Introduction

Environmental pollution and global energy shortages are the two most serious current threats to human life. To rival the fossil energy, solar energy must be utilized in a cost-effective way. In fact, silicon solar cells have obtained power conversion efficiency (PCE) of up to 25 % in the case of single-crystal silicon [1], and 20.4 % in the case of multi-crystalline silicon [2]. The production of such materials, however, requires relatively expensive production lines and energetically demanding processes. Fortunately, thin film solar cells can reduce energy payback time and material costs, for example, DSSCs. It stands for a new class of electrochemical solar cells which includes sensitized mesoscopic titanium oxide (TiO2), consisting of a catalyst layer and a liquid electrolyte in a sandwich-like architecture. Recently, the highest efficiency of this type of device may achieve 12.3 % [3]. However, there are still problems for the commercial DSSC, such as the potential leakage of liquid electrolytes and aging stability [4–6].

To avoid those problems, solid state DSSCs with a hole transport material (HTM) replacing the liquid electrolyte have been proposed, In this kind of device, the organic–inorganic lead halide perovskite compounds are used as a light-harvesting active layer. As light is absorbed in the perovskite, holes would be transferred to the HTM. Meanwhile, electrons would be transferred to the TiO2. Figure 1 illustrates the primary structure for the emerged cell, i.e., pervoskite solar cells.

Fig. 1
figure 1

Primary structure of fluorine-doped tin oxide (FTO)/TiO2/perovskite/HTM/Ag solar cells

Full size image

Figure 2 presents a diagram of the recent development of perovskite solar cells. Perovskite materials have been well known for many years, but the first incorporation into a solar cell was reported by Kojima et al. [7]. Their group first attempted to use perovskite compounds as a sensitizer and iodine redox in conventional DSSCs. The perovskite nanocrystalline particles self-organized on TiO2 showed a photovoltaic effect, with a high photo voltage of 0.96 V and an efficiency of 3.8 % based on perovskite CH3NH3PbX3 cells. Note that the voltage of such pervoskite solar cells is higher than that of DSSCs. Two years later, Im et al. [8] reported a higher efficiency of 6.54 % achieved based on CH3NH3PbI3 semiconductor quantum dots (QDs) cells. The QDs were on a mesoporous TiO2 surface as absorbers. In reality, configuration of the semiconductor quantum-sensitized solar cells is analogous to the DSSCs, while the dye is replaced by 2–3-nm sized CH3NH3PbI3 QDs with no liquid electrolytes involved.

Fig. 2
figure 2

Significant advances in perovskite solar cells since 2009

Full size image

In 2013, important progress on the preparation method and structure was made for the emerged pervoskite solar cells. One example was the replacement of the one-step coating procedure with a two-step coating method. The latter method used CH3NH3I intercalation following layered PbI2 deposition to improve perovskite quality. As a result, the PCE was enhanced up to 15 % [9]. Another example was a co-evaporation technique was first proposed to form a perovskite layer for planar heterojunction (PHJ) perovskite solar cells. This method exhibited a PCE as high as 15.4 % [10]. In 2014, Yang’s group exploited many efforts to manipulate carrier behavior in PHJ perovskite solar cells: (1) reducing the work function of the indium tin oxide (ITO) layer to enhance the carrier concentration, and (2) doping yttrium into TiO2 electron transport layers to improve the electron transport channel. Those efforts offered an efficient carrier pathway in the perovskite film and relevant interfaces, and finally gave rise to a very pronounceable efficiency of 19.3 % [11].

More details regarding the preparation methods and functional layers in the emerged perovskite solar cells are addressed in the following sections.

General preparation methods of perovskite solar cells

A typical preparation process of perovskite solar cells includes electrode etching, compact layer deposition, mesoporous and perovskite layer coating, as well as counter electrode deposition. These steps are generally illustrated in Fig. 3.

Fig. 3
figure 3

A typical preparation process of perovskite solar cells. ALD atomic layer deposition

Full size image

Firstly, hydrochloric acid (HCl) and zinc (Zn) powder is used to etch the FTO substrate to achieve the required electrode pattern. A compact TiO2 layer is formed by spray pyrolysis or spin-coating of titanium diisopropoxidebis (acetylacetonate) solution in isopropanol. Then, TiO2 paste is spin-coated onto the compact TiO2 layer to form the mesoporous TiO2 film. The combination film is annealed at 455–500 °C. Afterwards, the film is treated by TiCl4 for 10–30 min at 70 °C and then sintered at 455–500 °C.

For the synthesis of CH3NH3PbI3 on the TiO2 surface, earlier work applied one-step to complete the step. For example, equimolar amounts of CH3NH3I and PbI2 were dissolved in dimethyl formamide (DMF) at 70 °C to prepare CH3NH3PbI3 precursor solution. CH3NH3PbI3 film is induced by spin-coating and annealed at 100 °C. Recently, an effective method, the so called two-step deposition technique, was developed. For the two-step coating method, the solution of PbI2 is first spin-coated onto TiO2 film. In order to remove DMF solvent, the film should be heated to 70–100 °C. After that, the PbI2 film intercalates in a 10 mg mL−1 solution of CH3NH3I, and is then dried at 70–100 °C to form the crystalline phase [6]. Finally, the HTM solution is spin-coated on the CH3NH3PbI3 film. Au/Au thermals evaporate above the HTM layer to form the counter electrode.

It should be noted that the above procedures may imply more significant concerning issues, as follows: (1) the perovskite is formed, which is indicated by the color of the CH3NH3PbI3 layer turning dark brown, (2) moisture should be controlled, since perovskite easily decomposes in the presence of moisture [11].

Development of the composition and processing of the compact layer

The perovskite solar cell fabrication process begins with deposition of a dense compact layer on top of the FTO substrate. The dense compact layer prevents direct contact between the FTO and the hole conductor layer. Such a dense compact layer is also called the “blocking layer”. Normally, TiO2 can be used as a dense, compact layer material. The thickness of the dense, compact layer is a key parameter, varying in the range of 50–120 nm. Too thick or too thin a layer will reduce the efficiency or fill factor.

Wu et al. [12] reported that atomic layer deposition (ALD) technology can prevent pinholes from forming. Nanoscale pinholes with higher density will reduce V oc and increase charge recombination; the latter will decrease the short-circuit current density (J sc). Compared with spin-coating and the spray pyrolysis method, the ALD fabrication process achieved high shunting resistance as well as a high PCE. Hu et al. [13] demonstrated that the ALD–TiO2-based device presents a significant advantage. The surface morphology of the compact TiO2 layer and platinum nanoparticle size may be easily moderated via the number of ALD cycles.

In contrast to the ALD–TiO2-based device, the spin-coating and spray pyrolysis methods obtained even higher efficiency in other research groups, as summarized in Table 1.

Table 1 Photovoltaic parameters of perovskite solar cells depending on the device structure and blocking layer formation
Full size table

TiO2 as the compact layer material can produce excellent performance. However, to date, the most efficient perovskite solar cells still demand a 500 °C sintering step, which may limit substrate options. Fortunately, several low-temperature processable materials have been found. ZnO is one viable alternative for the compact layer in pervoskite solar cells. Besides, ZnO has high electron mobility with low recombination losses as compared to TiO2. It is easy to process into complex nanostructures with a low-temperature sintering step [25, 26]. Xu et al. [24] reported that a high PCE of 13.1 % was achieved from perovskite solar cells based on a FTO/compact ZnO/Al2O3/CH3NH3PbI3/spiro-MeOTAD/Ag structure. They demonstrated that compact ZnO layers were deposited by the ALD method at 70 °C.

Advances in mesoporous layer composition

TiO2 and Al2O3 films for a mesoporous layer

Normally, the mesoporous TiO2 film is coated on the dense compact layer. The perovskite film is then deposited on the surface of mesoporous TiO2. The thickness of the perovskite film is a key parameter. If the film is too thin, this region fails to absorb sufficient sun light. If the film is too thick, there may be a significant chance that the film thickness will be longer than the electron and hole diffusion length [10].

In perovskite solar cells, TiO2 is the ideal material for the electron transport layer. To date, TiO2 nanoparticles are the most commonly used as the efficient electron transport layer in both DSSCs and perovskite solar cells. The highest PCE of 19.3 % was achieved in the TiO2 nanoparticles/CH3NH3PbI3/spiro-MeOTAD device [11].

The electron charge transport processes for mesoporous TiO2 -based perovskite solar cells are shown in Fig. 4a. As one of the cheap film production methods, spin-coating is widely used in solution-processed solar cells to fabricate mesoporous TiO2 film. However, this method is not suitable for preparing an homogeneous perovskite layer with a large area and uniform thickness [27].

Fig. 4
figure 4

a Schematic illustration of the electron behaviour in a perovskite TiO2 solar cell, and b an insulating Al2O3 -based solar cell

Full size image

It is also suggested that TiO2 can be substituted by an other oxide to form a mesoporous layer, such as Al2O3. Lee et al. [28] obtained an efficiency of up to 10.9 % utilizing mesoporous Al2O3. It is hard for the electrons emerged in the absorber layer to flow within the mesoporous Al2O3 to the electrode. Since the conduction band edge of mesoporous Al2O3 is higher than that of CH3NH3PbI3, it is demonstrated that perovskite could function as a hole conductor and light harvester. Figure 4b shows the electron charge transport processes within Al2O3 mesoporous materials, in contrast to the case of TiO2.

As the Al2O3 acts as a “scaffold” for the perovskite, i.e., meso-superstructured solar cells (MSSCs), the CH3NH3PbI3 precursor particles filtrate into the pores of the Al2O3 films, then give rise to uniform precursor distribution. Remarkably, it is possible to apply the temperature lower than 150 °C, to deposit the Al2O3 nanoparticles from a simple binder-free colloid to form the meso-superstructured scaffold [29]. This is unlike the TiO2 nanoparticles which typically require a sintering temperature as high as 500 °C. Such a reduction in processing temperature will lead to an Al2O3–based perovskite cell with certain advantages, such as rapid production, lower cost, and the applicability to plastic substrate-based devices.

One dimensional (1-D) TiO2 nanostructures for the mesoporous layer

Several approaches develop one dimensional nanostructures (such as TiO2 nanofibers, nanotubes, nanobelts and so on) to replace TiO2 nanoparticles. Moreover, compared to nanoparticles, 1-D nanostructures exhibit two advantages in liquid DSSCs: (1) with less grain boundary effects and an extended random walk network, electrons within 1-D nanostructures may slow recombination; (2) 1-D nanostructures can provide “highways” for electrons flowing to the electrode [30–32].

Figure 5 shows the nanofiber-based perovskite solar cells, where the mesoporous nanofibers replace the mesoporous nanoparticles used in conventional perovskite solar cells. To obtain higher performance, two requirements should be taken into account: (1) smaller diameter nanofibers which form less closed structures result in more perovskite loading, which can improve the light harvesting; (2) A thicker nanofiber film may be not benefit the charge transport but will load more perovskite. It is clear that controlling the photoanode thicknesses and diameters of the nanofibers are the major goals [33].

Fig. 5
figure 5

a Schematic illustration of nanofiber-based perovskite solar cells, b field emission scanning electron microscopy (FESEM) images of a TiO2 nanofiber. Reprinted from Ref. [33]

Full size image

The efficiency of perovskite-based devices still has much room for improvement, if only considering light harvesting efficiency. This is because nanoparticle electrode-based pervoskite solar cells show poor absorption of long-range wavelength light. Gao et al. [34] reported that applying a TiO2 nanotube was regarded as a promising way to enhance light absorption in the visible region, if nanotube arrays can enhance light trapping ability. The UV–Vis absorption spectra of TiO2 nanoparticle and TiO2 nanotube-based perovskite solar cells are shown in Fig. 6a. The TiO2 nanotube-based cells demonstrate stronger absorption at 350–800 nm than the TiO2 nanoparticlebased cells.

Fig. 6
figure 6

a UV–Vis absorption spectra of TiO2 nanoparticle and TiO2 nanotube-based perovskite solar cells, b normalized adsorption spectra of TiO2 nanotube length compared with TiO2 nanoparticle-based perovskite solar cells. Reprinted from Ref. [34]

Full size image

In order to improve the efficiency of TiO2 nanotube-based perovskite solar cells, the length of the nanotube must be appropriate. When the length of a nanotube is increased, the electron diffusion pathway is prolonged, resulting in a higher recombination. Additionally, the TiO2 nanotube length is 2.3, 4.8 and 9.4 μm; the absorption is almost the same, as show in Fig. 6b. As a result, the efficiency decreased from 6.52 to 3.26 % when the TiO2 nanotube length changes from 2.3 to 9.4 μm.

Emerged structure of an organic/inorganic perovskite layer

Development of perovskite structure and bandgap

Perovskites are organic–inorganic hybrid layered materials with a chemical formula ABX3, in which X is an anion that bonds to A and B. The structure of perovskite can be described as an octahedral structure of BX6. Generally, the A represents an organic ion (CH3NH3 +, CH3CH2NH3 +) which sites at the interstices. The B component represents a metal cation (Pb2+, Sn2+) which is in the middle of the octahedra. The X sites are occupied by halide anions (Cl−, Br−, I−) located in the corner [35]. The three dimensional structure of perovskite is shown in Fig. 7a. The terminal cationic groups and larger molecules of perovskite form a simple MNMN (where M and N are alternating sheets of organic and inorganic material) layered structure. These layers are connected by Van der Waals’ forces. Figure 7b shows the layered structure of organic/inorganic perovskite.

Fig. 7
figure 7

The crystal structure (a) and layered structure (b) of the organo-lead halide perovskite. Note that a is reprinted from Ref. [10]

Full size image

The distortion and stability of the organic/inorganic perovskite structure is decisively determined by the value of tolerance factor t. The tolerance factor t is expressed as: \(t = (r_{\text{A}} + r_{\text{X}} )/\sqrt 2 (r_{\text{B}} + r_{\text{X}} )\), where r A, r B and r X correspond to the ionic radii of A, B and X, respectively. Generally, the perovskite structure can be stabilized if t dose not largely deviate from 1. As such, it can tailor the chemistry of the components in perovskite structures to regulate perovskite's intrinsic properties.

The electrical and optical characteristics of pervoskite are highly correlated with their crystalline structure. This structure can be transformed by changing the alkyl group, metal atom and halide. In order to tune the band gap, researchers have tested substituting the chemistry of the organic and inorganic components, as summarized in Table 2. These may be classified as three routes, as below.

Table 2 Energy levels for different materials acting as absorbers in perovskite solar cells
Full size table

(1) One attempt is to tailor the chemistry of the halide components. For example, as the x is turned, the bandgap of CH3NH3Pb(I1−x Br x )3(0 ≤ x ≤ 1) can be changed in a wide range from 1.5 to ~2.3 eV. Noh et al. [36] demonstrated that the V oc increased from 0.87 to 1.13 V in CH3NH3Pb(I1−x Br x )3(0 ≤ x ≤ 1)-based solar cells, which attributed to broadening of the bandgap with an increasing x. (2) The other strategy is to replace the A site cation. The whole lattice of pervoskite can be expanded or contracted by a smaller or larger A cation, so the bandgap can be tuned by controlling the size of A cation [37]. As in CH3CH2NH3PbI3, Im et al. [38] replaced CH3NH3 + with a long chain CH3CH2NH3 +, increasing the band gap to 2.2 eV. (3) The last effort is to modify the metal cation. The optical band gap of the current CH3NH3PbX3 perovskite is 1.55 eV, indicating a poor absorption in the 400–800 nm range. Therefore, incorporation of stable low-band-gap perovskite along with broadening of the light absorption to the near infrared spectrum is a plausible way. Hao et al. [39] reported the bandgap of CH3NH3Pb1−x Sn x I3 perovskite absorber ranged between 1.1 and 1.7 eV by varying x, extending light absorption into the near-infrared (1.1 eV or 1050 nm).

Planar heterojunction perovskite solar cells

In the case of structure, there are two basic types of the perovskite solar cells namely: PHJ solar cells (PHJSCs) and mesoporous scaffold solar cells (MSSCs). PHJSCs have the advantage of a simple architecture with a thin layer of perovskite sandwiched between the hole and electron transport layers. Liu et al. [10] reported that a PHJSC with a device structure consisting of FTO/compact TiO2/CH3NH3Pb(I1−x Cl x )3 (0 < x < 1)/spiro-MeOTAD/Au can lead to a PCE of 15.4 %, a high performance. High-quality perovskite film was fabricated by vacuum evaporation, and this film exhibited the highest thin film coverage and the most homogeneous morphology, as showed in Fig. 8.

Fig. 8
figure 8

Cross-sectional SEM images of perovskite film by a vacuum evaporation and b solution processed. Reprinted from Ref. [10]

Full size image

However, the vacuum evaporation technique demands energy to obtain a high vacuum, which hinders mass production. Yang et al. [11] demonstrated a solution-processed PHJSC with dense Yttrium-doped TiO2 as an electron transport layers and CH3NH3PbI3−x Cl x as an absorber layer. Figure 9a shows perovskite film was conducted from solution. The perovskite film was prepared by the one-step deposition method: PbCl2 and CH3NH3I were mixed in a 1:3 molar ratio in anhydrous DMF to produce a precursor solution, and then the solution was spin-coated onto TiO2-coated ITO substrates. The fabrication of their PHJSC obtained an extremely high V oc of 1.15 V, an FF of 0.76, a J sc of 22.80 mA/cm2, and a PCE of 19.3 %, as shown in Fig. 9b.

Fig. 9
figure 9

a Cross-sectional SEM images of perovskite solar cells, b current–voltage curve and performance characteristics for perovskite solar cells. Reprinted from Ref. [11]

Full size image

Compared to MSSCs, the PHJSC doesn’t need high-temperature processing to sinter mesoporous structures. Besides, there is also no complex nanostructure in the PHJSC. Those two aspects pave the way for future industrial-scale production of solar cells using a roll-to-roll continuous manufacturing process as demonstrated in organic photovoltaic devices in industry [44–46]. Recently, plenty of techniques have been evaluated as possible means of industrial-scale fabrication. For example, Barrows et al. [47] used spin-coating to create perovskite films for the fabrication of PHJ CH3NH3PbI3−x Cl x perovskite solar cells. The PCEs oreached 11 %.

Development of the composition of the hole-transport layer

HTM, which acts as the electron-blocking layer, should be infiltrated into the pores of the mesoporous TiO2 layer. The ideal HTM should have good thermal characteristics, hole mobility, UV stability, and must have a compatible highest occupied molecular orbital (HOMO) energy level relative to the perovskite. So far, only a few materials are used as an effective HTM. It has been reported that spiro-OMeTAD (2,2′,7,7′-tetrakis-(N,N-di-p-methoxyphenylamine) 9,90-spirobif-luorene) is the most successful organic HTM for fabricating perovskite solar cells. The spiro-OMeTAD shows good pore filling of the TiO2 film, low recombination rates and efficient charge transport. Recently, a perovskite solar cell efficiency greater than 19 % was obtained with spiro-OMeTAD.

However, spiro-OMeTAD is tedious and expensive to obtain. For the sake of lower-cost materials, several polymer HTMs having already demonstrated the possibility of replacing spiro-OMeTAD in perovskite solar cells have been investigated, such as polytriarylamine (PTAA), poly(3-hexylthiophene) (P3HT), poly[2,1,3-benzothiadiazole-4,7-diyl[4,4-bis(2-ethylhexyl)-4H-cyclopenta[2,1-b:3,4-b′]dithiophene-2,6-diyl]] (PCPDTBT) and so on. Typically, the perovskite solar cells with P3HT achieved a PCE of 12.4 % and PTAA exhibited a 12 % PCE [48–52]. Moreover, amorphous small-molecule HTMs are also promising candidates due to their low cost, mechanical flexibility and easily prepared crystalline thin films. Song et al. [53] demonstrated the synthesis of several TPB-based low molecular weight HTMs, and obtained a PCE up to 13.10 % with TPBC as the HTM in perovskite solar cells. In principle, for HTMs to be suitable for perovskite solar cells, it should primarily fulfill following requirements: (1) high hole mobility, (2) low-cost synthesis, (3) HOMO energy level well-matched to the perovskite [54]. Figure 10 shows cost reductions and simplifications of the synthetic HTM methodology.

Fig. 10
figure 10

Research progress of HTMs: cost reductions and simplified synthetic methodologies

Full size image

In general, most of the HTMs are polymers whose purification and synthesis are often tedious and expensive. It is useful and necessary to develop and improve the performance of HTM-free solar cells. More importantly, the other advantage HTM-free solar cells is an improved stability. The working principle of the HTM-free solar cells can be expressed as: on light illumination, the light absorber injects holes into the gold contact while electron transport to the TiO2 layer occurs [58]. Recently, Etgar et al. [42] reported that the hole conductor-free CH3NH3PbI n Br3−n perovskite solar cells showed very good stability, with the PCE as high as 8.54 %. The hybrid PbI2 + MABr/MAI 1:2 perovskite hole conductor-free solar cells showed the highest stability after 80 days without affecting their efficiency. The structure and energy level is presented in Fig. 11.

Fig. 11
figure 11

Structure and energy level of a hole conductor-free perovskite solar cell. Reprinted from Ref. [42]

Full size image

Conclusion and outlook

In the present work, the photovoltaic mechanisms, device structures, processing methods and applied materials for the emerging perovskite solar cells were reviewed. Significant progress has been made on the perovskite solar cells, as evidenced by frequent enhancements in PCE to over 15 %, up to the highest of 19.3 % in the past 2 years. This demonstrates the immense potential for commercial application.

Further research is still important and expected. Firstly, Al2O3 as the mesoporous film and ZnO as the compact film can be prepared by an evaporation process or low-temperature solution deposition. It is, thus, believed that industrial-scale production of perovskite solar cells through a roll-to-roll processing method is possible. Secondly, environmentally friendly materials are highly desired. It is necessary to investigate alternative metal ions, such as tin (Sn). Thirdly, in view of reducing cost, a carbon layer has been introduced to replace Au or Ag for charge collection and transport. Meanwhile, polymer and inorganic HTMs like CuSCN, even HTM-free solar cells, have also been developed. Last but not least, the perovskite CH3NH3PbX3 tends to decompose in moisture or oxygen. Hence, it is urgent that the long-term stability of perovskite solar cells is increased.

Clearly, with further improvement in performance, the emerging perovskite solar cells appears to be a remarkable step towards the realization of low-cost, high-efficiency, environmentally benign, next-generation, solid-state solar cells.