Forum Article pubs.acs.org/IC
Main-Group Halide Semiconductors Derived from Perovskite: Distinguishing Chemical, Structural, and Electronic Aspects Douglas H. Fabini,†,‡ John G. Labram,§ Anna J. Lehner,†,⊥,∥ Jonathon S. Bechtel,‡ Hayden A. Evans,†,¶ Anton Van der Ven,‡ Fred Wudl,†,‡,¶ Michael L. Chabinyc,†,‡,§ and Ram Seshadri*,†,‡,¶ †
Materials Research Laboratory, ‡Materials Department, §California Nanosystems Institute, and ¶Department of Chemistry and Biochemistry, University of California, Santa Barbara, California 93106, United States ⊥ Institute for Applied Materials, Karlsruhe Institute of Technology, 76131 Karlsruhe, Germany ∥ Fraunhofer-Institut für Werkstoffmechanik, 79108 Freiburg, Germany ABSTRACT: Main-group halide perovskites have generated much excitement of late because of their remarkable optoelectronic properties, ease of preparation, and abundant constituent elements, but these curious and promising materials differ in important respects from traditional semiconductors. The distinguishing chemical, structural, and electronic features of these materials present the key to understanding the origins of the optoelectronic performance of the well-studied hybrid organic−inorganic lead halides and provide a starting point for the design and preparation of new functional materials. Here we review and discuss these distinguishing features, among them a defect-tolerant electronic structure, proximal lattice instabilities, labile defect migration, and, in the case of hybrid perovskites, disordered molecular cations. Additionally, we discuss the preparation and characterization of some alternatives to the lead halide perovskites, including lead-free bismuth halides and hybrid materials with optically and electronically active organic constituents.
■
INTRODUCTION Main-group halide perovskites, particularly hybrid organic− inorganic lead halides, have attracted immense attention of late for their remarkable optoelectronic properties, ease of preparation, and abundant constituent elements.1−7 Such compounds, of which methylammonium lead iodide (CH3NH3PbI3) has been the most extensively studied, are strong optical absorbers,8 bear high charge-carrier mobilities,9 exhibit exciton−diffusion lengths on the order of microns,10 display an unusually low concentration11 of energetically shallow12 trap states, and demonstrate evidence of photon recycling.13 Like their well-studied oxide analogues, the halide perovskites exhibit great chemical flexibility, leading to rich and varied structure and function. To date, these materials have demonstrated their utility in photovoltaic devices,1−3 lightemitting diodes,14,15 lasers,16,17 radiation detectors,18 field-effect transistors (FETs) and phototransistors,19−23 and nonlinear optics24 and have been proposed as high-performance thermoelectrics25 and even topological insulators.26−29 Despite their comparable performance by some metrics, the main-group halide perovskites and related phases are chemically, structurally, and electronically quite distinct from traditional group IV, III−V, and II−VI semiconductors. The crystal and electronic structures and properties of these materials are intimately linked to the chemistry of the heavy main-group metals and heavy halogens. Substantial structural complexity derives from mixed ionic−covalent bonding, mechanical indeterminacy of the perovskite structure,30,31 © XXXX American Chemical Society
lone-pair electrons on the main-group dication, and the presence of disordered molecular cations in the hybrids. Together, the chemistry and crystal structure of the halide perovskites give rise to an unusual “orbital-inverted” electronic structure,32,33 which, in turn, imbues them with unique properties related to defect tolerance, optical absorption, and lattice dynamics. In this Forum, we summarize some recent work from authors in this area, with updated discussion and an emphasis on the distinguishing chemical, structural, and electronic features of these materials. First, we discuss the electronic structure of perovskite and 1D polymorphs of APbI3 (A = Rb, Cs) as a reference case for understanding these materials at large and establish the level of theory required to accurately describe these compounds within density functional theory (DFT). Next, drawing inspiration from the lead halides, we review work on bismuth halides derived from perovskite as promising leadfree, air-stable optoelectronic materials via computation, experiment, and the fabrication of photovoltaic devices. Additionally, we describe a new layered hybrid lead iodide complex with electronically and optically active organic radical cations and charge transfer between the inorganic and organic sublattices. Moving on to the prototypical hybrid perovskite Special Issue: Halide Perovskites Received: June 30, 2016
A
DOI: 10.1021/acs.inorgchem.6b01539 Inorg. Chem. XXXX, XXX, XXX−XXX
Forum Article
Inorganic Chemistry lead iodides, we discuss temperature-dependent polarization of CH3NH3PbI3 thin films observed in FETs, photovoltaic devices, and capacitors and relate this behavior to the device functionality and the low barriers for defect formation and migration. Focusing on the dynamics of the organic cations and their impact on the structure, dielectric screening, and charge separation, we describe glassy behavior in CH3NH3PbI3 and HC(NH2)2PbI3 at low temperatures that offers clues about room temperature molecule−cage interactions and demonstrates the delicate balance between octahedral rotations, molecular motion, and hydrogen-bonding tendencies in these materials. Finally, we discuss the energy landscape for molecular rotation and translation in cubic CH3NH3PbI3 from first principles and illustrate coupling between the molecular motion, local distortions of the inorganic sublattice, and electronic structure.
Figure 1. Crystal structures of (a) PbI2 and (b and c) polymorphs of APbI3 (A = Rb, Cs). Small purple spheres are I−, large blue spheres are countercations A+, and spheres inside octahedra are Pb2+. Reprinted with permission from ref 41. Copyright 2014 American Chemical Society.
■
RESULTS AND DISCUSSION 1. Electronic Structure of Main-Group Halide Perovskites. Because of the presence of lone-pair s2 electrons on the divalent metal cation (Sn2+ and Pb2+) in the main-group halide perovskites, these materials bear more electronic similarity to rock-salt chalcogenides of the heavier carbon-group elements than to traditional tetrahedral semiconductors. Where III−V and II−VI semiconductors exhibit a band gap between occupied bonding states with primarily anion character and unoccupied antibonding states with primarily cation character, the band gap of the halide perovskites is between occupied (cation s and anion p) and unoccupied (cation p and anion p) antibonding states.32,33 Early work by Watson, Seshadri, and others on the stereochemistry of lone-pair electrons in extended solids provides a useful context for understanding the electronic structure and lattice dynamics of main-group halide perovskites.34−39 Additionally, an excellent tutorial on the relationship between the chemical bonding and electronic structure in electronically similar rock-salt chalcogenides has recently been provided by Zeier and co-workers.40 The initial work from our group on main-group halide perovskites, by Brgoch and co-workers, sought to systematically establish the level of theory required to accurately describe the electronic structure and dielectric properties of complex lead iodide semiconductors within the framework of ab initio calculations based on DFT.41 Calculations employing the generalized gradient approximation (GGA) as well as screened hybrid functionals and omitting or including spin−orbit coupling (SOC) were performed on the layered binary PbI2 (P3m ̅ 1) and the results compared with experiment. PbI2 was chosen as a model system because of its similar Pb−I coordination and iodine close-packing with that of the α phases (perovskite, Pm3̅m, corner-sharing PbI6 octahedra in 3D) and δ phases (Pnma, ribbons of edge-sharing PbI6 octahedra) of APbI3 (A = Rb, Cs). These structures are presented in Figure 1. The importance of SOC in similar heavy halides is well established,42−44 and indeed inclusion of this interaction was required to accurately reproduce the experimental band gap and ionization energy of PbI2. Knowledge of the ionization energy, which expresses the position of the valence-band maximum (VBM) relative to the vacuum level, is essential for the energetic alignment of interfacial layers in devices. It was determined that the GGA functional of Perdew, Burke, and Ernzerhof45 with spin−orbit coupling (PBE+SOC) suffices to accurately reproduce experimental ionization energies using
slab calculations.46 Because of the shortcomings of the GGA in accurately capturing excited states, the screened hybrid functional of Heyd, Scuseria, and Ernzerhof47 with 25% exact exchange and with SOC included (HSE06+SOC) was required to reproduce the experimental band gap of PbI2. The calculated electronic structures of the polymorphs of APbI3 are given in Figure 2.41 Band gaps are strongly correlated to structure: The perovskite phases exhibit narrow band gaps due to strong orbital overlap and 3D connectivity, while those of the δ phases are similar to that of PbI2, reflecting their similar connectivity of edge-sharing Pb−I octahedra. Recent reports on
Figure 2. Calculated electronic band structures (HSE06) for polymorphs of (a and c) RbPbI3 and (b and d) CsPbI3 with and without SOC. Because of computational expense, only the highsymmetry paths proximate to the band gap are presented. Reprinted with permission from ref 41. Copyright 2014 American Chemical Society. B
DOI: 10.1021/acs.inorgchem.6b01539 Inorg. Chem. XXXX, XXX, XXX−XXX
Forum Article
Inorganic Chemistry the preparation of α-CsPbI3 in bulk48 and in thin-film solar cells49,50 and α-CsPbI3:Cl nanocrystals51 have appeared that find the band gap to be ∼1.7 eV at ambient temperature, larger than the 0.6 eV calculated by Brgoch and co-workers.41 The unconventional “orbital-inverted”52 nature of the electronic structure,32,33 together with the extremely large thermal expansion coefficients,53 causes the band gaps of main-group halide perovskites to widen rapidly with temperature,54,55 as they do in divalent group IV chalcogenides.40 Additionally, a recent report finds that band-gap renormalization due to electron−phonon interactions widens the gap by >100 meV at ambient temperature in isostructural cubic CsSnI3,56 further underscoring the importance of finite temperature effects in these unusual systems. Together, the ionization energy and band gap determine the absolute positions of the VBM and conduction-band minimum (CBM), which are presented in Figure 3 for the compounds
instability and the manner in which this behavior may be exploited to elicit desirable properties. Additionally, recent work has found that in the structurally complex hybrid perovskites, the motion of the molecular cation couples to local structural distortions of the inorganic network, thus indirectly modifying the electronic structure (vide infra).61 2. Alternative Chemistries I: Lead-Free Inorganic Bismuth Halides. Because of concerns about lead toxicity for practical application of lead halide perovskite photovoltaics,62−64 significant effort has been directed toward pursuing Pb-free alternatives that share the favorable optoelectronic performance of CH3NH3PbI3. Early attention focused on the substitution of Pb2+ with isovalent Sn2+,65,66 but these materials are prone to oxidation because of the reduced relativistic stabilization of the 5s2 lone pair. The unique electronic features and defect tolerance of systems with lonepair-bearing metals discussed above pushed us to pursue halides of other heavy main-group elements in their lower oxidation states. While the isoelectronic period 5 ions (In+, Sn2+, and Sb3+) are prone to oxidation and Tl+ is highly toxic, Bi3+ is stable against oxidation, relatively nontoxic, and reasonably abundant. As such, we investigated the structure and optoelectronic properties of binary and complex bismuth iodides, BiI3 and A3Bi2I9 (A = K, Rb, Cs).67,68 The layered BiI3 has been investigated for hard radiation detectors,69−72 Xray imaging,73−75 and hole-transport layers (HTLs) in solar cells76 but had not been explored as a photovoltaic absorber prior to the work by Lehner and co-workers that is described here.67 While the crystal structures of some complex bismuth iodides have been reported,77−81 few reports have addressed their optical properties.82−84 In particular, the suitability of the alkali-metal complex bismuth iodides for optoelectronic applications was minimally explored85 prior to our work. In our report, we described facile preparation routes for single crystals, bulk materials, and solution-processed thin films of the A3Bi2I9 compounds.68 The reported crystal structures of K3Bi2I979 and Cs3Bi2I981 were verified, and that of Rb3Bi2I980 was corrected on the basis of single-crystal and powder X-ray diffraction (XRD). 1D magic-angle-spinning (MAS) and triplequantum MAS 87Rb NMR experiments and complementary calculations of NMR observables using DFT confirmed the revised P21/n structure model. Further details of the crystal structure investigation of Rb3Bi2I9 may be obtained from FIZ Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany, upon quoting the deposition number CSD 430179. The structures of the binary and complex bismuth iodides are given in Figure 4. Detailed structure descriptions, including their relationship to the perovskite aristotype, are provided in our recent reports,67,68 and we note here for brevity only the distinct connectivity and slight distortions of the Bi−I octahedral anions, which have important consequences for the electronic structure. The electronic structures for BiI3, Rb3Bi2I9, and Cs3Bi2I9 were calculated using DFT, and the representative case of BiI3 is shown in Figure 5 (energy levels are referenced to the vacuum, vide infra).67,68 Agreement with band gaps extracted from Kubelka−Munk-transformed diffuse-reflectance spectroscopy (Figure 6)67,68 is favorable. BiI3 exhibits a slightly indirect gap near A of 1.9 eV, which compares well with our experimental value of 1.8 eV.67 The greater dispersion of the conduction bands is consistent with reports of an electron mobility of 30 times that for holes (μe = 600 cm2 V−1 s−1 and μh = 20 cm2 V−1 s−1).87 The flatter band dispersions compared with lead iodide
Figure 3. Calculated PBE+SOC absolute levels of the VBM and CBM of PbI2, polymorphs of RbPbI3 and CsPbI3, and anatase TiO2. Reprinted with permission from ref 41. Copyright 2014 American Chemical Society.
studied by Brgoch and co-workers.41 There is a correlation between the dimensionality and ionization energy, with the VBMs of the δ phases of 1D ribbons substantially deeper than those of the cubic perovskites. The deep VBMs and wide band gaps of the δ phases suggest that they are unsuitable for many optoelectronic applications, while the shallower VBM of αCsPbI3, together with its experimental band gap, suggests its suitability for photovoltaics49,50 if it can be appropriately stabilized.48,51 High Born effective charges imply a large static dielectric response, which is proposed to enhance charge transport through the screening of charged defects.57 Indeed, in the doped complex oxide KTa1−xNbxO3:Ca, the peak in the Hall mobility tracks the dielectric response associated with the pseudo-Jahn−Teller distortion58 due to the d0 transition metals.59 The calculated optical dielectric tensors and Born effective charge tensors, Zij*, of the APbI3 polymorphs are an indicator of such a potential lattice instability, another consequence of the unusual electronic structure that results from the lone-pair electrons on Pb2+.38 The Born charges of Pb are elevated relative to the nominal 2+ charge in all cases but are particularly high for the perovskite phases (Z* = 5.4 e and 5.2 e for Rb and Cs, respectively) and only just shy of those for PbTe,38,60 a well-known incipient ferroelectric. Ongoing work seeks to elucidate the role of the main-group metal lone-pair electrons in driving an usual form of lattice C
DOI: 10.1021/acs.inorgchem.6b01539 Inorg. Chem. XXXX, XXX, XXX−XXX
Forum Article
Inorganic Chemistry
Figure 6. UV−vis diffuse-reflectance spectra for powders of BiI3 and A3Bi2I9, Kubelka−Munk-transformed to absorbance. Band gaps were extracted assuming direct gaps for K3Bi2I9 and Rb3Bi2I9 and indirect gaps for the others, in accordance with DFT calculations. Reprinted with permission from ref 68. Copyright 2015 American Chemical Society.
Figure 4. Crystal structures of (a) BiI3 (R3)̅ ,86 (b) Cs3Bi2I9 (P63/ mmc),81 and (c) Rb3Bi2I9 (P21/n).68 The relationship to the perovskite structure is evident in the inset of part c, with thermal ellipsoids at the 80% level. Small purple spheres are I−, large orange spheres are countercations A+, and spheres inside the octahedra are Bi3+. Reprinted with permission from ref 68. Copyright 2015 American Chemical Society.
Despite the modest carrier effective masses, the electronic structures and dielectric properties of the binary and complex bismuth iodides are indicative of defect tolerance. As in the case of lead halide perovskites, the electronic structures of these materials are orbital-inverted, with antibonding (Bi s and I p) states constituting the top of the valence band and antibonding (Bi p and I p) states the lowest conduction bands. These features have been correlated with the suppressed formation of deep trap states in the lead halides and other semiconductors.85,90−92 As for Pb2+ in RbPbI3 and CsPbI3 above, the Born effective charges of Bi3+ in the layered compounds are significantly elevated,67,68 suggesting a proximal lattice instability, enhanced ionic dielectric response, and accordant screening of charged defects.57,59 The absolute band positions, which are essential for the energetic alignment of interfacial layers in devices, were determined both computationally and from ultraviolet photoemission spectroscopy, with excellent agreement (Figure 7).67,68 Supercell slab calculations93 were performed on BiI3, and the ionization energy was determined from the difference in the electrostatic potentials within the crystal slab and in the vacuum region. For the complex bismuth iodides, deep, contracted Bi s-like states were aligned with those in BiI3 to establish the energy relative to the vacuum. The success of this method, with intensive computations required only on a simpler reference compound, suggests its future applicability as a screening tool for complex materials. As in the case of layered lead iodides,94 the VBM is sensitive to the dimensionality of the Bi−I network.68 The VBMs are deep relative to the lead iodides, a consequence of the more stabilized atomic 6s levels and reduced orbital overlap of these spatially contracted states with those of the ligands. On the basis of the suitable band gap, reasonably small carrier effective masses, and defect-tolerance signatures of BiI3, we fabricated photovoltaic devices with this material as the absorber, dense TiO2 as the electron-transport layer, and two different HTLs. The devices and characterization results are presented in Figure 8.67 The external quantum efficiency exhibits a steep Urbach tail and indicates absorption across the spectrum. The deep VBM of BiI3 presents a challenge for hole
Figure 5. Calculated electronic band structure (PBE+SOC) and orbital-projected density of states (HSE06+SOC) of BiI3. The energies are referenced to the vacuum using slab calculations. Reprinted with permission from ref 67. Copyright 2015 AIP Publishing.
perovskites reflect the greater electronegativity difference of Bi and I, as well as the influence of the slight octahedral distortions that reduce orbital overlap.88 Interestingly, among the layered compounds, the countercation size appears to have a minimal effect on the band gap compared to the Bi−I−Bi angles and connectivity.67,68 The sharp absorption onsets of the layered compounds (Figure 6)68 suggest a lack of deep trap states and a small opencircuit voltage loss in photovoltaic applications.11 The calculated electronic structures also point to the existence of p−p transitions within the visible and near-UV range, which have been implicated in the strong optical absorption of the lead halide perovskites.89 Considering the shape of the absorption onset for Cs3Bi2I9, it is conceivable that the discrepancy between calculated and experimental band gaps (indirect, 2.3 and 1.9 eV, respectively)68 can be explained by excitonic effects, which might be expected for a system with isolated [Bi2I9]3− anions.42 D
DOI: 10.1021/acs.inorgchem.6b01539 Inorg. Chem. XXXX, XXX, XXX−XXX
Forum Article
Inorganic Chemistry
bismuth halides are promising candidates for optoelectronic applications. Concurrent with and since our reports, a number of theoretical and experimental studies have appeared that address the optoelectronic properties and applications of bismuth and antimony halides: BiI 3 , 9 6 Cs 3 Bi 2 I 9 , 9 7 Cs 3 Sb 2 I 9 , 9 8 (CH3NH3)3Bi2I9,97,99−103 (CH3NH3)3Sb2I9,104 (NH4)3Bi2I9,105 bismuth oxyhalides and chalcohalides,106−108 ordered bismuth halide elpasolites,109−114 and minimally explored phases with the supposed compositions of AgBi2I7115 and CsBi3I10.116 While most of these investigations have revealed charge-transport properties and photovoltaic performances thus-far inferior to those of the lead halide analogues, many of these compounds appear to exhibit superior chemical stability, and the recent report of an unusually long carrier recombination lifetime of 670 ns for powders of the indirect band-gap elpasolite Cs2AgBiBr6110 is a testament to the fact that the optoelectronic potential of heavy main-group halides is far from fully explored. 3. Alternative Chemistries II: Hybrid Halides with Functional Organic Cations. Hybrid organic−inorganic materials present an opportunity to combine the usually disparate areas of organic and inorganic chemistry in a synergistic manner. While the hybrid-halide perovskites have recently demonstrated remarkable performance in optoelectronic applications, their commercial viability is questionable because of their modest chemical and thermal stability and photostability. Additionally, while the organic cations of the hybrid perovskites have indirect impacts on various physical properties, they do not appear to participate directly in light absorption or electronic transport. Chemical modification of these promising materials by the incorporation of organic cations other than the small amines employed in the prototypical hybrid perovskites presents a pathway to enhancing stability and tailoring functionality. The incorporation of monovalent organic cations of heterogeneous size into main-group halides can result in layered structures, as has been demonstrated for decylammonium,42 phenyl ethylammonium,19,117 butylammonium,94,118,119 and various chromophores (vide infra). These materials, which typically adopt the perovskite-derived structures of the Ruddlesden−Popper phases, exhibit highly anisotropic properties and constitute self-assembled quantum wells because of the vastly different dielectric environments of the organic and inorganic slabs. In most systems, the inorganic slab thickness is easily tuned via the precursor stoichiometry, offering a facile means of tailoring optical, electronic, and thermodynamic properties. Of particular consequence here, recent reports have found greatly enhanced stability for these systems in photovoltaic applications relative to their 3D analogues,94,117,119 perhaps because of the hydrophobicity of the alkyl chains, and one recent report has demonstrated a 12.5% photovoltaic conversion efficiency due to the unexpected morphological preference for the inorganic slabs to align orthogonally to the substrate when cast from solution.119 These layered perovskite derivatives have been extensively reviewed, with emphasis on their structural design,120,121 electronic structure and functionality,121,122 and stability.123 An alternative approach to achieving layered main-group halides is to terminate the inorganic Pb−I slabs by replacing one apical halide with a pseudohalide, such as SCN−,124 although the optical properties and stability of the resulting compound125,126 appear to be less favorable than originally suggested.127
Figure 7. Calculated PBE+SOC VBM and CBM for binary and complex bismuth iodides (indicated by wide, colored bars) compared to the experimental values (presented with narrow, black error bars). The reported experimental values for PTAA95 and calculated values for anatase TiO241 are also displayed. Reprinted with permission from ref 68. Copyright 2015 American Chemical Society.
Figure 8. (a) SEM micrograph and schematic of BiI3 photovoltaic devices using PTAA or PIDT-DFBT as the HTL. (b) External quantum efficiency and (c) JV curves of the assembled devices. Reprinted with permission from ref 67. Copyright 2015 AIP Publishing.
injection: While the photovoltaic conversion efficiencies of these unoptimized devices are sub-1%, replacement of the poly(triarylamine) (PTAA) HTL with the deeper HOMO poly(indacenodithiophene difluorobenzothiadiazole) (PIDTDFBT) substantially improves the open-circuit voltage and fill factor. In addition to optimization of film formation, the performance of BiI3 devices could be substantially improved by the design or selection of HTLs with deeper, more suitably aligned frontier energy levels. These challenges notwithstanding, our computational and experimental results suggest that E
DOI: 10.1021/acs.inorgchem.6b01539 Inorg. Chem. XXXX, XXX, XXX−XXX
Forum Article
Inorganic Chemistry
other monovalent TTF•+ salts are closer to 10−6 S cm−1.144 This further suggests the synergistic relationship between the radical cation system and Pb−I network, which improves charge-carrier transport throughout the material. Radical cations could potentially serve as spacers for stable layered perovskite derivatives, and ongoing work in this area seeks to expand this growing class of design-oriented layered hybrid materials. Additionally, in the course of exploring hybrid-halide materials for optoelectronic applications, some of us have prepared a new pyrroloperylene−iodine complex with the first crystalline homoatomic polymer chain.145 This metallic compound, with a possible Peierls distortion below room temperature,145 illustrates that electronically interesting hybrid halides are possible beyond those containing main-group metals. 4. Hybrid Perovskites I: Indicators of Ionic Mobility in CH3NH3PbI3 Devices. While it has long been known that ionic semiconductors exhibit electronic characteristics that are broadly distinct from covalent systems,146 the impact of structural instabilities on the device and electronic properties remains under debate147,148 and is now a highly pertinent topic in the field of hybrid-halide perovskites.149 Hybrid organic− inorganic perovskite solar cells are notable for their various instabilities and measurement-dependent behavior.148,150−156 Ionic conductivity has been widely observed in many families of perovskite compounds,157−161 and compelling evidence that these processes take place in hybrid-halide perovskites has recently been provided.148,150,156,162−165 The motion of various ionic species in hybrid-halide perovskite compounds has been studied using a diverse range of electronic device measurements.23,148,155 Of these devices, thin-film transistors (TFTs) are of particular significance because their geometry and operation principle imply a considerable sensitivity to ionic motion in the active layer. Despite previous reports of TFTs employing tin-based perovskites as the active layer,19 until recently20−23 there has been a notable absence of TFTs based on compounds such as CH3NH3PbI3 from the literature. As part of our recent investigation by Labram and co-workers,23 we observed that CH3NH3PbI3 TFTs do not operate at room temperature. As the temperature of the device is reduced to below 220 K, however, the device becomes functional (Figure 10a). Soci and co-workers reported the first demonstration of ambipolar and light-emitting CH3NH3PbI3 transistors that also exhibited a similar temperature dependence.21 These characteristics are attributed to ionic screening effects, whereby mobile ions move under the influence of an applied gate field and screen the channel. As the temperature of the perovskite film is reduced, the ionic mobility is believed to fall,23 resulting in a mitigation of ionic screening behavior and consequential observation of a field-modulated source−drain current. The mobility of charge carriers can be estimated using the gradual-channel approximation166 and is plotted as a function of the temperature for our study in Figure 10b. Unlike contactless characterization techniques such as time-resolved microwave conductivity,12,167 the large static electric field present in TFT measurements is likely to disturb mobile ions, potentially rendering the gradual-channel approximation inappropriate and resulting in an underestimate of the mobility being extracted using this technique. Time-dependent transistor measurements were conducted, wherein the magnitude of the current flowing between the source and drain terminals of a CH3NH3PbI3 TFT could be
The incorporation of organic chromophores into main-group hybrid halides has also been extensively explored because of the promise of strong light absorption and possible charge transfer between the organic and inorganic sublattices. Systems based on derivatives of pyrene,128 benzyl-, naphthyl-, and anthrylammonium,129 quaterthiophene,130 bithiophene,131 tetrathiafulvalene (TTF),132,133 tetraselenafulvalene,132 and tropylium134 have been prepared and studied for their structure and optoelectronic properties. Owing to the chemical and structural tunability of these systems, they have offered a fertile platform for understanding charge transfer and light emission in hybrid systems. In this vein, we recently reported an air-stable monovalent TTF135,136 lead iodide hybrid, (TTF)Pb2I5, which displays favorable optoelectronic properties. Other metal halide complexes that incorporate TTF and its derivatives have been reported,132,133,137−139 although these were primarily studied for their magnetic or superconducting rather than their optoelectronic functionality. (TTF)Pb2I5 consists of sheets of edge- and corner-connected Pb−I octahedra charge-balanced by infinite stacks of TTF•+dimers (Figure 9a)140 and is found
Figure 9. (a) Crystal structure of (TTF)Pb2I5, viewed down the b axis. (b) Diffuse-reflectance spectra, Kubelka−Munk-transformed to absorbance, of (TTF)Pb2I5 and reagent (TTF)3(BF4)2, highlighting the presence of neutral to charged TTF transitions near hν = 0.6 eV. Reprinted with permission from ref 140. Copyright 2016 American Chemical Society.
to display back charge transfer between the Pb−I and TTF•+layers, in a manner similar to the charge transfer observed in the N-methylphenazinium 7,7,8,8-tetracyanoquinodimethanide (NMP-TCNQ) system.141 This back charge transfer is clear from the UV−vis spectrum (Figure 9b),140 where the absorption features near 0.6 eV are indicative of TTF0 to TTF•+transitions, although no crystallographic TTF0 is found in (TTF)Pb 2 I 5. 142,143 Furthermore, sustained absorption above 2.5 eV is observed due to the Pb−I matrix. Conductivity values on a pressed pellet of (TTF)Pb2I5 were found to be close to 0.01 S cm−1 at room temperature, whereas F
DOI: 10.1021/acs.inorgchem.6b01539 Inorg. Chem. XXXX, XXX, XXX−XXX
Forum Article
Inorganic Chemistry
Figure 10. (a) Transfer curves (drain current as a function of the gate voltage) of a CH3NH3PbI3 TFT measured between 150 and 250 K. Inset: Schematic representation of the FET structure employed in this study (Si gate electrode, 150 nm SiO2 dielectric, CH3NH3PbI3 semiconductor layer, and Al source and drain electrodes). (b) Field-effect mobility evaluated using the gradual-channel approximation for a CH3NH3PbI3 TFT. Above 215 K, the measured drain current was comparable to or less than the gate current. Reprinted with permission from ref 23. Copyright 2015 American Chemical Society.
Figure 11. (a) Schematic representation of the experimental setup for low-temperature, time-dependent FET measurements. A constant voltage (VD) is applied to the drain terminal, a pulsed voltage is applied to the gate terminal (VG), and the source−drain current (IS) is monitored at the source terminal. (b) Gate voltage and (c) source−drain current as a function of time in a CH3NH3PbI3 TFT, measured at 150 K. (d and e) Source− drain current as a function of time, measured at various temperatures. Reprinted with permission from ref 23. Copyright 2015 American Chemical Society.
mechanism,171,172 these traps will become more influential at lower temperatures. If ionic processes provide some benefit in these PVs, these advantages must be balanced against the adverse effects related to the device stability and reproducibility.148,151,154,156,169 While the existence of mobile ions is well established in some all-inorganic halide perovskites157 and is increasingly accepted in hybrid systems such as CH3NH3PbI3,149,163 the species, nature, and impact of the ionic contribution to conduction are under substantial debate.163,165,173−176 At present, reports favor the halide as the major ionic species,163,165,173,176,177 yet arguments involving the methylammonium cation,175,176 lead,175 and protons174 have also been put forward. While this debate is still in its infancy, with the majority of reports being theoretical 165,170,173−176 rather than experimental,163,165,177,178 there is a distinct possibility that a number of processes are taking place simultaneously and that the system may be much more complex than previously believed. We briefly discuss the state of the field below. Yang and co-workers provided experimental evidence that iodine is the major mobile species using electrical measurements of solid-state CH3NH3PbI3-based electrochemical cells and energy-dispersive X-ray spectroscopy.163 They reported
monitored as a function of time after a gate voltage is applied. Figure 11a shows the experimental setup used to conduct this study, and parts b−e of Figure 11 show the results obtained at various temperatures. The source−drain current is observed to rapidly increase upon application of the gate field and then slowly fall as ionic species screen the channel from the gate field. This screening mechanism is faster at higher temperatures, in accordance with a higher ionic mobility. Impedance spectroscopy measurements revealed a temperature- and frequency-dependent dielectric constant that is consistent with our interpretation of TFT data and the literature.150,152,153 By conducting temperature-dependent measurements of solar cells based on CH3NH3PbI3 (omitted here for brevity), we have additionally observed a reduction in the relative power conversion efficiency with reduced temperature,23 a result accordant with other reports.168,169 One interpretation of this result is that the motion of ionic species in CH3NH3PbI3 is a prerequisite for the high open-circuit voltage synonymous with hybrid-halide perovskite solar cells, via the formation of favorable internal electric field gradients in the perovskite.156,165,170 An alternative explanation is provided by the shallow carrier traps present in the film, reported to be ∼10 meV.12 If charge transport is dominated by a hopping-based G
DOI: 10.1021/acs.inorgchem.6b01539 Inorg. Chem. XXXX, XXX, XXX−XXX
Forum Article
Inorganic Chemistry diffusion coefficients between 10−8 and 10−7 cm−2 s−1 for the mixed-cation compound (MA)1−x(FA)xPbI3, where MA and FA stand for methylammonium and formamidinium, respectively. In another study by Meloni and co-workers,177 it was again argued that the halide is the species relevant for the observed hysteresis in hybrid-halide cells. Using electrical measurements of CH3NH3PbI3 and CH3NH3PbBr3 solar cells, activation energies of 168 ± 43 and 333 ± 47 meV were determined, respectively. Li and co-workers used a combination of electrical measurements, electroabsorption spectroscopy, and X-ray photoemission spectroscopy to argue that the characteristic hysteresis in CH3NH3PbI3 solar cells is due to modification of the energetic interfaces in these systems, through the migration of iodine ions and vacancies.178 Using a compact TiO2 holeblocking layer and a spiro-OMeTAD electron-blocking layer, they demonstrated how the movement of iodine ions and vacancies to these respective interfaces under positive and negative biases modified the work function of the interfaces and, consequently, the open-circuit voltage during measurement. A number of theory papers have also been published on this subject. Delugas and co-workers used molecular dynamic simulations to show that iodine is the dominant ionic conductor in methylammonium lead halides. They calculated room temperature diffusion coefficients higher than those reported experimentally by Yang et al.163 of 7 × 10−7 and 4 × 10−6 cm−2 s−1 for interstitial iodine and iodine vacancies, respectively, and corresponding activation energies of 240 and 100 meV.173 Eames and co-workers have shown that the activation energy for the migration of iodine vacancies is 600 meV both experimentally via chronophotoamperometry and computationally via DFT.165 The activation energies of lead and methylammonium vacancies were calculated to be 2.3 eV and 800 meV, respectively, and the diffusion coefficients of iodine and methylammonium ions were calculated at room temperature to be 10−12 and 10−16 cm−2 s−1, respectively. Azpiroz and co-workers used first-principles calculations to evaluate the activation energies of iodine vacancies and interstitials, methylammonium vacancies, and lead vacancies to be ∼100, ∼500, and ∼800 meV, respectively. They suggested that iodine migration is too fast and that it was, in fact, more likely that methylammonium or lead vacancies are responsible for the time-dependent behavior observed in hybrid-halide solar cells on the order of seconds.175 Egger and co-workers focused their first-principles study on interstitial hydrogen in CH3NH3PbI3.174 They determined that the energy for proton migration was between 300 and 500 meV depending on the conditions, and under device-relevant conditions, it was likely hydrogen that plays a nonnegligible role in electronic measurements. Other theoretical works have focused on the nature of point defects in general,170 and an in-depth first-principles study of the direction dependence of iodine, methylammonium, and formamidinium migration by Haruyama and co-workers evaluated values of the activation energy between 300 and 900 meV, depending on the mobile species and direction.176 Despite the broad range of hypotheses on the subject, the ionic conductivity in hybrid-halide perovskites is clearly an influential phenomenon, with implications not only for device stability but also for understanding basic operation and the origins of high performance. The extremely sharp band
edges11,179 in hybrid-halide perovskites illustrate that these are not electronically active defects,180 yet their indirect effect on the device performance is critical. For hybrid-halide electronic devices to become a pervasive technology, this phenomenon must be both well understood and controllable, a challenge that at present appears to be far from resolved. 5. Hybrid Perovskites II: Low-Temperature Molecule− Cage Interactions in CH3NH3PbI3 and HC(NH2)2PbI3. The presence of isolated, weakly interacting molecular cations introduces significant structural and dynamical complexity to the main-group hybrid-halide perovskites and places them conceptually in the realm of “plastic crystals”.181,182 Plastic crystals are molecular or partially molecular solids such as C60, alkali-metal cyanides, and ammonium halides, wherein the molecules display translational periodicity but orientational disorder. At high temperatures, molecular orientations are dynamically disordered, while at low temperatures, they are ordered or disordered statically, depending on the system.183 The role of the molecular cation in phase transitions of hybrid perovskites is well established.48,184−188 Because of its impact on the dielectric properties and local electric fields, the motion of the organic cation and its interaction with the surrounding inorganic framework have been probed extensively for CH3NH3PbI3 since it was first prepared in 1978.189 In particular, studies employing dielectric spectroscopy,162,190 NMR and nuclear quadrupolar resonance,191−193 calorimetry,194 IR spectroscopy,194,195 quasi-elastic neutron scattering,196,197 neutron powder diffraction,188 and ab initio calculations61,181,198−200 have drawn various conclusions regarding the degree of motion and disorder of the CH3NH3+ cation, particularly in the orthorhombic phase (T < 160 K). The homologous HC(NH2)2PbI3 has only recently been prepared,48 and its structural description is incomplete,48,201 despite the high HC(NH2)2+ content in the current highest-performing perovskite photovoltaic alloys.3 The reported structures of the cubic phases of both compounds and the orthorhombic phase of CH3NH3PbI3 are given in Figure 12. Recently, we observed signatures of glassy behavior upon cooling of CH3NH3PbI3 and HC(NH2)2PbI3 below 100 K, with HC(NH2)2PbI3 exhibiting greater glass fragility.182 The dielectric response of these materials is directly influenced by
Figure 12. (a) High-temperature cubic perovskite (Pm3̅m) structure of CH3NH3PbI3 at 352 K188 and HC(NH2)2PbI3 at 300 K201 with a dynamically disordered A-site cation. (b) Low-temperature orthorhombic (Pnma) structure of CH3NH3PbI3 at 100 K.188 Lowtemperature structures (
