5 .
Figure 5. Manipulating the interface between perovskite and c-Si subcells.a) Schematics of C60-anchored α-NbO x .Reproduced with permission.[79]Copyright 2021, Royal Society of Chemistry.b) The FF of PSCs correlated to nonradiative interface loss.Reproduced with permission.[82]Copyright 2020, American Association for the Advancement of Science (AAAS).c) Quasi-Fermi-level splitting (QFLS) mapping for the perovskite, and perovskite/MgF x /C 60 samples on a 2PACz coated Si cell.[80]Copyright 2022, American Association for the Advancement of Science (AAAS).d) Schematic illustration of a fully-textured perovskite/c-Si TSC with the addition of CsBr layer.[83]Copyright 2021, Wiley-VCH.e) The addition of excess CsCl regulated the perovskite crystal, preventing the formation of PbI 2 À x Br x at the interface.Reproduced with permission.[84]Copyright 2022, Wiley-VCH.

Figure 6 .
Figure 6.Tunnel junction-based ICL in perovskite/c-Si TSCs.a) Schematic diagram of a silicon-based tunnel junction 2T perovskite/c-Si TSCs and b) the charge transport mechanism in the tunnel junction.Reproduced with permission.[8]Copyright 2015, American Institute of Physics.c) Nanocrystalline silicon tunnel junction-based tandem cells.Reproduced with permission.[93]Copyright 2018, Wiley-VCH.d) The cross-section of perovskite front cell deposited on the SHJ rear cell.Reproduced with permission.[18]Copyright 2018, Nature Publishing Group.e) Schematic device design of interface-layer-free perovskite/ silicon homojunction cells.Reproduced with permission.[94]Copyright 2018, Royal Society of Chemistry.

Figure 7 .
Figure 7. Recombination layer-based ICL in perovskite/c-Si TSCs.a) Schematic diagram of perovskite/c-Si TSCs and b) EQE curves of the device with different thickness ZTO layers.Reproduced with permission.[99]Copyright 2016, American Institute of Physics.c) Perovskite/c-Si device with ITO recombination layer.Reproduced with permission.[44]Copyright 2017, Nature Publishing Group.

Figure 8 .
Figure 8. Photon management in perovskite/c-Si TSCs.a) Sketch of light absorption in a bifacial perovskite/c-Si TSCs.Reproduced with permission.[21]Copyright 2021, Nature Publishing Group.b) Cross-section through a meshed unit cell of a tandem device with nanotextures between perovskite and silicon subcells, as used for simulations with the finite element method.Reproduced with permission.[109]Copyright 2022, Nature Publishing Group.c) Schematics of device structures for simulation in the p-i-n orientations (the arrows indicate the direction of incident light) and d) light distribution for the p-i-n perovskite/c-Si TSCs.Reproduced with permission.[110]Copyright 2016, Optica Publishing Group.e) Perovskite/c-Si TSCs with and without texture.[111]