Lucas
Scalon
a,
Flavio Santos
Freitas
b,
Francisco das Chagas
Marques
*c and
Ana Flávia
Nogueira
*a
aInstitute of Chemistry, University of Campinas, Campinas, São Paulo 13083-970, Brazil. E-mail: anafla@unicamp.br
bCentro Federal de Educação Tecnológica de Minas Gerais, Minas Gerais 30421-169, Brazil
cInstituto de Física Gleb Wataghin, University of Campinas, São Paulo 13083-970, Brazil
First published on 20th December 2022
Perovskites are in the hotspot of material science and technology. Outstanding properties have been discovered, fundamental mechanisms of defect formation and degradation elucidated, and applications in a wide variety of optoelectronic devices demonstrated. Advances through adjusting the bulk-perovskite composition, as well as the integration of layered and nanostructured perovskites in the devices, allowed improvement in performance and stability. Recently, efforts have been devoted to investigating the effects of quantum confinement in perovskite nanocrystals (PNCs) aiming to fabricate optoelectronic devices based solely on these nanoparticles. In general, the applications are focused on light-emitting diodes, especially because of the high color purity and high fluorescence quantum yield obtained in PNCs. Likewise, they present important characteristics featured for photovoltaic applications, highlighting the possibility of stabilizing photoactive phases that are unstable in their bulk analog, the fine control of the bandgap through size change, low defect density, and compatibility with large-scale deposition techniques. Despite the progress made in the last years towards the improvement in the performance and stability of PNCs-based solar cells, their efficiency is still much lower than that obtained with bulk perovskite, and discussions about upscaling of this technology are scarce. In light of this, we address in this review recent routes towards efficiency improvement and the up-scaling of PNC solar cells, emphasizing synthesis management and strategies for solar cell fabrication.
Despite the benefits, the integration of PNCs in solar cells is recent. The first report dates from 2016,26 and the efficiency of these devices still lack far behind the conventional perovskite bulk-based solar cells. As a matter of comparison, the record efficiency for bulk-based devices is 25.7%,27 whereas for PNC-based ones is ∼16%.28 The progress in this area toward improved efficiency and further upscaling requires advances in terms of synthesis protocols, comprehension of surface chemistry, the understanding of the fundamental mechanisms of nanocrystal growth and degradation, as well as device manufacture. When we consider upscaling strategies, several studies and reviews have addressed bulk perovskite-based devices13,15–17,19,20,22,29,30 and PNCs in LED applications31–34 or perspectives regarding their use in solar cells,18,25 leaving a gap in scientific reports about these materials applied in large-area solar cells. Here, we contribute to this discussion by emphasizing the application of PNCs in solar cells and approaches to integrating them into large-area modules. To reach this goal, we start to discuss the main properties of PNCs, the state-of-art of synthetic routes, and strategies to fabricate efficient and stable photovoltaic devices. This background information provides a basis to draw a path toward the upscaling of PNC-based solar cells.
Another method to control the MHP's properties is changing its dimensionality rather than composition. Going from bulk 3D to 0D quantum dots (QD) the energy band structure changes from continuous to discrete (i.e., quantized) (Fig. 1b),41–43 in which the electron can only occupy specific energies in the system. The effect of the quantization can be visualized in the electronic absorption spectra of the QDs as sharp and well-defined peaks,44,45 a feature that becomes less prominent as the size increases and the material assumes a bulk-like behavior. Another particularity is the dependence of the bandgap (Eg) on the NC size. This happens because as the material's size becomes smaller than the Bohr radius (the average separation length between electron–hole pairs generated by photoexcitation), the de Broglie wavelength of a free electron (or hole) becomes highly confined inside the particle (Fig. 1c),46 making the Eg of the QD dependent on the size: decreasing the size, increases the Eg.47 The explanation for this phenomenon lies in the solution of the Schrödinger equation for a particle confined in a three-dimensional system (case of a 0D QD), which tell us that the energy of the levels is inversely proportional to the square of the wall thickness, i.e., the nanoparticle size. In other words, narrow walls have more wide-spaced levels, resulting in higher Eg.48,49 In this sense, the variation of the NC size can be used to control the confinement regime, which ultimately affects the material's bandgap. Let us take the CsPbBr3 NC as an example. Its exciton Bohr radius is about 7 nm.50 As size decreases to <7 nm, the electron energy levels assume a discrete rather than continuous energy landscape (see Fig. 1b), causing the material to fall into the quantum confinement regime.51 Based on the size, we can classify these materials in two different regimes: a high confinement regime when the NC dimensions are ≪7 nm, and a low confinement regime when the NC dimensions are ≫7 nm,48,52 which will affect the optoelectronic properties of the material.
Because of its small size, the surface/volume ratio of NC is large. Considering a spheric NC, its surface area can be expressed as 4πr2, and its volume 4/3πr3; here r is the radius of the sphere. Therefore, the surface/volume ratio is 3/r. In this sense, a 7 nm diameter particle will have a surface area equal to 154 nm2, and the surface/volume ratio is ∼8 × 108 m2/m3. For this reason, there are much more atoms on the surface than in the interior of the NC, which is opposite to bulk materials, in which the surface is only a tiny fraction of the whole.53 As a consequence, the chemical potential (i.e., reactivity) of the surface is very high. To decrease the surface's reactivity, the NC tends to agglomerate with time to form large domains, which causes loss of the quantum confinement regime (note that, in this situation, the material's size will increase). To avoid this natural tendency of agglomeration, capping ligands are required to stabilize the material by forming micelles, as schematized in Fig. 1d. Commonly, alkylamines, carboxylic acids, and phosphonic-based ligands are used (Fig. 1e).54–58 These molecules are composed of a polar head that binds to the NC surface, and a long hydrocarbon tail that interacts with the solvents.
Fig. 2 (a) Dependence of the Stokes shift with the variation of NC size from three compositions of PNC: CsPbCl3, CsPbBr3, and CsPbI3.62 (b) Simulated XRD of α-, β-, and γ-phase of CsPbI3 PNC; (c) experimental XRD of CsPbI3 with different NC phases.66 (d) Calculated (empty symbols) and experimental (filled symbols) charge separation (CS) and charge recombination (CR) rates of CsPbI3 QD-rhodamine and their variation with the NC size.68 (a) Reproduced with permission from ref. 62. Copyright 2020 American Chemical Society. (b and c) Reproduced with permission from ref. 66. Copyright 2019 American Chemical Society. (d) Reproduced with permission from ref. 68. Copyright 2019 AIP Publishing. |
Additionally, the size was revealed to impact the lattice strain and crystal phase of the PNC.66 For instance, as the CsPbI3 size increases from 5.7 to 15.3 nm, it loses the α-phase cubic symmetry, approaching γ-phase symmetry with an orthorhombic structure for larger NC, as can be seen from the XRD patterns in Fig. 2b and c. The results also showed that small NC experiences higher tensile strain and revealed octahedra tilt as the size increases. Small NC also seems to be less prone to cubic-to-tetragonal phase transitions at low temperatures (120 K), whereas the large ones suffer from phase transition because of changes in electrical polarization.67
Furthermore, Shang et al. revealed a size dependence on the charge separation (CS) and charge recombination (CR) rates from CsPbI3 QD to rhodamine (electron acceptor) by varying the NC size from 11.8 to 6.5 nm, as depicted in Fig. 2d.68 In this regime of weak to moderate quantum confinement regime, the authors found that the size dependence of the rates is mainly controlled by the orbital overlap between the QD and the rhodamine.69 As a result, the electron transfer is not dependent on the energy level matching between the donor (in this case the QD) and the acceptor (rhodamine) lowest-unoccupied molecular orbital (LUMO), as observed for QD with a strong quantum confinement regime (i.e., small NC size).69 These findings are particularly interesting for optoelectronic applications, in which charge separation and charge recombination dictate the main working mechanism of the devices. It should be highlighted that this study investigated the dynamics of charge transfer in the solution state. Translating those findings to the thin-film state should be analyzed carefully, since the solvent may play a role in the charge transfer mechanism.
In addition to the halide anion, the A-site cation may also influence the nature and properties of the PNCs. In the APbBr3 series, in which A: Cs, MA, and FA, the NC size is larger for the ones based on the organic cations.78Fig. 3a shows TEM images of CsPbBr3, MAPbBr3, and FAPbBr3. The NC size goes from 9.2 nm for the inorganic-NC, to 17.3 and 12.9 nm for MA-, and FA-based NC, respectively, which directly impacts the absorption and emission features.78 The energy separation between dark and bright states decreases from Cl to Br and I, and with the replacement of Cs by FA (Fig. 3b).79 This is important because the thermalization of the excitons towards dark states reduces the light emission efficiency, mainly because of the spin and momentum-forbidden transition to the ground state.80 In this sense, having a large energy separation between the bright and the dark excitonic states should, in principle, improve the radiative recombination of the NC. However, the impact of the dark exciton on the performance of optoelectronic devices based on MHP NC is still understudied.
Fig. 3 (a) TEM images, and absorption and emission features of a series of APbBr3 PNC (A = Cs, MA or FA).78 (b) Schematic illustration of energy separation between the dark (spin-forbidden transition to ground state) and bright (allowed the transition to ground state) and its dependence on the X-site cation specie.79 (a) Reproduced with permission from ref. 78. Copyright 2018 American Chemical Society. (b) Reproduced with permission from ref. 79. Copyright 2018 American Chemical Society. |
Regarding the defect tolerance, ten Brinck et al.82 found that the interstitials defects in CsPbBr3 nanocrystals are energetically favored and the PNC tends to accommodate ionic pairs (i.e. CsBr or PbBr2) in the lattice. The authors noticed that midgap states can arise only from halide species independently of the interstitial defect, either cationic or anionic. These localized states triggered by the halide ion only cause trap states if the anion occupies surface positions that point outward the surface. The excess of halide species against the available halide surface sites exemplifies this situation, where its nonbonding lone electron pairs produce trap states. Since surface passivation can form halide-hydrogen bonds on the surface, this treatment, beyond avoiding agglomeration, could inhibit the capture of the excited electrons by the surface traps.87,88 Vacancy defects by removing ions from either the surface or the core have negligible densities since the energy required to create vacancies is in the range of 1.4 to 2.3 eV. It means that ionic extraction must be induced from species around PNC, which is more effective at the surface, and denotes how surface chemistry plays an important role in the NCs’ stability.82
As will be addressed in the next sections, surface ligands are molecules that bind to the surface and drive the synthesis through tunable sizes, dimensionalities, morphologies, and the related aspects of PNC including stability and optoelectronic properties. Stability is improved when surface passivation is carefully selected, although an effective surface shelling may inhibit the charge transport and deteriorate the device's performance.78,88–90 Furthermore, surface ligands are labile and the detachment of these species can be promoted by organic solvents used in the synthesis or purification due to moderate solubility, and also through neutralization, e.g., oleylammonium bromide being converted into oleylamine and HBr. The detachment mechanisms lead to the elimination of ion pairs from the PNC surface, leaving vacancies and exposing undercoordinated halide ions, which are responsible to trap states and the reduced optical response.90
An elegant way to reduce surface defects associated with an appropriated surface coverage is to interconnect the PNCs to form crosslinks. This strategy not only enhances the PNC stability but also influences optical and solubility properties.81 Li et al.,91 for example, used butylphosphonic acid and 4-ammonium chloride in MAPbI3 NCs, and demonstrated that hydrogen bonds through N–H⋯I and P-OH⋯I are intermolecular interactions sufficient to link PNCs. The solution was then spin-coated onto TiO2 and annealed at 100 °C giving rise to a homogeneous layer and solar cells with an efficiency of 16.55%.91 The use of ligands able to form stronger bonds, i.e. covalent bonds, between PNCs is another interesting approach to reaching/improving PNC-PNC connections. Liu et al.,92 used nitrene radicals from the photochemistry of bisazide under UV irradiation to form covalent C–N bonds with long alkyl chains from ligands on neighboring PNCs, creating PNC networks and retaining their morphological properties. This method for direct optical patterning of perovskite NCs with ligand cross-linkers (DOPPLCER) does not require ligand exchange and is universal to PNCs with various compositions, sizes, surface chemistry and synthesized by different methods. The multicolored patterning of various PNCs via DOPPLCER, retained their size-defined, and emission features before and after patterning (CsPbBr3 and CsPbCl3−xBrx, except for the unstable red-emitting CsPbI3−xBrx).92 Post-patterning anion exchange with iodide salts converts CsPbBr3 NCs to CsPbI3−xBrx patterns and suggests that the PNC cores remain reactive and accessible probably due to the low degree of cross-linking, in contrast to cross-linked ligands via X-ray irradiation or with polymeric ligands.93–95 Furthermore, the method shows the preservation of optical absorption and emission characteristics before and after cross-linking at mild conditions92 without a high-energy source or a ligand exchange process. UV irradiation has promoted crosslinking reactions between PNCs using a polymer backbone containing a cinnamoyl group95 or 2,2-dimethoxy-2-phenylacetophenone (DMPA).96
From computational calculation based on MAPbI3 NCs, the VBM and CBM are characterized by the same k-vector, giving to PNCs a direct bandgap nature with an abrupt absorption onset at the bandgap energy, Eg. Moreover, the bands around the VBM and CBM exhibit a similar curvature. Upon photoexcitation, electron–hole pairs are generated and remain coulombically bounded forming an exciton that required activation energy on the order of the thermal energy, kBT, to be dissociated into mobile charges.99 The single exciton formation involves an electron in the conduction band coulombically bound to a hole in the valence band with four temporal regimes. In the first three, the electron stays within the CB in intraband processes (coherent, non-thermal, and hot excitation regime), followed by one last isothermal regime which involves an interband transition from CB to VB, the exciton recombination.
Prior to electron–hole recombination, a strong spin–orbit interaction, or L–S coupling, between the spin and the orbital motion of the carriers caused by the presence of heavy atoms like Pb in their crystal structure results in the triplet and singlet exciton states, with the last having the lowest energy. However, when L–S coupling is combined with the crystal or inversion symmetry breaking (i.e., Rashba effect), it reduces the triplet state energy to below the singlet state, leading to bright triplet states with high photoluminescence quantum yield (PLQY) and fast radiative lifetimes. These low excitonic states are nearly mixed and not distinguishable at room temperature, but experimentally observed at cryogenic temperatures.11,100,101 Therefore, the energetic order of bright and dark states is still under debate, since bright-to-dark relaxation can be induced by magnetic field, leading to the interpretation that the bright-to-dark spin relaxation is inhibited due to the faster radiative decay from the bright state than their relaxation to the dark state.102
PNC composition influences exciton binding energy since the atomic orbital of the species contributes to both valence and conduction bands. As the halide species changes from Cl−, to Br− and to I− in PNCs, due to contribution to the VBM, the exciton binding energy decreases, while the exciton recombination, carrier trapping, and hot carrier cooling times become longer. When the A-site cation is changed from FA to MA, and Cs, an increase in the carrier cooling time due to the decrease in the strength of the carrier-phonon interaction is observed. Quantum confinement also affects exciton dynamics, on account of their small Bohr exciton diameters,102 estimated to be 5, 7, and 12 nm for CsPbX3 NCs, respectively to Cl, Br and I.2 Thus, the photo-excited electrons and holes in PNC are physical manifestations of “particles in a box”, which promotes fast and efficient radiative recombination due to spatial confinement within the NC.103
The time-resolved spectroscopy studies provide a comprehensive understanding of photoinduced processes, such as the excited-state lifetimes and charge carrier transfer kinetics. Especially, transient absorption spectroscopy (TAS) shows that the radiative lifetimes of CsPbX3 go from 1 to 29 ns with accelerated decays from I-based toward Cl-based PNCs. For ligand-capped PNCs, a long-lived exciton lifetime is attributed to a lower density of trap states.81
In addition to single excitons, the generation of multiexcitons is expected to boost the photocurrent and performance of photovoltaic devices, overcoming the Shockley–Queisser efficiency limit. Bi/multiexcitons can be produced by an energy excess released by a photon with energy at least twice the bandgap, creating additional electron–hole pairs instead of being dissipated through thermalization. However, these bi/multiexctions recombine non-radiatively through the Auger process, in which the recombination energy of one exciton is transferred to another charge carrier of the same PNC, therefore their extraction using an electron/hole acceptor prior to the Auger recombination is mandatory.104 If a multiple exciton generation is observed for the PNC, all of the four temporal regimes previously discussed may overlap. Thus, the mechanism of multiple exciton generation is not an instantaneous process, that is, the energy excess of the hot carrier is not transferred directly to the lowest state. It starts with the excitation energies just above the energy limit of 2Eg. Then, the additional carriers appear after those initially generated upon photon absorption in a state with longer cooling time, probed by studies with FAPbI3 NCs with sizes of 7.5 and 9.8 nm, and CsPbI3 with 11.5 nm, using an excitation energy of ∼2.25Eg and ∼2.4Eg, respectively.102
When the presence of mixed halides is evaluated in PNCs with Cl doping (CsPbI3−yCly and FAxCs1−xPbI3−yCly) the average photoluminescence (PL) lifetime was found to be ∼3 times longer compared to CsPbI3. The authors attributed the increased PL lifetime to the return of the photogenerated charge carriers to the band-edge state by thermal excitation after the decay to the trap states. In this work, suppression of the Auger recombination rate and consequent lengthening of biexciton lifetime was demonstrated using C60 as an electron acceptor, with single exciton dissociation followed by electron transfer from CsPbI3−yCly to C60 revealed by a biexponential dynamics with a fast component of 18–45 ps. The multiple exciton transfer was observed for CsPbI3−yCly and FAxCs1−xPbI3−yCly NCs, and it was found that the biexciton lifetime is long enough (∼195–205 ps) to be extracted by fullerene as well.104 Studies with PNC in contact with redox-active molecules, e.g. the CsPbBr3-rhodamine B, showed that the Föster mechanism (related to energy transfer over larger distances (up to 12 nm)) is dominant, in agreement with the expected and measured rate constant of energy transfer (kET). For Cl-rich nanocrystals, the measured kET is higher, indicating that the Dexter mechanism over shorter distances (1–2 nm) in these systems becomes dominant. These results demonstrate that PNCs present a unique susceptibility to chemical transformations able to access a range of energy and electron transfer process.100 Different hetero and homojunction architectures could be employed to facilitate efficient charge extraction, e.g. PEDOT:PSS and/or metal oxides.105 Terahertz kinetics of MAPbI3 show that, in presence of a metal oxide with a high electron affinity like TiO2, the charge mobility is ∼7.5 cm2 V−1 s−1, lower than in neat MAPbI3 (∼20 cm2 V−1 s−1), supporting the electron injection from MAPbI3 to TiO2.106
Fig. 4 (a) Schematic illustration of the hot-injection method used to produce perovskite quantum dots (QD). HPB is for high-boiling point solvents. (b) Dependence of the PLQY with the dielectric constant of different solvents used to wash the NC in the purification step.125 (c) Variation on the emission peak and photoluminescence quantum efficiency (PLQE) with the concentration of octyl phosphonic acid (OPA) at a fixed reaction time (300 s).126 (d) Variation of nanocrystal size and PLQE with the reaction time at a fixed concentration of OPA (0.3 M).125,126 (a and b) Adapted with permission from ref. 125 Copyright 2018 American Chemical Society. (c and d) Adapted with permission from ref.126 Copyright 2021 American Chemical Society. |
Purification is the next step after synthesis. It is required to remove unreacted precursors and the excess ligands present on the surface, which otherwise can act as trap centers for photogenerated charge carriers.127 Usually, the purification is performed by redispersing the NC in polar solvents (e.g. methanol, acetone, or ethyl acetate), known as “poor solvents”, which do not have the ability to solubilize the NC. This dispersion is then centrifuged to recover the powder – the centrifugation speed can be adjusted to collect the NCs based on their size. These solvents will solubilize the unreacted precursors whereas the NC will precipitate after centrifugation. The number of washing steps should be optimized since it can lead to ligand desorption, compromising the colloidal stability, and elemental composition changes, modifying the optoelectronic properties of the NC.128 The choice of solvent is also important since it influences the optical properties of the PNC. Fig. 4b correlates the PLQY of CsPbBr3 NC dispersed in toluene after washing it with solvents of varied dielectric constant (ε).125 As can be seen, the region with the highest PLQY values is between 5 < ε < 10, and, among the solvents used, the best one was diglyme (ε = 7.23).125 The authors found that poor solvents with low dielectric constant can allow multiple steps of NC washing without compromising the properties of the material. An interesting alternative to the centrifuge purification method was proposed by Tong et al.129 They used lauryl methacrylate (LMA) as a solvent during the synthesis of CsPbBr3 NC. It possesses a high boiling point, and a hydrophobic chain to stabilize the colloidal material. Adding polyurethane acrylate oligomer, and applying UV-light it can be polymerized, yielding a green-emissive composite material comprised of the CsPbBr3 NC embedded into the polymeric network with PLQE as high as 85%, which keeps stable for 100 h.
One of the main challenges of the hot-injection method is the high temperature required for the synthesis, which makes the scalability difficult and very energetically expensive. Nevertheless, room-temperature alternatives to the hot injection method have been studied.126,130,131 For example, a room temperature strategy for synthesizing CsPbBr3 NC was proposed by using octyl phosphonic acid (OPA) and didodecyldimethylammonium bromide.126 The first can control the nucleation of the NC, while the last is employed to quench the growth. By controlling the concentration of OPA and the reaction time, it was possible to tune the size of the NC, as shown in Fig. 4c and d.126 Another alternative is reducing the solvent's boiling point through pressure reduction, which decreases the synthesis temperature by ∼40 °C.132 All of these alternatives facilitate the process's scalability and the industrial manufacture of PNC in solar cells and other optoelectronic devices.
Fig. 5 (a) Scheme of the ligand-assisted reprecipitation (LARP) method: a solution containing the PNC precursors dissolved in polar solvents is dropped into a solution of the capping ligands, leading to the precipitation of the NC, which is recovered and purified by centrifugation.133 (b) Schematic illustration of the NC size variation through the tuning of FA/Pb ratio, and the different morphologies obtained for low and high FA/Pb ratio.139 (c) Enhancement of the solubility of MAPbI3 perovskite in acetonitrile (ACN) by bubbling into methylamine (MA) gas, resulting in a clear yellow solution.140 (d)–(f) CsPbBr3 nanocubes obtained from a synthesis using PbBr2 dissolved in toluene. The complete dissolution of this salt in the apolar solvent was assisted by tetraoctylammonium bromide (TOAB), yielding high fluorescent PNC.141 (g) Illustration of the method used for the in situ synthesis of FAPbBr3 NC directly into the desired substrate.142 (a) Reprinted with permission from ref. 133 Copyright 2015 American Chemical Society. (c) Reprinted with permission from ref. 140 Copyright 2019 American Chemical Society. (d–f) Reprinted with permission from ref. 141. Copyright 2017 American Chemical Society. (g) Reprinted with permission from ref. 142. Copyright 2018 American Chemical Society. |
The heart of this technique lies in the different solubility of the perovskite precursors in the solvent mixture. Strictly speaking, the precursors are highly stable (soluble) in the polar solvents, but when in contact with a poor apolar solvent it precipitates immediately, forming the NC stabilized by the capping ligands that coordinate Pb-sites on the surface.134 To achieve small NCs with expressive quantum confinement effect, the chain length of the capping ligands should be optimized.133,135 Capping ligands with a very small chain, for instance, will cause NC aggregation, thereby compromising the quantum confinement-dependent properties. On the other side, long-chain molecules will preclude the NC-to-NC electronic coupling and difficult charge extraction in the system. In addition to capping ligands, several other parameters will impact the final properties of the NCs, including the synthesis temperature, ligand/precursor concentration, and choice of solvents.134–137
Typically, the purification is performed by multiple centrifugation steps of the solids suspended in the poor solvent, which is analogous to the described in the hot injection method. This allows the removal, at least partially, of the unreacted capping ligands, A- and B-site precursors. The choice of the solvent in this step is also crucial since it can change the emission profile and, as a consequence, the properties of the PNC.138 The purification with toluene and chlorobenzene, for instance, results in a broad and multiple peak emission due to a large size distribution. In contrast, hexane, a low polarity solvent (polarity equals 0.06), yields a single and narrow emission peak as a result of the narrow size distribution.
One of the major issues of this technique is the use of highly coordinated polar solvents to solubilize the perovskite precursors.130,134,140,143 The strong coordination ability between those solvents and PbX2 precursors results in the crystallization of the perovskite phases from PbX2-solvent complex intermediates, causing the solvent molecules to remain in the final product.144 These defective perovskites are more prone to degrade in the presence of water or oxygen. Strategies to circumvent this issue have been reported. For example, PbX2 (X = Br, I) was dissolved in toluene with the addition of small amounts of oleic acid (OA) and oleylamine (OLA).139 Concomitantly, the A-site source (FA-acetate) was dissolved in OA using ultrasonication. By mixing both solutions, the crystallization of the NC occurs spontaneously at room temperature, and the size can be controlled by adjusting the ratio between the A-site cation and the Pb2+ precursors (Fig. 5b). Acetonitrile saturated with methylamine has also been used to solubilize the MAPbI3 precursors (Fig. 5c).140 This solution was then added into a mixture of toluene, OA, and OLA, yielding cubic NC of MAPbI3 with 10–15 nm size, and PLQY >90%. This was possible because methylamine can form a liquid perovskite phase with the MAPbI3 perovskite with the chemical formula CH3NH3PbI3·xCH3NH2, improving the solubility of the perovskite in the acetonitrile (ACN), which is a less coordinated solvent when compared to DMF and DMSO, for instance.145 Besides, PbBr2 was found to be completely soluble in toluene in the presence of tetraoctylammonium bromide (TOAB).141 The subsequent swift injection of this solution into a Cs2CO3 solution in oleic acid and toluene resulted in the formation of highly emissive CsPbBr3 nanocubes with ∼13 nm (Fig. 5d–f).
Using the LARP method it is also possible to fabricate in situ PNC films. The process is quite similar to the ex situ mentioned in the lasts paragraphs, the difference is that the perovskite precursor solution containing the capping ligand is coated into a substrate, and then the poor solvent is dropped, causing the supersaturation of the film due to the removal of the polar good solvent and the consequent precipitation of the NC (Fig. 5g).142,146
Fig. 6 (a) Schematic illustration of the PNC synthesis, and (b) absorption and emission spectra of the different compositions obtained using the ultrasonication method.147 (c) Variation of the energy gap with the number of layers of CsPbI3 NC prepared from ultrasonication, achieved by controlling the CsCO3/PbI2 ratio.147 (d) Variation of the PL intensity from various times of ultrasonication; the time required to achieve the high PL intensity is dependent on the NC composition.148 (a–c) Reproduced with permission from ref. 147. Copyright 2016 Wiley-VCH. (d) Reproduced with permission from ref. 148. Copyright 2016 Royal Chemical Society. |
When compared to the hot injection method, the ultrasonication procedure uses room temperature for the synthesis. This is possible because ultrasonication promotes the formation of bubbles with a high temperature inside.148 Those bubbles help the dissolution of the precursor's salts and cause supersaturation, nucleation, and consequent growth of the PNC. Another advantage is that it is not necessary to use an inert atmosphere during the synthesis to obtain high-quality materials.149 In addition, it can be used to synthesize PNC without the use of polar solvents, such as octadecene and dimethylformamide (DMF). For instance, liquid paraffin was used to obtain CsPbBr3 PNC with shapes varying from 3D nanocubes to 0D QD by adjusting the ratio of the capping ligands and the ultrasound power and time.150,151 Furthermore, the PNCs can be prepared directly from the perovskite powders (e.g., MAPbBr, MAPbI) and PbX2 using OLA and OA as both capping ligands and solvents. The ultrasound breaks the bulk perovskite into NC which, in the presence of the capping ligands, stabilizes the small particles formed.152–154 Although promising, the scaling-up of this process is challenging. The tip of the sonicator has a small size (<1 cm), which could difficult the homogeneity of the reaction, decrease the reaction yield and purity, and affect the NC properties. The scaling up of the ultrasonication method would require, for instance, multiples ultrasound tips into the reaction reservoir to allow a synthesis with appreciable yield.
The most common ligands used for the synthesis of PNCs are oleic acid and oleylamine. When this acid–base pair is mixed, a chemical equilibrium takes place, with the carboxylic acid protonating the amine to ammonium according to the reaction: R-COOH + R-NH2 ⇌ R-COO− + R-NH3+. The –NH3+ group is able to interact with the A-site of the PNCs at the surface, and stabilizes the NC by the formation of three hydrogen bonds with the X− ions.160 On the other hand, OA does not bind to the surface of the NC, being present as a free ligand in the solution even after the washing steps of purification.160,161 Since the binding between the ligand and the PNC surface is very dynamic,162 the presence of free OA helps to maintain the colloidal stability of the system. It should be highlighted that despite the role of OLA being well understood by the scientific community, the role of OA is still under constant debate, and it is not clear the mechanisms for the colloidal stability conferred by this molecule.
Several types of surface ligands have been used. We can divide them into monodentate, bidentate, branched, and zwitterions (Fig. 7). The monodentate ligands, which include both OA and OLA, are comprised of one polar head that can interact with the PNC or its precursors, and an aliphatic or aromatic moiety that interacts with the solvent. Bidentate ligands, such as 2,2′-iminodibenzoic acid and 12-aminododecanoic acid, have two polar heads, allowing them to bind at two NCs at the same time. They have been reported to reduce structural disorder at the NC surface,163 and lead to a beneficial passivating effect.164 In these molecules, tailoring the alkyl/aromatic chain is a good alternative to improve the optoelectronic properties of the material, benefiting charge injection and transport.165,166 Another class of ligands are the branched ones, e.g. 2-hexyldecanoic acid, trioctylphosphine oxide (TOPO) and (3-aminopropyl)triethoxysilane (APTES). They are bulky molecules with a large steric hindrance, which is suggested to benefit the formation of small and more uniform NCs.54,158,167 Finally, zwitterionic ligands such as amino acids and phosphocholine have at least two functional groups in their structure: one with a positive charge and another with a negative charge. The use of these molecules as ligands in PNCs has been reported to increase chemical stability and PLQY.168–170
Fig. 7 Classes of ligands to be used during the synthesis of PNCs: monodentate, bidentate, branched, and zwitterionic ligands. |
As can be seen, there is a plethora of molecules that can be used as surface ligands in PNCs, and it is quite hard to state universal rules for the choice of a proper ligand. In general, what is aimed is a stable and cheap ligand, capable of passivating the surface defects, and conferring colloidal stability to the NCs. In this sense, it is important to know the surface composition of the NC in order to perform modulations in the interaction strength between the ligand and the NC surface, which can be achieved by inserting electron donating or withdrawing groups to tune the electron density over the interacting polar head. This type of modification can also tune the van der Waals interactions among the ligand molecules, which can facilitate the NC-to-NC electronic coupling. The electronic coupling between the nanocrystals is an important aspect to be considered for their application in solar cells. This is because a long hydrocarbon chain can hinder proper electronic coupling and also imposes issues for charge transfer, because of its insulating nature. Therefore, short alkyl/aryl chains are preferred. However, it should be highlighted using these short-chain ligands it is more difficult to obtain small-size NC since they are less effective to prevent aggregation when compared to long alkyl/aryl chains. Therefore, the selection of the proper ligand depends on a plethora of aspects, including the synthetic methodology used, the composition of the NC surface, the density of defects, the morphology aimed, and the desired application for the material.
There is a huge floor for the exploration of new ligands, in particular, the use of functionalized molecules endowed with groups that can properly interact and passivate surface defects on the NC should be considered; we will return to this discussion later on. However, as we will discuss in section 5, attention should be taken to selecting the proper ligand when applying the NC to solar cells is aimed. The reason for this is that the insulating nature of the vast majority of capping ligands decreases the NC-to-NC electronic coupling, hampering the efficient charge separation in the system. At this point, it is clear that the ligand's nature is a crucial part of the integration of PNC into solar cells. Just to mention, short-chain molecules are reported to have a long exciton diffusion length compared to those with long-chain,171 which directly impacts the charge extraction efficiency of the PNC solar cells. However, direct synthesis using short alkyl chain molecules can be challenging, especially because of the decreased solubility conferred by these molecules when compared to the long-chain ones. To circumvent this issue, post-synthetic strategies are used to replace the long-chain molecules that came from the synthesis with small ones, a process known as “ligand exchange reaction”.
Regarding the ligands, it was found that the number of carbons in the alkylic chain of the carboxylic acid and amine ligands can influence the size and shape of PNCs at a given temperature.165,172,173 For instance, amines with a long alkylic chain in combination with oleic acid results in nanoplatelets with larger length compared to short amines.173,174 The explanation for this phenomenon lies in the van der Waals interactions between the ligands, which are stronger for molecules with a long alkylic chain.174 As a consequence of this interaction, the surface energy of the NC is higher for long-chain molecules, resulting in a natural tendency of the NCs to grow with a large size in order to minimize their surface-to-volume ratio and have better stability.174 Another aspect to be considered is the higher solubility of long alkylic chain amines, which facilitates the growth of large-size NCs.175 In addition to this chemical aspect regarding the nature of the ligands, the concentration of the ligands and Pb-source is also capable of modulating the size of the NCs.134,139,176 Considering PNCs prepared from the LARP method, as the concentration of the Pb-source increases, more precursors are available in the reaction medium, allowing the formation of large NCs. Conversely, at a low concentration regime, the Pb-source becomes soluble in both polar and apolar solvents, so it is not observed in the precipitation of NCs. In the case of the ligands, a high concentration slows down the growth of the NC due to a decrease in the reactivity of the precursors since the ligand forms complexes with Pb2+.134
The temperature of the synthesis also impacts the size, with higher temperatures resulting in larger NC. For instance, Otero-Martínez and collaborators showed that from 175 °C to 100 °C the MHP NC shape changes from 3D, with bulk-like properties, to 0D nanocubes, with a strong quantum confinement effect (Fig. 8a).177 This is also valid for other NC compositions, such as FAPbBr3, and CsPbX3 (X = Cl, Br, I). As a consequence of the NC size variation, different emission peaks (corresponding to different Eg) were obtained.177 Because of this intimate relationship between the NC size and the reaction temperature, controlling this parameter during the synthesis is crucial and it can be one of the causes of batch-to-batch non-reproducibility. The size dependence on the temperature can be explained in terms of the solubility of the PNCs precursors: as the temperature increases, their solubility increases, facilitating the formation of bigger nanocrystals due to reducing the supersaturation degree.134 Other works also reported that temperature impacts the products obtained, the morphology, crystal phase, and purity of the PNC,178–180 reinforcing the importance of fine control of the reaction temperature.
Fig. 8 (a) The dependence of the emission peak on the thickness/size of the NP can be controlled through the synthesis temperature.177 (b) Influence of ZnBr2 treatment on the NC size and size distribution for CsPbBr3 PNC.181 (c) Schematic illustration, scanning electron microscopy (SEM) image, and photography of the APbX3 NC powder grown inside mesoporous SiO2, that acts as a template for the NC grow.182 (d) Schematic illustration of the NC growth inside the cavities of metal–organic framework (MOF).184 In both (c) and (d) the size of the NC is controlled by the pore size of the template used, here mp-SiO2 and MOF. (e) TEM images and size distribution diagram of the CsPbBrxCl3−x NC obtained at different cooling methods.185 (f) Schematic illustration of the purification using gel permeation chromatography. The sample and the solvent are added in the stationary phase (ii). As the solvent evolves, products A, B, and C are separated (iii) due to the differences in their interaction with the mobile/stationary phases. (b) Reproduced with permission from ref. 181. Copyright 2018 American Chemical Society. (c) Reproduced with permission from ref. 182. Copyright 2016 American Chemical Society. (d) Reproduced with permission from ref. 184. Copyright 2016 American Chemical Society. (e) Reproduced with permission from ref. 147. Copyright 2016 Wiley-VCH. |
In addition to controlling the size and shape, the application of PNCs in solar cells also depends on the size distribution of the NCs. Having a sample with a broad size distribution means the existence of NCs with different bandgaps, which can result in charge funneling from the higher to the lower bandgap material and trap centers, which decreases the maximum photovoltage delivered by the device. Several strategies have been reported to control PNCs size distribution. Dong et al.,181 for instance, introduced ZnBr2 as an additional source of bromide and demonstrated that Br−-rich condition in the synthesis of CsPbX3 NC through the hot-injection method helps to obtain narrow size distribution (<1 nm) and small size (<7 nm), as shown in Fig. 8b. Furthermore, an approach to prepare fixed-size NCs is growing them inside porous materials. Several porous materials have been tested, including silica,182 nanoporous silicon, alumina,183 and metal–organic frameworks184 (Fig. 8c and d). In addition, the cooling rate of the reaction prepared from the hot-injection method can also impact the size distribution. The slow cooling of the reaction at room temperature, for instance, can prolong the nucleation and growth process, yielding wide-size distribution (Fig. 8e). To circumvent this issue, Luo et al. introduced an ultrafast thermodynamic control (UTC) strategy for CsPbBrxCl3−x NCs.185 This method consists of quickly cooling the reaction using liquid nitrogen at 77 K inserted directly into the reaction flask, immediately ceasing the NC growth, which allows a narrow size distribution, as shown in Fig. 8e. The authors combined the UTC strategy with the defect passivation of PbX64− octahedra using Pb(BrCl)2 salt to obtain strongly blue-emitting CsPbBrxCl3−x NC with PLQY of 98%.
An elegant way to obtain samples of PNCs with a controlled size distribution is by employing gel permeation chromatography (GPC) (Fig. 8f). This method is routinely used in organic chemistry laboratories to separate mixtures of different products obtained from a reaction. A stationary phase, such as SiO2 or Al2O3, and a mobile phase, e.g. chloroform, ethyl acetate, toluene, and diethyl ether, are used to provide the separation of the reaction's products. Each compound from the reaction mixture will interact differently with the stationary and mobile phases. Those that interact strongly with the stationary phase will be more retained, and will eluate from the column at later times. Conversely, those that interact strongly with the solvent, i.e. less interaction with the stationary phase, will eluate in early times. By collecting small aliquots of the solvent from the column it is possible to separate the mixture into (more) pure products. Translating this to PNC, GPC allows the narrowing of the polydispersity in a mixture of NC of different sizes, because of the different interactions that the small and the large NCs have with the stationary/mobile phase. This strategy has also been successfully used for the removal of unreacted materials from PNC synthesis,186 the optimization of ligand binding strength with the outer shell of the NC, and ligand exchange.187
Fig. 9 (a), (b), and (c) Show the transmission microscopy images, absorption features, and PL decay curves, respectively, of CsPbBr3 NCs prepared with different amounts of water.191 (d) and (e) TEM images of CsPbBr3 and the CsPbI3 NC obtained after the halide ion exchange reaction with PbI2.193 (a–c) Reproduced with permission from ref. 147. Copyright 2018 Wiley-VCH. (d and e) Reproduced with permission from ref. 193 Copyright 2018 Wiley-VCH. |
The shape control is also possible through the adjustment of reaction temperature and the amount of Cs-precursor in the case of CsPbBr3 PNC.192 By increasing the Cs:Pb ratio, the NC shape evolves from 2D nanoplatelets to 0D nanocrystals, with dimensions of (8.6 × 10) nm and a diameter of 17.4 nm, respectively. Fixing the Cs:Pb ratio and increasing the reaction temperature from 120 to 185 °C, the morphology goes from thin nanoplatelets with (7.5 × 2.5) nm dimension to a cubic shape with 10.5 nm length. Also, the anion exchange reaction was demonstrated to change the morphology of the NC.193 Anion exchange reaction is used to change the composition of the pre-synthesized NC and occurs through the diffusion from an ion in solution to the interior of the material, a process controlled by the diffusion both outwards and inwards, and the ion exchange rate.194 Generally, this methodology retains the morphology of the MHP NC,195,196 but this was not what Tong and collaborators demonstrated.193 By replacing Br− with I−, or Cl− in CsPbBr3 NC, the authors verified that morphology changes from nanowires with 1–2 μm and width of 12 nm, to nanorods with a low aspect ratio (Fig. 9d and e). The morphology change causes a blueshift in the absorption and emission spectra, and also increases the PLQY from 10% in the CsPbBr3 nanowires to 47% in CsPbI3 nanorods. This shape variation is attributed to the excess of ligands present in the PbX2 precursor solution, which causes the random fragmentation of the nanowires forming the nanorods.
Fig. 10 (a) Schematic illustration of a PNC solar cell with a nip-type architecture. Here, TCO is the transparent conductive oxide (e.g., ITO or FTO); ETM is the electron transport material (e.g., SnO2, TiO2); and HTM is the hole transport material (e.g., Spiro-OMeTAD, PTAA). (b) Scheme of the PNCs deposition by spin-coating for solar cell fabrication. In the first step, a solution containing the PNC is deposited onto the desired substrate. Following this, the ligand exchange step is realized. (c) Scheme of a PNC solar cell and the resultant current density vs. voltage (JV curve) obtained after 1, 2, 3, 4, and 5 successive cycles of solution deposition and ligand exchange.214 (d) Proposed mechanism of the effect of the MeOAc treatment on the PNC surface.214 (c and d) Reproduced with permission from ref. 214 Copyright 2021 American Chemical Society. |
The first report on PNCs applied in solar cells dates from 2016, when Swarnkar et al.26 used 9 nm cubic α-CsPbI3 QDs to fabricate a nip-type solar cell with TiO2, as the ETM, and Spiro-OMeTAD, as the HTM. The device delivered an efficiency of 10.77%, with an open-circuit voltage (VOC) of 1.23 V, short-circuit current (JSC) of 13.47 mA cm−2, and fill factor (FF) of 65.0%. Since then, numerous strategies have been proposed to improve solar cell efficiency, either by management of the PNC synthesis and deposition or by passivation of device interfaces. In Table 1 we provide a summary of the recent achievement in PNC solar cells together with the method used to prepare the NC and the solar cell architecture, and the bandgap of the active layer, which will influence the photon absorption. We can see that most of the devices are prepared using the regular nip-type architecture, with TiO2 and Spiro-OMeTAD being the ETM and HTM materials, respectively. Also, the NCs of the photoactive layer are usually synthesized from the hot injection method, probably because this is an established procedure for NC synthesis, and allows versatility in terms of NC composition and shape, which can influence the optoelectronic properties of the material and, consequently, the photovoltaic efficiency of the solar cell. Among the works reported in Table 1, we highlight the efforts made by Hao et al.197 and by Li et al.198 obtaining efficiencies of 16.6% and 16.2%, respectively, for nip-type architecture with Cs0.5FA0.5PbI3 and CsPbI3 QD as the active layer, respectively. To the best of our knowledge, these are the highest efficiencies for PNC solar cells reported so far. Compared to the first report in 2016,26 the performance gain in only six years is superior to 60%.
Solar cell architecture | Synthesis method | E g (eV) | V OC (V) | J SC (mA cm−2) | FF (%) | PCE (%) | Year | Ref. |
---|---|---|---|---|---|---|---|---|
a Active layer prepared with NC with different sizes. | ||||||||
FTO/TiO2/CsPbI3 NC/PTB7/MoO3/Ag | Hot injection | 1.73 | 1.27 | 12.39 | 80.0 | 12.55 | 2018 | 199 |
ITO/SnO2/FAPbI3 NC/Spiro-OMeTAD/Au | Hot injection | 1.55 | 1.10 | 11.83 | 64.42 | 8.38 | 2018 | 200 |
ITO/ZnO/Ba(OH)2/FAPbI3 NC + conductive polymer/MoOx/Ag | LARP | — | 0.56 | 15.22 | 65.41 | 5.51 | 2018 | 201 |
FTO/TiO2/α-CsPbI3 + FAPbI3 NCs/PTAA/MoO3/Ag | Hot injection | 1.30 (α-CsPbI3) 1.36 (FAPbI3) | 1.22 | 17.26 | 74.0 | 15.6 | 2019 | 198 |
FTO/TiO2/CsPbBr3 NC/Spiro-OMeTAD/Au | Hot injection | 2.30 | 1.34 | 9.41 | 36.0 | 4.57 | 2020 | 202 |
FTO/TiO2/CsPbI3 NC/PTAA/MoO3/Ag | Hot injection | 1.29 | 1.22 | 75.0 | 15.1 | 13.8 | 2020 | 203 |
FTO/TiO2/FAPbI3 NC/PTAA/MoO3/Ag | Hot injection | 1.37 | 1.12 | 16.7 | 71.0 | 13.2 | 2020 | 203 |
FTO/TiO2/CsPbI3 NC/PTAA/MoO3/Ag | Hot injection | — | 1.24 | 15.84 | 75.50 | 14.9 | 2020 | 204 |
ITO/SnO2/CsPbI3 NC/Spiro-OMeTAD/Ag | Hot injection | ∼1.75 | 1.22 | 17.66 | 63.38 | 13.66 | 2020 | 205 |
FTO/TiO2/α-CsPbBr3 NC/PTAA/MoO3/Ag | Hot injection | 2.38 | 1.54 | 4.49 | 72.45 | 5.01 | 2020 | 206 |
FTO/NiOx/CsPbI3 NC/C60/ZnO/Ag | Hot injection | 1.76 | 1.19 | 14.25 | 77.6 | 13.10 | 2020 | 207 |
ITO/SnO2/Cs0.5FA0.5PbI3 QD/Spiro-OMeTAD/Au | Hot injection | 1.64 | 1.17 | 18.3 | 78.3 | 16.6 | 2020 | 197 |
FTO/TiO2/CsPbI3 NC/Spiro-OMeTAD/MoO3/Ag | Hot injection | — | 1.27 | 16.1 | 74.8 | 15.3 | 2021 | 208 |
ITO/SnO2/PCBM@CsPbI3/CsPbI3 NC/PTB7/MoO3/Ag | Hot injection | 1.77 | 1.26 | 15.2 | 78.0 | 15.1 | 2021 | 209 |
FTO/TiO2/CsPbI3 NC/PTAA/MoO3/Ag | Hot injection | 1.77 | 1.26 | 15.81 | 75.3 | 15.05 | 2021 | 210 |
FTO/TiO2/CsPbI3 NC/CsPbI3 NC:F6TCNNQ/PTAA/MoO3/Ag | Hot injection | 1.77 | 1.25 | 16.90 | 71.0 | 15.01 | 2021 | 211 |
FTO/TiO2/MAPbI3 NC@SiO2/Spiro-OMeTAD/Au | In situ synthesis @SiO2 matrix | — | 1.02 | 16.42 | 56.1 | 9.3 | 2021 | 212 |
FTO/TiO2/CsPbI3 NC/Spiro-OMeTAD/Ag | Hot injection | 1.80, 1.79, 1.78a | 1.25 | 14.9 | 71.0 | 13.2 | 2021 | 213 |
FTO/TiO2/CsPbI3 NC/Spiro-OMeTAD/MoO3/Ag | Hot injection | — | 1.19 | 14.59 | 73.85 | 12.85 | 2021 | 214 |
FTO/m-TiO2 + CsPbI2.25Br0.75 QD/CsPbI2.25Br0.75 QD/Spiro-OMeTAD/Au | Hot injection | 1.87 | 1.20 | 14.21 | 72.1 | 12.31 | 2021 | 215 |
ITO/SnO2/CsPbI3 QD/Spiro-OMeTAD/Ag | Hot injection | — | 1.27 | 17.71 | 72.0 | 16.21 | 2021 | 216 |
PET/ITO/SnO2/CsPbI3 QD/PTB7/MoO3/Ag | Hot injection | 1.77 | 1.24 | 13.6 | 73.0 | 12.3 | 2021 | 209 |
FTO/TiO2/Ag:CsPbBr1.5I1,5/Spiro-OMeTAD/MoO3/Ag | Hot injection | 2.12 | 1.04 | 12.51 | 74.0 | 9.67 | 2021 | 217 |
ITO/PEDOT:PSS/α-CsPbI3 NC/C60/BCP/Al | Colloidal synthesis | 1.93 | 0.98 | 16.21 | 67.3 | 10.06 | 2022 | 218 |
FTO/c-TiO2/CsPbI3 QD/Spiro-OMeTAD/MoOx/Ag | Hot-injection method | ∼1.77 | 1.24 | 15.4 | 71.3 | 13.7 | 2022 | 219 |
ITO/SnO2/CsPbI3 QD/Spiro-OMeTAD/Ag | Modified hot-injection method | 1.77 | 1.23 | 17.73 | 74.5 | 16.25 | 2022 | 28 |
ITO/PEDOT:PSS/FAPbI3 QD/PCBM/BCP/Ag | Modified hot-injection method | 1.55 | 0.89 | 14.56 | 89.0 | 10.13 | 2022 | 220 |
FTO/TiO2/Cs0.5FA0.5PbI3 QD/PTAA/MoO3/Ag | Hot-injection method | ∼1.63 | 1.19 | 16.82 | 72.4 | 14.58 | 2022 | 221 |
FTO/c-TiO2/m-TiO2/CsPbI3 QD/PMMA/Cu12Sb4S13 QD/Au | Hot injection | — | 1.14 | 16.14 | 53.1 | 10.99 | 2022 | 222 |
There are several effects of this treatment on solar cell performance. From one side the short alkyl chain replaces the long ligands coming from the synthesis, improving the electronic coupling between the NCs, which in turn improves the charge separation in the device, allowing higher values of VOC and JSC.229,230 On the other side, lead salts are a source of Pb2+ ions, which passivates Lewis base defects present on the NC surface.26 Using this protocol, devices based on CsPbI3 NC with efficiencies of ∼10% can be obtained.26 The sequential repetition of the spin-coating deposition and washing steps can yields thin films with thicknesses varying from 100 to 400 nm, conferring the possibility to fabricate films thin enough to be integrated into flexible and semitransparent devices.
Although the ligand exchange improves the efficiency of the PNC solar cell when compared to the device fabricated using NC with large alkyl chains, it is necessary to control the number of MeOAc dipping steps.214 As shown in Fig. 10c and d, the number of washing steps can significantly impact the device's performance. For instance, the current density versus voltage (JV) curve in Fig. 10c shows that only one step of washing (1-cycle) results in low VOC, JSC, and FF compared to 3-cycles. Further increase in the number of washing cycles above three causes the decrease of all photovoltaic parameters. A proposed explanation is that a few steps will not remove an appreciable amount of long-chain ligands, while excessive steps can cause the formation of the yellow perovskite phase;214 both will compromise the device's efficiency and stability. In this sense, a balance is necessary, and three washing steps seem to be the most appropriate to replace the long capping ligands with short ones and improve the PNC solar cell performance.
As can be inferred from the last paragraphs, the ligand exchange is one of the most important steps in the PNC solar cell assembly, as it will influence the coupling between the NC, impact the defect density of states on the surface, and affect the device efficiency. Given its importance, several strategies based on ligand exchange modifications have been used to improve the performance and stability of the PNC solar cell, with the possibility to add different molecules to confer passivation of the NC surface; some examples of the molecules are shown in Fig. 11a. It should be highlighted that the choice of the molecule and type of treatment realized on the NC to replace the ligands will depend on the nature of the capping ligands, and on the composition of the NC since it influences the binding strengths between the ligand-NC surface.
Fig. 11 (a) Molecular structure of some passivating molecules used for the surface passivation of PNC. (b) Schematic illustration of the defect passivation ability of amino acids vs. methyl acetate (MeOAc) used during ligand exchange.205 (c) Comparison of JV curves of control PNC solar cell using Pb(NO3)2 in MeOAc and using amino acids dissolved in MeOAc.205 (d) Evolution of the PNC solar cell performance with time in an environment with 20–30% of humidity and temperature of 20–30 °C. Here is shown the effect of the mercapto group position in the mercaptopyridine (MP) molecule used during the ligand exchange step.232 (b and c) Reproduced with permission from ref. 205. Copyright 2020 Wiley-VCH. (d) Reproduced with permission from ref. 232. Copyright 2020 American Chemical Society. |
An alternative to the conventional Pb(NO3)2/MeOAc mixture was demonstrated using hexane/ethyl acetate (EtOAc).231 This solvent mixture was able to remove ∼57% of the long capping ligands, causing an increase in the NC size from 7.8 to 10.1 nm, and leading to a PCE improvement from 5.9% to 12.2%, respectively. This is possible since the coordinating ability of the ethyl acetate molecule, which is capable of replacing some ligands from the nanocrystal surface. In addition, it is possible to use functionalized ligands, such as amino acids. These bidentate ligands are shorter than OA and OLA ligands, causing a better electronic coupling between the NC and conferring a dual passivation effect on the surface due to the presence of carboxyl (–COOH) and ammonium (–NH3+) groups in the same molecule.205Fig. 11b shows a schematic illustration of this passivation effect conferred by the amino acids when compared to MeOAc, in which we can see the passivation of Cs+ and I−-related defects. Applying these molecules during the ligand exchange process resulted in PNC solar cells with the architecture ITO/SnO2/CsPbI3 NC/Spiro-OMeTAD/Ag with efficiency up to 13.7%, presenting improvements in VOC, JSC, and FF when compared to the ligand exchange performed with Pb(NO3)2, as indicated in Fig. 11c.205 This value is 2% higher than the obtained for the ligand exchange using Pb(NO3)2/MeOAc mixture, which evidences the key role of the short and functionalized ligands. Another alternative is the use of pyridine derivatives, such as mercaptopyridine (MP) in MeOAc solution.232 This approach resulted in solar cells (FTO/TiO2/CsPbI3 NC/PTAA/MoO3/Ag) with efficiencies of ∼14% (versus 12% for the control devices), and negligible hysteresis. It is interesting to stress that the position of the mercapto (–SH) group in MP impacts both device performance and stability: in the ortho position, –SH experiences hysterical hindrance to binding on the NC surface, difficulting the defect passivation and negatively impacting the device's stability (Fig. 11d) and efficiency (which decreases to 13.09%). Conversely, when the mercapto group is in the para position both efficiency and stability were increased, highlighting the importance of molecular engineering strategies in the design of new passivating candidates. This type of treatment, which includes both improvements in performance and stability, is of central importance, since, with time, the ligand desorption in the thin film can difficult the maintenance of stable photoactive phases,161,233 induces aggregation, and consequent loss (or attenuation) of the quantum confinement effects, compromising the device performance. Therefore, it is important to look carefully not only at the nature of the functional group present on the molecule and its isolated interaction strength with the NC surface defects, but also at its position on the molecule and the steric hindrance for the interaction with the NC. Other functionalized molecules can also be explored to be used during the ligand exchange step, including chiral ligands,234 methoxy silane-based molecules,235 potassium bromide,236 zinc methacrylate (ZnMA), trioctylphosphine oxide (TOPO) co-passivation,237 and zwitterions, as sulfobetaine and phosphocholine.
A different approach for the photoactive layer was proposed by Zhao et al.227 They used the versatility of PNCs to be synthesized in a giant plethora of compositions to prepare an active layer consisting of a QD heterojunction. That is, each layer was deposited with a different ratio of CsPbI3 and FAPbI3 QD, forming Cs1−xFAxPbI3 QD. The controlled composition of each layer allows a boosting of the charge separation extraction inside the QD layer, achieving efficiencies of nearly 17%. Similar control of the band alignment on the PNC layers was reported by Yuan et al.213 The authors used layer-by-layer deposition of CsPbI3 NC with different sizes to prepare the photoactive PNC layer, which allows gradient band alignment, improving the device performance from 10.3 to 13.2%, with VOC of 1.25 V.
Fig. 12 (a) Schematic illustration of the effect of CsOAc treatment on mp-TiO2 layer to facilitate the penetration of the PNC.238 (b) Space-charge limited current (SCLC) measurements of electron-only devices with a structure of ITO/TiO2/CsPbI3 PNC/PCBM/Ag.238 (c) Distribution of the ligand's energy level and energetic position of the respective HOMO and LUMO. (3-Aminopropyl)triethoxysilane (APTES), oleic acid (OA), oleylamine (OAm), naphthoic acid (NCA).240 (d) Schematic diagram of the PNC solar cell assembled with the PCBM:PNC blend and the respective JV curve.209 (e) Comparison between the energy level of the CsPbI3 QD and a series of Y6 molecules. On the right side, it is shown a comparison between the J–V curves obtained for QD:Y6 blend and the control device.210 (a and b) Reproduced with permission from ref. 238. Copyright 2020 American Chemical Society. (c) Reproduced with permission from ref. 240. Copyright 2021 De Gruyter. (d) Reproduced with permission from ref. 209. Copyright 2021 Nature. (e) Reproduced with permission from ref. 210. Copyright 2021 Wiley-VCH. |
Tin dioxide (SnO2) is an alternative to TiO2. It offers a good energy level match with the perovskite layer, high electron mobility, and conductivity, and less photocatalytic activity, which can help improve the device's stability.241 Usually, this layer is deposited onto the conductive transparent oxide, e.g. indium tin oxide (ITO) or fluorine-doped tin oxide (FTO), from a SnO2 QD dispersion in water. The Cl-doping of SnO2 QD solution through the use of Cl-based salts, for example, KCl, was found to promote the passivation of oxygen vacancies on the film and improve the cubic-phase stability of CsPbI3 QD that is deposited on top due to reducing VOC loss.242 The benefits of the Cl-passivated SnO2 layer resulted in a champion efficiency of 14.5% (against 13.8% of the TiO2-based PNC solar cell), which can keep 80% of the initial efficiency after 8 h of light soaking in an environment with 50% of relative humidity. Another alternative for SnO2 passivation is the use of hybrid interfacial architecture. Hu et. al,209 deposited a blend of CsPbI3 QD (10 nm size) and PCBM electron acceptor onto the SnO2 layer. The integration of this blend in a solar cell based on ITO/SnO2/PCBM@CsPbI3-QD/CsPbI3-QD/PTB7/MoO3/Ag caused an efficiency improvement from 12% (without the blended active layer) to 15.1% (Fig. 12d).209 The reason behind the improvement is the coordinating effect of the ester groups present on the PCBM, allowing the passivation of Pb2+ defects on the surface. Moreover, the energy level of the small molecule provides an energy level cascade, helping the charge extraction to the ETM. Also, it should be considered that PCBM has an electron acceptor character, which can help the electron extraction from the NC layer to the SnO2. The same strategy was also demonstrated to be efficient in TiO2/PNC interface using non-fullerene acceptors (e.g., Y6 (Fig. 12e)).210 It was found that by blending CsPbI3 with Y6 a type-II energy level alignment is obtained, and the organic molecule passivates Pb2+ defects on the surface of the NC. PSC based on the architecture FTO/TiO2/blended active layer/thin QD layer/PTAA/MoO3/Ag yielded efficiencies as high as 15%, with main improvements observed in JSC and VOC, outperforming the 13.2% efficiency obtained for the device based only in CsPbI3 QD.
Post-passivation of the PNC layer has also been demonstrated to passivate defects. This strategy consists in modifying the surface of the PNC layer after its deposition, being efficient to improve the PLQY,243 photostability,244 and external quantum efficiency (EQE).245 In general, the mechanism of passivation conferred by this approach lies in the improved electronic coupling between the NCs and surface passivation conferred by surface-layer post-treatment.246 For example, cesium salts, such as cesium acetate, in EtOAc solution, can increase the PNC solar cell efficiency from 12.6% (control) to 14.1% when used as a post-passivation at the NC layer.247 This improved performance is due to a synergetic effect of both Cs-salt and EtOAc: the first fill Cs-vacancies on the surface caused during the ligand-exchange step, and the second helps in the additional remotion of capping ligands on the surface of the QD. Another strategy is to use hydrophobic conjugated molecules on the surface, such as phenylethylamine and its derivatives, which will replace some ligands on the NC surface.219,248,249 These molecules are capable of improving the charge carrier delocalization in the PNC, which in turn improves the charge separation efficiency.250 Also, AX salts (A = guanidinium, formamidinium) have been used in the post-passivation of the PNC layer.207,251 The efficiency improvement achieved by using these salts comes from an improved charge extraction, enhanced electronic coupling, and their ability to maintain the cubic-NC phase unaltered.
Regarding VOC, small losses have been observed for higher bandgap (>1.7 eV) materials, which benefits the achievement of high open-circuit voltages; the mechanism behind will be explained later on. The widening of the bandgap can be obtained, for instance, by alloying Br− with I− in the X-site of PNC (Fig. 13a) yielding VOC as high as 1.31 V, and delivering 5.3% efficiency.253 However, the inclusion of bromide usually increases the non-radiative recombination loss in the system. To circumvent this issue it was found that the concomitant incorporation of FA+ and Br− in CsPbI3 NC is beneficial to decrease photovoltage loss (Fig. 13b).215,254,255 Devices assembled with the architecture FTO/TiO2/FAxCs1−xPb(I1−xBrx)3 PNC/Spiro-OMeTAD/MoOx/Al yielded VOC as high as 1.25 V and 1.27 V, for Eg equals 1.82 eV and 2.03 eV, respectively, and maximum PCE of 5.3% and 2.5%, respectively.254 Additional advances were obtained for CsPbBr3 NC-based solar cells; this perovskite composition can be prepared with bandgaps in the range of 2.38 to 2.46 eV, depending on the synthesis temperature employed.256 For CsPbBr3 NCs, Cho et al.257 reported a solvent engineering strategy during the ligand exchange step consisting in mixing carboxylate solvents, such as EtOAc and butyl acetate (BuOAc), that resulted in an improved electronic coupling between the QDs. This approach yielded PNC solar cells with an efficiency of 4.2%, and an impressive 1.6 V of VOC. As a matter of comparison, the VOC of bulk CsPbI3 or CsPbI3−xBrx-based solar cells is typically <1.10 V. The possibility of obtaining wide bandgaps makes PNCs particularly interesting for application as the bottom layer in the perovskite-perovskite, or perovskite-silicon tandem solar cells.258 One of the origins of the high VOC in PNC solar cells is the formation of surface traps during the ligand exchange step.255 These traps increase the background free charge carrier in the QD, which maximizes the quasi-Fermi level splitting, decreasing VOC losses.255 However, it should be highlighted that the bandgap opening is a double-edged sword: if from one side it increases the maximum VOC delivered by the solar cell, it dislocated the material's absorption towards the high-energy region of the solar spectrum, causing consequent JSC loss.254
Fig. 13 (a) Variation of the PNC bandgap by alloying Br− and I− X-site,253 and (b) by alloying concomitantly Br− and I− in the X-site and FA+ and Cs+ in the A-site.254 (c) Spontaneous conversion of the photoactive to the non-photoactive phase of CsPbI3.269 (a) Reproduced with permission from ref. 253. Copyright 2018 American Chemical Society. (b) Reproduced with permission from ref. 254. Copyright 2019 American Chemical Society. (c) Reproduced with permission from ref. 269. Copyright 2019 American Chemical Society. |
Concerning phase stability, the use of NC with composition analog to perovskite bulk materials allows stabilization of given phases otherwise unstable. For instance, the photoactive (α-phase) and the non-photoactive (δ-phase) of 3D CsPbI3 are metastable at room temperature, which means that these two phases coexist and are interchangeable (Fig. 13c).254,259 The stabilization of the desired bulk α-CsPbI3 phase is only achieved at high temperatures (>150 °C),260–262 and this material is prone to oxygen and moisture degradation, which can be attributed to its low tolerance fact or (0.803).263 For that reason, it is quite hard to obtain long-term stability in bulk CsPbI3-based solar cells, and the exposure of the unencapsulated device to ambient air results in the irreversible formation of the yellow non-photoactive phase.264,265 On the other hand, PSCs prepared using CsPbI3 NCs (Eg ∼ 1.72 eV, excellent for top-cell in tandem solar cells), are not only more stable under ambient conditions compared to the bulk analog, but also allow stabilization of the photoactive phase at room temperature.266,267 A decrease in the NC size accompanied by an increase in the CsI content during the synthesis can help further stabilize the α- concerning the δ-phase.252 This improved stability is attributed to the high surface/volume ratio, which increases the contribution of the surface energy to the total Gibbs-free energy of the material.268 Since the surface is passivated in the NC, improved phase stability is expected compared to the bulk material.
In section 4.2 we also mentioned the growth of PNCs inside porous matrixes – an approach that used the porosity of material to physically restrict the growth. MAPbI3 NC was grown in a porous SiO2 matrix through the impregnation of a precursor solution in a thin film of SiO2 NP of 30 nm average size.212 This was further used as the active layer of a solar cell (FTO/TiO2/PNC embedded in SiO2 matrix/Spiro-OMeTAD/Au). The best efficiency was achieved when 45% of the SiO2 pores were filled (NC size equals 10 nm), yielding VOC, JSC, FF, and PCE equal to 1.02 V, 16.42 mA cm−2, 56.1%, and 9.3%, respectively, being stable for 5 h under illumination. It is important to note that using a porous template to grow the NCs can avoid the use of capping ligands, and save steps of purification and ligand exchange that are necessary for the proper operation of the conventional colloidal NC-based solar cells.
Some recent works demonstrated the fabrication of pin-type PNC solar cells. In 2019, Tavakoli et al.,272 used the inverted architecture based on ITO/PTAA/CsPbI3 NC/C60/BCP/graphene (Fig. 14a) to prepare solar cells with 6.8% efficiency. This work showed the chemical vapor deposition of graphene to act as a transparent electrode, being a possible alternative to expensive metals such as silver and gold, commonly used as electrodes, and offering the possibility to fabricate all-transparent solar cells. In 2020, Shivarudraiah et al.,207 used the device architecture FTO/NiOx/CsPbI3 NC/C60/ZnO/Ag (Fig. 14b) and obtained efficiencies up to 13%. They used NiOx as the HTM, which is, in principle, more stable than PTAA, and the electron transport layer was a combination of C60 and ZnO. The authors also introduced a ligand exchange step using FAI in MeOAc, instead of Pb(NO3)2/MeOAc, which they found to benefit the NC-NC electronic coupling and cause surface passivation. Recently, Mahato et al.218 verified that adding 5% of DMSO into PEDOT:PSS HTM can increase the contact potential difference (CPD) of the PNC film, benefiting the hole extraction. The effect of DMSO-doping of PEDOT:PSS layer on the solar cell was verified by assembling an inverted device with the architecture: ITO/PEDOT:PSS/CsPbI3 NC/C60/BCP/Al. By increasing the amount of DMSO from 2.5 to 5%, VOC improves from 0.908 to 0.978 V, and the PCE goes from 7.02 to 10.2%, respectively. The author attributed this improvement to the better interchain coupling between PEDOT units and better π–π stacking of PSS, which improves film conductivity.273
Fig. 14 Schematic diagram, cross-section scanning electron microscopy, and photography of the correspondent inverted pin-type PNC solar cells fabricated by (a) Tavakoli et al., in 2019,272 and (b) Shivarudraiah et al., in 2020.207 (a) Reproduced with permission from ref. 272. Copyright 2019 Wiley-VCH. (b) Reproduced with permission from ref. 207. Copyright 2020 American Chemical Society. |
While different deposition methods, including evaporation, have already been applied to produce perovskite layers in bulk PSC, only solution-based methods are compatible with PNC solar cells. These methods include spin coating, blade coating, slot-die coating, bar coating, spray coating, inkjet printing, and screen printing. Nevertheless, spin coating is not the ideal method for PSC scaling up because of its incompatibility with roll-to-roll systems. Blade coating, slot-die coating, bar coating, and screen printing are suitable for PNC deposition and they can be described as a tool that spreads the solution onto the substrate while removing the excess. All these methods were already explored in bulk-perovskite-based PSM,16 however, other PSM techniques like spray coating and inkjet printing, drive a better prospectus to be extended to PNC solar cells.
The upscaling of photovoltaic devices based on perovskite materials is relatively new looking back in the timeline of perovskite technology, as demonstrated in Fig. 15. Large area (>1 cm2) bulk-perovskite-based cells were first fabricated in 2015,16 the same year that first monodisperse PNC was prepared by Protesescu et al.2 Only one year later, a PNC solar cell was reported by Swarnkar et al.,26 while PSM has been already reported in the literature for bulk PSC.16 Since then, PNC in large-area devices was only reported using PNC as an interlayer for bulk-based PSM to improve photovoltaic (PV) parameters, but not as a light-harvesting layer.17,23 To our knowledge, PNC solar cells have only been fabricated in small areas as lab cells and probed some techniques that might be compatible with large-area deposition.18 However, it is interesting to note that advances in fabrication processes from bulk PSC and quantum dot solar cells (QDSC), especially with PbS and PbSe QD, are expected to benefit the upscaling of PNC solar cells.
Although PNC has not been applied over a large area as an active layer, it has been successfully applied as an interlayer between the bulk-based perovskite film and the HTM, as introduced before. Large-area devices in a nip-type architecture have been demonstrated by Cheng et al.17 and Mali et al.,23 showing that PNC plays a significant role in the surface passivation of bulk perovskite films.280–283 Cheng et al.17 reported a PSM with (6 × 6) cm2 dimensions comprised of eight separated cells with an active area of 18.0 cm2 (Fig. 16a, b, and c). They used spin-coated perovskite QDs as interlayers in the device architecture FTO/c-TiO2/mp-TiO2/MAPbI3/PNC/o-HTM/Au (o-HTM is P3HT, PTAA or Spiro-OMeTAD). The device was encapsulated and stored for 24 h to improve the sealing effect. Different PNC compositions were optimized firstly in small devices, including CsPbI2Br, CsPbI1.85Br1.15, CsPbI1.5Br1.5, CsPbI1.15Br1.85, and CsPbIBr2, with a maximum PCE of 21.1% obtained with CsPbI1.85Br1.15 NC as an interlayer between bulk-MAPbI3 film and dopant-free-P3HT.17 All PNCs capped by OLA and OA in octane solution were synthesized according to the one-pot hot-injection method in α-phase, uniform in size, nearly cubic shape, and demonstrated stability in ambient air with 30–40% relative humidity, maintaining 46.1% of PLQY after five months. Using the optimized conditions obtained for the small-area devices (0.12 cm2), an 18.0 cm2 lab module connected in series with the architecture FTO/c-TiO2/mp-TiO2/MAPbI3/CsPbI1.85Br1.15/P3HT/Au was constructed, yielding an efficiency of 17.6% using P3HT as o-HTM (Fig. 16c). It is worth mentioning that P3HT is preferable in front of Spiro-OMeTAD and PTAA, because of its lower cost and improved oxygen and moisture stability.
Fig. 16 (a) Schematic device architecture and (b) the cross-section SEM images of P3HT-based PSC with PNC interlayer. (c) J–V curve showing the PCE of dopant-free P3HT-based PSMs with and without a PNC interlayer at an active area of 18.0 cm2. (d) Stable output measurement at maximum power point under constant 1-sun illumination for the P3HT-based PSMs with and without a PNC interlayer.17 (a–d) Reproduced with permission from ref. 17. Copyright 2020 American Chemical Society. |
PNCs were able to passivate the bulk-perovskite surface, promoting hole extraction from MAPbI3 to HTM by forming a cascade of the energy levels (valence band alignment) and regulating polymer/molecule orientation for improved hole mobility without dopants in the HTM layer. This successful outcome of the PNCs multifunctional interlayer was probed by atomic force microscopy (AFM) images and grazing incidence X-ray diffraction (GIXRD).17 It was found that the P3HT film deposited onto the perovskite/QD layer has a smooth morphology, a decreased pinhole density, and improved crystallinity. This translates into a boosted efficiency and stability for more than 10000 h under constant 1-sun illumination (Fig. 16d). As a matter of comparison, the module without the surface treatment with PNC showed an initial PCE equal to 11.6%, with an expressive efficiency drop in the first 2000 h over 1-sun illumination. The presence of PNC is believed to reduce tail and trap states, promoting an increase of VOC and FF, leading to an enhancement of more than 50% in PCE. Compared with single cells (active area = 0.12 cm2), PSM has preserved more than 80% of PCE. By changing P3HT to PTAA and Spiro-OMeTAD, PCE of 15.9 and 16.0% was obtained in the modules, respectively. Also, the modules with PTAA and Spiro-OMeTAD maintained 80% of PCE when compared to small area devices (0.12 cm2).17
In another work, Mali et al.23 used a similar strategy with CsPb0.95Tb0.05I2Br QDs to stabilize the bulk and surface of CsPbI2Br film. In the bulk passivation, PNC was spin-coated onto the as-deposited CsPbI2Br film before thermal annealing, causing the mixing between the PNC and the γ-CsPbI2Br phase during its formation. Then, a surface treatment with CsPb0.95Tb0.05I2Br QDs was realized by spin-coating a solution of this material onto the CsPbI2Br-CsPb0.95Tb0.05I2Br film after the annealing. Subsequently, this film was quickly dipped in ethyl acetate. Un-encapsulated PSM with the architecture FTO/c-TiO2/mp-TiO2/CsPbI2Br + CsPb0.95Tb0.05I2Br/CsPb0.95Tb0.05I2Br/P3HT/Au and an active area of 19.80 cm2 was constructed with series connection, reaching a PCE of 10.94%. Moreover, 80% of its initial efficiency was kept over a 30 days stability test (720 h) under continuous illumination (100 mW cm−2) in the open atmosphere. The improved performance is because the PNC with Tb3+ is able to improve charge carrier lifetime and mobility under ambient conditions. Density functional theory (DFT) calculations suggested that Tb3+ doping benefits CsPbI2Br stability by lowering the binding energy. Also, the Tb d-orbital contributes to the conduction band edge of the material, shifting the Fermi level closer to the CB and introducing donor states at the bottom of the CB.23 It corroborates the assumption that PNCs as an interlayer are able to improve charge mobility either through cascade energy levels17 or by tuning the Fermi level.23
Spray-coated technology has been shown as an important deposition method for large areas in PSC. Yuan et al.21 reported the spray-coated deposition of monodisperse CsPbI3 PNC films (average size of 10 nm, synthesized by the conventional hot-injection method). The long-chain ligands from synthesis were partially removed using phenyltrimethylammonium bromide (PTABr) previously. The ligand exchange reaction was conducted by adding PTABr powder to the PNC in the hexane solution under stirring, then the residual PTABr was removed by centrifugation. The authors demonstrated an automated spray-coating process in the air (Fig. 17) where two nitrogen pipelines are responsible for atomizing and spraying PNC dispersed in octane, respectively. A uniform and pinhole-free PNC film on a (10 × 10) cm2 substrate was obtained and the homogeneity was confirmed by overlapping UV-vis absorption spectra obtained for different regions of the film. The authors presented a proof-of-concept that spray-coating can cover large areas with a PNC solution, removing insulating ligands, passivating defects using Pb(NO3)2 in methyl acetate solution, and aiding in solvent drying of QD films.21
Fig. 17 Process of an automated spray-coating deposition developed by Yuan et al.21 Adapted with permission from ref. 21. Copyright 2019 Wiley-VCH. |
An extension of the automated spray-coating process to an all-sprayed process, in which all the layers are deposited from spray-based techniques, is expected to speed up PSM to industrial scale-up. A proposal that addresses this approach has been idealized in modules for bulk-based PSC, as proposed by Lee et al.;22 this technology can be transplanted to PNC solar cells. The authors demonstrated a PCE of 10.08% for an all-sprayed PSM with the architecture FTO/c-TiO2/CsPbI2Br/MoO2-PTAA/Carbon black, and an active area of 25 cm2 (the small-area cells, with 0.12 cm2, showed PCE of 14.25%). Spray-coating technology as much as inkjet printing seems to be a very interesting way to scale up PNC solar cells.
Additionally, reports about the long-term stability of PNC solar cells are scarce, and very little is known about the intrinsic mechanism that leads to device degradation, which includes not only the PNC but also the other device's layers. This degradation can be triggered by both external and internal factors. As external we can mention moisture, oxygen, and light exposure. As intrinsic, the loss of capping ligands, aggregation, and phase conversion plays an important role. Despite knowing what causes the PNC solar cell degradation, we do not know how the device interfaces behave in front of these extrinsic and intrinsic factors. Knowing about the interface dynamics may allow the discovery of methods to retard or prevent the efficiency loss of the device. Further, there is no consensus about a standard protocol for evaluating the degradation of PNC solar cells, as we have for organic284 and bulk-based perovskite solar cells.285 The existence of a standard protocol can facilitate the literature data analysis, providing a way to compare how different approaches can impact the shelf life of the PNC solar cell. Along with it, it is imperative to probe the power conversion efficiency of the device over time by tracking the maximum power point, allowing identify the stabilized efficiency of the device.
Furthermore, there is room for improvement in the device's performance and fabrication process. Using short-chain organic molecules during the ligand exchange step endowed by different functional groups seems to be an alternative to improve efficiency. In this context, the design of new molecules plays an important role. By identifying the type of defect present on the NC surface, by means of DFT, for instance, and using Lewis acid–base chemistry it is possible to design and synthesize appropriate passivating molecules. This approach has been used for bulk-based perovskite solar cells, demonstrating meaningful results.286–288
In addition, new strategies for PNC solar cell assembly are necessary. Firstly, it is urgent to find substitutes for the commonly used HTM for nip-type devices: Spiro-OMeTAD and PTAA. The cost of these materials is very high, and not compatible with industrial manufacturing. In addition, Spiro-OMeTAD requires dopants agents (e.g. tert-butylpyridine and lithium bis(trifluoromethanesulfonyl)imide) to be properly oxidized and perform well as an HTM. However, these dopants can accelerate the degradation due to their hydrophobic nature, and compromise the long-term stability of the device. Replacing those HTMs with cheaper ones, such as P3HT and polyaniline, is necessary for the manufacturing viability of any perovskite-based technology. However, P3HT and polyaniline and their derivatives have been less applied in PNC-based solar cells. Furthermore, thinking about scalability, the complexity introduced by the deposition of subsequent PNC layers and the ligand exchange reaction, can enhance the fabrication cost when compared to bulk-based perovskite solar cells, in which no ligand-exchange process is required. As discussed, in situ methods to grow the NCs directly into the conductive substrate can be an alternative, but this approach is in a very early development stage. Moreover, production costs have been cited as a challenging step in the industrialization of this technology. In this sense, research focusing on improving the synthesis yield, solvent recycling, and the use of environmentally friendly and less expansive reactants225 are encouraged to push forward the upscaling of PNC solar cells. Cost studies related to higher production volumes are believed to be more cost-effective, but the impact of lower-grade solvents and precursors is expected to help the attractiveness of this technology. Furthermore, efforts are needed in the development of lead-free PNCs-based solar cells. Recent advances are obtained for the replacement of lead with tin, germanium, bismuth, and antimony; however, the efficiency is far behind the Pb-based-PNC solar cells.289 Beyond PNC, other lead-free materials have been investigated, for instance, silver bismuth sulfide (AgBiS2) demonstrating a record efficiency of 9.17%.290
As can be seen in this Review, there is a long way toward the commercialization of PNC solar cells, and several questions about the fundamental properties of the material and the device are still open. We hope that the discussions brought here encourage the community to find suitable solutions for the up-scaling of PNC-based solar cells.
This journal is © The Royal Society of Chemistry 2023 |