1. Introduction

Organic–inorganic metal halide perovskites have become attractive light absorber materials for solar cells due to their high optical absorption coefficient, long carrier diffusion length, low trap density, tunable bandgap, and simple processing^[1–8]. The power conversion efficiency (PCE) of single-junction perovskite solar cells (PSCs) has increased from 3.8% to 25.2% in just a decade^[9–11].

Fabricating tandem solar cells by combining subcells with different bandgaps offers an avenue to go beyond the Shockley-Queisser limit^[12] of single-junction solar cells. The highest PCE of 47.1% among all photovoltaic cells has been achieved by a multi-junction solar cell with six subcells made of III–V compound semiconductors^{[10, 11, 13]}. Nonetheless, these devices are not commercially viable for terrestrial applications because of their complex manufacturing and extremely high costs^{[14, 15]}. Organic tandem solar cells offer potential of much lower fabrication costs because of their simple solution processability. However, the PCEs of organic photovoltaics are far below those available in inorganic devices^[16–19]. Metal halide perovskites have become ideal candidates for fabricating tandem solar cells because they have demonstrated both high PCEs and low costs in single-junction PSCs.

The bandgap tunability enables metal halide perovskites to construct perovskite-perovskite (all-perovskite) tandem solar cells^[20]. A wide-bandgap perovskite is used for the top subcell while a narrow-bandgap perovskite is used for the bottom subcell. The bandgap of Pb halide perovskite can be continuously tuned from 1.5 to 2.5 eV by adjusting the I/Br ratio^[21–24], making them suitable for top subcells in tandem solar cells. Meanwhile, the bandgap of mixed Pb–Sn perovskites can be lowered to ~1.2 eV by adjusting the composition of the B-site metal cation^[25–31]. In addition to the application on all-perovskite tandem solar cells, wide-bandgap perovskites have been deployed in perovskite/silicon^[32–36], perovskite/Cu (In, Ga) Se₂ (CIGS)^[37–42], and other tandem configurations^[43–45]. Overall, all-perovskite tandem solar cells have the potential of offering lower fabrication cost, less environmental impact, and higher PCE than other perovskite-based tandem structures^[46–51].

In the early stage, all-perovskite tandem solar cells showed much lower performance compared to single-junction PSCs^[52]. With improvement in the growth of mixed Pb–Sn narrow-bandgap perovskites, Snaith et al. firstly demonstrated a PCE of 16.9% by monolithically combining a narrow-bandgap perovskite (FA_0.75Cs_0.25Sn_0.5Pb_0.5I₃) with a wide-bandgap perovskite^[53]. Very recently, our group has developed a strategy to improve the performance and stability of all-perovskite tandem solar cells via a comproportionation reaction for the growth of mixed Pb–Sn perovskites. This enabled us to achieve high PCEs of 24.8% and 22.1% for small-area (0.073 cm²) and large-area (1.048 cm²) tandem solar cells, respectively^[54]. The PCE evolution of all-perovskite tandem solar cells in the past five years is presented in Fig. 1. The PCE of all-perovskite tandem solar cells is now approaching that of best-performing single-junction PSCs (24.8%^[54] versus 25.2%^[10]).

Here we give a short review on recent research progress in improving the efficiency of all-perovskite tandem solar cells. In Section 2, we present the structure and working mechanism of all-perovskite tandem solar cell. Fig. 1 and Table 1 show the PCE evolution of all-perovskite tandem solar cells in the past few years. In Section 3, we summarize recent progress in improving the efficiency of all-perovskite tandem solar cells in three aspects: tunnel recombination junction, wide-bandgap top subcell, and narrow-bandgap bottom subcell. Finally, we summarize in Section 4 the remaining issues that constrain the performance of all-perovskite tandem solar cells. We thereby provide an outlook and perspective for future development of all-perovskite tandem solar cells.

2. Device structure and working mechanism

2.1 Device structure and working mechanism of all-perovskite tandem solar cell

Unless otherwise stated, tandem solar cells are referred to monolithic two-terminal devices in this review article. The monolithic all-perovskite tandem solar cell includes two subcells: the top subcell made of a wide-bandgap perovskite and the bottom subcell using a narrow-bandgap perovskite, as shown in Fig. 2(a). The subcells are connected with a tunnel recombination junction (also called interconnecting layer or charge recombination junction). The current in subcells should be matched to deliver optimal performance. The photocurrent of tandem solar cells follows the Kirchhoff's law^[55], depending on the minimum current in two subcells.

Sunlight consists of a broad energy spectrum with varying intensity, ranging from ultraviolet (photon energy higher than 3 eV) to infrared (photon energy lower than 1.7 eV to about 0.5 eV). When light is incident on a semiconductor, light with energy higher than the bandgap is absorbed, and light with energy lower than the bandgap is passed through without absorption, as shown in Fig. 2(b). The excess energy of the photon (above the bandgap energy) is lost as heat due to electron thermalization. The trade-off between harvesting more photons and minimizing thermalization loss in single-junction photovoltaic devices limits their theoretical maximum PCE (Shockley-Queisser limit)^{[12, 56]}.

The working principle of tandem solar cells is to combine two subcells with different bandgaps, thereby absorbing a region of sunlight in each subcell to improve performance by reducing thermalization loss^[56]. The schematic of the different spectral distribution is shown in Fig. 2(c). In the tandem solar cell, photons with higher energy are absorbed in the top subcell, whereas the remaining lower energy photons enter the device and are absorbed by the bottom subcells. Consequently, the combination of perovskites layers with different bandgaps enables maximum utilization of the solar spectrum. Consequently, the theoretical PCE limits are increased to 44.3% for double-junction cells and even higher than 50% for triple-junction cells^[57].

Theoretically, the open-circuit voltage (V_oc) in a tandem solar cell is nearly the sum of the V_oc of two subcells, where the V_oc loss is mainly caused by the tunnel recombination junction. The short-circuit current density (J_sc) value in the device is the smaller one of two subcells. Furthermore, the thickness of subcells has a significant effect on light capturing and thus the resulting J_sc. Both bandgap and absorber thickness matching are required in all-perovskite tandem solar cells to achieve optimal performance. Fig. 2(d) shows the maximum PCE of tandem solar cells with different bandgap matching. For all-perovskite tandem solar cells, the best bandgap matching is considered to be that the bottom subcell has a bandgap around 1.2 eV and the top subcell has a bandgap around 1.8 eV^[58]. Performance and bandgap matching in all-perovskite tandem solar cells are summarized in Table 1.

2.2 Bandgap tuning of perovskite materials

Perovskite materials have a general structural formula: ABX₃, A⁺ is usually an organic cation, B²⁺ is a metal ion, and X^- is a halogen ion. We list the different substitution types in Fig. 3(a). The bandgap can be continuously tune between 1.2 and 2.5 eV by adjusting the composition^{[21–24, 27–31, 59]}.

2.2.1   Pure lead wide-bandgap perovskite

This is the most common strategy to obtain a perovskite material with a wider bandgap by partial substitution of I^– with Br^–. For a typical MAPbX₃ perovskite, the bandgap can be continuously widened from 1.58 eV (0%-Br) to 2.28 eV (100%-Br) by increasing the proportion of Br^- at X-site. It can be seen from Fig. 3(b) that when the bromine content in MAPb(I_1–xBr_x)₃ is gradually increased, the onset absorption band is tuned from 786 to 544 nm^[60]. The same tendency is observed in FA-based (Fig. 3(c))^[24] and Cs-based (Fig. 3(d))^[21] lead halide perovskites. The difference is that FAPbI₃ (1.48 eV) has a slightly lower bandgap than MAPbI₃ (1.57 eV), while CsPbI₃ (1.73 eV) has a slightly wider bandgap than MAPbI₃. At the same time, the all-inorganic wide-bandgap perovskite shows the potential of processing stable tandem solar cells^{[61, 62]}.

Partial substitution of the A-site using DMA⁺ (DMA is dimethylammonium) also leads to widening of the bandgap (Fig. 3(e))^[63]. It has been reported that halide segregation in a wide-bandgap perovskite can be effectively suppressed by adjusting the ratio of DMA⁺ at the A-site, thereby improving stability of PSCs. Recent studies have shown that alloying Rb⁺/Cs⁺/MA⁺/FA⁺ at A-site are helpful to improve the photo-stability under illumination^[64–67].

2.2.2   Mixed Pb–Sn narrow-bandgap perovskite

Partial substitution of Sn²⁺ with Pb²⁺ at B-site can further lower the bandgap of tin halide perovskites^[25–31]. Freeman et al. observed that the MASn_xPb_1–xI₃ perovskites have a minimum bandgap of 1.17 eV (determined by the absorption onset at 1060 nm, the actual optical bandgap is ~1.25 eV)^[69]. Snaith et al. found the same trend in FASn_xPb_1–xI₃ perovskites. For compositions with > 50% Sn content, a special type of the Pb–Sn positions allowed the bandgap to shift lower than the end point, causing the bowing trend^[53]. The bandgap tuning by varying the tin content in FA and FA/MA perovskites is shown in Figs. 3(f) and 3(g).

Additionally, the radius of the A-site ion can tune the bandgap of the perovskite by affecting the structure of the lattice^[70]. The substitution of MA⁺ with FA⁺ and Cs⁺ at A-site also affects the bandgap. A narrow-bandgap perovskite can be obtained by introducing FA⁺ at the A-site, possibly because FA⁺ have a slightly larger ionic radius than MA^+[71]. Jen et al. achieved improved stability while maintaining high performance for narrow-bandgap Sn-based PSCs by introducing different amounts of FA⁺ into MAPb_0.75Sn_0.25I₃^[29]. McGehee et al. observed an abnormal phenomenon that larger ratio of Cs⁺ can slightly reduce the bandgap of FA_1–xCs_xPb_ySn_1–yI₃ narrow-bandgap perovskites when Sn content is higher than 50%^[70].

3. Recent progress in monolithic all-perovskite tandem solar cells

We can divide the structure of all-perovskite tandem solar cells into three crucial parts: the top subcell, the bottom subcell, and the tunnel recombination junction (interconnecting layer). We next summarize the progress in all-perovskite tandem solar cells in these three parts. Details of all-perovskite tandem solar cells developed so far can be found in Table 1.

3.1 Tunnel recombination junction

One challenge in fabricating tandem solar cells is how to protect the top subcells when depositing the bottom subcells using solution processing. A tunnel recombination junction (also known as interconnecting layer) is required for electrical series connection between subcells in all-perovskite tandem solar cells. The tunnel junction must meet following requirements: (1) It must form ohmic contact with the charge extraction layers, promote the recombination of electrons and holes from subcells, and lead to as small series resistance loss as possible. (2) It must be compact enough to protect the top subcell from damage when depositing the bottom subcell. (3) It must have sufficient optical transparence to ensure low parasitic absorption.

The structure of tunnel recombination junction is continuously evolving in the development of all-perovskite tandem solar cells. At present, conductive transparent materials are deployed in tunnel junction for tandem solar cells, such as N4,N4,N4”, N4”-tetra([1,1’-biphenyl]-4-yl)-[1,1’:4’,1”-terphenyl]-4’,4”-diamine(TaTm) doped with 2,2’-(perflfluor-onaphthalene-2,6-diylidene) dimalononitrile(TaTm:F6-TCNNQ), poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS), aluminum doped zinc oxide (AZO), and indium tin oxide (ITO). Deposition techniques include vacuum deposition^[72], sputter coating^[73] and film transfer lamination^[74], thermal evaporation^[54] and atomic layer deposition^[63].

There are two strategies to avoid damaging the top subcell when the bottom subcell is deposited on top: (1) process the bottom subcell using non-solution approach; and (2) deploy a dense tunnel recombination junction to prevent the solvent from damaging the top subcell. We will introduce the recent progress of tunnel recombination junction used in all-perovskite tandem solar cells from following three aspects.

3.1.1   Lamination for connecting subcells

Im et al. used a 2-micron hole-transport layer (PEDOT: PSS) as a recombination layer in their early work to assemble two independent subcells by lamination, as shown in Fig. 4(a)^[52]. Because of the lack of dense tunnel recombination junction to protect the top subcell, non-orthogonal solvents (specifically, dimethyl formamide (DMF) and dimethyl sulfoxide (DMSO)) cannot be used to deposit the bottom subcell without damaging the top subcell.

3.1.2   Tunnel junction for bottom subcell processed by non-solution methods

Vacuum deposition is a useful way to process perovskite films. Snaith et al. used spin-coated PEDOT:PSS as the tunnel recombination junction, as shown in Fig. 4(b). PbI₂ films were deposited by thermal evaporation on top of the front subcell, and then converted to perovskite by reacting with MAI in isopropanol solution through interdiffusion^[79]. However, the thickness of the perovskite layer is only about 200 nm due to the limited penetration of MAI.

Other tunnel recombination junctions have also been reported to protect top subcells when processing bottom subcells using vacuum deposition. Sessolo et al. developed the N4, N4, N4″, N4″ -tetra([1,1′-biphenyl]-4-yl)-[1,1′: 4′, 1″ -terphenyl] -4,4″ -diamine(TaTm) as the charge recombination layer, whose conductivity can be increased by two orders of magnitude after doping with 2,2′-(Perfluoronaphthalene-2,6-diylidene) dimalononitrile (F6-TCNNQ)^[72]. The structure of this device is shown in Fig. 4(c). Vacuum deposition was used to deposit MAPbI₃ onto the wider bandgap perovskite subcell with a composition of Cs_0.15FA_0.85Pb (I_0.3Br_0.7)₃. The champion device achieved a PCE of 18% with a relatively low matched J_sc value of only 9.2 mA/cm². In another work by Bolink et al., F6-TCNNQ-doped TaTm was also chose as the charge recombination layer. The thickness of the top and the bottom CH₃NH₃PbI₃ layer were relatively different (95 and 420 nm) by controlling the vacuum deposition process^{[80, 81]}. The tandem solar cell achieved a high PCE of 18.02% but a low J_sc value of 9.84 mA/cm². When vacuum deposition is used to process the bottom cell, a particularly dense tunnel recombination junction is not required because there is no potential solvent to damage the top cell. This advantage makes it a potential strategy to fabricate all-perovskite tandem solar cells. However, efficient mixed Pb–Sn narrow-bandgap PSCs are challenging to process using vacuum deposition approach.

3.1.3   Dense tunnel recombination junction for full solution processing of perovskites

Processing a dense tunnel recombination junction is considered to be the best strategy to protect the top subcell from solvents in subsequent processing. In 2015, Zhou et al. firstly demonstrated bottom-up solution-processed all-perovskite tandem solar cells via designing a novel tunnel recombination junction: Spiro-OMeTAD/PEDOT:PSS/PEI/PCBM:PEI, as shown in Fig. 4(d). This tunnel recombination junction is efficient to collect electrons and holes from its top and bottom subcells, and robust enough to protect the top perovskite film during the bottom perovskite film deposition^[74]. The use of sputtered ITO offers another candidate for fabrication of tandem solar cells.

Sputtered ITO is widely used in semitransparent solar cells as the transparent electrode due to its high optical transparency in the visible and near-infrared (NIR) regions^{[82, 83]}. Low resistivity and high optical transparency of sputtered ITO make it suitable as the tunnel recombination junction in tandem solar cells as well. In 2016, Snaith et al. firstly chose sputtering-deposited ITO as the charge recombination layer and ALD-deposited tin oxide (SnO₂) as the buffer layer, as shown in Fig. 4(e)^[53]. The ITO layer was mainly used as the charge recombination layer and to protect the top subcell from being dissolved during subsequent perovskite processing. Here the SnO₂ layer was mainly deposited to protect the top subcell from sputtering. They obtained tandem solar cells with a PCE of 16.9% and a V_oc over 1.65 V. However, the J_sc values exhibited by the top and bottom subcells were not an ideal match (14.1 and 15.8 mA/cm²).

Since then, ITO based tunnel recombination junction has been widely used in p–i–n structured all-perovskite tandem solar cells. The difference is that in order to reduce the damage to the top subcell during the sputtering process, different protective buffer layers are searched. Jen et al. sputtered ITO as the charge recombination layer on C₆₀/Bis-C₆₀ (Fig. 4(f)). By combining a narrow-bandgap (1.2 eV) perovskite and a wide-bandgap (1.8 eV) perovskite, using indene-C₆₀ bis-adduct (IC₆₀BA) as the electron transport layer, They achieved a V_oc of 1.98 V, which reached 80% of the theoretical limit^[73]. Multi-layer structures as tunnel recombination junction are an effective way to improve device stability. Yan et al. designed a vacuum-processed multiple stack of ultrathin Ag (1 nm), MoO_x (3 nm) and ITO (~120 nm) as the tunnel recombination junction to construct a configuration of ITO/PTAA/FA_0.8Cs_0.2Pb(I_0.7Br_0.3)₃/C₆₀/BCP/Ag/MoO_x/ITO/PEDOT:PSS/(FASnI₃)_0.6(MAPbI₃)_0.4/PCBM/BCP/Ag all-perovskite tandem solar cell^[76]. A schematic of the tandem solar cell is shown in Fig. 4(g). Ag and MoO_x layers were deposited by thermal evaporation, and then ITO was sputtered on the MoO_x layer. The MoO_x layer prevented damaging to the BCP/Ag layers caused by the sputtering of ITO, and ensured good contact between Ag and ITO. The MoO_x layers make the top subcell film very smooth, forming a smooth, pinhole-free tunnel recombination junction that protects the top subcell from potential solvents during bottom subcell depositing processing, even the deposition of water-based PEDOT: PSS. In addition, the tunnel recombination junction had a transmittance of more than 70% in the range of 720 nm to nearly 900 nm, which ensured great light absorption of the bottom subcell. The champion device achieved a PCE of 20.6%. Moreover, better device performance cannot be achieved when the Ag layer or MoO_x layer is removed separately. Following that, Zhu et al. used the same structure to combine different components of perovskites, and achieved a PCE more than 23%^[77]. They concluded that the PCE of tandem solar cells is mainly limited by the V_oc loss in the wide-bandgap perovskite and the low J_sc value in the device, and the optical loss of the tunnel recombination junction is the main reason for the low J_sc. The high-power deposition process and the large optical loss caused by the large refractive index of the ITO film are important factors that limit the further improvement of tandem device performance.

Atomic layer deposition (ALD) is the key technology for constructing a compact buffer layer. The use of a nucleation layer to improve the compactness of the tunnel recombination junction has recently been demonstrated by Moore et al.^[63]. The schematic of the tandem solar cell is shown in Fig. 4(h). This strategy makes the ALD layer itself a barrier to effectively prevent sputtering and solvent damage.

During the same time, our group fabricated all-perovskite tandem solar cells with a configuration of glass/ITO/PTAA (poly(triarylamine)/wide-E_g perovskite/C₆₀/ALD-SnO₂/Au(~1 nm)/PEDOT:PSS/low-E_g perovskite/C₆₀/BCP/Cu^[54]. The schematic about the tandem solar cell is shown in Fig. 4(i). This compact and robust ALD-SnO₂ layer (~20 nm) can efficiently prevent damage to the deposited-underlying top subcell during the solution processing of the bottom subcell, allowing the achievement of fabricating all-perovskite tandem solar cells without the ITO layer. The SnO₂ layer provides excellent electron extraction as well, resulting in a slight improvement in the performance of the wide-bandgap top subcell. An ultra-thin Au layer (~1 nm) was deposited on SnO₂ by thermal evaporation to promote electron-hole recombination. The champion tandem solar cell achieved the highest PCE of 24.8% reported so far, with a high V_oc of 1.965 V, a J_sc value of 15.6 mA/cm² and a high FF of 81%.

3.2 Progress in wide-bandgap perovskites

The performance of tandem solar cells closely depends on the performance of each. Compared with narrow-bandgap perovskites, wide-bandgap perovskites have been studies more extensively^{[84, 85]}. In the previous works, wide-bandgap perovskites are used for tandem solar cells in combination with mature narrow-bandgap semiconductors, such as silicon^[32–36], and chalcogenide^[37–42].

In the early stages of developing all-perovskite tandem solar cells, Im et al. and Ho-Baillie et al. used CH₃NH₃PbBr₃ (2.3 eV) as top subcell^{[52, 79]}. The PCEs have been limited to ~10% due to the nonideal bandgap matching between subcells. In order to search a top subcell to match MAPbI₃ (1.55 eV) as the bottom subcell, Sessolo et al. synthesized a wide-bandgap perovskite of Cs_0.15FA_0.85Pb(I_0.3Br_0.7)₃ (2 eV). The tandem solar cell achieved a PCE of 15.6%^[72].

According to the bandgap matching rules in tandem solar cells, researchers often choose the combination of wide-bandgap (~1.8 eV) perovskites and narrow-bandgap (~1.2 eV) perovskites to fabricate tandem solar cells. In 2016, Snaith et al. achieved a high-efficiency, stable perovskite of FA_0.83Cs_0.17Pb(I_0.5Br_0.5)₃ (1.8 eV) by using a mixture of FA⁺ and Cs⁺ cations and adjusting the Br^–/I^– ratio^[53]. In a tandem solar cell using a perovskite layer with a 1.8 eV bandgap as the top subcell, they achieved a PCE of 16.9%, with a V_oc of 1.66 V, a J_sc value of 14.5 mA/cm² and a FF of 70%, as shown in Fig. 5(a).

In tandem solar cells, the V_oc is more provided by the wide-bandgap subcell^[84]. The V_oc loss due to halide phase segregation is a common challenge for wide-bandgap perovskites. Introducing strain into the crystal lattice by ion doping, tilting of [BX₆] octahedron, introducing low-dimensional perovskite structures are strategies to widen the bandgap and to reduce the V_oc loss^{[85, 86]}. Jen et al. alleviated the halide segregation in wide-bandgap perovskite by adding Cs⁺ into MAPb(I_0.6Br_0.4)₃ and obtained a more stable composition MA_0.9Cs_0.1Pb(I_0.6Br_0.4), as shown in Fig. 5(b)^[73]. They obtained a subcell with a top transparent electrode by using a wide-bandgap (1.8 eV) perovskite, showing a V_oc of 1.22 V. In the all-perovskite tandem solar cell, a V_oc of 1.98 V was achieved, which reached 80% of the theoretical limit.

As a photostable perovskite component, FA_0.8Cs_0.2Pb(I_0.7Br_0.3)₃ was also used as the top subcell in all-perovskite tandem solar cells^[76]. Moore et al. fabricated single-junction PSCs using DMA_0.1FA_0.6Cs_0.3PbI_2.4Br_0.6, and achieved a PCE higher than 19% and a V_oc of 1.2 V, while the single-junction PSC without DMA only achieved a V_oc of 1.15 V. They fabricated tandem solar cells with a PCE of 23.1% by combining this wide-bandgap perovskite with a narrow-bandgap perovskite of FA_0.75Cs_0.25Sn_0.5Pb_0.5I₃, as shown in Fig. 5(c)^[63].

3.3 Progress in narrow-bandgap perovskites

The PCE of all-perovskite tandem solar cells is still limited by the performance of mixed Pb–Sn narrow-bandgap PSCs. A thick absorber is required in mixed Pb–Sn PSCs to completely absorb the infrared light passing through the wide-bandgap cell^[87]. The progress in mixed Pb–Sn narrow-bandgap perovskites paved the way for achieving highly efficient all-perovskite tandem solar cells. Snaith et al. firstly reported an all-perovskite tandem solar cell using FA_0.75Cs_0.25Sn_0.5Pb_0.5I₃ as the bottom subcell^[53]. They used a strategy called precursor-phase anti-solvent immersion (PAI) to make a smooth, pinhole-free perovskite film, which took a low vapor pressure solvent to hinder crystallization and an anti-solvent bath to crystallize the film with only gentle heating^{[88, 89]}. They achieved a stable, 14.8% efficient PSC based on a 1.2 eV bandgap FA_0.75Cs_0.25Pb_0.5Sn_0.5I₃ perovskite. Consequently, the PCE of the all-perovskite tandem solar cell they achieved was 16.9%, together with a V_oc of 1.66 V, and a J_sc value of 14.8 mA/cm².

Low V_oc is one of the main efficiency losses in narrow-bandgap PSCs. Considering the mismatch between the minimum conduction bands of MAPb_0.5Sn_0.5I₃ and C₆₀, Jen et al. used IC₆₀BA, an alternate fullerene derivative, as the electron transport layer to achieve better energy level alignment^[73].

The fast crystallization of mixed Pb–Sn perovskites makes it difficult to achieve good film uniformity and good photoelectric performance, resulting in short carrier lifetimes of photo-generated s (typically one the order of ns)^[53]. McGehee et al. reported a post-treatment using methylammonium chloride vapor to grow larger grains and to repair cracks in the deposited film, thereby increasing the V_oc and FF. After MACl post-processing and the addition of formic acid, the carrier lifetime was extended to nearly 0.5 µs, and the single-junction narrow-bandgap PSC achieved a stable PCE of 15.6%^[75]. Tandem solar cells with a stabilized PCE of 19.1% was obtained. The top and bottom subcells showed J_sc values of 15.5 and 14.8 mA/cm², respectively. The EQE and J–V cures of this tandem solar cell are shown in Figs. 6(a) and 6(b).

Reducing non-radiative recombination loss in bulk perovskite is an important strategy to improve the performance of narrow-bandgap PSCs. Yan et al. introduced lead chloride in the precursor solution to expand the grain size, increases crystallinity and carrier mobility, and reduce the electronic disorder^[76]. This enhancement allowed them to successfully process a high-efficiency, 700 nm-thick, 1.25 eV narrow-bandgap PSCs. The cells achieved a best PCE of 18.40% with a V_oc of 0.842 V, a J_sc value of 29.40 mA/cm² and a FF of 74.4%. With a 1.75 eV wide-bandgap perovskite top subcell, they achieved a PCE of 21% and a V_oc of 1.922 V in tandem solar cells. The EQE and J–V cures are shown in Fig. 6(c) and 6(d).

Huang et al. revealed that the charge collection efficiency in mixed Pb–Sn PSCs is mainly limited by a short diffusion length of electron^[90]. They reduced the hole concentration and electron trap density by adding 0.03 mol% of Cd²⁺ into the precursor solution, and the electron diffusion length increased to 2.72 µm. They fabricated an all-perovskite tandem solar cell with a configuration of ITO/PTAA/FA_0.6Cs_0.4Pb(I_0.65Br_0.35)₃/C₆₀/SnO₂/ITO/PEDOT:PSS/PTAA/Cd-FA_0.5MA_0.45Cs_0.05Pb_0.5Sn_0.5I₃/C₆₀/BCP/Cu, achieving a stable PCE of 22.7% in tandem cells. The EQE spectra showed that the J_sc values of the top and bottom subcells is 15.2 and 15.1 mA/cm², respectively, as shown in Figs. 6(e) and 6(f). Zhu et al. used guanidinium thiocyanate (GuaSCN) to improve the carrier diffusion length of narrow-bandgap mixed Pb–Sn perovskites to 2.5 μm^[77]. Narrow-bandgap single-junction solar cells exhibited PCEs of 20.5% and 20.4% in forward and reverse scans, respectively. Following this strategy, the tandem solar cell achieved a PCE of 23.4%, with a V_oc of 1.942 V, a J_sc value of 15.01 mA/cm², and a FF of 80.31%. The EQE spectra showed that the J_sc values of the top and bottom subcells are 14.85 and 14.67 mA/cm², respectively (Fig. 6(g)).

Recently, our group improved the carrier diffusion length up to 3 μm, and achieved a PCE of 21.1% in single-junction narrow-bandgap PSC with a bandgap of 1.22 eV^[54]. We achieved this by reducing Sn vacancies caused by Sn²⁺ oxidation via a comproportionation reaction by adding Sn powders in the precursor solution. By increasing the thickness of the narrow-bandgap perovskite to 860 nm, we achieved maximum J_sc value above 32 mA/cm². The devices had high EQE values in the near infrared region. By combining a 1.77 eV wide-bandgap perovskite (about 300 nm), we fabricated all-perovskite tandem solar cells with a configuration of glass/ITO/PTAA/wide-E_g perovskite/C₆₀/ALD-SnO₂/Au (~1 nm)/PEDOT:PSS/low-E_g perovskite/C₆₀/BCP/Cu. We achieved a high PCE of 24.8%, with a V_oc of 1.965 V, a J_sc value of 15.6 mA/cm², and a high FF of 81% in a small-area tandem solar cell. The integrated J_sc values of the top and bottom subcells from EQE curves are 15.7 and 15.5 mA/cm², respectively. The EQE and J–V cures are shown in Figs. 6(h) and 6(i). The tandem solar cell retained 90% of initial performance following 463 h of operation at the maximum power point (MPP) under full 1-sun illumination (Fig. 6(j)). We also fabricated a large area all-perovskite tandem solar cell following the same process and achieved a high PEC of 22.3%.

4. Conclusion and perspective

The past five years have witnessed an impressively rapid progress in the development of all-perovskite tandem solar cells. The PCE has achieved a boost from less than 10% at the beginning to nearly 25% at present. However, there remain several challenges that need to be addressed to achieve higher PCEs beyond 30%. We suggest that researchers in the community shall pay close attention to following issues to obtain better performed all-perovskite tandem solar cells.

(1) Tunnel recombination junction is a crucial component in all-perovskite tandem solar cells. A tunnel recombination junction should have good optical transmission, low series resistance, sufficient robustness, large area processability, and low cost. Developing new tunnel recombination junctions is important to further reduce the parasitic absorption and V_oc loss.

(2) Large V_oc loss in wide-bandgap PSCs remains a considerable challenge for all-perovskite tandem solar cells. Phase segregation in wide-bandgap perovskite is one of the reasons for such large V_oc loss. In addition, interfacial recombination is another cause of V_oc loss.

(3) Achieving efficient tandem solar cell relies on high-performance mixed Pb–Sn PSCs. Oxidation of mixed Pb–Sn perovskites remains to be addressed for commercial viability.

In addition, current matching in tandem solar cells is one of the key factors that can affect the overall efficiency of the device. It is important to establish current matching by adjusting both the bandgaps and the thickness of perovskite subcells. Developing suitable encapsulation is another key factor ensure excellent environmental stability for commercial viability.

Acknowledegments

This work is financially supported by the National Key R&D Program of China (2018YFB1500102), National Natural Science Foundation of China (61974063), Natural Science Foundation of Jiangsu Province (BK20190315, BZ2018008), Program for Innovative Talents and Entrepreneur in Jiangsu, and Thousand Talent Program for Young Outstanding Scientists in China.