Creation of Highly Efficient and Durable Organic and Perovskite Solar Cells Using Nanocarbon Materials^#

Abstract

This accounts article describes examples of improving power conversion efficiency and stability of organic and perovskite solar cells by using nanocarbon nanotubes such as fullerene derivatives, endohedral fullerenes, and carbon nanotubes. Appropriately doped carbon nanotubes can replace indium-tin-oxide transparent electrodes and evaporated metal electrodes to produce stable and flexible solar cells. Properly designed fullerene derivatives can serve as electron transporting layer through passivation of defects at the interface between perovskite crystals and an inorganic charge selective layer. Lithium-ion endohedral fullerene can dope organic semiconducting molecules and carbon nanotubes for improving not only power conversion efficiency but also stability because neutral lithium endohedral fullerene forms to scavenge intruding oxygen. The author suggests creation of carbon-rich solar cells using such nanocarbon materials for further development of practical organic solar cells.

1. Introduction

Acquiring more natural energy from solar cells is a major social issue for humankind in this century in order to preserve the beautiful environment of the Earth. At the end of this century at the latest, photovoltaic power generation will become the main energy generation source, and it is expected that the future will be supplemented by thermal, wind, and hydro power generations, etc., during the time when solar cells are not generating electricity. A society is envisioned in which limited resources such as fossil fuels are utilized as chemical resources as much as possible, and electric power is produced by directly or indirectly utilizing the energy of the Sun. Against this background of the times, research on solar cells is being actively conducted in this century. In addition to researching the materials and device structures used to improve energy conversion efficiency, many studies have been conducted with the aim of improving stability and low-cost production for practical use.

In order to reduce the manufacturing cost of solar cells, research on organic solar cells represented by organic thin-film and perovskite solar cells is being conducted all over the world.^1–7 In these solar cells, although they are organic, indium tin oxide (ITO) transparent electrodes and backside metal electrodes using gold, silver, aluminum, etc., use a lot of inorganic materials. What cannot be ignored is that the ratio of the cost of electrodes such as ITO electrodes to the total cost of organic solar cells is very high. In other words, if this is made cheaper, the cost of the entire organic solar cell can be further reduced. From this point of view, we are conducting research and development of organic solar cells in which inorganic electrodes are replaced with nanocarbon materials as much as possible. Carbon exists universally on the earth as an element, and it is not necessary to use a limited element such as indium. This accounts article introduces the research and development of devices that use nanocarbon electrodes instead of inorganic electrodes for organic thin-film and perovskite solar cells.

2. Organic Thin-Film Solar Cells Using Transparent Carbon Nanotube Electrodes

In ordinary organic solar cells, indium tin oxide (ITO) electrodes are used as transparent electrodes, and metal electrodes such as gold, silver, and aluminum are used as backside electrodes that do not transmit light. Since ITO contains indium, which is a rare metal, there is a concern that prices may fluctuate due to international circumstances, and gold and silver are of course expensive. In organic solar cells, the ITO electrode is usually the most vulnerable in terms of flexibility, and the organic active layer and the vapor-deposited back electrode are resistant to bending. This is because an ITO thin film sputtered on a plastic substrate will crack the polycrystalline structure of ITO if it is bent too much. In terms of the cost of organic solar cells, the back metal electrode, which requires vacuum deposition, is the most expensive. Such problems can be solved by using carbon nanotube electrodes, which have excellent elasticity and exist infinitely as carbon elements.

Carbon nanotubes are a new cylindrical material formed only of carbon as an element, and have excellent mechanical properties, electrical selectivity, excellent field emission ability, and highly efficient hydrogen storage medium characteristics. Their high conductivity and transparency mean that they can replace ITO electrodes and metal electrodes in photoelectric conversion devices such as solar cells. Compared to graphene, carbon nanotubes are considered to be advantageous in terms of industry and practical use because they have fewer defects, are easier to handle in large areas, and have superior mechanical properties. Among carbon nanotubes, single-walled carbon nanotubes are known to have the highest conductivity.

This section firstly introduces organic thin-film solar cells^8–10 that use carbon nanotube thin films as transparent electrodes.^11–13 Carbon nanotube thin films are obtained by filtering aerosol-like carbon nanotubes grown by floating iron nanoparticle catalysts in vapor phase and carbon monoxide as a carbon source. Although the carbon nanotube bundle cannot be unwound if it is collected as a solid powder, a high-quality carbon nanotube thin film can be obtained by collecting it directly on a filter from the aerosol. By pressing the carbon nanotube thin film on the filter directly onto a glass or plastic substrate, or onto organic or perovskite thin film from the back of the filter, transfer can be easily performed without the need for a solvent-based coating process or vacuum deposition process. However, as with all nanocarbon electrodes, doping is necessary to increase conductivity. There have been problems such as doping stability and obstructing the electronic current flow in the device. For example, nitric acid (HNO₃) dopants, which show a high doping effect and are often used, have the disadvantage of emitting corrosive gas and having a short-time doping effect. Relatively safe gold chloride (AuCl) dopants also have the problem of reducing electrical performance. The authors applied MoO₃ thermal doping,¹⁴ which exceeds these two restrictions, to carbon nanotube transparent electrodes used in organic solar cells for the first time (Figure 1).¹⁵ MoO₃ is deposited on carbon nanotubes, and by heat-treating this, electrons are transferred from carbon nanotubes to MoO₃, and the carbon nanotubes are hole-doped and MoO_x that receives electrons and is partially reduced plays the role of a hole transport layer. As a result, we were able to construct a carbon nanotube transparent electrode that selectively collects holes. The energy level of MoO_x itself is suitable for the hole transport layer, and since MoO_x does not volatilize, doping with excellent stability and safety was realized. This technology makes it possible to construct a stable and highly efficient flexible organic thin-film solar cell (Figure 1, bottom), and at the time of our report, it has the highest power conversion efficiency (6.0%) among organic thin-film solar cells using carbon nanotube transparent electrodes.

Although the MoO_x doping introduced above is effective and stable, it has the major disadvantage of requiring vacuum deposition of MoO₃ and high temperature thermal annealing. The thermal vapor deposition method in vacuum requires a special instrument for vacuum, which limits not only the installation cost but also the large area. In addition, it is not perfect from the viewpoint of flexible application because it is limited to plastic substrates that can withstand thermal annealing. Therefore, dopants that can be coated are preferred, but as mentioned above, acidic dopants have safety issues and have a short doping effect lifetime. As a solution to this, the authors applied an acidic polymer as a dopant to carbon nanotubes, updated the highest energy conversion efficiency in organic solar cells using carbon nanotube electrodes, and achieved the longest doping effect lifetime (Figure 2).¹⁶ Unlike conventional acidic dopants, Nafion, which is a polymer acid, does not volatilize because it has a high molecular weight, and the polymer main chain wraps around carbon nanotubes and has good affinity with carbon nanotubes. Therefore, the p-doped carbon nanotube thin film is stably maintained. In addition, since Nafion itself has low acidity and chemical reactivity, it has a low effect on the human body and maintains stability even in extreme environments. For reliable stability evaluation, the authors used a stable electron donor called PBTZT-stat-BDTT-8 in the active layer to fabricate indium-free organic thin-film solar cells. The carbon nanotube organic thin-film solar cells to which Nafion doping was applied showed a high conversion efficiency of 8.0% and retained up to 7.0% after 2 months. On the other hand, in the case of carbon nanotube organic thin-film solar cells using conventional acidic doping, the conversion efficiency was reduced to nearly half in the same time. Interestingly, the reference device using the ITO electrode and the same electron donor showed a conversion efficiency of 9% immediately after device fabrication, but two months later, the conversion efficiency was lower than that of carbon nanotube organic thin-film solar cells using Nafion doping. We consider that the reason for this is that acidic PEDOT:PSS etches ITO and causes ion migration, which reduces the efficiency of organic thin-film solar cells using ITO.

3. Perovskite Solar Cells Using Transparent Carbon Nanotube Electrodes

In the same year we reported a carbon nanotubes-based organic thin-film solar cell, we also reported a perovskite solar cell using a carbon nanotube transparent electrode in which the photoelectric conversion active layer was changed from an organic semiconductor to an organometallic perovskite (Figure 3).¹⁷ First, we tried doping with MoO₃ as in the case of organic thin-film solar cells, but it failed because both perovskite and MoO₃ are crystalline substances with mismatched energy levels. Therefore, we decided to adopt nitric acid doping, which was the first option for hole doping of carbon nanotubes at that time. The fuming nitric acid used is a chemically very dangerous reagent, and even a small amount of vapor touching the skin can turn it yellow. As a result of examining the concentration of nitric acid by using dilute nitric acid, it was confirmed that the effect of hole doping to carbon nanotubes can be seen even when diluted to 35 v/v%. The next issue was that carbon nanotubes are hydrophobic in nature and the perovskite layer was hydrophilic. There is also a concern that the water contained in the remaining nitric acid may damage the perovskite layer. To solve these problems, a PEDOT:PSS layer was applied between the carbon nanotube thin film and the perovskite layer. In addition, in order to successfully form hydrophilic PEDOT:PSS on the carbon nanotube thin film, we added isopropanol and a small amount of surfactant to the aqueous dispersion of PEDOT:PSS to give it some hydrophobicity. The carbon nanotube thin film itself was also doped with nitric acid to increase its hydrophilicity. In this way, the adhesion of the carbon nanotube thin film/PEDOT: PSS layer/perovskite layer was secured, and finally PC₆₁BM was over-coated as an electron transport layer to fabricate an inverted (p-i-n) device. The energy conversion efficiency was only 6.3% on the glass substrate and 5.4% on the flexible PET substrate, but this was the first report of a perovskite solar cell using an indium-free carbon nanotube transparent electrode.

Next, we tried to replace back metal electrodes with carbon nanotube electrodes. Various studies were conducted to improve the energy conversion efficiency of the perovskite solar cell using the carbon nanotubes and carbon-based materials. As a result, the perovskite layer was sandwiched between two types of carbon allotropes, namely C₆₀ and carbon nanotubes. It was found that the device provides high stability, excellent hysteresis characteristics, and high conversion efficiency (Figure 4).¹⁸ The device structure is a normal (n-i-p) device, and ITO glass is used for the substrate. C₆₀ was vapor-deposited on an ITO glass substrate to form an electron transport layer, and after a perovskite layer was formed, a carbon nanotube thin film was transferred onto it. If nitric acid was doped here, it would damage the perovskite layer, so the energy conversion efficiency in that state was 13.2% without doping. The carbon nanotube thin film has a mesoporous structure and is highly transparent, but it also has a certain degree of water vapor permeability. For the purpose of improving the conversion efficiency and the barrier property against the permeation of water vapor, solutions of various organic hole transport materials were applied to fill the nano-sized space in the network structure of carbon nanotubes. When P3HT was used, the lifetime of the device without encapsulation was improved by more than 6 times. When the water vapor transmittance of the carbon nanotube thin film interpenetrated with P3HT was measured, the barrier performance was significantly improved from 34,000 g/m²d to 1,200 g/m²d, which is the reason for the longer lifetime of the unsealed device. In addition, by filling the spatial defect that creates capacitance, the hysteresis, which was originally not very large, has become even smaller. By electron microscopic observation, it was observed that P3HT covered the surface of the carbon nanotubes. When spiro-MeOTAD was infiltrated as a hole transport material, the highest conversion efficiency of 17.0% was obtained. However, the lifetime was inferior to that of the device infiltrated with P3HT. In addition, the device interpenetrated with PTAA showed an intermediate between the P3HT device and the spiro-MeOTAD device in terms of parameters such as conversion efficiency and lifetime. When the lifetime of the encapsulated device was evaluated by the maximum power point tracking (MPPT) method for the device with P3HT, the decrease in conversion efficiency was only 20% in 2200 hours (about 90 days). It was shown that sandwiching the perovskite layer with carbon allotropes such as fullerenes and carbon nanotubes improves the lifetime and hysteresis characteristics of perovskite solar cells.

When the perovskite photoactive layer absorbs sunlight, a pair of positively charged holes and negatively charged electrons is generated. It is necessary to separate the pair into holes and electrons, and collect only the holes and only the electrons at the upper and lower electrodes, respectively, of the active layer. As mentioned above, it is relatively easy to realize carbon nanotube electrodes collecting holes (anode receiving electrons from external circuit). The next challenge was to construct a carbon nanotube electrode that collects only electrons (cathode). The authors constructed a cathode with carbon nanotubes by infiltrating a network of carbon nanotubes with a fullerene derivative (PC₆₁BM). This was the key, and we were able to fabricate a perovskite solar cell using carbon nanotube thin films at both electrodes (Figure 5).¹⁹

The procedure for manufacturing a perovskite solar cell using both sides as carbon nanotube electrodes is as follows. A P3HT solution is interpenetrated into the carbon nanotube thin films on glass or plastic substrates to prepare the transparent anode. Subsequently, PEDOT:PSS is applied as a hole transport material for the purpose of flattening the surface of the carbon nanotube network. A perovskite layer is applied on top of it to form a photoelectric conversion active layer. Finally, the carbon nanotube thin film is transferred again on it, and a solution of the fullerene derivative is interpenetrated therein to form a cathode. Both carbon nanotube electrodes perovskite solar cell fabricated on the plastic substrate in this way is resistant to mechanical bending. Deterioration of the characteristics was considerably smaller even when bent compared to a perovskite solar cell using an ITO electrode and a metal back electrode. In addition, it was confirmed by electron microscopy that the fullerene derivative exists in the gap of the carbon nanotube network structure (rectangular frame, Figure 5), inside thick carbon nanotubes (square frame, Figure 5), and on the surface of carbon nanotubes (arrow, Figure 5) at the cathode. It is thought that the fullerene derivative attached to such a surface gives charge selectivity on the surface of the carbon nanotubes, instead of injecting electrons into the carbon nanotube to n-dope the carbon nanotube.²⁰

This two carbon nanotube electrodes-based perovskite solar cell has a relatively simple layer structure and does not require a vacuum process in fabrication, so it is expected to lead to technology that reduces the cost of the fabrication process in the future (Figure 6). A solar cell is manufactured by feeding out a rolled plastic substrate and fabricating thin films of necessary materials by transfer or coating, and it is possible to avoid a costly vacuum process. First, a carbon nanotube anode is transferred on the plastic film, and P3HT is interpenetrated on it. Subsequently, PEDOT:PSS is applied for flatness of the carbon nanotubes network and dried with a dryer. Then, the photoactive perovskite layer is applied, the reprecipitation solvent is applied, and dried, and then the carbon nanotube thin film is transferred on it and the fullerene derivative is permeated. Finally, the device is laminated and cut. By replacing both the metal and metal oxide electrodes with carbon nanotubes, we consider that we can approach a new solar cell containing carbon materials as the main components.

Various research groups have reported perovskite solar cells that use single-walled carbon nanotube films or graphene as transparent conductive electrodes, and there has been much debate about which is better, single-walled carbon nanotube membranes or graphene. Therefore, we attempted a direct comparison of these using the flexible perovskite solar cells (Figure 7).²¹ For the single-walled carbon nanotube film, a film made by Prof. Esko Kauppinen of Aalto University, which has the highest transmittance and conductivity at present, was used, and high-quality CVD graphene made by Hiroki Ago of Kyushu University with a grain size of mm class was used as the graphene. As a result, the photoelectric conversion efficiencies of the device using the single-walled carbon nanotube film and graphene were 13% and 14%, respectively, and graphene was slightly advantageous. The cause of this was considered to be the high light transmittance and surface smoothness of single-layer graphene. On the other hand, in the performance evaluation of mechanical properties by repeating bending more than 1,000 times, the solar cell to which the transparent electrode of single-walled carbon nanotube was applied surpassed the solar cell to which the transparent electrode of graphene was applied. As a matter of course, the ITO electrode device that we tried at the same time deteriorated in performance immediately. This cause can be easily verified from the thin film tensile test. By attaching a single-walled carbon nanotube film and graphene to dimethylpolysiloxane (PDMS), applying tension, and performing Raman spectroscopy, and comparing the shifts of the G band and 2D bands, the deformation of the 6-membered carbon ring of graphene became larger than that of carbon nanotubes. It is intuitively understandable that the single-walled carbon nanotubes thin film, which has a structure entwined in various directions, is advantageous for tensioning.

4. Perovskite Solar Cells Using a Novel Fullerene Derivative in the Electron Transport Layer

Carbon nanotubes are essentially promising as a material for extracting and transporting holes, but fullerenes and their derivatives are promising as carbon materials for transporting electrons with electron acceptor properties. In inverted (p-i-n) perovskite solar cells, PC₆₁BM is used as the fullerene derivative in an electron transport layer to be applied on the perovskite layer because of its availability and good solubility. Are there any other fullerene derivatives that give superior properties?²² The Prof. Hin-Lap Yip group with us used methano-indene-fullerenes (MIF, C₆₀(CH₂)(Ind)) instead of PC₆₁BM as the electron transport material in the inverted perovskite solar cell with the configuration of glass/ITO/NiO/DEA/MAPbI₃/fullerene/PN₄N/Ag.²³ The NiO layer formed by the sol-gel method was used as the hole transport layer, and was modified with a diethanolamine (DEA) self-assembled monolayer to optimize the interface. The work function is adjusted by using PN₄N, which is a polyamine similar to PFN, for the silver electrode on the back metal surface. Comparing devices using PC₆₁BM and MIF with the same device structure, the device using MIF gave a higher open circuit voltage (V_OC = 1.13 V). This is because MIF has an essentially higher LUMO level, and MIF is a mixture of positional isomers and is highly amorphous, which makes it difficult for pinholes to form in the electron transport layer. In fact, when comparing the leakage currents, the device using MIF was smaller. As a result, a high fill factor (FF = 80.0) could be achieved. In addition, the short-circuit current density (J_SC = 20.4 mA/cm²) and EQE did not decrease. As a whole, the device using MIF showed higher conversion efficiency (maximum 18.1%, average 17.9%), and showed stable characteristics even at MPPT at a voltage of 0.98 V.

When a nanocarbon material is used as an electron transport layer in a normal (n-i-p) perovskite solar cell, C₆₀ is usually deposited on a transparent electrode such as ITO or FTO, and a perovskite layer is formed on it. On the other hand, the authors have proposed a simple film formation method for the electron transport layer.²⁴ In order to deposit C₆₀ between the transparent electrode substrate and the perovskite layer, it takes time because the substrate must first be placed in the vapor deposition instrument and the applied voltage must be adjusted while looking at the film thickness meter. The cost of the vapor deposition instrument is also high. It is easy to form a film by solution coating, but when C₆₀ film is formed by spin-coating, C₆₀ crystallizes in the solution and a uniform film cannot be obtained. To avoid this problem, we used an o-dichlorobenzene solution of mix-fullerene containing a small amount of C₇₀ added to C₆₀. A solution with a mixing ratio of C₆₀:C₇₀ of 9:1 was spin-coated on ITO and heated at 100 °C for 10 min under vacuum (0.01 MPa) to form a uniform electron transport layer. A conversion efficiency of 18.0% was obtained in a normal-structured perovskite solar cell with MAPbI₃ as the perovskite material and spiro-MeOTAD as the hole transport material. This value was higher than the conversion efficiency (16.7%) of the reference device using the vapor-deposited C₆₀ as the electron transport layer with the same device configuration. In addition, the optimized device showed no hysteresis. When the mixing ratio of C₆₀:C₇₀ was changed to C₆₀ only, 9:1, 1:1, 1:9, and C₇₀ only, the performance was specifically high only for the device with a mixing ratio of 9:1. When the UV-Vis spectra of mix-fullerene films with various mixing ratios were evaluated, absorption around 450 nm due to the HOMO-LUMO transition between adjacent molecules in the solid of C₆₀ was lowered at near this mixing ratio. From this result, it is considered that the intermolecular interaction between C₆₀ was suppressed by mixing a small amount of C₇₀, and a uniform film was obtained. In addition, when thermal annealing at 100 °C, thermal annealing at 200 °C, and thermal annealing at 100 °C under reduced pressure were examined, the latter gave the highest performance. We consider this is because C₆₀ cannot be crystallized after this treatment because o-dichlorobenzene solution is removed.

Another interesting example of improving the efficiency of a solar cell by using a novel fullerene derivative is FIF (five-membered carbon-fused indano[60]fullerene) for the passivation (surface stabilization treatment) of the tin oxide (SnO₂) electron transport layer of a perovskite solar cell (Figure 8).²⁵FIF is a novel fullerene derivative having a 5-membered ring made of carbon atoms on the fullerene skeleton. It has a methoxy group (-OCH₃) on the 5-membered ring. A fullerene cation intermediate^26,27 is used to form a 5-membered ring of carbon on fullerenes. When divalent copper (CuBr₂), which is an oxidant, is generated in the fullerene anionic intermediate, the carbon atom nucleophilically attacks the fullerene cation and cyclizes it, resulting in a 5-membered ring product.

For perovskite solar cells with a basic structure (ITO/SnO₂/MAPbI₃/spiro-MeOTAD/Au; MAPbI₃ = CH₃NH₃PbI₃), modifying the junction between the inorganic oxide SnO₂ and the perovskite layer MAPbI₃ improves conversion efficiency. SnO₂ and MAPbI₃ are crystalline inorganic and organometallic hybrid compounds, respectively. If there is a defect such as a pinhole in SnO₂, it is considered that the efficiency decreases due to charge recombination. In addition, if defects such as voids are formed between the layers of both compounds in the process of crystal growth of MAPbI₃ on SnO₂, there is a concern that the hysteresis will increase and the characteristics will deteriorate. Therefore, we decided to perform surface stabilization treatment of SnO₂ using FIF to suppress the formation of defects at this interface. FIF has a methoxy group, and the oxygen atom of the methoxy group can be coordinated to a metal ion. As a result, it is considered that FIF stays on the defect of SnO₂ with the methoxy group facing the SnO₂ side and C₆₀ facing the outside. In addition, the methoxy group of FIF is also coordinated to the Pb²⁺ ion of MAPbI₃, and MAPbI₃ can also be passivated. The coordination of the methoxy group to the Pb²⁺ ion was confirmed by infrared absorption spectroscopy. In addition, FIF can fill the voids that are formed between SnO₂ and MAPbI₃ to suppress capacitance at the interface. In this way, a perovskite solar cell exhibiting high power conversion efficiency (PCE = 20.7%, open circuit voltage V_OC = 1.09 V, short circuit current density J_SC = 23.8 mA/cm⁻², fill factor FF = 0.79) is successfully manufactured.

The size of the crystal grains of MAPbI₃ formed on SnO₂ was investigated by scanning electron microscope (SEM) observation (lower part of Figure 8). It was found that the crystals of MAPbI₃ grow larger when the defects on SnO₂ are eliminated by FIF. As a result, the grain boundaries in MAPbI₃ are reduced, the charge transport efficiency is improved, the charge recombination is suppressed, and the characteristics of the solar cell using FIF are considered to be improved.

Figure 9 shows the reference data without FIF and the data when C₆₀ and other fullerene derivatives are used instead of FIF. When C₆₀ is used, the performance is generally improved compared to when it is not used. On the other hand, when PCBM, which is used as an electron acceptor material for organic thin-film solar cells, is used, the overall characteristics deteriorate. When FIF is used, the open circuit voltage, short circuit current density, and fill factor all improve, and the characteristics improve. The other fullerene derivatives in Figure 9 have the methoxy group of FIF replaced with long-chain alkoxy groups, but their properties are lower than when FIF is used.

5. Stable Perovskite Solar Cells with Lithium-Ion-Endohedral Fullerenes

Although perovskite solar cells show a power conversion efficiency of 25% or more,¹ they still have major problems in durability. Organometallic perovskite used in the photoactive layer of solar cells is unstable to water and oxygen, and research to overcome this problem has been widely conducted. At the first stage of research of perovskite solar cells, the efficiency has been greatly improved by using organic semiconductors such as spiro-MeOTAD for the hole transport material, which is used in the charge selective layer. However, the characteristics of transporting holes in the organic semiconductor itself are not sufficient, and it is necessary to mix a lithium salt or a cobalt complex with the organic semiconductor to extract electrons from the organic semiconductor (doping the holes). The lithium salt used has hygroscopicity and attracts water. Also, when lithium salts are used, oxygen is required to extract electrons from spiro-MeOTAD. There was a contradiction that perovskite solar cells, which should avoid water and oxygen, needed hygroscopic materials and oxygen.

A perovskite solar cell was fabricated using an endohedral fullerene lithium salt (lithium ion-endohedral fullerene, [Li⁺@C₆₀][TFSI⁻]; TFSI⁻ = bis (trifluoromethanesulfonyl) imide)²⁸ in which lithium ion (Li⁺) is located in a fullerene C₆₀ cage instead of the use of conventional lithium salt (Li⁺TFSI⁻). Durability of an unencapsulated device was found to improve 10 times (Figure 10).²⁹ Lithium-ion-containing fullerenes have low hygroscopicity because lithium ions are inside the hydrophobic C₆₀, and have high electron affinity because a positive ion is inside fullerene. It was possible to quickly extract electrons from spiro-MeOTAD simply by mixing them without exposure to oxygen to extract electrons from spiro-MeOTAD. Electron transfer smoothly occurs from spiro-MeOTAD to lithium-ion-containing fullerene, resulting in hole-doped spiro-MeOTAD and neutral lithium-containing fullerene (Li⁺@C₆₀^•− = Li@C₆₀) (Figure 10a).^30,31 This electron transfer was confirmed by the characteristic light absorption spectrum of the radical cation of spiro-MeOTAD and the neutral lithium-containing fullerene (Figure 10c). The neutral lithium-containing fullerene produced at this time has an antioxidant ability and has the effect of removing a small amount of oxygen intruding to the solar cell.

Because of the hydrophobicity of lithium-ion-containing fullerenes and the antioxidant ability of lithium-containing fullerenes, stable perovskite solar cells against water and oxygen (Figure 11) were obtained. The unencapsulated device showed a unique behavior in the lifetime evaluation (Figure 11a). In conventional perovskite solar cells, the organic semiconductor layer containing a hygroscopic lithium salt, attracts surrounding water molecules, and the device stops working in 50 hours. With the unencapsulated device using lithium-ion-containing fullerenes, the conversion efficiency slowly increased over about 50 hours, and decreased over about 500 hours from the highest efficiency point. This is because oxygen is blocked by neutral lithium-containing fullerenes, which slows the formation of holes in spiro-MeOTAD, which results in an induction period, while blocking water and oxygen also slows deterioration. The lifetime of unencapsulated devices was about 10 times longer. The energy conversion efficiency at the highest point was 16.8% (Figure 11b). In addition, the encapsulated device showed decrease in performance was within 10% after 1000 hours under continuous simulated sunlight irradiation, which satisfies requirement of a guideline for practical use research (Figure 11c). Lithium-ion-containing fullerenes are thought to contribute not only to improving the durability of perovskite solar cells, but also to improving the functionality and stability of organic electronics materials.

For the purpose of producing a perovskite solar cell that does not use a metal back electrode, we performed research using a single-walled carbon nanotubes thin film as the back electrode. The charge selectivity and charge mobility of the single-walled carbon nanotubes thin film itself are insufficient, and doping is necessary to improve conductivity and charge carrier selectivity. In this research, the single-walled carbon nanotubes thin film is interpenetrated with a mixture of the organic semiconductor molecule spiro-MeOTAD and the oxidizing agent lithium ion-containing fullerene salt ([Li⁺@C₆₀][TFSI⁻]). The hole selectivity and hole mobility of the carbon nanotube electrode were increased, and the stability of the perovskite solar cell was also improved (Figure 12).³²

When spiro-MeOTAD, [Li⁺@C₆₀][TFSI⁻], and tert-butylpyridine (t-BP) are mixed in a chlorobenzene or o-dichlorobenzene solvent for a predetermined time, electron transfer occurs from spiro-MeOTAD to Li⁺@C₆₀, and hole doping occurs. Doped spiro-MeOTAD (spiro-MeOTAD⁺TFSI⁻) and neutral lithium-containing fullerene (Li⁺@C₆₀^•−) are produced. The solution is interpenetrated into the single-walled carbon nanotubes thin film, which is the top electrode of the perovskite solar cell. Spiro-MeOTAD⁺ can penetrate the porous network of the single-walled carbon nanotubes thin film, while Li⁺@C₆₀^•− tends to aggregate and stays on the single-walled carbon nanotubes thin film. In this situation, single-walled carbon nanotubes is hole-doped and becomes a back electrode that selectively collects only the holes. In addition, Li⁺@C₆₀^•− that stays on the surface captures oxygen that enters the device and contributes to improving the durability of the device. The energy conversion efficiency of the device was 17.2%.

Figure 13 shows comparative data on device durability. After encapsulation of the devices, durability evaluation was performed under accelerated conditions of temperature 60 °C and humidity 70%. The conversion efficiency of a perovskite solar cell with a normal gold electrode as the back electrode was lost in about 300 hours. Then, a carbon nanotube electrode is used instead of a gold electrode, and carbon nanotubes are functionalized with spiro-MeOTAD, Li⁺TFSI⁻, and t-BP without using an endohedral fullerene. Compared to the reference device using a gold electrode, its stability was significantly improved. The highest stability was observed in the device in which carbon nanotubes were functionalized using Li⁺@C₆₀, and almost no decrease in efficiency was observed in the measurement for 1,100 hours.

6. Perovskite Solar Cell Using Double-Walled Carbon Nanotube Thin Film as an Electrode

We introduce a perovskite solar cell using double-walled carbon nanotubes as a transparent electrode (Figure 14).³³ Double-walled carbon nanotubes tend to have higher mobility due to the inner tube with less defects and higher solubility due to the outer tube with defects, and have excellent properties for forming carbon nanotube thin film electrodes in the coating process. Double-walled carbon nanotube film was fabricated by solution coating, and PTAA (= poly(triarylamine)), perovskite layer (MA_0.6FA_0.4PbI_2.9Br_0.1; MA = methylammonium; FA = formamidinium), electron transport layer of C₆₀ and BCP (bathocuproine), and copper electrode were deposited on it to prepare a perovskite solar cell. When undoped double-walled carbon nanotubes were used, the energy conversion efficiency of the device was 15.6%. When double-walled carbon nanotubes were doped with nitric acid (HNO₃), the conversion efficiency improved to 16.7%. When double-walled carbon nanotubes were doped with trifluoromethanesulfonic acid (TFMS),^34,35 the conversion efficiency was 17.2%.

When the fluorescence quenching from the perovskite photoactive layer was confirmed, remarkable quenching was observed when double-walled carbon nanotubes was doped with HNO₃ or TFMS. This indicates that the efficiency of charge collection by double-walled carbon nanotubes is significantly improved by doping. In double-walled carbon nanotubes, the outer tube is susceptible to modification and doping and is considered to be useful.

7. Perovskite Solar Cells Using Semiconducting Carbon Nanotubes

Finally, examples of using semiconducting single-walled carbon nanotubes (s-SWNT) in the photoactive layer of perovskite solar cells are described.³⁶ s-SWNT was separated by agarose gel chromatography. SWNT is dispersed by sodium dodecyl sulfate (SDS) and adsorbed on the gel, and then s-SWNT is selectively eluted by sodium deoxycholate (DOC). Therefore, DOC is attached to s-SWNT, and water molecules are also attached due to the hydrophilicity of DOC. The separated s-SWNT dispersion is added to a perovskite precursor solution (CH₃NH₃I, PbI₂, anhydrous dimethyl sulfoxide (molar ratio 1:1:1) mixed in dimethylformamide (concentration 50 wt%). In the perovskite crystal, s-SWNT (DOC, water molecules) is extruded to the grain boundaries of the perovskite crystal. In other words, s-SWNT is embedded in the grain boundary of the perovskite crystal to enhance performance in solar cells.

The maximum power conversion efficiency of the reference device with nothing added to the perovskite layer was 18.1% (J_SC = 23.1 mA/cm⁻², V_OC = 1.06 V, FF = 0.74). It has been previously reported that the efficiency is improved by adding water to the perovskite precursor solution. In the reported papers, the optimum value of the concentration of water added was different, but in the author's examination, 2 wt% was the optimum value. The highest conversion efficiency at that time was 18.7% (J_SC = 22.9 mA/cm⁻², V_OC = 1.08 V, FF = 0.76). Then, s-SWNT was further added to the perovskite active later, showing higher power conversion efficiency of up to 19.5% (J_SC = 23.7 mA/cm⁻², V_OC = 1.14 V, FF = 0.72). Further improvement of the performance was also performed by using specially designed anthracene-based surfactant to disperse s-SWNT in water.³⁷

8. Conclusion

The author discussed nanocarbon materials such as appropriately designed fullerene derivatives, endohedral fullerenes, and carbon nanotubes improve performance of organic and perovskite solar cells through various reasons. Carbon nanotubes can be doped with nitric acid, trifluoromethanesulfonic acid, and molybdenum oxide as well as modified with a variety of organic semiconductors such as spiro-MeOTAD, conjugating polymers such as P3HT and PTAA, and lithium-ion containing fullerene to improve conductively and charge selectivity. Single-walled and double walled carbon nanotubes are used in this way as cathode and anodes. Highly flexible perovskite solar cells with carbon nanotubes electrodes at both sides can also be realized by using neither transparent conductive metal oxide nor thermally evaporated metal electrodes. The author felt through research these nanocarbon materials have good compatibility with perovskite photoactive layer to decrease hysteresis in perovskite solar cells for improving performance and stability.

Carbon has no restrictions as an element, and because it is light, it is also suitable for constructing flexible lightweight devices. Currently, the authors are working on research to replace various components used in organic and perovskite solar cells with nanocarbon materials as much as possible (Figure 15). In addition, we hope such research will lead to the development of solar cells that belong to a new category whose main components are carbon materials.

Acknowledgment

The author sincerely expresses appreciation to co-workers listed in the reference section, especially, Prof. Il Jeon (Pusan National University), Prof. Shigeo Maruyama (The University of Tokyo), and Prof. Esko Kauppinen (Aalto University). The author also thanks Strategic International Research Cooperative Program (SICORP, Grant Number JPMJSC18H1), Japan Science and Technology Agency (JST).

References

Yutaka Matsuo

Yutaka Matsuo received his PhD from Osaka University in 2001 under the supervision of Profs. Kazuhide Tani and Kazushi Mashima. After that, he moved to the Department of Chemistry, The University of Tokyo to work in Prof. Eiichi Nakamura's group as an assistant professor. In 2004, he joined Nakamura Functional Carbon Cluster Project, ERATO, Japan Science and Technology Agency as a group leader. He was awarded The Chemical Society of Japan Award for Young Chemists for 2005, and so on. From 2019, he was appointed at Nagoya University as a professor. His current research interests focus on the creation of carbon-based functional materials and solar cells.

Author notes