Letter pubs.acs.org/JPCL
Impedance Spectroscopic Indication for Solid State Electrochemical Reaction in (CH3NH3)PbI3 Films Arava Zohar,† Nir Kedem,† Igal Levine, Dorin Zohar, Ayelet Vilan, David Ehre, Gary Hodes, and David Cahen* Department of Materials and Interfaces, Weizmann Institute of Science, Rehovot 76100, Israel S Supporting Information *
ABSTRACT: Halide perovskite-based solar cells still have limited reproducibility, stability, and incomplete understanding of how they work. We track electronic processes in [CH3NH3]PbI3(Cl) (“perovskite”) films in vacuo, and in N2, air, and O2, using impedance spectroscopy (IS), contact potential difference, and surface photovoltage measurements, providing direct evidence for perovskite sensitivity to the ambient environment. Two major characteristics of the perovskite IS response change with ambient environment, viz. -1appearance of negative capacitance in vacuo or post-vacuo N2 exposure, indicating for the first time an electrochemical process in the perovskite, and -2- orders of magnitude decrease in the film resistance upon transferring the film from O2-rich ambient atmosphere to vacuum. The same change in ambient conditions also results in a 0.5 V decrease in the material work function. We suggest that facile adsorption of oxygen onto the film dedopes it from n-type toward intrinsic. These effects influence any material characterization, i.e., results may be ambient-dependent due to changes in the material’s electrical properties and electrochemical reactivity, which can also affect material stability. ithin just a few years, solar cells based on thin films of hybrid organic−inorganic halide perovskites have reached impressive efficiencies1,2 and show potential for a technology complementary to silicon. The perovskite materials that form the basis of these cells are composed of earthabundant and readily available elements and can be produced using low-cost processes, and with low defect densities.3,4 The materials have high optical absorption coefficients;5 a ∼ 0.3 μm thick layer of the most studied member of this family, CH3NH3PbI3, absorbs 90% of the solar spectrum above the ∼1.55−1.6 eV absorption edge (∼40% of the total solar spectrum). While the efficiency/reproducibility/size of devices that contain these films continue to improve as a result of optimization efforts, understanding of the chemistry and physics of these perovskites, including charge transport mechanism(s) and chemical (in)stability, is far from complete. Understanding the electronic transport and doping mechanisms and the nature of the dopants are critical for improving the performance (efficiency and stability) of perovskite solar cells. A common hypothesis pertaining to the devices’ electrical transport is that it involves mixed ionic and electronic conduction within the perovskite lattice. There is, however, no agreement as of yet on this issue and especially on the identity of the ion(s) involved. Because of vastly different time constants for transport, the frequency response of the two types of transport mechanisms should be well resolved using frequency-dependent measurements. The ability of impedance spectroscopy (IS) to cover a wide range of frequencies allows distinguishing between various conduction processes. A model of the electronic transport frequency dependence can be fitted
W
© XXXX American Chemical Society
to three regimes, each representing different physical processes with a distinct time constant. The high frequency response (>10s of kHz) can be attributed to a high density of free carriers, such as occur in a highly doped semiconductor. At intermediate frequencies (100 Hz to 10 kHz), the material’s electronic (electron/hole) transport dominates the IS response as well as carrier capture by, and release from, shallow defect states. The low-frequency response is commonly attributed to carrier capture by, and release from, deep traps and ionic transport,6 with distinct features of the latter in the impedance spectrum. In certain systems, ionic conduction can be accompanied by an electrochemical reaction,7−9 depending on the mobile ion species and the reaction’s activation energy. IS data are commonly presented in the form of the real and imaginary impedance (Zre, Zim), corresponding to processes in and out of phase with the frequency modulated driving signal (commonly referred to as AC bias). IS analyses of halide perovskite-based solar (photovoltaic, PV) cells were previously reported for a number of PV device configurations with different AC and DC biases.10−14 In all those reports, similar spectra for full devices are presented; when illuminated, all yield an RC (R = resistance; C = capacitance) semicircle for the high frequency region and an additional arc in the low frequency range. The low-frequency response was attributed to various effects, such as frequency-induced changes in the resistance of the electron/hole transport layer,13 ion transport11,12 and Received: November 24, 2015 Accepted: December 19, 2015
191
DOI: 10.1021/acs.jpclett.5b02618 J. Phys. Chem. Lett. 2016, 7, 191−197
Letter
The Journal of Physical Chemistry Letters trapping/detrapping.14 IS of perovskite-based solar cells, subjected to various environments, revealed that their electronic properties are influenced by the cell’s storage conditions.15 By correlating IS measurement and current− voltage (I−V) performance, it was found that dry air is the most suitable storage condition for retaining high photovoltaic energy conversion efficiency. In situ environmental conditions, such as those used for XPS (i.e., UHV), PL, optical absorption or I−V at low temperatures (high vacuum), in turn, affect the electrical properties of the material and may influence, or even dictate, the results. A downside of IS analysis is that it can easily be affected by extraneous factors. When measuring an electrically rectifying interface, for example, where the current response is not linear, additional impedance will be introduced. As modeling of the IS response relies on linear response, this impedance interpretation will not represent real processes in the sample. A similar impediment occurs if the voltage drops dominantly over (one of) the interfaces, in which case the impedance data will reflect only the interface(s) and not the bulk properties, as described by, for example, Juarez-Perez et al.13 Thus, study of basic material properties by IS is preferably done using the simplest system, which will require minimal modeling for the data. Studies of electronic properties such as electron mobility and diffusion coefficient in single-crystal16,4 and pellet17 samples of MAPbI3(Cl) were reported recently. An exceptionally low trap-state density (1010 cm−3) and high charge carrier diffusion length (10 μm)4 was found in MAPbI3 single crystals, and a redox reaction was suggested to occur at the interface between an MAPbI3 pellet and a Pb electrode, involving a remarkably high chemical diffusion coefficient at room temperature (2.4 × 10−8 cm2 s−1) for the I− ion. We focus here on MAPbI3(Cl) films, deposited in a manner, similar to what is commonly used for PV devices, because morphology, grain size, and thickness can play an important role in the films’ charge transport properties. In addition, we use symmetric, nonrectifying, contacts for our measurements to allow us, in principle, to probe the materials properties without interference from interfaces with electronically active materials. We chose a lateral transport configuration to allow in situ (i.e., during the measurements) exposure of the material to various atmospheres. We find positive imaginary impedance (negative capacitance) during IS measurements after MAPbI3(Cl) films were subjected to vacuum, bias, and illumination in O2-free environment. This feature has been observed in other systems such as a solid-state dye-sensitized solar cells, fuel cells, Schottky diodes, and polymer light-emitting diodes.18 Even though different physical interpretations for negative capacitance were given, the end result is the same: at low frequency additional carriers become available for transport. Negative capacitance indicates that the current in the device lags behind the alternating driving voltage. Possible explanations for this phenomenon include carrier accumulation at the electrode interface, transport via a quantum well, ion/vacancy drift and accumulation, and an electrochemical reaction. Carrier accumulation requires the system to have a high density of states at the interface between the electrode and semiconductor and that the states’ occupation decreases with increasing applied potential. To achieve that, the system needs to be far from thermal equilibrium (high bias) and transport between the intermediate layers is governed by the occupation of the states. Accumulation of carriers is then
explained as a kinetic process in which highly unsymmetrical filling/discharging of the states take place.19,20,18 Ion/vacancy accumulation at a rectifying contact can lead to a change in barrier height, width, or both, such as in the case of a Schottky diode with a mixed ionic-electronic conductor or a metal−insulator−metal capacitor with high ionic conductivity. In the former case, at low frequency, carrier injection into and from the electrode will depend on the ion/vacancy concentration. The occurrence of negative capacitance is then the result of more efficient charge transport with increasing accumulation at low frequency.21 An electrochemical reaction,7,8 where electron transfer is involved in the reaction, i.e., a redox reaction, will also show negative capacitance in the IS spectrum. Here, an additional charge is generated during reaction, decreasing the resistance of the material. The amount of generated charge will depend on the reactants’ mobilities and applied frequency. As the frequency decreases, more carriers are generated. The IS response of fuel cells, for example, in which an electrochemical reaction occurs, as well as preceding ion diffusion and electron transport to the electrodes after the reaction takes place, is normally modeled by negative capacitance.7 We find that occurrence of negative capacitance in the halide perovskite layers depends on the environment in which the measurement is done. More specifically, in an O2-free environment, the lateral resistance of the film decreases along with the appearance of the negative capacitance. The functionality of the (relatively large area) layers was tested in separate, different projects, in FTO/TiO2 /MAPbI3/spiro/Au solar cells, with reasonable average efficiencies (∼8.5% Figure S9). Kelvin-probe contact potential difference (CPD) and surface photovoltage (SPV) measurements were performed on the films, in addition to IS, to assess the contribution of oxygen to the material’s work function and its change upon illumination. O2 was found to dedope MAPbI3(Cl) from highly n-type to nearly intrinsic, compared to what is the case for samples exposed to vacuum or N2 after vacuum. We interpret a shift in the measured surface potential of O2-exposed samples upon illumination as an increase in minority carrier lifetime, which results in selective capturing of holes at the HOPG back contact. Results of CPD and SPV measurements, using a Kelvinprobe, of MAPbI3(Cl) deposited on highly oriented pyrolytic graphite (HOPG) under various environmental conditions are presented in Table 1. Measurements in air were performed in a relative humidity (RH) of typically ∼60 (±10)%. The vacuum environment corresponded to a residual pressure of 10−3 mbar. The N2 used was purified to
