We have recently witnessed a breakthrough in highly efficient solar cells, where the organolead halide perovskite, CH3NH3PbI3, was used as an absorber of sunlight. The perovskite was used to sensitize mesoporous TiO2 films in a solid-state mesoscopic solar cell to deliver a power conversion efficiency (PCE) of up to 9.7% [1]. A short period later, mesoporous Al2O3 was employed as a scaffold to support the formation of continuous thin films of a mixed halide perovskite, CH3NH3PbI3-xClx, to form nonsensitized solar cells. This so-called “meso-superstructure” perovskite solar cell had a PCE as high as 10.9% [2]. This study showed that the mixed halide perovskite could function as both a light absorber and an electron transporter. The PCE of the perovskite solar cell was further boosted to 12% by infiltrating the pores of the mesoporous TiO2 films with CH3NH3PbI3 along with overlayers of the perovskite on top [3]. The perovskite functioned as both a light harvester and a hole transporter in this study. A significant improvement in the fabrication of the organolead halide perovskite solar cells was recently achieved using a two-step sequential deposition of PbI2 followed by a reaction with CH3NH3I, resulting in a device with a PCE of 15% [4]. Remarkably, planar heterojunction p-i-n solar cells without the mesoporous scaffolds (such as TiO2 and Al2O3) can be made using CH3NH3PbI3-xClx as the light absorber, a compact layer of n-type TiO2 as an electron-collecting layer, and spiro-MeOTAD [2,2’,7,7’-tetrakis-(N,N-di-p-methoxyphenylamine)9,9’-spirobifluorene] as a p-type hole conductor using dual-source vapor deposition, to yield a PCE of 15.4% [5]. To date, the record PCE of perovskite solar cells was 16.2% (certified by NREL) reported by the Korean Research Institute of Chemical Technology (http://www.nrel.gov/ncpv/images/efficiencychart.jpg). Despite progress in the field, there is still room to further optimize perovskite solar cells by creating cells that are able to better harvest the light, using interface engineering and by modifying the structure. Therefore, perovskite solar cells are promising as highly efficient alternatives to traditional solar cells and can be made at low cost. This makes perovskite solar cells the “next big thing in photovoltaics”, triggering extensive interest worldwide. Herein I present my perspective of perovskite solar cells. A comprehensive review of the literature on this topic is beyond the scope of this paper.
Perovskite materials have similar crystal structures to CaTiO3 and can be represented by the simple building block AMX3, where M is the metal cation and X is an oxide or halide anion. In the case of the organolead halide perovskite, the smaller lead cations (Pb2+) are surrounded by six halide anions (I−) at the corners, forming six-fold coordinated octahedrons that extend to a three-dimensional network by connecting the corners. The larger cationic alkylammonium (CH3NH3+) groups balance the charge between the octahedron layers (Fig. 1). The stability and the distortion of the perovskite structure are dependent on the tolerance factor (TF), which is the ratio of the (A−X) distance to the (B−X) distance. TF is given by TF = (RA + RB)/[21/2(RB + RX)], where RA, RB, and RX correspond to the atomic radius of A, B and X, respectively.
Ideal perovskites have a cubic geometry if TF = 1. Perovskites have a pseudocubic or distorted cubic structure for stability when the TF deviates from 1. Such a distortion will affect the electronic, optical, and dielectric properties of perovskite materials [6]. Both the sizes of the organic cation and the metal ions influence the fundamental properties of the perovskite materials, which will diversify them for applications in solar cells.
The first perovskite solar cells were fabricated using the configuration of the classic mesoscopic dye-sensitized solar cells (DSSCs), where perovskite nanocrystals were used as the light absorber in lieu of the dye in a liquid electrolyte-based device. This gave rise to a PCE of 3.8% and 3.1% using CH3NH3PbI3 and CH3NH3PbBr3 as sensitizers, respectively [7]. Subsequently, Im et al. [8] improved the process to deposit CH3NH3PbI3 nanocrystals onto the mesoporous TiO2 films, resulting in liquid electrolyte cells with an efficiency of up to 6.5%. However, the performance of these electrolyte-based perovskite solar cells degraded within only a few minutes because of the rapid decomposition of the organolead halide perovskite in the electrolyte. Therefore, to improve the device stability, the liquid electrolyte was replaced with a solid-state hole transport material to form a solid-state sensitized solar cell. Kim et al. [1] developed the first highly efficient solid-state perovskite-sensitized solar cells employing spiro-MeOTAD as the hole transport material (HTM), nanoporous TiO2 film as the electron transport material (ETM) and CH3NH3PbI3 as a sensitizer that was deposited using a one-step process. This device architecture had a PCE of 9.7%, which was a breakthrough in the development of solid-state sensitized solar cells. Subsequently, we have adopted a new p-type polymer, PCBTDPP, poly[N-9-hepta-decanyl-2,7-carbazole-alt-3,6-bis- (hiophene-5- yl)-2,5-dioctyl-2,5-di-hydropyrrolo[3, ]pyrrole-1,4-dione], as a HTM to fabricate perovskite-sensitized hybrid solar cells [9]. The CH3NH3PbBr3-based devices had an open-circuit voltage (Voc) of ~1.15 V and a PCE of ~5.55%. The high hole mobility of the polymer and the match of the highest occupied molecular orbital (HOMO) energy to that of the perovskite were suggested to be responsible for low-energy losses in these devices (Fig. 2).
At almost the same time, another p-type polymer, poly- triarylamine (PTAA), was demonstrated to be an excellent HTM for perovskite-sensitized solar cells. When PTAA was used, a PCE of up to 12% was achieved [3]. Kwon and co-workers [10] recently synthesized a diketopyrrolopyrrole- containing polymer, PDPPDBTE (poly[2,5- bis(2-decyldodecyl) pyrrolo [3,4-c]pyrrole-1,4(2H,5H)-dione-(E)-1,2-di(2,20-bithiophen-5-yl)ethane]), for use as a HTM for perovskite solar cells. PDPPDBTE has an optimal HOMO energy and an excellent charge carrier mobility, resulting in solar cells with an efficiency of up to 9.2%. The performance is reported to be better than the cells fabricated using the model HTM, spiro-MeOTAD. To develop HTMs with a superior device performance, it is straightforward to modify the well-developed spiro-MeOTAD [11] or to synthesize materials with spiro-MeOTAD-like structures and properties. With this in mind, pyrene-core arylamine derivatives [12] and 3,4-ethylenedioxythiophene (H101) [13] were synthesized and confirmed to be good HTMs for perovskite solar cells. Using these alternative HTMs, devices with efficiencies that are comparable to spiro-MeOTAD-based devices can be fabricated using a simpler synthetic route to produce the HTM. A new HTM for perovskite cells must have a high hole mobility. This could reduce the series resistance and improve the fill factor of the resulting cells. Another consideration is that the energy of the HOMO of the HTM must match the organolead halide perovskite and so must be in the range of −5.2 to approximately −5.4 eV for CH3NH3PbI3. A higher HOMO would decrease the open-circuit voltage while a deeper HOMO level would reduce the hole extraction ability from the perovskite to the HMT. This would enhance the charge recombination within the devices.
Although widely adopted in the early study of perovskite solar cells, the one-step deposition of perovskite nanocrystals is limited in efficiency and devices fabricated using this method usually have a broad distribution in performance. This is because the precipitation of perovskite using a one-step process is uncontrollable and there are wide variations in the morphology of the perovskite crystals. This limitation was overcome using a sequential deposition method. In this method, PbI2 was first deposited on the nanoporous TiO2 film and was subsequently converted into the perovskite by reacting with CH3NH3I. In these cells, the perovskite materials fully infiltrate the pores of the nanoporous TiO2 films, forming three- dimensional (3D) TiO2/perovskite bulk-heterojunction structures (BHJ). The solar cells made using the two-step sequential deposition approach have a PCE as high as 15% and a certified value of 14.1% with high reproducibility [4]. This is a considerable step toward the fabrication of high-performance perovskite solar cells using a solution deposition method. Figure 3 displays the device structures of a classic mesoscopic sensitized solar cell and a 3D BHJ sensitized solar cell. It was observed that the perovskite formed a continuous film across the whole mesoporous layer in 3D BHJ sensitized solar cells.
In the case of the classic sensitized solar cell, excited electrons are transferred into the TiO2 mesoporous layer and then diffuse to the compact TiO2/transparent conducting oxide (TCO) anode. A significant fraction of the electron transport could occur through the organolead perovskite phase itself because perovskites are significantly more conducting than TiO2. The electron mobility in the perovskite is ~ 25 cm2 V−1 s−1, which is 3-4 times higher than the electron mobility in the TiO2 (7.5 cm2 V−1 s−1) [14]. Therefore, the 3D BHJ sensitized solar cells are a combination of the classic sensitized cell and a p-i-n junction cell. This will be further discussed in a later part of this perspective.
At present, there has been little investigation into the improvement of the electron transport material (ETM) in perovskite solar cells. Since TiO2 matches the organolead halide perovskites energetically, it is currently the most useful ETM in high-performing cells. TiO2 is usually processed according to preparation methods that were already in use for solid-state DSSCs. Nevertheless, modified processing of the TiO2 layer as an ETM or as a blocking layer at low temperature was found to be essential to fabricate perovskite solar cells and especially for the development of flexible devices [15, 16, 17]. Also, ZnO has been successfully employed as an ETM in the perovskite solar cells in a few cases [17, 18]. Other types of ETM are necessary to optimize the perovskite solar cells with good photovoltaic characteristics and ease of processing.
Since CH3NH3PbI3 has ambipolar characteristics and is slightly more p-type than n-type, it is reasonable to make p-n junction-like devices without a HTM, known as HTM-free photovoltaic cells. Etgar et al. [19] demonstrated that CH3NH3PbI3 could act as both light harvester and hole transporter in a CH3NH3PbI3/mesoporous TiO2 heterojunction device with a PCE of 5.5%. The device architecture is shown in Fig. 4. The PCE could be further improved to over 8% by optimization of the deposition of the perovskite layer [20]. Mott-Schottky analysis confirmed that there was a depletion zone between the mesoporous TiO2 and the CH3NH3PbI3. Such a device was thus referred to as a depleted heterojunction solar cell. After further optimization of the device processing, Shi et al. [21] have recently improved HTM-free perovskite solar cells to achieve a PCE of 10.4%. In general, HTM-free perovskite solar cells had a poor fill factor and a low Voc in comparison with those with a HTM, which is associated with the larger shunt current as well as a lower IPCE for these devices.
It was initially a surprise to find that the n-type conducting mesoporous TiO2 can be completely replaced by the insulating nanoporous Al2O3 to fabricate perovskite solar cells with an efficiency of 10.9% in a configuration of TiO2 blocking layer (bl-TiO2)/nanoporous Al2O3/CH3NH3PbI2Cl/spiro-OMeTAD [2]. This configuration is very similar to the TiO2-based classic mesoscopic sensitized cells (Fig. 5(a)). Because of the insulating nature of Al2O3, the photoexcited electrons cannot inject into it and so it was assumed that the electrons were transported to the bl-TiO2/FTO anode within perovskite itself. The spiro-MeOTAD was used to extract the holes and transport them to cathode. The mixed halide perovskite served as both a light absorber and an electron transport material. Such a device was known as a “meso-superstructured solar cell”. Interestingly, the Voc obtained using the insulating Al2O3 scaffold was ~200 mV higher than the Voc with a TiO2-based sensitized device. This is indicative of lower fundamental energy losses. The meso-superstructured perovskite solar cells operate in a similar way to the classic p-i-n junction solar cells with the exception that they are in a three-dimensional configuration. This implies that it is possible to use the perovskite in a thin-film planar device configuration. Indeed, Liu et al. [5] successfully made such a planar junction device with CH3NH3PbI3-xClx as the light absorber, a compact layer of n-type TiO2 as the ETM and spiro-MeOTAD as the HTM using a dual-source vapor deposition approach. The perovskite thin films were extremely uniform without any pinholes, which was apparently the main reason for the achievement of a photovoltaic efficiency of over 15% (Fig. 5(b)). This study demonstrates that perovskites can function well in simplified device architectures without the need for a complex nanoporous scaffold. On the basis of the planar p-i-n concept, Liu et al. [18] fabricated perovskite solar cells at a low temperature using solution-deposited ZnO nanoparticles as the ETM and CH3NH3PbI3 as the light harvester. This device structure delivered a PCE as high as 15.7%. A flexible device with an efficiency of 10.2% was also reported.
In less than 2 years, perovskite solar cells have achieved a record 16.2% efficiency and it is expected that ~20% efficiency is achievable using the present techniques. The underlying reasons for the high performance of perovskite solar cells could be attributed to many factors. There are three important points that make them distinct from other types of semiconductors and dyes used as the light harvesters in solar cells. These will be further discussed.
As light absorbers, the primary advantage of the organoloead halide perovskites over dyes is that they have a stronger absorbance over the entire visible to near infrared range. For instance, the molar extinction coefficient of CH3NH3PbI3 is about 1.5 × 105 (mol/L)−1 cm−1 at 550 nm, which is 2-3 times higher than those of organic dyes widely adopted in solid-state DSSCs [22]. This enables the perovskite to have complete light absorption in films as thin as 500-600 nm to overcome the thickness limitations (of ~2 μm) of the classic solid-state DSSCs.
Traditionally, organolead halide perovskites were believed to exhibit p-type behavior. Etgar’s demonstration of a simple nanoporous TiO2/CH3NH3PbI3 solar cell (with ~5.5% efficiency) presents direct evidence of this [19]. However, n-type behavior of CH3NH3PbI3 was clearly shown in nonsensitized cells consisting of mesoporous ZrO2/CH3NH3PbI3/spiro- MeOTAD configuration that achieved an efficiency of 10.8% [23]. Hence, CH3NH3PbI3 must exhibit ambipolar charge transport properties. In Snaith’s mesoscopic Al2O3/mixed halide CH3NH3PbI3-xClx device [2] and the planar heterojunction perovskite cells with efficiencies in the range of 10%-15.4% [5], the mixed halide perovskites were demonstrated to have n-type charge transport. Therefore, the organolead halide perovskites are able to exhibit ambipolar charge transport, which means that they have the potential to be excellent materials for use in solar cells with various different configurations to achieve the required photovoltaic performance.
Organolead halide perovskite have Wannier-type excitons under illumination, which means that these materials should have better charge transport properties than the widely studied organic solar cells with Frenkel excitons. Experimentally, Xing et al. [24] and Stranks et al. [25] independently studied the transport properties of CH3NH3PbI3 and CH3NH3PbI3−xClx, respectively, by performing time-resolved photoluminescence (TRPL) and transient absorption spectroscopy. It was shown that the diffusion lengths of the electrons and holes are ~130 nm and ~100 nm in CH3NH3PbI3, and ~1100 nm and ~1200 nm, respectively, in CH3NH3PbI3−xClx. Such long charge diffusion lengths that are approximately balanced can facilitate good charge transport in perovskites. This is critical for device design. For instance, since the absorption depth of organolead halide perovskites is 500-600 nm, the mesoscopic sensitized devices are preferred if the diffusion length of the charge carriers is shorter than the depth of the light absorption. This is the case for the sensitized devices using CH3NH3PbI3 as a light absorber that have a maximum efficiency of 15% [4]. In contrast, the planar junction structured devices are more suitable if the diffusion length of the charge carriers is longer than the depth of the light absorption, as demonstrated in the CH3NH3PbI3−xClx- based planar cells that have an efficiency of 15.4% [5].
Organolead halide perovskite-based solar cells have ultrafast charge generation, high carrier mobilities, balanced charge carrier mobilities, and slow recombination rates. It is therefore relatively easy to achieve good device performance [14]. Also, solution processing of the perovskite is facile and the resulting materials have a high crystallinity and are defect tolerant. These characteristics make the study of perovskite solar cells simpler than other photovoltaic systems such as CZTS.
Organolead halide perovskites meet all three of the main prerequisites for a well-performing solar cell. They have strong light absorption, a high carrier generation ability, and excellent electron and hole transport properties. The unique combination of the strong, panchromatic absorbance and the ambipolar charge transport properties make perovskites ideal light harvesters in solar cells. Since organolead halide perovskites are more conductive (10−3 S cm−1) than the HTMs currently used, high-efficiency perovskite solar cells require a capping layer of HTM to prevent contact between the perovskite and metal cathode. However, this increases the series resistance and decreases the fill factor of the resulting cells, which limits the device performance. There is a trade-off between the series and the shunt resistance for perovskite solar cells. Currently, HTMs are a limiting factor that hampers further improvement of the power conversion efficiency for perovskite solar cells. It is necessary to investigate new HTMs to replace spiro-OMeTAD and to obtain better photovoltaic performance. The ideal characteristics of HTMs are a high hole mobility, a good thermal and UV stability and a well-matched HOMO energy to the organolead halide perovskites.
Solar cells with practical applications should have a good long-term stability. Unfortunately, perovskite solar cells are unstable when exposed to irradiation under UV-light. The origin of this instability was not attributed to the decomposition of the perovskite but was instead attributed to the TiO2 [26]. In contrast, TiO2-free perovskite solar cells had better long-term stability. Therefore, a more suitable alternative material to TiO2 should be investigated for use as an ETM, or a more effective approach to protect against UV instability should be developed. Also, interface engineering of the ETM was demonstrated to improve the performance of organic solar cells. Similar studies of the interface engineering of the ETM and the perovskite are few [27, 28, 29]. Such a study should be carried out to create intimate contact between the perovskite and the anode to enhance charge transfer and to reduce the charge recombination probability. This will increase the fill factor to a value approaching 0.8. Currently, processing of perovskite solar cells is suboptimal, resulting in film inhomogeneity and uncontrollable device performance. Despite the achievement of an efficiency of over 15%, optimization of the processing parameters could further boost the efficiency of the perovskite solar cells to perhaps 20%. Theoretically, a perovskite with a band gap of 1.5 eV could give rise to a short circuit current (Jsc) of 28 mA/cm2. A photocurrent density of 24 mA/cm2 is thermodynamically achievable when taking into account losses of 15% caused by reflection at the TCO. Considering that a fill factor of 0.8 [30] and a Voc above 1.1 V [2, 9] have been obtained, ~20% efficiency for a single- junction cell is a practical goal for perovskite solar cells based on the present techniques.
Another strategy for the development of high-performance perovskite solar cells is to develop multi-junction tandem devices, where there is a top subcell with a wider band gap to absorb the high-energy photons while the low-energy photons pass through the top subcell and are absorbed by the bottom subcell that has a narrower band gap semiconductor. The organolead halide perovskite materials have band gaps that can be tuned in the range of 1.5-2.2 eV by tailoring their chemical species [31]. This makes them suitable for use as the top subcell to match with crystalline silicon (that has a band gap of 1.1 eV) and other thin-film technologies such as CIGS and CZTSSe (0.9-1.4 eV) in hybrid tandem photovoltaics. An efficiency approaching 30% was estimated for this design.
The use of Pb is believed to be one of the main limitations for future large-scale fabrication of organolead halide perovskite solar cells from environmental and economic views. However, this should not be an insurmountable problem considering that millions of tons of lead are consumed every year around the world in lead-acid batteries. Less than 1000 tons of lead can cater for a product capacity of 1000 GW solar cells per year [32]. The sufficient supply of lead along with solution processing make perovskite solar cells low cost. Nevertheless, the investigation of more environmentally friendly alternatives, such as tin, is needed to create perovskite solar cells with high efficiency and protect the environment and human health.
A breakthrough in solar cells was made using organolead halide perovskites as light harvesters with low-cost fabrication techniques. Perovskites are capable of a high light absorbance, have excellent charge transport properties and have tunable band gaps. PCEs of over 15% have been achieved for both mesoscopic sensitized and planar heterojunction devices. A further increase in PCE to 20% for a single-junction perovskite solar cell and approaching 30% PCE for a tandem configuration with a silicon or CIGS solar cell is anticipated. It is not unrealistic to expect practical production of these cells on a small scale in the near future
This work was partly supported by the National Natural Science Foundation of China (20873141) and the “Hundred Talents Program” of the Chinese Academy of Sciences.
2014, Vol. 35 Issue (7): 983-988
PDF (457KB)
2014, Vol. 35 Issue (7): 983-988
PDF (457KB)