Thermal stability of mobility in methylammonium lead iodide

Metal halide perovskites (MHPs) are a fascinating class of photovoltaic materials; possessing distinctive optoelectronic properties and simple processing routes. The most significant remaining barrier to commercialization is their poor stability under ambient conditions. While the stability of electronic parameters in this class of material has been studied extensively, to date the overwhelming majority of such studies have been carried out using PV devices. The presence of electrodes and transport layers in this approach involves both implicit encapsulation, and modification of interface properties. To develop an extensive understanding of environmental stability of electronic properties in MHPs, it is crucial to study the electronic properties of the material in isolation, rather than in a finished device. In this work, we have studied the thermal stability of electronic properties of solution processed methylammonium lead iodide (MAPbI3). MAPbI3 thin films were subjected to extended periods of elevated temperatures before their electronic properties were probed using time-resolved microwave conductivity (TRMC), a contactless technique enabling extraction of a proxy for the material’s mobility, without the need to form a device. The films were analysed with x-ray absorption spectroscopy (XAS) to study the impact of temperature on film microstructure. We observed an increase in average Pb-I bond length with increased annealing temperature.


Introduction
Photovoltaic devices based on metal halide perovskites (MHPs) have drawn significant attention due to their outstanding optoelectronic properties [1] and simple processing routes [2]. As opposed to silicon, MHPs are processable from solution, at low temperature (100°C) and at ambient pressure. These processing routes enable reduced costs and make MHPs good candidates for utility scale photovoltaic power generation [3]. The record power conversion efficiency (PCE) of laboratory-scale MHP solar cells has surpassed polycrystalline silicon based solar cells [4][5][6][7][8][9][10]. The most notable remaining hurdle that must be overcome for these compounds to be viable UPV is their stability to environmental factors [11][12][13]. MHPs and related compounds are known to degrade upon exposure to moisture [14], oxygen [15], ultraviolet radiation [16] and excessive heat [17]. For utility scale photovoltaic power generation to be economically viable, solar cells need to remain operational for >25 years in ambient conditions, and they need to be stable at temperatures ranging from −4°C to 85°C [18,19].
While the environmental stability of MHPs is a widely-studied topic [20], to date the overwhelming majority of such studies are based on electrical measurements of finished PV devices [4,12,21], or structural [22][23][24], or optical [25][26][27] studies. Only a handful of transport [28] stability studies have been carried out on individual thin films. While valuable, the implicit encapsulation provided by electrical contacts and hole-/electron-blocking layers in devices obscures information on the stability of the material itself. Additionally, the electronic properties of interfaces between MHPs and transport layers may not necessarily be representative of the material itself [29]. Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence.
Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
In this report we use the contactless characterisation technique: time resolved microwave conductivity (TRMC) to evaluate the electronic stability of the prototypical MHP compound: methylammonium lead iodide (MAPbI 3 ) to extended periods of elevated temperature. While methylammonium-free [30] and mixed-cation [31] MHPs are known to be more stable to environmental conditions, we here focus on MAPbI 3 as it is the most well-studied and well-understood MHP compound [32]. TRMC allows one to evaluate a proxy for charge carrier mobility: f m f m m S = + ( ) e h of a semiconductor, where f is the fraction of electrons-hole pairs generated per absorbed photon (between 0 and 1), m e is the average electron mobility and m h is the average hole mobility of carriers in the sample. Because the technique does not require electrical contacts or transport layers, it can be carried out on isolated thin films, discontinuous films or even powders. The absence of physical contacts removes contributions from external interfaces [29,33,34], and enables us to make unambiguous statements on the thermal stability of the compound itself. For MHPs such as MAPbI 3 the exciton binding energy is known to be very small [35], and we can approximate f » 1. We can therefore interpret f m S in a similar way to how one would interpret the sum of electron and hole mobilities: The PCE of solar cells is directly related to the mobility of charge carriers in the active material [36,37]; i.e. the ability of photo-generated charges to be extracted from the active layer under normal operation. Thus, evaluating and understanding the electronic stability of MHP compounds, under different conditions, such as exposure to elevated temperature, are crucial for future commercialization efforts [20].

Experimental
2.1. Thin film preparation 2.1.1. Methylammonium lead iodide (MAPbI 3 ) thin films Lead iodide (PbI 2 ), methylammonium iodide (CH 3 NH 3 I) and dimethyl sulfoxide (DMSO) were mixed in a 1:1:1 molar ratio then dissolved in dimethylformamide (DMF). Films were spin-cast under atmospheric-pressure onto quartz substrates and ether was dripped a few seconds before the spin-casting ends. The films were annealed at various temperatures for 96 h using a computer-controlled Dolomite IKA hotplate. Film preparation and annealing was carried out inside an N 2 -filled glovebox. Three samples were prepared for each annealing condition and averaged. Six different annealing conditions were studied, CH 3 NH 3 I was purchased from Greatcell Solar and PbI 2 was purchased from Sigma Aldrich.

Lead iodide (PbI 2 ) thin films
A mixture of lead iodide (PbI 2 ) and DMSO in a 1:1 molar ratio was dissolved in DMF. Films were spin-cast under atmospheric-pressure onto quartz substrate without any antisolvent. Films were then heated at 100°C for 10 min to evaporate the solvent.

Time-resolved microwave conductivity (TRMC)
A schematic representation of the TRMC system employed in this study is shown in figure 1. A microwavefrequency oscillatory electric signal is generated using a Sivers IMA VO4280X/00 voltage-controlled oscillator (VCO). The signal has an approximate power of 16 dBm and a tunable frequency between 8 GHz and 15 GHz. The VCO is powered with an NNS1512 TDK-Lambda constant 12 V power supply, and the output frequency is controlled by a Stahl Electronics BSA-Series voltage source. The oscillatory signal is incident on an antenna inside a WR90 copper-alloy waveguide. The microwaves emitted from the antenna pass through an isolator and an attenuator before they are incident on a circulator (Microwave Communication Laboratory Inc. CSW-3). The circulator acts as a uni-directional device in which signals entering from port 1 exit through port 2 and signals entering from port 2 exit through port 3. The incident microwaves pass through a fixed iris (6.35 mm diameter) into a sample cavity. The cavity supports a TE 103 mode standing wave and consists of an ITO-coated glass window that allows optical access to the sample. The sample is mounted inside the cavity at a maximum of the electric-field component of the standing microwaves, using a 3D-printed PLA sample holder. Microwaves reflected from the cavity are then incident on port 2 of the circulator, exiting through port 3, directed through an isolator, and onto a zero-bias Schottky diode detector (Fairview Microwave SMD0218). The detector outputs a voltage which is linearly proportional to the amplitude of the incident microwaves. The detected voltage signal is amplified by a Femto HAS-X-1-40 high-speed amplifier (gain=×100). The amplified detector voltage is measured as a function of time by a Textronix TDS 3032C digital oscilloscope. A Continuum Minilite II pulsed neodymium-doped yttrium aluminium garnet (Nd:YAG) laser is used to illuminate the sample. The laser pulse has a wavelength of 532 nm, a full width at half-maxima of approximately 5 ns and a maximum fluence incident on the sample of ∼10 15 photons/cm 2 /pulse. An external trigger link is employed to trigger the oscilloscope before the laser fires. The photoconductance was evaluated from changes in the detector voltage using standard analysis [38][39][40]. All measurements were conducted in air, without encapsulation, in the under-coupled regime 2.3. X-ray absorption spectroscopy (XAS) X-ray absorption spectra of the samples were obtained at beamline 5BM-D, Advanced Photon Source, Argonne National Laboratory. The samples were mounted onto adhesive tape and placed in the beam path. Spectra were recorded in the Pb L III edge (13 035 eV) in fluorescence mode, using a Hitachi Vortex ME4 silicon drift detector equipped with four Si elements. Three scans were averaged to obtain data of a satisfactory signal/noise ratio.
The data were processed using the Athena and Artemis programs of the Demeter suite. Extended x-ray absorption fine structure (EXAFS) data were fitted in a k range from 2 to 9 Å −1 and an R range of 2.2 to 3.6 Å, using a first-shell Pb-I scattering path with a degeneracy set equal to 6. The amplitude reduction factor was obtained from a fit of a Pb foil.

Scanning electron microscopy (SEM)
SEM measurements were performed on the samples using a Quanta 600 reduced pressure SEM at the Oregon State University Electron Microscopy Facility. The samples were sputter coated with Au/Pd particles and mounted on sample holders using double sided conductive copper tape. The images were then taken under vacuum at a working distance of 10.5 mm in Secondary Electrons mode. The beam typically had a voltage of 15 kV. Two images of each sample were taken. One image was at a lower magnification than the other. The low magnification images had a horizontal field width (HFW) of around 15 μm and the higher magnification images had a HFW of around 2 μm.

Results and discussion
Thin films of MAPbI 3 and PbI 2 were deposited onto quartz substrates using standard solution-deposition protocols [2,41]. MAPbI 3 films were then annealed for 96 h (4 days) under ambient pressure N 2 , at temperatures of 60°C, 70°C, 80°C, 90°C, 100°C, 110°C, 120°C. Samples annealed at temperatures 110°C and over yielded photoconductance (DG) values below measurable limits. For each compound/annealing condition, three samples were prepared to enable a mean and standard deviation to be evaluated.
A TRMC measurement involves measuring the transient photoconductance (DG) of a semiconductor upon exposure to an applied pulse of light (typically from a ∼ns pulsed laser) [39]. The form of the TRMC transient is a convolution of the sample photoconductance and the instrument response function (IRF) [42,43], in turn determined by the time constant of the microwave cavity: t . RC In our case t » 20 RC ns in all cases. The lifetime as determined by TRMC depends on the carrier concentration, and hence the relative contribution from bimolecular and Auger recombination [44], but values in the range of a few 100 ns are common for polycrystalline MHP films at typical TRMC fluences [38,45].   Figure 2(c) shows an example transient of a film of lead iodide (PbI 2 ), prepared using previously reported solution-deposition protocols [41], and not subjected to extended annealing. PbI 2 is one of the main reported decomposition by-products of MAPbI 3 [46] and should possess some character of a fully-decomposed MAPbI 3 film.
In supporting information section S3 we present scanning electron microscope (SEM) images of films processed under similar conditions. At annealing temperatures of 100°C and above, the coverage of MAPbI 3 films appears to be reduced, suggesting some material is sublimated. The MAPbI 3 precursor methylammonium iodide (MAI) is known [47] to sublime at temperatures as low as 70°C, so this is not unreasonable. The fact that we could not detect a photoconductance for samples annealed at 110°C and over, but could for pristine PbI 2 , suggests that the PbI 2 sheets are likely to be disrupted by other decomposition products, in addition to reduced coverage. Unlike the 3-dimensional perovskite structure formed of corner connected octahedra, PbI 2 is a layered compound, suggesting its in-plane mobility is likely to be substantially different from its out of plane mobility [38,48,49]. Cavity-based TRMC measures the electronic properties of charge carriers in the plane of the sample. It is hence important to emphasize that we are measuring the in-plane mobility of our samples. The Here, b is the ratio of the internal dimensions of the cavity, where the denominator/numerator depend on the polarization of the microwaves. In our case b=2. 25. e is the magnitude of the fundamental unit of charge, F A is the fractional absorption of photons at the excitation wavelength (l=532 nm) measured with ultravioletvisible spectroscopy (data not shown). M is a parameter we define as the 'masking parameter' and is the fraction of the cross-sectional area of the cavity that is exposed to the incident light. In our case M=0.25.
Using equation (1), f m S was extracted as a function of fluence for all measured samples. Figure 3(a) shows the average TRMC figure of merit for three pristine MAPbI 3 films, as a function of incident laser fluence. The error bars represent standard deviations in measured f m S between the three samples. When plotting extracted f m S as a function of fluence, two regimes are commonly observed [44]. At low fluence (10 13 cm −2 ) f m S is observed to be broadly independent of fluence. At higher fluence (10 13 cm −2 ), f m S decreases with increasing fluence. This behaviour is attributed to bimolecular and Auger recombination during the finite duration of the laser pulse, reducing the measurable maximum photoconductance [44]. Average extracted f m S as a function of fluence can be fitted to a model accounting for recombination. The model is described briefly in supplementary information section S2 (available online at stacks.iop.org/JPMATER/3/014003/mmedia), and in more detail elsewhere [44]. Figures 3(b) and (c) show the average f m S as a function of fluence for a MAPbI 3 film that has been annealed for 4 days at 70°C, and pristine PbI 2 , respectively. For this study, the characteristic f m S for each material/annealing condition was taken from these fits. It should be noted that due to its layered two-dimensional nature, the binding energy of PbI 2 is expected to be larger than MAPbI 3 [52], suggesting that f<1 in this case. Since TRMC measures the mobility of carriers in plane only, any significant out of plane orientation in PbI 2 sheets is anticipated to lead to a significant reduction in measured average f m S [48]. While we studied eight temperatures between 50°C to 120°C, we were not able to extract mobility for those above 100°C, as the signal was below measurable noise limits. Figure 4(b) shows photographs of each sample after thermal annealing. Samples annealed at temperatures above 100°C turned yellow within a day while those exposed to temperatures below 80°C remained brown over the course of 96 h. The TRMC data shows a stable f m S up to roughly 70°C, and f m S falls for temperature 80°C. This temperature is in agreement with the phase transition temperature of MAPbI 3 [53,54]. As the temperature increases, the symmetry of crystalline structure increases and transforms to cubic near 350 K and MA cations become disordered [19,20,[55][56][57]. It is however noteworthy that even when the samples are visibly yellow (e.g. 90°C or 100°C) they still exhibit carrier mobilities in excess of those of PbI 2 . This suggests that a strong in-plane percolation pathway exists even as a mixed phase. Figure S12 of the supplementary information shows the XRD spectra of thin-films prepared under identical conditions. We observe that extended annealing at moderate temperatures of 50°C to 80°C leads to mixed PbI 2 and MAPbI 3 phases. Above 90°C, the XRD spectra show more PbI 2 -type character consistent with the photographs in figure 4(b). While at moderate temperatures (70°C) the mobility is relatively unaffected by thermal treatment, sample-to-sample variation is increased (see error bars in figure 3). We interpret this as decrease in sample homogeneity, and increase in complexity of in-plane percolation pathways.
In disordered semiconductors, charges are generally described as being transported via a variable range hopping (VRH) mechanism [58], or a temperature-activated multiple trapping and release (MTR) model [59]. For MHPs however, there are multiple reports of an inverse correlation between mobility and temperature, even  in polycrystalline thin films [45,[60][61][62][63]. These observations cannot be reconciled with the models of VRH or MTR in their current form, and the theory of charge transport in MHPs is at present incomplete.
While the mechanism(s) responsible for charge-transport in this class of materials are not known, it is clear that thin-films are polycrystalline (see SEM images in section S3 for example). Experimental evidence exists that f m S increases with increasing grain size in this class of materials [64]. We therefore hypothesize that an exponential reduction in the average carrier mobility with increasing distance between transport sites should be valid [65][66][67].
In particular, the shortest path that charges can travel in PbI 2 structure is the Pb-I scattering path with R Pb-I =3.23 Å [68]. To probe the average distance between transport sites (Pb to I), we performed XAS measurements on similarly-processed thin films of MAPbI 3 and PbI 2 , subjected to identical thermal treatment. Figure 5 shows the dependence of average first-shell Pb-I distance on the sample annealing conditions. While the near-edge (XANES) part of the spectra is indistinguishable from sample to sample (see supporting information figure S13), suggesting no metallic Pb formation, there are differences in the EXAFS part of the spectrum, suggesting differences in bonding. More specifically, as the annealing temperature increases, the first-shell Pb-I distance increases from 3.119 Å in the case of MAPbI 3 to 3.133 Å for the sample annealed at 70°C, all the way to 3.147 Å for PbI 2 . This increase in the nearest neighbour distance is interpreted as the reason behind the f m S extracted from TRMC measurements.

Conclusions
In summary, we have studied the effect of extended periods (96 h) of elevated temperature on the electronic properties of solution processed MAPbI 3 thin films using time-resolved microwave conductivity (TRMC). Because we have not used electrical contacts or hole-/electron-blocking layers we were able to make statements on the effect of the material itself without ambiguity due to interfaces. The TRMC figure of merit (f m S ), a proxy for average electron and hole carrier mobility, was measured as a function of annealing temperature for films of MAPbI 3 subjected to 96 h of thermal treatment. Our results show that as temperature increases, f m S values approach, but do not reach, that of PbI 2 . The mobility is stable up to 70°C, below a known structural phase transition temperature in MAPbI 3 . A decrease in mobility is observed at annealing temperatures above 80°C. Samples annealed at 110°C and 120°C were below measurable limits, yet pristine PbI 2 was measurable. From x-ray absorption spectroscopy (XAS) data, we observed the impact of extended thermal treatment on intertransport-site distance. An increase in the Pb-I distance with temperature was observed. We hence conclude that the reduction in f m S is due to an increase in average charge transport distance.