Design rules for the preparation of low-cost hole transporting materials for perovskite solar cells with moisture barrier properties

Michiel L. Petrus *a, Arif Music a, Anna C. Closs a, Johan C. Bijleveld b, Maximilian T. Sirtl a, Yinghong Hu a, Theo J. Dingemans bc, Thomas Bein a and Pablo Docampo *ad
aDepartment of Chemistry, Center for NanoScience (CeNS), University of Munich (LMU), Butenandtstr. 11, 81377 Munich, Germany. E-mail: michiel.petrus@lmu.de
bDelft University of Technology, Faculty of Aerospace Engineering, Kluyverweg 1, 2629 HS Delft, The Netherlands
cUniversity of North Carolina, Department of Applied Physical Sciences, 1113 Murray Hall, Chapel Hill, NC 27599-3050, USA
dNewcastle University, School of Electrical and Electronic Engineering, NE1 7RU Newcastle upon Tyne, UK. E-mail: pablo.docampo@newcastle.ac.uk

Received 23rd July 2017 , Accepted 10th November 2017

First published on 11th November 2017


The current state-of-the-art hole transporting materials (HTM) for perovskite solar cells are generally synthesized via cross-coupling reactions that require expensive catalysts, inert reaction conditions and extensive product purification, resulting in high costs and therefore limiting large-scale commercialisation. Here we describe a series of HTMs prepared via simple and clean Schiff-base condensation chemistry with an estimated material cost in the range of 4–54 $ per g. The optoelectronic and thermal properties of the materials are linked to the changes in the chemical structure of the HTMs, which allow us to extract design rules for new materials, supported by density functional theory calculations. Charge transport measurements show hole mobilities in the range of 10−5 to 10−7 cm2 V−1 s−1. Upon addition of LiTFSI the HTMs can be oxidized, resulting in a large increase in the conductivity of the hole transporting layer (HTL). When employed as HTL in perovskite solar cells, power conversion efficiencies close to those of spiro-OMeTAD are obtained. In particular, devices prepared with Diazo-OMeTPA show a higher open-circuit voltage. Furthermore, we show that azomethine-based HTMs can act as effective moisture barriers, resulting in a significant increase in the stability of the underlying perovskite film. We assign the improved properties to the presence of a dipole in our molecules which promotes a close molecular packing and thus leads to a high density of the as-formed HTM films, preventing the ingress of water. This work shows that HTMs prepared via condensation chemistry are not only a low-cost alternative to spiro-OMeTAD, but also act as a functional barrier against moisture-induced degradation in perovskite solar cells.


Introduction

In the last decades, a large variety of conjugated organic molecules with hole-transporting properties have been synthesized for diverse optoelectronic applications such as organic light emitting diodes (OLED), organic field effect transistors (OFET) and organic, dye-sensitized and perovskite solar cells (OPV, DSSC and PKSC). Among these, 2,2′,7,7′-tetrakis(N,N-di-p-methoxyphenylamine)-9,9′-spirobifluorene (spiro-OMeTAD) is one of the most well-known hole-transporting materials (HTM).1 This small-molecule was initially developed for solid-state dye-sensitized solar cells, where it was used as a replacement for liquid electrolytes. More recently, spiro-OMeTAD has been the preferred HTM for perovskite-based solar cells as a result of its good energy level alignment with the valence band of methylammonium lead iodide (CH3NH3PbI3) and related perovskite structures, leading to excellent photovoltaic performance.2–4

However, spiro-OMeTAD is very expensive to produce as a result of its multistep synthesis, which requires inert reaction conditions, transition metal catalysts and extensive purification.5 The high materials cost in turn results in a significant cost contribution in perovskite solar cells,6 making this HTM industrially less interesting. Recently, several research groups have focussed on tuning the structure to reduce the cost,7,8 without losing the photovoltaic performance in perovskite solar cells.2,9–13 The majority of these materials contain methoxy-substituted triphenylamine side groups, which are connected via aryl–aryl bonds to a conjugated core.9,10,14 These materials are generally synthesized using transition metal catalysed reactions that require extensive product purification.15–20 This is due to the fact that these reactions are prone to produce side-products and leave (metal) catalyst residues that can affect the performance of the resulting devices.21 In our previous work, we have shown that condensation chemistry offers an excellent alternative by connecting the core moiety to the triphenylamine via an azomethine-bond (–CH[double bond, length as m-dash]N–, also known as imine or Schiff-base).5 More recently, other groups have also shown that condensation chemistry offers an excellent alternative to produce low-cost hole transporting materials.22,23 This chemistry has the advantage that it can be performed under ambient conditions without the use of expensive catalysts and with water as the only side-product. The absence of metal catalysts and side reactions results in a simple product workup, if necessary at all.24 Although the azomethine bond in aliphatic or semi-aromatic compounds is prone to hydrolysis, studies have shown that in highly conjugated (all-aromatic) compounds the azomethine bond is robust and resists hydrolysis.25,26 Furthermore, Koole et al. recently showed that the azomethine bond exhibits good electrical conductance, which is expected to be beneficial for charge transport and shows the potential of this chemistry to produce high-quality organic semiconductors.27

In this work we extend our study of azomethine-based HTMs and correlate the effect of the changes to the chemical structure on the material properties. We emphasize that the simple and clean synthesis allows us to prepare a large series of materials, thus enabling the study of a variety of different cores. We find that all these materials can be synthesized in good yields and at low cost. Our computational studies indicate that the energy levels of the small-molecules can be fine-tuned by changing the core, which is in good agreement with our experimental results. In order to investigate the role of the azomethine-linkage, we also synthesized two reference molecules where the azomethine bond was inverted. We observe that the photovoltaic performance is strongly influenced by the core of the HTMs and that the best values are obtained for HTMs that have a strong dipole within the core. The reversed orientation of the azomethine bond slightly lowers the HOMO energy levels, which also influences the energy alignment with the perovskite. Furthermore, we performed in situ X-ray diffraction measurements to study moisture-induced degradation of a perovskite film covered by a thin layer of these HTMs, which shows the superior function of these materials as a barrier against moisture-induced degradation of the perovskite.

Synthesis

For this study, we designed and synthesized a series of azomethine-based HTMs based on the previously reported EDOT-OMeTPA (Fig. 1 and 2). We replaced the EDOT core with different (aromatic) groups in order to study their influence on the optoelectronic and charge transport properties. For example, the core was substituted by a phenyl ring in order to further reduce the cost and simplify the molecular structure (Ph-OMeTPA). We also synthesized the thiophene (Th-OMeTPA) analogue in order to study the effect of the cyclic ether in EDOT, while the use of furan (Fu-OMeTPA) results in a more sustainable alternative, allowing us to study the effect of the hetero atom.28 By introducing a triphenylamine (TPA-OMeTPA) as the core, an azomethine-based small-molecule analogue of the conductive polymer PTAA (polytriarylamine) was synthesized. All these materials were synthesized in a low-cost condensation reaction between 4-amino-4′,4′′-dimethoxytriphenylamine and the dialdehyde analogue of the different cores (Fig. 1).
image file: c7ta06452g-f1.tif
Fig. 1 Reaction scheme of the Schiff-base condensation reaction to prepare the isomers.

image file: c7ta06452g-f2.tif
Fig. 2 Chemical structure of the different azomethine-based small-molecules, including the isolated yields and estimated materials cost.

Furthermore, we inverted the azomethine bond in order to study the influence of the position of the nitrogen in the π-bridged linker moiety. This was done by coupling 1,4-phenylenediamine with 4-formyl-4′,4′′-dimethoxytriphenylamine to synthesize the analogue of Ph-OMeTPA, which was termed Ph-inv-OMeTPA. By reacting 4-formyl-4′,4′′-dimethoxytriphenylamine with hydrazine, a simplified analogue with a reversed bond sequence was synthesized that does not contain any core, but just a double azomethine linkage (Diazo-OMeTPA). This material has already been published by Ma et al., where the authors reported a relatively poor photovoltaic performance compared to spiro-OMeTAD.29 The authors attributed this to the relatively low mobility of the material and the resulting high series resistance.

All these materials were synthesized in high yields (>80%) and purified with a simple washing step. Fu-OMeTPA has a relatively high solubility and is soluble in isopropanol, it is therefore important to use stoichiometric amount of starting materials, simplifying the purification.

Cost estimation and environmental impact

Since material cost will have a major influence on the commercial viability of these materials as HTMs in perovskite solar cells, a cost estimation was performed following our previously published method.5 The cost was calculated by quantifying material costs of their published small-scale synthetic routes (more details in the ESI). The estimated material costs for the azomethine-based small-molecules are in the range of 4–54 $ per g (Fig. 2). The lowest costs are obtained for the small-molecules prepared from 4-amino-4′,4′′-dimethoxytriphenylamine, as this starting material can be synthesized at an estimated material cost of only $2.28 per g (Scheme S2 and Table S4). In contrast, the HTMs with the inverted azomethine-bond prepared from 4-formyl-4′,4′′-dimethoxytriphenylamine show a significantly higher material cost, which is the result of the less efficient synthetic pathways available for this aldehyde. However, we believe that by optimisation of the synthesis this starting material can also be prepared at a comparable cost.

As expected, the Ph-OMeTPA shows the lowest material cost at $4 per g as a result of the very low cost of the core. It is worth noting that all other HTMs described here also show a significantly lower material cost than the state-of-the-art HTM spiro-OMeTAD (estimated 91 $ per g),5 which stems from the straightforward chemistry and purification of the azomethine-based materials. Additionally, the high yields and simple purification leads to the use of fewer chemicals, which is also beneficial from an environmental point of view (Table S2). For a better understanding of the environmental impact of a synthetic route, Sheldon introduced the term E factor, which is defined as kg waste per kg product.30 Based on our small-scale synthesis route, our materials with the standard azomethine bond show values of E = 400–700, while the materials with an inverted azomethine bond reach E = 1800–2800 (Table S2). In comparison, spiro-OMeTAD has an E factor of 3600, which gives a clear indication of the environmental impact of the synthesis and the benefit of using simple condensation chemistry.

As a result of the good film-forming properties of the azomethine-based HTMs, relatively thin layers of the HTM can be used in devices (50 nm compared to 250 nm for spiro-OMeTAD), which additionally reduces the cost contribution and environmental impact by a factor of five for a typical device. In this way, the material cost of the hole transporter becomes negligible compared to the total device cost.6

Thermal properties

Materials with low melting points or glass transition temperatures are expected to have poor stability when used in photovoltaic devices,31,32 which can reach temperatures of 80 °C under operational conditions. The thermal properties of our new HTM materials were studied by thermogravimetric analysis (TGA, Fig. S2) and differential scanning calorimetry (DSC, Table 1 and Fig. S3). All materials possess excellent thermal stability with degradation temperatures under dynamic conditions exceeding 325 °C, which is well above photovoltaic operation temperatures. The high degradation temperatures are attributed to the conjugated backbone of all these molecules. The reported degradation temperature for Diazo-OMeTPA (Td10% = 320 °C) is in good agreement with our observations.29
Table 1 Thermal properties of the azomethine-based HTMs obtained from DSC, melting point apparatus and TGA measurements
Compound T g (°C) T m (°C) T d 5% (°C)
EDOT-OMeTPA 105 241 359
Ph-OMeTPA 90 145 390
Th-OMeTPA 100 117 371
Fu-OMeTPA 70 106 325
TPA-OMeTPA 89 127 397
Ph-inv-OMeTPA 84 106 375
Diazo-OMeTPA 120 208 340


After the synthesis, most materials are in an amorphous state and only EDOT-OMeTPA and Ph-OMeTPA show a melting transition with DSC. The phenyl core in Ph-OMeTPA slightly reduces the glass transition temperature (Tg) and melting point (Tm) compared to EDOT-OMeTPA, while inverting the azomethine bond further reduces the Tg and Tm. The lower Tm for furan compared to thiophene is in agreement with literature and results from the smaller heteroatom.33–35 Both melting points are significantly lower than that of EDOT-OMeTPA, which is ascribed to the large dipole moment in EDOT-OMeTPA. TPA-OMeTPA also shows a relatively low melting point, which originates from the bulky triphenylamine core. In contrast, Diazo-OMeTPA shows a high Tg and Tm which is ascribed to the rigid and planar backbone.36 Comparing the thermal properties of these azomethine-based HTMs with the reported thermal data of their fully conjugated analogues, we find that the thermal properties for our materials are superior. More specifically, the Tg of EDOT-OMeTPA is 32 °C higher than that of its analogue H101; the Tg and Td of Th-OMeTPA are respectively 32 and 88 °C higher than those of its analogue Th101; and the Tm of Diazo-OMeTPA is 28 °C higher than that of its fully conjugated analogue, demonstrating the excellent thermal stability of the azomethine-based HTMs.10,37,38

Optoelectronic properties

We evaluated the absorption and transmittance of the HTMs, as a low absorption in the hole transporting layer is important in order to minimize the optical losses. Light absorption spectra of the different materials were recorded in solution and in the solid state. The samples were dissolved in chlorobenzene at a concentration of 20 mg L−1 and measured in a wavelength range of 350 to 800 nm (Table S12, Fig. 3 and S4). All materials show a clear absorption maximum in the measured range and upon increasing the electron donating properties of the core, a redshift was observed as expected from theory.39,40
image file: c7ta06452g-f3.tif
Fig. 3 Normalized UV-vis absorption spectra of the azomethine-based HTMs. The materials were dissolved in chlorobenzene at a concentration of 20 mg mL−1.

The absorption spectra of the materials in the solid state showed small bathochromic shifts of 4–10 nm compared to the solution spectra (Table S12), which we assign to molecular packing in the solid state. The band gaps (Eg) of the HTMs were estimated from the onset of the absorption spectrum of each of the films. The prepared HTM films are only around 50 nm thick, increasing the transmittance of the films compared to films of spiro-OMeTAD, which is generally around 250 nm thick. In the spectral range of 350 to 800 nm, films prepared from our materials show a transmittance of more than 50% at their absorption maximum, while spiro-OMeTAD absorbs more than 90% of the light at its absorption maximum (Fig. S4b). The high transmittance in the visible range makes these materials interesting, especially for tandem applications,41 as it would reduce the parasitic absorption. However, as our materials have a narrower bandgap than spiro-OMeTAD, their absorption maxima appear at higher wavelengths, and therefore increasing the band gap of our HTMs would be desirable to further reduce optical losses.

Inverting the azomethine bond significantly increases the bandgap of the material, as is clear from the comparison of Ph-OMeTPA with Ph-inv-OMeTPA (Fig. 3). Similar observations have been reported by Sek et al. and Hindson et al.40,42 In contrast, the introduction of the heteroatom into the core decreases the bandgap, as is clearly shown for EDOT-OMeTPA and Th-OMeTPA. The smaller bandgap of these materials is consistent with NMR spectroscopic results, where a down-field shift of the azomethine proton is observed, indicating a higher degree of conjugation (Table S1).42 This hypothesis is also supported by FTIR measurements, which show that the azomethine stretching vibrations are shifted to lower wavenumbers (Table S1).42 Our results demonstrate that by varying the core and inverting the azomethine bond, the bandgap of the HTM can be easily and substantially tuned.

A good alignment of the HOMO energy level of the HTM with the valance band of the perovskite is essential in order to obtain high open-circuit voltages in photovoltaic devices.43 In order to study the energy level alignment, the oxidation potentials of the HTMs were measured using cyclic voltammetry (Fig. 4, S6 and Table 2). Very comparable oxidation potentials were found for all materials, resulting in a small spread in HOMO energy levels. These energy levels are dominated by the electron-donating part of the molecules, which in this series of HTMs is always the triphenylamine side group, hence explaining the similar HOMO energy levels.44 Inverting the azomethine bond slightly lowers the HOMO energy level since the electronegative nitrogen in the azomethine-bond is not directly connected to the triphenylamine moiety. Between the two phenyl-based HTMs (Ph-OMeTPA and Ph-inv-OMeTPA), we observe a difference of 90 meV as a result of inverting the azomethine-bond. The HOMO energy levels are close to the valance band of CH3NH3PbI3 (5.43 eV), which is expected to minimize the energy losses in photovoltaic devices and should enable high open-circuit voltages.


image file: c7ta06452g-f4.tif
Fig. 4 Electrostatic surface potential, Frontier molecular orbital distributions and energies. Kohn–Sham HOMO and LUMO at B3LYP 6-31G(d,p) level and the calculated HOMO energy level following the procedure of Chi et al. (red dotted line).45 Grey box and black lines indicate the experimentally determined energy levels and bandgap. The blue line indicates the valence and conduction band of CH3NH3PbI3.
Table 2 HOMO energy levels obtained from DFT calculations and cyclic voltammetry
Compound HOMOCVa (eV) HOMOcalcb (eV) HOMODFT,vacuum (eV) HOMODFT,DCM (eV)
a HOMO energy level experimentally obtained from cyclic voltammetry measurements. DFT HOMO energy levels from B3LYP/6-31G(d,p) level in vacuum and DCM. b Calculated following the procedure by Chi et al.45 where HOMOcalc = HOMODFT,DCM + 0.624 eV.
EDOT-OMeTPA −5.28 −5.24 −4.37 −4.62
Ph-OMeTPA −5.30 −5.45 −4.51 −4.83
Th-OMeTPA −5.30 −5.31 −4.47 −4.68
Fu-OMeTPA −5.30 −5.29 −4.44 −4.66
TPA-OMeTPA −5.28 −5.28 −4.44 −4.65
Ph-inv-OMeTPA −5.39 −5.45 −4.59 −4.83
Diazo-OMeTPA −5.36 −5.36 −4.49 −4.74


The LUMO energy levels were estimated by adding the optical bandgap to the HOMO energy levels; they mainly depend on the core of the small-molecules, resulting in a wide range of energies for these levels.44 In all cases, the LUMO levels are higher than the conduction band of CH3NH3PbI3, making them good electron blocking materials.

Computational study

To gain insights into the geometric and electronic structure of the different small-molecules, density functional theory (DFT) geometry optimisations (B3LYP/6-31G(d,p) level) were performed in vacuum and in dichloromethane as the solvent, by means of the conductor-like polarizable continuum model (CPCM) as implemented in the Gaussian 09 program package. Except for TPA-OMeTPA, the HOMO is distributed over the entire molecule (Fig. 4). For TPA-OMeTPA we observed that the HOMO is more localized on the terminal triphenylamines, which we ascribe to the disruption of the conjugation of the central triphenylamine.46,47 The disturbance of the conjugation in the triphenylamine-based azomethines was experimentally observed by Skene and co-workers in a previous study.47

HOMO energy levels obtained from computational chemistry for single molecules in vacuum generally strongly differ from the experimentally results. Chi et al. showed that it is possible to get a good estimate of the experimental HOMO energy levels from the HOMO energy levels obtained from DFT calculations (B3LYP/6-31G(d,p) in dichloromethane)45 using the following formula:

HOMOcalc = HOMODFT,DCM + 0.624 eV

In our case, applying this simple correction resulted in calculated HOMO energy levels that are in excellent agreement with the experimental values and shows that this method is a valuable tool for designing novel azomethine-based HTMs (Fig. 4 and Table 2).

The calculated electrostatic potential map is depicted in Fig. 4. We observe for all small-molecules that the most electronegative part is near the nitrogen of the azomethine bond, while the more electropositive part is much more spread out over the molecule, except for EDOT-OMeTPA, where the cyclic ether in the core shows to be the most electropositive. This leads to a strong dipole in the core of EDOT-OMeTPA. Similarly, we also observe a clear directional dipole in Th-OMeTPA, and Fu-OMeTPA. The strong dipole in the core could facilitate close molecular packing in the solid, which is expected to enhance the intermolecular charge transport. Despite the anticipated close molecular packing, thin films prepared via spincoating are amorphous according to thin film XRD measurements.

Charge transport properties

Besides good energy level alignment, the charge transport properties of the HTMs are important in order to obtain an efficient hole-transporting layer. We studied the hole mobility by preparing hole-only devices using the pristine materials. The devices were built using a common device architecture: indium-doped tin oxide (ITO)/MoOx/HTM/MoOx/Au.39,48,49 The HTM layer was prepared by spincoating the small-molecules from chlorobenzene at a concentration of 10 mg mL−1. JV curves were recorded in the dark, with the current assumed to be space-charge limited at higher voltages. From this, the charge carrier mobility can be estimated using the Mott–Gurney equation (more details in the ESI).50 The two molecules with the inverted azomethine-bond, Diazo-OMeTPA and Ph-inv-OMeTPA showed a good fit to this equation with the extracted mobility value for Diazo-OMeTPA in good agreement with the one reported by Ma et al.29 The other small-molecules exhibited significant field dependence and the mobilities were extracted using a Poole–Frenkel-type field dependence.51 The zero-field mobility (μ0), field coefficient (γ) and the mobility at a field of 5 × 105 V cm−1 (μ(E)) are listed in Table 3. We note that the orientation of the azomethine bond seems to have a strong influence in cases where the mobility shows a field dependence. Fu-OMeTPA showed relatively low mobilities on the order of 10−7 cm2 V−1 s−1, which is significantly lower than its thiophene analogue. This is believed to be the result of the enhanced diene character of the furan compared to the more aromatic character of the thiophene, and is in agreement with a systematic study done by Bijleveld et al.52 The other materials show significantly higher hole mobilities, on the order of 10−5 to 10−6 cm2 V−1 s−1, which are among the highest reported mobilities for azomethine-based small-molecules (Fig. S7 and S8).39,53–55 These mobilities are comparable to values extracted from spiro-OMeTAD.56
Table 3 Hole mobility of the pristine and the conductivity for the pristine and doped HTMs
Compound μ 0 (cm2 V−1 s−1) γ (m1/2 V−1/2) μ (cm2 V−1 s−1) σ pristine (S cm−1) σ doped (S cm−1)
a HTM solutions were doped with 30 μL mL−1 of a 170 mg mL−1 LiTFSI solution in acetonitrile before spincoating, and the films were allowed to oxidize overnight.
EDOT-OMeTPA 1 × 10−10 3 × 10−4 5 × 10−6 6 × 10−8 2 × 10−6
Ph-OMeTPA 5 × 10−11 4 × 10−4 4 × 10−6 1 × 10−8 5 × 10−6
Th-OMeTPA 3 × 10−10 3 × 10−4 1 × 10−5 2 × 10−8 2 × 10−6
Fu-OMeTPA 3 × 10−12 5 × 10−4 5 × 10−7 2 × 10−8 1 × 10−6
TPA-OMeTPA 2 × 10−11 4 × 10−4 1 × 10−6 1 × 10−8 1 × 10−6
Ph-inv-OMeTPA 6 × 10−6 7 × 10−9 1 × 10−7
Diazo-OMeTPA 1 × 10−6 2 × 10−8 3 × 10−6
Spiro-OMeTAD 4 × 10−5 9 × 10−8 3 × 10−5


In order to increase the charge transport in the hole-transporting layer, the HTM is generally oxidized using LiTFSI. Abate et al. showed the influence of the addition of LiTFSI to spiro-OMeTAD on the conductivity (σ).57 In our previous work we showed that the addition of LiTFSI is necessary to obtain efficient devices when EDOT-OMeTPA is used.5 Although there are several studies regarding electro- and chemical oxidation of azomethine-based dyes,58–61 the effect of the oxidation by LiTFSI on azomethine-based HTMs and the effect on the conductivity have not been reported.

The oxidation of the small-molecules upon addition of LiTFSI was studied by light absorption measurements, showing an additional absorption band for the oxidized species (ca. 550 nm for Diazo-OMeTPA). The fraction of the oxidized species increases upon addition of LiTFSI, while we observe bleaching of the signature of the neutral species as expected. The absorbance of the oxidized species also depends on the HTM and especially for Ph-inv-OMeTPA, a strong increase of the oxidized species was observed (Fig. 5 and S5).


image file: c7ta06452g-f5.tif
Fig. 5 The effect of doping with LiTFSI on the optical properties of Diazo-OMeTPA (left) and conductivities of the HTMs as a function of LiTFSI addition (right). The solid grey bar was added as a guide to the eye.

The impact of the presence of the oxidized species on the conductivity was measured as a function of the amount of added LiTFSI. The pristine materials show a low conductivity on the order of 10−8 S cm−1, which is comparable to the conductivity of pristine spiro-OMeTAD (Fig. 5).

An increase of around two orders of magnitude in the conductivity was observed for all materials upon adding 30 μL mL−1 of a 170 mg mL−1 LiTFSI solution in acetonitrile to the HTM casting solutions. This demonstrates that LiTFSI can be used as an efficient oxidizer for azomethine-based HTMs, resulting in conductivities that are among the highest reported for azomethine-based semiconductors.62–66 However, in contrast to spiro-OMeTAD, where a maximum in conductivity was reported for around 20 mol% doping,57 the azomethine-based HTMs require significantly higher amounts of LiTFSI to reach comparably high conductivities. Since the azomethine-based HTMs do not show a maximum in conductivity in the range from 0 to 250 mol%, the conductivity can be increased upon addition of LiTFSI and generally exceeds the conductivity of spiro-OMeTAD when approximately 100 mol% of LiTFSI is added to the azomethine-based HTMs. The significant difference in oxidation behaviour of the azomethine-based HTMs compared to spiro-OMeTAD indicates that for new classes of hole transporters, it is crucial to optimize the additives and their concentration in order to obtain a comparable performance to the state-of-the-art materials.

Photovoltaics properties

The photovoltaic performance of perovskite solar cells is strongly influenced by the quality of the perovskite layer. New synthetic routes for the preparation of perovskite films are frequently reported, leading to a wide spread of the power conversion efficiencies (PCE) in the field and making it difficult to compare the influence of different parameters between different publications.23 This poses a significant challenge to compare different HTMs with each other and, therefore, developing design rules is difficult. Here we present an entire series of small-molecules that we apply as HTMs in perovskite photovoltaic devices generated using the same preparation method and in the same batch. This allows for the direct comparison of the performance and assists in the extraction of design rules regarding the core of the HTM, and the influence of the orientation of the azomethine bond.

We prepared solar cells employing a common device architecture (fluorine doped tin oxide/TiO2/perovskite/HTM/Au) with a planar compact TiO2 hole-blocking layer (Fig. S9–S13). Smooth methylammonium lead iodide (CH3NH3PbI3) perovskite films were prepared using our previously published protocol.67 The HTMs were spincoated from chlorobenzene at a concentration of 10 mg mL−1 following our previously optimized procedure,5 which results in layers with thicknesses in the 50 nm range. The azomethine-based HTMs generally form uniform, pinhole-free films even though the concentrations in the spincoating solutions are relatively low (Fig. S14). In addition to saving material, the thin HTM layer is particularly interesting for tandem solar cell applications, since the parasitic absorption can be reduced. In a direct comparison, we found that preparing spiro-OMeTAD films with this thickness results in shunting behaviour and very low device reproducibility.5

Power conversion efficiencies (PCEs) in the range of 9.3 to 14.4% were obtained for devices prepared with the different azomethine-based HTMs (Table 4), marginally lower than those made with spiro-OMeTAD showing a PCE of 15.1% on a side-to-side comparison. Besides the previously published EDOT-OMeTPA, Diazo-OMeTPA also shows a good PCE of 14.4%. Both materials exhibit a stabilized power output that is comparable to devices made with spiro-OMeTAD (Fig. S11). More specifically, Diazo-OMeTPA shows a high open-circuit voltage (Voc) of 1.10 V, which is slightly higher than the voltage we obtained for spiro-OMeTAD (1.06 V), and higher than the voltages we obtained with all other azomethine-based HTMs. The high Voc can be ascribed to the relatively deep HOMO energy level of the Diazo-OMeTPA, resulting in an optimal alignment with the valence band of the perovskite. Additionally, we expect that the interface between the perovskite and Diazo-OMeTPA exhibits lower surface recombination than EDOT-OMeTPA, as was recently shown,68 which could explain the high voltages. In contrast, further lowering the HOMO energy level, as in Ph-inv-OMeTPA, results in an injection barrier and prevents efficient extraction of the charges, hence explaining the very poor photovoltaic performance of this material.

Table 4 Device performance of perovskite solar cells containing different HTMs
Compound J sc (mA cm−2) V oc (V) FF (%) PCE (%)
EDOT-OMeTPA 19.2 1.01 73 14.4
Ph-OMeTPA 18.6 0.89 55 9.3
Th-OMeTPA 18.9 0.99 70 13.2
Fu-OMeTPA 17.9 0.98 72 13.0
TPA-OMeTPA 17.4 0.93 59 9.6
Ph-inv-OMeTPA 1.4 0.62 26 0.2
Diazo-OMeTPA 18.9 1.10 69 14.4
Spiro-OMeTAD 18.7 1.06 76 15.1


The HTMs with a heterocycle in the core (EDOT-OMeTPA, Th-OMeTPA and Fu-OMeTPA) show significantly better performance than the HTMs without a heterocycle in the core (Ph-OMeTPA and TPA-OMeTPA). We hypothesize that the high electron density in EDOT-OMeTPA, Th-OMeTPA and Fu-OMeTPA likely results in stronger interactions with the perovskite interface and/or enhances the packing of the material, as both azomethine bonds face the same direction, resulting in a directional dipole that potentially enhances the interaction with the perovskite interface. Diazo-OMeTPA also shows good photovoltaic performance, which could be the result of the flat structure of the core,29 enhancing the packing efficiency. In addition, we note that the best performing materials also have the highest melting points, which might also hint at tight molecular packing in the solid.

Despite the significant differences in the mobility of the different HTMs, we do not observe large differences in the short-circuit current (Jsc), or a correlation between the hole mobility and the fill factor, which could be related to the relatively thin HTM layer. However, we do observe that the fill factor for all azomethine-based HTMs is lower than for spiro-OMeTAD. This could originate from the slightly lower mobility and conductivity, although it is more likely the result of the relatively high surface recombination rates between the perovskite and the HTM, as was shown for the MAPbI3/EDOT-OMeTPA interface.68

Moisture-induced degradation

The rather poor stability of CH3NH3PbI3 perovskite, especially towards moisture, is a major drawback for the potential commercialisation of the technology. Various groups have studied the degradation of the perovskite and found several decomposition pathways. Frost et al. reported that the presence of water can degrade the perovskite to HI, PbI2 and CH3NH2.69 Leguy et al. showed that humidity can also degrade the perovskite resulting in the formation of CH3NH3PbI3·H2O and (CH3NH3)4PbI6·2H2O.70 The formed degradation products depend on several factors like humidity and exposure time, but also on the chemical composition and precursor excess in the perovskite.67,70,71 Organic HTMs are generally hydrophobic and are therefore expected to protect the perovskite against degradation. However, if there is a weak adhesion of the HTM with the perovskite, the interface can still be destabilized, leading to large-scale delamination and therefore device failure.72 Recent work by Lee et al. showed that the adhesion of the HTM with the perovskite at the interface depends on the HTM.73 Moreover, pinholes in the HTM layer can play an important role as well and are known as a starting point for moisture-induced degradation.74

Our azomethine-based HTMs show good film-forming properties and we are able to form relatively thin (∼50 nm compared to 250 nm for spiro-OMeTAD) pinhole-free films of the HTMs, even upon addition of LiTFSI and tBP (Fig. S14). To study the moisture barrier properties of the HTMs, perovskite films covered with a layer of the HTMs (with additives) were stored under a high humidity environment with a controlled relative humidity (RH) of 75% in air. The edges of the films were covered with epoxy glue to slow down moisture-induced degradation initiating from the edge of the substrate (schematic in Fig. S16). Despite the significantly thicker layer of spiro-OMeTAD, the perovskite covered with the latter started to degrade within several hours, which is in agreement with literature.73 In contrast, all our azomethine-based HTMs protected the perovskite significantly better, as can be observed in the photos in Fig. 6 and S15, and even after 14 days the perovskite did not appear to degrade as a result of hydration, demonstrating the excellent barrier properties of our HTMs (Fig. S18).


image file: c7ta06452g-f6.tif
Fig. 6 (a) Photos of perovskite films covered with a layer of the HTM and stored at 75% RH. (b, c, d) XRD patterns obtained from in situ hydration measurements of the perovskite:HTM bilayers at 90% RH. The reflection from PbI2, which is the degradation product of the perovskite, is indicated with an asterisk. All perovskites were prepared using 5% PbI2 excess in the precursor solution, explaining the initial small PbI2 peak observed in all samples.

The moisture-induced degradation of the perovskite was confirmed by in situ XRD measurements at 90% RH for perovskite/HTM bilayers with EDOT-OMeTPA and Diazo-OMeTPA. We compared the results to pristine perovskite films and perovskite/spiro-OMeTAD bilayers as reference (Fig. 6 and S17). The XRD patterns show the influence of water through the formation of PbI2 as a degradation product for the reference devices within hours,67 while the perovskites covered with the azomethine-based HTMs show no signs of degradation over a period of 8 hours. Additionally, bilayers that have been exposed for 30 days at 75% RH do not show the formation of degradation products. In contrast, the spiro-OMeTAD based reference device, which lost its colour, shows diffraction peaks for the monohydrate species of the perovskite and no peak assigned to the perovskite was observed any longer, indicating complete degradation of the perovskite (Fig. S18).

Dense and pinhole-free HTM layers are key for a good barrier against the infiltration of moisture.74 Microscopy images show that the azomethine-based HTMs exhibit good film-forming properties with few pinholes, but more importantly, these films have a significantly higher density than spiro-OMeTAD. We determined the density for EDOT-OMeTPA and Diazo-OMeTPA to be 1.24 and 1.27 g cm−3 respectively (details in ESI), which is significantly higher than the density of 1.03 g cm−3 for spiro-OMeTAD.75 The spiro-core in spiro-OMeTAD results in a large molecule with a large free volume and also makes it difficult to form a tight molecular packing, thus explaining the relatively low density. The large free volume could facilitate the diffusion of water molecules through the film and degrade the underlying perovskite. Reducing the free volume in the HTM could therefore be a suitable design route to prepare HTMs with good moisture barrier properties.

Considering an additional point, Lee et al. report that the adhesion between the HTM and the perovskite makes a significant contribution to the degradation.73 It is possible that azomethine-based HTMs also have an improved adhesion to the perovskite, explaining the improved stability towards moisture. All the azomethine-based HTMs have in common that they contain heteroatoms with an accessible free electron pair that could interact with the perovskite, and thereby passivate the surface. This is in agreement with the work of Saliba et al. who reported simulations showing the thiophene–iodine interaction between a thiophene-containing HTM and the perovskite.9 Further studies are required to study possible interactions between the azomethine-bond and the perovskite, however, our results clearly demonstrate the superior performance of these azomethine-based HTMs.

Conclusions

In conclusion, we have prepared a series of azomethine-based HTMs in which we changed the core and the direction of the azomethine bond to study their influence on the optoelectronic properties and photovoltaic performance of perovskite-based solar cells. We show that our materials can be synthesized at low-cost with strongly reduced environmental impact, while changing the chemical structure allows the fine tuning of the energy levels of the HTMs, resulting in a good energy level alignment with the valence band of CH3NH3PbI3. The HTMs can be efficiently doped with LiTFSI, leading to a significant increase in conductivity of up to 10−3 S cm−1 when 60 μL mL−1 of the LiTFSI solution is added, while the materials reach hole mobilities comparable to that of well-known state-of-the-art HTMs. As hole transporting layer in perovskite solar cells, especially Diazo-OMeTPA shows an impressive high Voc of 1.10 V, which outperforms that of spiro-OMeTAD (1.06 V). The orientation of the azomethine bond influences the optoelectronic and charge transport properties, and with both orientations high PCE can be obtained when combined with the appropriate core as demonstrated for EDOT-OMeTPA and Diazo-OMeTPA. The best photovoltaic performance was achieved with HTMs having heterocycles in the core, rather than non-heterocyclic aromatic rings, which we relate to the strong dipole in the core of the HTM and to close molecular packing. In contrast to spiro-OMeTAD, all the azomethine-based HTMs prove to be excellent barriers against moisture-induced degradation of the underlying perovskite. We attribute the superior moisture barrier properties to the relatively high density of the HTM films, ∼1.25 g cm−3, compared to spiro-OMeTAD (1.03 g cm−3). We propose that dense films with close molecular packing make it difficult for water to diffuse through the HTM layer and thereby protect the underlying perovskite against degradation. Our work demonstrates that HTMs prepared via condensation chemistry does not only significantly decrease the fabrication cost, but can also significantly improve the stability of perovskite solar cells.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors acknowledge funding from the German Federal Ministry of Education and Research (BMBF) under the agreement number 03SF0516B, the Bavarian Ministry of the Environment and Consumer Protection, the Bavarian Network “Solar Technologies Go Hybrid”, and the DFG Excellence Cluster Nanosystems Initiative Munich (NIM). P. D. acknowledges support from the European Union through the award of a Marie Curie Intra-European Fellowship.

References

  1. U. Bach, D. Lupo, P. Comte, J. E. Moser, F. Weissörtel, J. Salbeck, H. Spreitzer and M. Grätzel, Nature, 1998, 395, 583–585 CrossRef CAS .
  2. T. Malinauskas, M. Saliba, T. Matsui, M. Daskeviciene, S. Urnikaite, P. Gratia, R. Send, H. Wonneberger, I. Bruder and M. Graetzel, et al. , Energy Environ. Sci., 2016, 9, 1681–1686 CAS .
  3. M. Li, Z.-K. Wang, Y.-G. Yang, Y. Hu, S.-L. Feng, J.-M. Wang, X.-Y. Gao and L.-S. Liao, Adv. Energy Mater., 2016, 6, 1601156 CrossRef .
  4. J. C. Brauer, Y. H. Lee, K. N. Mohammad and N. Banerji, J. Mater. Chem. C, 2016, 4, 5922–5931 RSC .
  5. M. L. Petrus, T. Bein, T. J. Dingemans and P. Docampo, J. Mater. Chem. A, 2015, 3, 12159–12162 CAS .
  6. A. Binek, M. L. Petrus, N. Huber, H. Bristow, Y. Hu, T. Bein and P. Docampo, ACS Appl. Mater. Interfaces, 2016, 8, 12881–12886 CAS .
  7. C. Zuo and L. Ding, Small, 2015, 11, 5528–5532 CrossRef CAS PubMed .
  8. C. Zuo and L. Ding, Adv. Energy Mater., 2017, 7, 1601193 CrossRef .
  9. M. Saliba, S. Orlandi, T. Matsui, S. Aghazada, M. Cavazzini, J.-P. Correa-Baena, P. Gao, R. Scopelliti, E. Mosconi and K.-H. Dahmen, et al. , Nat. Energy, 2016, 1, 15017 CrossRef CAS .
  10. H. Li, K. Fu, A. Hagfeldt, M. Grätzel, S. G. Mhaisalkar and A. C. Grimsdale, Angew. Chem., Int. Ed., 2014, 53, 4085–4088 CrossRef CAS PubMed .
  11. T. Malinauskas, M. Daskeviciene, G. Bubniene, I. Petrikyte, S. Raisys, K. Kazlauskas, V. Gaidelis, V. Jankauskas, R. Maldzius and S. Jursenas, et al. , Chemistry, 2013, 19, 15044–15056 CrossRef CAS PubMed .
  12. W. Qiao, Y. Chen, F. Li, X. Zong, Z. Sun, M. Liang and S. Xue, RSC Adv., 2017, 7, 482–492 RSC .
  13. X. Zhao, F. Zhang, C. Yi, D. Bi, X. Bi, P. Wei, J. Luo, X. Liu, S. Wang and X. Li, et al. , J. Mater. Chem. A, 2016, 4, 16330–16334 CAS .
  14. P. Agarwala and D. Kabra, J. Mater. Chem. A, 2017, 5, 1348–1373 CAS .
  15. Y.-K. Wang, Z.-C. Yuan, G.-Z. Shi, Y.-X. Li, Q. Li, F. Hui, B.-Q. Sun, Z.-Q. Jiang and L.-S. Liao, Adv. Funct. Mater., 2016, 26, 1375–1381 CrossRef CAS .
  16. D. Bi, B. Xu, P. Gao, L. Sun, M. Grätzel and A. Hagfeldt, Nano Energy, 2016, 23, 138–144 CrossRef CAS .
  17. K. Rakstys, S. Paek, M. Sohail, P. Gao, K. T. Cho, P. Gratia, Y. Lee, K. H. Dahmen and M. K. Nazeeruddin, J. Mater. Chem. A, 2016, 4, 18259–18264 CAS .
  18. S. Paek, M. A. Rub, H. Choi, S. A. Kosa, K. A. Alamry, J. W. Cho, P. Gao, J. Ko, A. M. Asiri and M. K. Nazeeruddin, Nanoscale, 2016, 8, 6335–6340 RSC .
  19. J. Wang, Y. Chen, F. Li, X. Zong, J. Guo, Z. Sun and S. Xue, Electrochim. Acta, 2016, 210, 673–680 CrossRef CAS .
  20. H.-C. Liao, T. L. D. Tam, P. Guo, Y. Wu, E. F. Manley, W. Huang, N. Zhou, C. M. M. Soe, B. Wang and M. R. Wasielewski, et al. , Adv. Energy Mater., 2016, 6, 1600502 CrossRef .
  21. K. H. Hendriks, W. Li, G. H. L. Heintges, G. W. P. van Pruissen, M. M. Wienk and R. A. J. Janssen, J. Am. Chem. Soc., 2014, 136, 11128–11133 CrossRef CAS PubMed .
  22. M. Daskeviciene, S. Paek, Z. Wang, T. Malinauskas, G. Jokubauskaite, K. Rakstys, K. T. Cho, A. Magomedov, V. Jankauskas and S. Ahmad, et al. , Nano Energy, 2017, 32, 551–557 CrossRef CAS .
  23. M. L. Petrus, J. Schlipf, C. Li, T. P. Gujar, N. Giesbrecht, P. Müller-Buschbaum, M. Thelakkat, T. Bein, S. Hüttner and P. Docampo, Adv. Energy Mater., 2017, 7, 1700264 CrossRef .
  24. M. L. Petrus, R. K. M. Bouwer, U. Lafont, S. Athanasopoulos, N. C. Greenham and T. J. Dingemans, J. Mater. Chem. A, 2014, 2, 9474–9477 CAS .
  25. M. Bourgeaux, S. A. Pérez Guarìn and W. G. Skene, J. Mater. Chem., 2007, 17, 972 RSC .
  26. D. Tsang, M. Bourgeaux and W. G. Skene, J. Photochem. Photobiol., A, 2007, 192, 122–129 CrossRef CAS .
  27. M. Koole, R. Frisenda, M. L. Petrus, M. L. Perrin, H. S. J. van der Zant and T. J. Dingemans, Org. Electron., 2016, 34, 38–41 CrossRef CAS .
  28. A. D. Hendsbee, J.-P. Sun, T. M. McCormick, I. G. Hill and G. C. Welch, Org. Electron., 2015, 18, 118–125 CrossRef CAS .
  29. B.-B. Ma, H. Zhang, Y. Wang, Y.-X. Peng, W. Huang, M.-K. Wang and Y. Shen, J. Mater. Chem. C, 2015, 3, 7748–7755 RSC .
  30. R. A. Sheldon, Chem. Commun., 2008, 3352–3365 RSC .
  31. Y. Fang, X. Wang, Q. Wang, J. Huang and T. Wu, Phys. Status Solidi A, 2014, 211, 2809–2816 CrossRef CAS .
  32. S. Bertho, I. Haeldermans, A. Swinnen, W. Moons, T. Martens, L. Lutsen, D. Vanderzande, J. Manca, A. Senes and A. Bonfiglio, Sol. Energy Mater. Sol. Cells, 2007, 91, 385–389 CrossRef CAS .
  33. E. Ripaud, D. Demeter, T. Rousseau, E. Boucard-Cétol, M. Allain, R. Po, P. Leriche and J. Roncali, Dyes Pigm., 2012, 95, 126–133 CrossRef CAS .
  34. S. Haid, A. Mishra, C. Uhrich, M. Pfeiffer and P. Bäuerle, Chem. Mater., 2011, 23, 4435–4444 CrossRef CAS .
  35. W. Li, M. Kelchtermans, M. M. Wienk and R. A. J. Janssen, J. Mater. Chem. A, 2013, 1, 15150–15157 CAS .
  36. J. Huang, X. Wang, X. Zhang, Z. Niu, Z. Lu, B. Jiang, Y. Sun, C. Zhan and J. Yao, ACS Appl. Mater. Interfaces, 2014, 6, 3853–3862 CAS .
  37. X. Liu, F. Kong, F. Guo, T. Cheng, W. Chen, T. Yu, J. Chen, Z. Tan and S. Dai, Dyes Pigm., 2017, 139, 129–135 CrossRef CAS .
  38. L. Cai, X. Qian, W. Song, T. Liu, X. Tao, W. Li and X. Xie, Tetrahedron, 2014, 70, 4754–4759 CrossRef CAS .
  39. M. L. Petrus, F. S. F. Morgenstern, A. Sadhanala, R. H. Friend, N. C. Greenham and T. J. Dingemans, Chem. Mater., 2015, 27, 2990–2997 CrossRef CAS .
  40. J. C. Hindson, B. Ulgut, R. H. Friend, N. C. Greenham, B. Norder, A. Kotlewski and T. J. Dingemans, J. Mater. Chem., 2010, 20, 937–944 RSC .
  41. C. Zuo, H. J. Bolink, H. Han, J. Huang, D. Cahen and L. Ding, Adv. Sci., 2016, 3, 1500324 CrossRef PubMed .
  42. D. Sek, A. Iwan, B. Jarzabek, B. Kaczmarczyk, J. Kasperczyk, Z. Mazurak, M. Domanski, K. Karon and M. Lapkowski, Macromolecules, 2008, 41, 6653–6663 CrossRef CAS .
  43. S. Ryu, J. H. Noh, N. J. Jeon, Y. Chan Kim, W. S. Yang, J. Seo and S. Il Seok, Energy Environ. Sci., 2014, 7, 2614–2618 CAS .
  44. H. Zhou, L. Yang, S. Stoneking and W. You, ACS Appl. Mater. Interfaces, 2010, 2, 1377–1383 CAS .
  45. W.-J. Chi, Q.-S. Li and Z.-S. Li, Nanoscale, 2016, 8, 6146–6154 RSC .
  46. M. L. Petrus, Azomethine-Based Donor Materials for Organic Solar Cells, Delft University Of Technology, Delft, 2014 Search PubMed .
  47. M.-H. Tremblay, T. Skalski, Y. Gautier, G. Pianezzola and W. G. Skene, J. Phys. Chem. C, 2016, 120, 9081–9087 CAS .
  48. H. T. Nicolai, G. A. H. Wetzelaer, M. Kuik, A. J. Kronemeijer, B. de Boer and P. W. M. Blom, Appl. Phys. Lett., 2010, 96, 172107 CrossRef .
  49. D. D. Medina, M. L. Petrus, A. N. Jumabekov, J. T. Margraf, S. Weinberger, J. M. Rotter, T. Clark and T. Bein, ACS Nano, 2017, 11, 2706–2713 CrossRef CAS PubMed .
  50. P. N. Murgatroyd, J. Phys. D: Appl. Phys., 1970, 3, 151–156 Search PubMed .
  51. P. W. M. Blom, M. J. M. De Jong and M. G. Van Munster, Phys. Rev. B, 1997, 55, 656–659 CrossRef .
  52. J. C. Bijleveld, B. P. Karsten, S. G. J. Mathijssen, M. M. Wienk, D. M. de Leeuw and R. A. J. Janssen, J. Mater. Chem., 2011, 21, 1600–1606 RSC .
  53. M. L. Petrus, R. K. M. Bouwer, U. Lafont, D. H. K. Murthy, R. J. P. Kist, M. L. Böhm, Y. Olivier, T. J. Savenije, L. D. A. Siebbeles and N. C. Greenham, et al. , Polym. Chem., 2013, 4, 4182–4191 RSC .
  54. D. Işık, C. Santato, S. Barik and W. G. Skene, Org. Electron., 2012, 13, 3022–3031 CrossRef .
  55. F. C. Krebs and M. Jørgensen, Synth. Met., 2004, 142, 181–185 CrossRef CAS .
  56. D. Shi, X. Qin, Y. Li, Y. He, C. Zhong, J. Pan, H. Dong, W. Xu, T. Li and W. Hu, et al. , Sci. Adv., 2016, 2, e1501491 Search PubMed .
  57. A. Abate, T. Leijtens, S. Pathak, J. Teuscher, R. Avolio, M. E. Errico, J. Kirkpatrik, J. M. Ball, P. Docampo and I. McPherson, et al. , Phys. Chem. Chem. Phys., 2013, 15, 2572–2579 RSC .
  58. A. Bolduc, C. Mallet and W. G. Skene, Sci. China: Chem., 2012, 56, 3–23 CrossRef .
  59. T. Tshibaka, S. Bishop, I. U. Roche, S. Dufresne, W. D. Lubell and W. G. Skene, Chemistry, 2011, 17, 10879–10888 CrossRef CAS PubMed .
  60. M. E. Mullholland, D. Navarathne, M. L. Petrus, T. J. Dingemans and W. G. Skene, J. Mater. Chem. C, 2014, 2, 9099–9108 RSC .
  61. H. Niu, H. Kang, J. Cai, C. Wang, X. Bai and W. Wang, Polym. Chem., 2011, 2, 2804–2817 RSC .
  62. C. Yang and S. A. Jenekhe, Chem. Mater., 1991, 3, 878–887 CrossRef CAS .
  63. X. Li, C. Li and S. Li, Synth. Met., 1993, 60, 285–288 CrossRef CAS .
  64. C. Wang, S. Shieh, E. Legoff and M. G. Kanatzidis, Macromolecules, 1996, 29, 3147–3156 CrossRef CAS .
  65. A. Iwan, M. Palewicz, A. Chuchmala, A. Sikora, L. Gorecki and D. Sek, High Perform. Polym., 2013, 25, 832–842 CrossRef .
  66. A. Iwan, M. Palewicz, A. Sikora, J. Chmielowiec, A. Hreniak, G. Pasciak and P. Bilski, Synth. Met., 2010, 160, 1856–1867 CrossRef CAS .
  67. M. L. Petrus, Y. Hu, D. Moia, P. Calado, A. M. A. Leguy, P. R. F. Barnes and P. Docampo, ChemSusChem, 2016, 9, 2699–2707 CrossRef CAS PubMed .
  68. E. M. Hutter, J.-J. Hofman, M. L. Petrus, M. Moes, R. D. Abellon, P. Docampo and T. J. Savenije, Adv. Energy Mater., 2017, 7, 1602349 CrossRef .
  69. J. M. Frost, K. T. Butler, F. Brivio, C. H. Hendon, M. van Schilfgaarde and A. Walsh, Nano Lett., 2014, 14, 2584–2590 CrossRef CAS PubMed .
  70. A. M. A. Leguy, Y. Hu, M. Campoy-Quiles, M. I. Alonso, O. J. Weber, P. Azarhoosh, M. van Schilfgaarde, M. T. Weller, T. Bein and J. Nelson, et al. , Chem. Mater., 2015, 27, 3397–3407 CrossRef CAS .
  71. Y. Hu, M. F. Aygüler, M. L. Petrus, T. Bein and P. Docampo, ACS Energy Lett., 2017, 2, 2212–2218 CrossRef CAS .
  72. J. H. Yun, I. Lee, T.-S. Kim, M. J. Ko, J. Y. Kim and H. J. Son, J. Mater. Chem. A, 2015, 3, 22176–22182 CAS .
  73. I. Lee, J. H. Yun, H. J. Son and T.-S. Kim, ACS Appl. Mater. Interfaces, 2017, 9, 7029–7035 CAS .
  74. L. K. Ono, S. R. Raga, M. Remeika, A. J. Winchester, A. Gabe and Y. Qi, J. Mater. Chem. A, 2015, 3, 15451–15456 CAS .
  75. P. Docampo, A. Hey, S. Guldin, R. Gunning, U. Steiner and H. J. Snaith, Adv. Funct. Mater., 2012, 22, 5010–5019 CrossRef CAS .

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c7ta06452g

This journal is © The Royal Society of Chemistry 2017