Abstract
Impressive performance of hybrid perovskite solar cells reported in recent years still awaits a comprehensive understanding of its microscopic origins. In this work, the intrinsic Hall mobility and photocarrier recombination coefficient are directly measured in these materials in steadystate transport studies. The results show that electronhole recombination and carrier trapping rates in hybrid perovskites are very low. The bimolecular recombination coefficient (10^{−11} to 10^{−10} cm^{3} s^{−1}) is found to be on par with that in the best directband inorganic semiconductors, even though the intrinsic Hall mobility in hybrid perovskites is considerably lower (up to 60 cm^{2} V^{−1} s^{−1}). Measured here, steadystate carrier lifetimes (of up to 3 ms) and diffusion lengths (as long as 650 μm) are significantly longer than those in highpurity crystalline inorganic semiconductors. We suggest that these experimental findings are consistent with the polaronic nature of charge carriers, resulting from an interaction of charges with methylammonium dipoles.
Introduction
Hybrid (organic–inorganic) perovskite solar cells represent the recent breakthrough in photovoltaic applications with reported power conversion efficiencies reaching 20% (refs 1, 2, 3). In addition to the abundance of the applied studies on this topic, there is a great interest in understanding the fundamental transport and photophysical properties of these materials, the picture of which is not firmly established yet, thereby calling for more reliable experimental studies. For instance, a possibility of high chargecarrier mobility has been considered as one important factor that contributes to the excellent photovoltaic performance of hybrid leadhalide perovskites. Yet, an unambiguous determination of the intrinsic mobility in these materials is missing. Existing experimental values in similar materials range from 0.6 to 50 cm^{2} V^{−1} s^{−1} (refs 4, 5, 6). However, most of the important transport and photophysical parameters, including the carrier mobilities, lifetimes and recombination rates, were so far determined either in materials different from those relevant for highperformance solar cells (for instance, in metallic tinhalide instead of insulating leadhalide perovskites, or perovskites that adopt 2D layered rather than 3D cubic structure) or obtained indirectly, under the conditions less relevant to applications (for instance, in ultrafast spectroscopic experiments, rather than steadystate transport measurements).
Here, we report artefactcorrected Hall effect and steadystate photoconductivity measurements carried out in a range of thin films and single crystals of exemplary hybrid perovskites, CH_{3}NH_{3}PbI_{3} and CH_{3}NH_{3}PbBr_{3}, of current interest for photovoltaic applications. Hall effect allows us to directly and independently address the density of photogenerated carriers, n_{Hall}, and the intrinsic carrier mobility, μ_{Hall}, without assumptions typical for other methods, such as in ultrafast spectroscopic techniques or spacechargelimited current measurements. We find that, in a wide range of illumination intensities, the dynamics of photocarriers is governed by bimolecular electronhole (eh) recombination with a very small recombination coefficient γ in the range of 10^{−11} to 10^{−10} cm^{3} s^{−1}, which is comparable to the values observed in the best singlecrystalline directband inorganic semiconductors, such as GaAs, even though the measured intrinsic Hall mobilities are moderate (μ_{Hall} of up to 60 cm^{2} V^{−1} s^{−1} in perovskite single crystals) and smaller than μ in typical inorganic semiconductors by 1–3 orders of magnitude. In addition, the carrier lifetime, τ, and diffusion length, l, directly measured in our steadystate transport experiments are found to be remarkably long (τ is up to 30 μs and l is up to 23 μm in polycrystalline films, and up to 3 ms and 650 μm in single crystals, respectively). Our experiment thus provides a direct steadystate measurement quantitatively revealing a lowrate photocarrier recombination and negligible trapping, as well as extremely long carrier lifetimes and diffusion lengths in hybrid perovskites. While in agreement with some of the recent theoretical predictions^{7,8,9,10}, these results accentuate important questions as of the physical origins of the found intrinsic carrier mobility, eh recombination and trapping rates in these materials synthesized via inexpensive vapour or solutionbased routes at temperatures close to room temperature. We propose rationalization of some of our findings based on the picture of reorganization of the methylammonium dipoles around the charge carriers. This interaction leads to carrier relaxation into polarons, whose properties differ from the bare band carriers^{11,12,13}.
Results
a.c. Hall effect measurements of chargecarrier mobility
It should be emphasized that unambiguous determination of the intrinsic (that is, not dominated by traps) chargecarrier mobility requires precise Hall effect measurements, which are quite challenging in highly resistive materials with relatively low μ, such as organic semiconductors or the hybrid perovskites studied here (see, for example, ref. 14). The major challenges are associated with a very high resistivity of pure stoichiometric perovskites (greater than GΩ), related to the negligible (in the dark) density of charge carriers, and a poor signaltonoise ratio in conventional d.c. Hall measurements of lowμ materials. To overcome these difficulties, here, we have developed a specialized highly sensitive a.c. Hall measurement technique, corrected for the Faradayinduction artefacts, in which a lowfrequency a.c. magnetic field, B, is applied perpendicular to the sample’s surface, while a d.c. current, I, is passed through the sample, and an a.c. Hall voltage, V_{Hall}, is detected across the channel by a phasesensitive lockin technique, which allows to markedly increase the signaltonoise ratio^{15}. A parasitic Faradayinduction electromotive force, occurring in a.c. Hall measurements at the same frequency as V_{Hall}, is usually comparable to the actual Hall voltage signal and can easily compromise these measurements. Therefore, Faradayinductioncorrected a.c. Hall measurements, as implemented here in perovskites, are absolutely necessary to obtain reliable data (see the ‘Methods’ section, Supplementary Fig. 1 and Supplementary Note 1)^{15}. Pure CH_{3}NH_{3}PbI_{3} samples are highly resistive in the dark (typical R100 GΩ), and thus we utilize a steadystate monochromatic photoexcitation (λ=465 nm) to generate a population of carriers and be able to measure Hall effect. In CH_{3}NH_{3}PbBr_{3} single crystals, we were able to measure a Hall effect in the dark as well, because these crystals are weakly conducting in the dark (at a level of 50 MΩ). We perform all our measurements in a 4probe/Hall bar geometry to account for contactresistance effects and ensure that channel conductivity, σ, as well as the Hall mobility and carrier density, μ_{Hall} and n_{Hall}, are determined correctly.
Steadystate photoconductivity vs. light intensity
Three types of hybrid perovskite samples were used in our study (Fig. 1). (a) Semi amorphous, solutiongrown thin films, (b) polycrystalline, vapourgrown thin films and (c) highly ordered, solutiongrown single crystals. The microscopy and Xray diffraction clearly show that the vapourprocessed films have a much better crystallinity than the solutionprocessed ones, and our single crystals have excellent quality (see Fig. 1 and see the ‘Methods’ section for details).
We have found that all our highquality (stoichiometric) samples exhibit a very low dark conductivity. Nevertheless, a significant photoconductivity is observed in all of them. Figure 2 shows typical dependences of a steadystate photoconductivity, σ_{PC}, on photoexcitation density, G, which always follows a power law, σ_{PC}∝G^{α}, with the exponent α=1 or ½ (linear or squareroot regimes). Photoexcitation density, G, is defined as the incident photon flux F (in cm^{−2} s^{−1}) divided by the absorption length of the material.
The observed σ_{PC}(G) dependence can be understood in terms of chargecarrier monomolecular decay (trapping in the linear regime) or bimolecular decay (eh recombination in the squareroot regime). The photocarrier density, n (n≡n_{e}≈n_{h}), in steadystate measurements is found from the following rate equation, with dn/dt=0 at dynamic equilibrium (see, for example,^{16}):
Here, κG is the rate of carrier generation via photon absorption resulting in production of free electrons and holes with probability κ per photon (the photocarriergeneration efficiency). The second and third terms represent the two channels of carrier decay: the trapping and eh recombination, where τ_{tr} is the traplimited carrier lifetime (an average time carriers diffuse before being trapped), and γ is the coefficient of eh recombination. In the carrier density range probed here, we do not see any experimental evidence of the thirdorder (Auger) processes, which are therefore excluded from equation (1). At low photoexcitation intensities, when the concentration of electrons and holes is small, the dominant process limiting the carrier lifetime is trapping, and γn^{2} term in equation (1) can be neglected, leading to a linear regime in photoconductivity: σ_{PC}≡eμn=eμκτ_{tr}G, where e is the elementary charge. With increasing excitation intensity, the bimolecular recombination eventually becomes dominant, resulting in a transition from the linear to a sublinear regime: σ_{PC}=eμ·(κ/γ)^{1/2}·G^{1/2}, obtained from equation (1) by neglecting the n/τ_{tr} term. Figure 2 shows that these two regimes are indeed what is observed in hybrid perovskites. Highly crystalline samples exhibit bimolecular recombination regime (α=1/2) in a wider range of excitation intensities, which is consistent with the higher σ_{PC} and μ_{Hall} in the crystal samples (see below).
Hall measurements under photoexcitation and in the dark
To decouple the carrier density and mobility in σ_{PC}=enμ and obtain the microscopic parameters describing the carrier dynamics, τ_{tr} and γ, from the rate equation for n (equation (1)) and experimental data, one needs to know the intrinsic chargecarrier mobility μ. For this purpose, we have performed Hall effect measurements, as discussed above, using 4probe/Hallbar device structures (Fig. 3a) and a variant of an a.c. Hall measurement technique (Fig. 3b) specifically adjusted for these materials (see the ‘Methods’ section, Supplementary Fig. 1 and Supplementary Note 1). A typical measurement result is shown in Fig. 3c: a very clear and quiet a.c. Hall signal as detected in a solutiongrown CH_{3}NH_{3}PbI_{3} thin film that has a Hall mobility of only μ_{Hall}=1.5 cm^{2} V^{−1} s^{−1}. Reliable Hall measurements with such an excellent signaltonoise ratio in highly resistive systems with carrier mobilities as low as 1 cm^{2} V^{−1} s^{−1} are unprecedented. For comparison, conventional d.c. Hall measurements performed in a perovskite single crystal with a much higher mobility, μ_{Hall}=11 cm^{2} V^{−1} s^{−1}, shown in Fig. 3d, evidently exhibit a much noisier signal. It is clear that Faradayinductioncorrected a.c. Hall measurements are by far superior to the conventional d.c. technique in terms of the signaltonoise ratio, even though it uses a smaller magnetic field (r.m.s. B=0.23 T).
As expected for pure undoped band insulators, our CH_{3}NH_{3}PbI_{3} samples are highly resistive in the dark (with a typical sample resistance >100 GΩ). Thus, we used a cw photoexcitation to generate photocarriers and perform steadystate photo a.c. Hall measurements. In a system with photogenerated electrons and holes (n_{e}≈n_{h}≡n) and a negligible concentration of dark carriers, Hall voltage is given by:
where W and L are the channel's width and length, respectively, V_{L} is the longitudinal voltage drop along the channel (corrected for contact effects by using the 4probe technique), and μ_{h}, μ_{e} are the mobilities of holes and electrons, respectively. Equation (2) shows that photo Hall effect measurements yield the difference between the electron and hole mobilities, rather than their absolute values. In the context of the perovskites under study, recent calculations showed that μ_{e} and μ_{h} in these materials should differ from each other by an amount comparable to the mobilities themselves^{7,17,18,19,20}. Indeed, theoretical calculations have consistently indicated that while effective masses of electrons, m_{e}, and holes, m_{h}, have the same order of magnitude, there is also a noticeable difference between the two, ranging from 20 to 200%, depending on the specific computational method used^{18}. A recent THz spectroscopy study also suggests a difference of a factor of 2 between the electron and hole mobilities^{21}.
Therefore, even though precise values of μ_{e} and μ_{h} cannot be obtained from photo Hall measurements, the difference, Δμ≡μ_{h}—μ_{e}∼μ, extracted from equation (2) would yield a faithful representation. In turn, the carrier density obtained from the photo Hall effect measurements is: n=n_{e}+n_{h}=σ_{PC}/(e·μ), which is also a good approximation for the actual density of photogenerated charges. The data presented below are analysed using this association. The dark Hall effect measurements possible in weakly conducting singlecrystal CH_{3}NH_{3}PbBr_{3} samples give further credence to the approach.
Figure 4 presents Hall effect data for a variety of hybrid perovskite samples. Fig. 4a shows that Hall mobility, μ_{Hall}=8±0.4 cm^{2} V^{−1} s^{−1}, measured in a polycrystalline CH_{3}NH_{3}PbI_{3} film remains almost constant over the range of nearly three orders of magnitude in light intensity (the error is defined by the fluctuations in Fig. 4a). The 4probe photoconductivity, σ_{PC}, and the density of photogenerated charge carriers, n_{Hall}, determined from the simultaneous longitudinal σ_{PC} and Hall effect measurements in these polycrystalline films are plotted in Fig. 4b. Within the entire measurement range, the carrier density dependence on the illumination intensity exhibits α=½ power law: n_{Hall}∝G^{1/2}. We emphasize that independent measurements of μ_{Hall} as a function of photoexcitation density are essential for obtaining n_{Hall}(G) dependence and therefore for determination of the microscopic transport parameters, τ_{tr} and γ, by using equation (1) to fit the data. With the density of photogenerated carriers determined experimentally, we can now interpret their dynamics with the help of equation (1). As pointed out above, under a steadystate photoexcitation, equation (1) gives n=(κ/γ)^{1/2}·G^{1/2} in the regime governed by a bimolecular recombination, that is, when the carrier trapping time τ_{tr} well exceeds the time of eh recombination τ_{r}, τ_{tr}>>τ_{r}≡(γn)^{−1}. Fitting the experimental Hall carrier density in Fig. 4b with this n(G) relationship yields the upper bound for the bimolecular eh recombination coefficient γ in polycrystalline perovskite films, γ≤3 × 10^{−11} cm^{3} s^{−1} (for the photocarriergeneration efficiency κ≤100%).
The measurements shown in Fig. 4 can also be used to directly estimate the steadystate chargecarrier lifetime, either τ_{tr} or τ_{r}, and diffusion length, l. Indeed, since bimolecular recombination (α=½) apparently dominates in the entire range of photon densities in Fig. 4b, we must have the condition n/τ_{tr}<<γn^{2} (or, τ_{tr}>>τ_{r}≡(γn)^{−1}) fulfilled for all the incident photon intensities, including the lowest one, at which n_{Hall}=9 × 10^{14} cm^{−3} (Fig. 4b). Therefore, the effective traplimited carrier lifetime τ_{tr} in the disordered polycrystalline film in Fig. 4b must be longer than the lifetime limited by bimolecular eh recombination, τ_{tr}>τ_{r}≈30 μs. This estimate shows that even in disordered CH_{3}NH_{3}PbI_{3} thin films, charge carriers have an extremely long lifetime, as far as trapping is concerned, and thus the effective density of the corresponding deep traps must be very low, or the traps must be electronically passivated. The corresponding lower bound of traplimited carrier diffusion lengths in these polycrystalline perovskite films is: l=(Dτ_{tr})^{1/2}∼23 μm. Here, we needed the Hall mobility again to calculate the diffusion coefficient, D∼μ_{Hall}k_{B}T/e, where k_{B} is the Boltzmann constant and T=300 K. Note that negligible trapping and a very long diffusion length, greater than the grain size in our polycrystalline films (Fig. 1b), are consistent with the recent theoretical studies predicting the presence of only shallow traps and benign grain boundaries that do not trap carriers in perovskites^{9,10}. Of course, given the dominant eh recombination, the actual carrier lifetime in the α=½ regime is a decreasing function of photoexcitation density G: τ_{r}=(γn)^{−1} (see also^{16}).
Figure 4c shows n_{Hall} and μ_{Hall} in six different solution and vapourgrown polycrystalline thin films. In general, vapourgrown samples have an appreciably higher μ_{Hall}, consistent with their better crystallinity. In sharp contrast, n_{Hall}(G) does not seem to correlate much with the preparation method and mobility. Indeed, while μ_{Hall} differs among the six samples by as much as a factor of 20, n_{Hall} varies only within a factor of two. Correspondingly, γ values are also very similar for all these samples, γ∼(1–5) × 10^{−11} cm^{3} s^{−1}. This suggests that while film morphology has a clear effect on charge transport, it has little effect on photocarrier generation and recombination. This is consistent with the notion that bimolecular recombination in this system is not governed by carrier diffusion, therefore we do not see a strong correlation between the recombination dynamics and chargecarrier mobility.
Finally, we have performed Hall measurements in CH_{3}NH_{3}PbBr_{3} single crystals (Fig. 4d). One important difference in this case is that these crystals are weakly conducting in the dark, which allows us to perform Hall effect measurements in the dark and obtain the mobility of holes (μ_{Hall}=60±5 cm^{2} V^{−1} s^{−1}), without having an ambiguity of charge compensation as in photo Hall measurements. Photoconductivity in these crystals is much higher than that in thin films and also exhibits a welldefined α=½ behaviour (Fig. 4d). A similar analysis of n_{Hall}(G) dependence shows that the eh recombination coefficient in these crystals is γ∼8 × 10^{−11} cm^{3} s^{−1} (for more details on the extraction procedure see Supplementary Note 2). By defining the effective recombinationlimited carrier lifetime again as τ_{r}=(γn)^{−1} and using the experimental carrier density from Hall effect measurements, we can determine the carrier lifetime and diffusion length in these single crystals. At the lowest incident photoexcitation flux in Fig. 4d, corresponding to the measured projected carrier density n_{Hall}=3 × 10^{11} cm^{−2} and the effective bulk carrier density n∼4.6 × 10^{12} cm^{−3} (Supplementary Note 2), we find: τ_{r}∼2.7 ms, and l≡(Dτ_{r})^{1/2}∼650 μm, which are remarkably long for a solutiongrown semiconductor. We emphasize that these values represent the lower limit for the traplimited carrier lifetime, τ_{tr}, and diffusion length, l_{tr}, since the condition τ_{tr}>>τ_{r} must be satisfied in the entire regime dominated by a bimolecular recombination, thus indicating again that trapping is strongly suppressed in these materials. We must add that σ_{PC}(G) in analogous lead iodide (CH_{3}NH_{3}PbI_{3}) single crystals (not shown here) are qualitatively similar, except that these crystals are highly insulating in the dark, and thus only photo Hall measurements were possible, yielding an estimate for μ∼few cm^{2} V^{−1} s^{−1}.
Theoretical estimates of eh recombination coefficients
Eh recombination in semiconductors is a fundamentally important process, and various approaches have been developed for assessing the corresponding kinetic coefficient γ. In one approach, for instance, γ is associated with the product sv of the Coulomb capture cross section s and carrier thermal velocity v (ref. 16). In the case of disordered organic and inorganic semiconductors, recombination of charge carriers is often described by the Langevin model^{22,23}, which leads to for the recombination coefficient, where ɛ_{0} and ɛ_{r} are the dielectric permittivities of vacuum and the material, respectively. Evaluating these expressions for our systems would yield estimates of γ on the order of 10^{−6} cm^{3} s^{−1} or higher, that is 4–5 orders of magnitudes greater than what we find experimentally. One, of course, realizes that the above models, evidently not applicable to our case, refer to the eh collision events rather than to the radiative recombination per se. A more appropriate approach that was successfully applied to the actual radiative recombination in inorganic semiconductors is based on the van Roosbroeck–Shockley’s theory rooted in the principle of detailed balance (for reviews, see, for instance, refs 24, 25). This theory, in particular, explains well the values of γ∼10^{−10} cm^{3} s^{−1} exhibited by highpurity directband inorganic semiconductors, such as, for instance, GaAs^{26,27} (see also http://www.ioffe.ru/SVA/NSM/Semicond/index.html), which, remarkably, are comparable to the values we extract from our observations in hybrid perovskites. In the van Roosbroeck–Shockley’s approach, the radiative recombination coefficient γ is established by the system’s properties in the thermal equilibrium (that is, in the dark):
where, R_{eq} and n_{eq} are the equilibrium (dark) recombination rate and concentration of electrons (equal to that of holes), respectively. Furthermore, R_{eq} is related thermodynamically to the optical properties of the system as^{25}:
utilizing the frequency ωdependent refraction, n_{r}(ω), and extinction, κ(ω), coefficients. An assessment of equation (3) can be made using model considerations (as illustrated in Supplementary Note 3). Even more attractively, one can use the actual experimental optical data to evaluate R_{eq} in equation (4), for which here we use the optical parameters (see Supplementary Fig. 2 and Supplementary Note 3) of CH_{3}NH_{3}PbI_{3} perovskite extracted from accurate ellipsometric measurements in ref. 28. On the other hand, the dark carrier concentration at the thermal equilibrium n_{eq} is not measured directly. If one were to use the standard textbook expression for a nondegenerate semiconductor with energy gap E_{g} separating two parabolic bands^{27}:
equation (3) then yields the recombination coefficient:
Here, the prefactor was calculated with T=300 K and m_{e}m_{h}=m=0.2m_{0} (m_{0} being the free electron mass). Equation (6) features the absorption onset parameter denoted E_{opt}, which is equal to 1.553 eV in the parameterization of ref. 28. The excitonic (eh attraction) effects^{11,25} are known to reduce the onset of optical absorption in comparison with the semiconductor bandgap in perovskites (see, for example, the discussion of excitons in ref. 29). One can approximately assess the resulting exponential factor in equation (6) from the corresponding exciton binding energy E_{X}. For m=0.2m_{0} and the relative permittivity equal to 5, for instance, E_{X}54 meV, indicating that the exponential factor is of the order of 10, which would make result (6) for γ larger than our experimental values by about one order of magnitude. The discrepancy here might result from the underestimate of n_{eq} by equation (5) for the conventional band carriers, and the larger values of n_{eq} would lead to a better agreement with the experiment. We suggest that larger concentrations n_{eq} might be actually realized in perovskites due to the interaction of band charge carriers with methylammonium dipoles, as illustrated in Fig. 5.
Accounting for polaronic effects
Theoretical calculations by Frost et al.^{7} demonstrate that the methylammonium dipoles in hybrid perovskites carry a significant dipole moment, P=2.29 D, and create a rough potential landscape at the nanoscale, but can be easily rotated or locally aligned by overcoming a small rotational energy barrier, U_{rot}≈1 kJ mol^{−1} (1.6 × 10^{−21} J or 10 meV per dipole). Here, we propose that a charge carrier moving through the perovskite lattice may itself induce a local orientational rearrangement of the surrounding methylammonium dipoles tending to align them along its electric field (Fig. 5), thus resulting in a type of a dipolar polaron, conceptually similar to polarons known in ionic crystals and polar semiconductors^{11,12,13}. The estimates outlined in Supplementary Note 3 show that such polarons in perovskites should be characterized as intermediatecoupling polarons^{12}, with the dimensionless electron–phonon coupling constant α_{e−ph}2.5 being a good representative value for the interaction of the band carriers with the longitudinal dipolar vibrational modes of energy ħω_{0}∼10 meV. The ‘dressing’ of a band carrier by a phonon cloud is known to change its properties^{11,12,13}: the standard band carrier dispersion E(k)=ħ^{2}k^{2}/2m would be modified to the polaronic energymomentum relation E_{p}(k). Two aspects of this modification are important here.
First, polaron formation is energetically favourable, resulting in the polaronic energy shift E_{p}(0)−α_{e−ph}ħω_{0}. The effective bandgap for the equilibrium concentration of electron and holepolarons is thus reduced: . With the estimates above, this reduction largely negates the effect of the exciton binding E_{X} and substantially decreases the exponential factor in equation (6). (One could also say that the thermal dissociation of an exciton into a polaron pair is more efficient than into a pair of band carriers.)
Second, polarons are ‘heavier’ than the band carriers. While this is qualitatively clear already at the level of the effective mass renormalization^{12}: m→m_{p}m(1+α_{e−ph}/6), the actual changes are even more significant, as the polaron dispersion becomes nonparabolic (see Fig. 6 with the accompanying caption, Supplementary Figs 3,4 and Supplementary Note 3)^{30,31,32}. This results in the increased density of the polaronic states and corresponding increase in the equilibrium carrier concentration n_{eq}. The data in Fig. 6 show that this effect can be substantial: in this illustration, the volume of phase space available for polarons with energies E_{p}(k)−E_{p}(0)≤ħω_{0} increases approximately by a factor of 1.54^{3}3.7 relative to the bare band carriers (compare the solid red and dashed black lines). Thereby the denominator in equation (3) could additionally increase by about an order of magnitude.
Discussion
While currently there is no analytical framework that would afford quantitatively reliable calculations of the interplay of excitonic and polaronic effects in the relevant parameter range (all energy parameters: k_{B}T, ħω_{0}, polaronic and excitonic bindings, are of the same order of magnitude), the above estimates and analysis demonstrate that the polaronic effects might provide an explanation that brings the experimental optical data and our measurements of the radiative recombination coefficient γ in a reasonably good agreement with each other. Indeed, the experimental estimates for γ derived above from the Hall effect measurements span the range of (1–8) × 10^{−11} cm^{3} s^{−1}. On the other hand, the polaronic effects we discussed evidently lead to a substantial decrease of the value in equation (6) evaluated without such effects and likely reducing this estimate below 10^{−10} cm^{3} s^{−1}. Given the fact that photo Hall measurements could precisely address only the difference of the hole and electron mobilities, and some uncertainty in theoretical estimates, the consistency of our results appears quite satisfactory. In addition, we note that the nonparabolic polaron dispersion displayed in Fig. 6 clearly features carrier group velocities v_{gr}(k)=E_{p}(k)/ħk considerably lower than those, ħk/m, of the bare band carriers. This, of course, leads to lower carrier mobilities as indeed observed in our experiments.
Another important experimental observation is that we have not found the monomolecular decay regime (trapping) down to rather low carrier densities, n≈10^{15} cm^{−3} in thin films, and 5 × 10^{12} cm^{−3} in single crystals (Fig. 4). If present, such a regime would manifest itself as a linear σ_{PC}(G) dependence (α=1). The absence of trapping is exactly the reason why bimolecular eh recombination dominates down to such a low carrier density, leading to a remarkably long carrier lifetime and diffusion length at diluted photoexcitation densities, consistent with prior observations^{33,34}. In directband inorganic semiconductors, even though similarly small values of γ=10^{−11}—10^{−10} cm^{3} s^{−1} are observed in highpurity crystalline samples, achieving a millisecond carrier lifetime and nearly a millimetrelong diffusion length is unheard of. Ordinarily, other recombination mechanisms such as trapping on recombination centres (or even Auger processes) would start to dominate at a higher crossover carrier density, given the rather small e–h recombination probability. In fact, in welloptimized highpurity single crystals of GaAs, InP or InAs, τ and l are a few μs and a few tens of μm at best^{26,27} (see also http://www.ioffe.ru/SVA/NSM/Semicond/index.html). This accentuates the question of why charge carriers in the (disordered) hybrid perovskites are less affected by trapping. The electronic aspects of the unusual defect physics in CH_{3}NH_{3}PbI_{3} perovskites have already been discussed^{10}. Here, we are wondering if the interaction of defects with methylammonium dipoles could also contribute to the suppression of trapping. Indeed, typical mediumenergy traps in semiconductors have energies δU_{tr}=0.1–0.3 eV relative to the band edge. Typical physical size of the traps is of the order of a lattice constant, δr_{tr}≈5 Å. The local rearrangement of methylammonium dipoles by the effective forces in the vicinity of a defect could then occur, if the potential barrier for the methylammonium dipole rotation, U_{rot}≈10 meV, is smaller than the energy gain associated with a dipoletrap interaction, for instance, if estimated as: ∼10–75 meV (for P=2.29 D). The rearranged dipoles would thus reduce the defect’s trapping cross section. Such a defect decoration would be reminiscent of the recently observed trap healing effect at twodimensional semiconductor/polymer interfaces^{35}, with the important distinction that in hybrid perovskites the functional (rotationally responsive) dipoles are naturally available throughout the entire bulk of the sample. The question arises if the dipole rearrangement effect could be greater than that afforded in the standard continuousdielectric electrostatics. Detailed microscopic computational studies are needed to clarify this issue.
To conclude, we have measured Hall effect in hybrid (organo–inorganic) perovskites films and single crystals and found Hall mobilities ranging from 0.5 to 60 cm^{2} V^{−1} s^{−1}, depending on the sample composition and crystallinity. Concurrent measurements of a steadystate photoconductivity and Hall carrier density have allowed to directly determine bimolecular recombination coefficients (as low as 10^{−10} to 10^{−11} cm^{3} s^{−1}), carrier lifetimes (up to 30 μs and 2.7 ms in polycrystalline films and single crystals, respectively) and diffusion lengths (up to 23 and 650 μm in films and crystals, respectively). These measurements provide a direct and conclusive evidence of low e–h recombination rates and remarkably weak charge trapping in hybrid perovskites. They show that photophysical properties of these materials are quite different from those of conventional organic or inorganic semiconductors. We emphasize that our study determines these important transport parameters directly, from steadystate transport measurements, relevant to practical applications. The dipolar polaron model has been found helpful to explain the observed relatively low intrinsic carrier mobilities and radiative recombination rates.
Methods
Growth of hybrid perovskite thin films and single crystals
For fabrication of thinfilm perovskite samples, we followed the published procedures^{36,37}. Solution of PbI_{2} in dimethylformamide was spincoated onto cleaned glass substrates, resulting in a homogeneous PbI_{2} films. For solutiongrown samples, PbI_{2} films were immersed in a solution of methylammonium iodide in isopropanol. For vapourgrown perovskite films, PbI_{2} was annealed in a saturated methylammonium iodide vapour at 190 °C. The resulting CH_{3}NH_{3}PbI_{3} thin films were uniform, largearea (centimetrescale) films on glass. CH_{3}NH_{3}PbBr_{3} single crystals were grown through antivapour diffusion process^{38}. In brief, 1:1 ratio of methylammonium bromide (CH_{3}NH_{3}Br, synthesized following ref. 39) and lead bromide (PbBr_{2}, Sigma Aldrich, 98%) were dissolved in N,Ndimethylformamide. This solution was filtered into an inner container. The inner container was then put into a bigger container with dichloromethane. The outer container was sealed and kept at room temperature. Square bulky CH_{3}NH_{3}PbBr_{3} single crystals (3–5 mm on a side) grew at the bottom of the inner container within a few days.
Device fabrication
Devices in Hall bar geometry were fabricated by depositing Au or Ti through a shadow mask on freshly grown films or crystals. After wiring, the devices were capped under vacuum (10^{−5} Torr) with a protective PFPE (perfluoropolyether) oil, which is a chemically and electrically inert perfluorinated polymer. We find that PFPE can effectively protect samples from degradation due to moisture and other environmental factors. Control tests showed that PFPE does not cause any qualitative changes in the electrical properties of hybrid perovskite samples.
Magnetotransport and photoconductivity measurements
All measurements in this work were carried out at room temperature. a.c. Hall measurements were performed in an a.c. magnetic field of B=0.23 T (r.m.s.), referenced to a Stanford Research lockin amplifier that measures a.c. V_{Hall}. To reduce the parasitic Faraday induction, frequencies below 1 Hz were typically used. More importantly, V_{Hall} generated across the sample at a nonzero d.c. excitation (I0) was always compared with that generated at zero current (I=0), which yields a pure, Faradayinductioncorrected, Hall voltage^{15}. We have verified that V_{Hall} in all our samples was independent of the frequency in the range 0.3 to 5 Hz, which confirms that undesirable Faradayinduction signal was eliminated. Keithley 6221 current source was used to drive a d.c. excitation current through the sample. Calibration of our a.c. Hall setup has been done by carrying out measurements of a control Si sample with known carrier density and mobility. Longitudinal (photo)conductivity was measured by 4probe technique, which ensured that ambiguities associated with contactresistance were eliminated. It is important to emphasize that although 2probe σ_{PC} also exhibits a sublinear power dependence (σ_{PC}∝G^{α}, with α<1), the power exponent α in 2probe measurements may vary in a wider range from sample to sample due to contact effects. In contrast, 4probe photoconductivity systematically shows α=½ at high illumination intensities, which signifies a regime governed by bimolecular recombination. Photoexcitation was achieved by illumination with a calibrated blue LED (max. power: 20 W, λ=465 nm) driven by a Keithley 2400 source meter. The highest photon flux used was close to that of one sun (integrating over the part of the spectrum absorbed by the perovskites). Thus, measurements in this work were performed within the range of light intensities relevant for solar cell applications. The errors in Hall mobility values obtained for polycrystalline and singlecrystal samples (μ_{Hall}=8±0.4 and 60±5 cm^{2} V^{−1} s^{−1}, respectively) are defined by standard deviation in V_{Hall} measurements.
Data availability
The data that support the findings of this study are available from the corresponding author on request.
Additional information
How to cite this article: Chen, Y. et al. Extended carrier lifetimes and diffusion in hybrid perovskites revealed by Hall effect and photoconductivity measurements. Nat. Commun. 7:12253 doi: 10.1038/ncomms12253 (2016).
References
 1.
Green, M. A., HoBaillie, A. & Snaith, H. J. The emergence of perovskite solar cells. Nat. Photon. 8, 506–514 (2014).
 2.
Sum, T. C. & Mathews, N. Advancement in perovskite solar cells: photophysics behind the photovoltaics. Energy Environ. Sci. 7, 2518–2534 (2014).
 3.
Kojima, A., Teshima, K., Shirai, Y. & Miyasaka, T. Organometal halide perovskites as visiblelight sensitizers for photovoltaic cells. J. Am. Chem. Soc. 131, 6050–6051 (2009).
 4.
Mitzi, D. B., Feild, C. A., Schlesinger, Z. & Laibowitz, R. B. Transport, optical and magnetic properties of the conducting halide perovskite CH_{3}NH_{3}SnI_{3}. J. Solid State Chem. 114, 159–163 (1995).
 5.
Kagan, C. R., Mitzi, D. B. & Dimitrakopoulos, C. D. Organicinorganic hybrid materials as semiconducting channels in thinfilm fieldeffect transistors. Science 286, 945–947 (1999).
 6.
Wehrenfennig, C., Eperon, G. E., Johnston, M. B., Snaith, H. J. & Herz, L. M. High charge carrier mobilities and lifetimes in organolead trihalide perovskites. Adv. Mater. 26, 1584–1589 (2014).
 7.
Frost, J. M. et al. Atomistic origins of highperformance in hybrid halide perovskite solar cells. Nano Lett. 14, 2584–2590 (2014).
 8.
Frost, J. M., Butler, K. T. & Walsh, A. Molecular ferroelectric contributions to anomalous hysteresis in hybrid perovskite solar cells. Appl. Mater. 2, 081506 (2014).
 9.
Yin, W.J., Shi, T. & Yan, Y. Unique properties of halide perovskites as possible origins of the superior solar cell performance. Adv. Mater. 26, 4653–4658 (2014).
 10.
Yin, W.J., Shi, T. & Yan, Y. Unusual defect physics in CH_{3}NH_{3}PbI_{3} perovskite solar cell absorber. Appl. Phys. Lett. 104, 063903 (2014).
 11.
Kuper C. G., Whitfield G. D. (eds) Polarons and Excitons Plenum Press (1963).
 12.
Appel, J. Polarons. in Solid State Physics Vol. 21, eds Seitz F.et al. 193–391Academic Press (1968).
 13.
Devreese J. T. (ed) Polarons in Ionic Crystals and Polar Semiconductors NorthHolland (1972).
 14.
Podzorov, V. Organic single crystals: addressing the fundamentals of organic electronics. MRS Bull. 38, 15–24 (2013).
 15.
Chen, Y., Yi, H. T. & Podzorov, V. High resolution ac measurements of the Hall effect in organic fieldeffect transistors. Phys. Rev. Appl. 5, 034008 (2016).
 16.
Bube, R. H. in Photoconductivity of Solids Ch. 3.4 and 11.1, (Wiley (1960).
 17.
Brivio, F., Butler, K. T., Walsh, A. & van Schilfgaarde, M. Relativistic quasiparticle selfconsistent electronic structure of hybrid halide perovskite photovoltaic absorbers. Phys. Rev. B 89, 155204 (2014).
 18.
Umari, P., Mosconi, E. & De Angelis, F. Relativistic GW calculations on CH_{3}NH_{3}PbI_{3} and CH_{3}NH_{3}SnI_{3} perovskites for solar cell applications. Sci. Rep. 4, 4467 (2014).
 19.
Feng, J. & Xiao, B. Crystal structures, optical properties, and effective mass tensors of CH_{3}NH_{3}PbX_{3} (X=I and Br) phases predicted from HSE06. J. Phys. Chem. Lett. 5, 1278–1282 (2014).
 20.
Giorgi, G., Fujisawa, J., Segawa, H. & Yamashita, K. Small photocarrier effective masses featuring ambipolar transport in methylammonium lead iodide perovskite: a density functional analysis. J. Phys. Chem. Lett. 4, 4213–4216 (2013).
 21.
Ponseca, C. S. et al. Organometal halide perovskite solar cell materials rationalized: ultrafast charge generation, high and microsecondlong balanced mobilities, and slow recombination. J. Am. Chem. Soc. 136, 5189–5192 (2014).
 22.
Pope, M. & Swenberg, C. E. Electronic Processes in Organic Crystals and Polymers Oxford University Press (1999).
 23.
Pivrikas, A., Neugebauer, H. & Sariciftci, N. S. Charge carrier lifetime and recombination in bulk heterojunction solar cells. IEEE J. Select. Top. Quant. Electron 16, 1746–1758 (2010).
 24.
Varshni, Y. P. Bandtoband radiative recombination in groups IV, VI and IIIV semiconductors (I). Phys. Status Solidi 19, 459 (1967).
 25.
Pankove, J. I. Optical Processes in Semiconductors Dover Publications (1971).
 26.
Levinstein, M., Rumyantsev, S. & Shur, M. Handbook Series on Semiconductor Parameters Vol. 1, (World Scientific (1996).
 27.
Piprek, J. Semiconductor Optoelectronic Devices—Introduction to Physics and Simulations Ch. 3.7, (Academic Press (2003).
 28.
Löper, P. et al. Complex refractive index spectra of CH_{3}NH_{3}PbI_{3} perovskite thin films determined by spectroscopic ellipsometry and spectrophotometry. J. Phys. Chem. Lett. 6, 66–71 (2015).
 29.
Sheng, C. et al. Exciton versus free carrier photogeneration in organometal trihalide perovskites probed by broadband ultrafast polarization memory dynamics. Phys. Rev. Lett. 114, 116601 (2015).
 30.
Lee, T. D., Low, F. E. & Pines, D. The motion of slow electrons in a polar crystal. Phys. Rev. 90, 297 (1953).
 31.
Larsen, D. M. Polaron energy spectrum. Phys. Rev. 144, 697–702 (1966).
 32.
Prokof’ev, N. V. & Svistunov, B. S. Polaron problem by diagrammatic quantum Monte Carlo. Phys. Rev. Lett. 81, 2514–2517 (1998).
 33.
deQuilettes, D. W. et al. Impact of microstructure on local carrier lifetime in perovskite solar cells. Sci. Express 348, 683–686 (2015).
 34.
Dong, Q. et al. Electronhole diffusion lengths >175 μm in solutiongrown CH_{3}NH_{3}PbI_{3} single crystals. Sci. Express 347, 967–970 (2015).
 35.
Lee, B. et al. Trap healing and ultralownoise Hall effect at the surface of organic semiconductors. Nat. Mater. 12, 1125–1129 (2013).
 36.
Burschka, J. et al. Sequential deposition as a route to highperformance perovskitesensitized solar cells. Nature 499, 316–320 (2014).
 37.
Chen, Q. et al. Planar heterojunction perovskite solar cells via vaporassisted solution process. J. Am. Chem. Soc. 136, 622–625 (2014).
 38.
Shi, D. et al. Low trapstate density and long carrier diffusion in organolead trihalide perovskite single crystals. Science 347, 519–522 (2015).
 39.
Kim, H.S. et al. Lead iodide perovskite sensitized allsolidstate submicron thin film mesoscopic solar cell with efficiency exceeding 9%. Sci. Rep. 2, 591 (2012).
Acknowledgements
We thank HangDong Lee and Torgny Gustafsson for their help with heliumion microscopy, Pavel Irkhin for his help with the calibration of light sources, SzuYing Wang for her help with thinfilm sample fabrication. Y.C., H.T.Y. and V.P. thank the National Science Foundation for the financial support of this work under the grant DMR1506609 and the Institute for Advanced Materials and Devices for Nanotechnology (IAMDN) of Rutgers University for providing necessary facilities. X.Y.Z. acknowledges support by the US Department of Energy, Office of Science—Basic Energy Sciences, Grant ER46980, A.Z. thanks the Increase Competitiveness Program of NUST «MISIS» (No. К22015014 ) for partial support and appreciates the Welch Foundation for their partial support under grant AT1617, Y.N.G. is grateful for support from the Department of Energy, Office of Basic Energy Science (DOE/OBES) grant DESC0010697.
Author information
Author notes
 Y. Chen
Present address: Department of Physics, South University of Science and Technology of China, Shenzhen, Guangdong, China
Affiliations
Department of Physics, Rutgers University, Piscataway, New Jersey 08854, USA
 Y. Chen
 , H. T. Yi
 & V. Podzorov
Department of Chemistry, Columbia University, New York, New York 10027, USA
 X. Wu
 & X. Y. Zhu
Department of Physics and NanoTech Institute, University of Texas at Dallas, Richardson, Texas 75080, USA
 R. Haroldson
 , Y. N. Gartstein
 & A. Zakhidov
The Institute for Theoretical and Applied Electrodynamics, The National University of Science and Technology, MISIS, Moscow 119049, Russia
 Y. I. Rodionov
 & A. Zakhidov
Landau Institute for Theoretical Physics, Moscow 119334, Russia
 K. S. Tikhonov
Institute for Adv. Mater. and Devices for Nanotech., Rutgers University, Piscataway, New Jersey 08854, USA
 V. Podzorov
Authors
Search for Y. Chen in:
Search for H. T. Yi in:
Search for X. Wu in:
Search for R. Haroldson in:
Search for Y. N. Gartstein in:
Search for Y. I. Rodionov in:
Search for K. S. Tikhonov in:
Search for A. Zakhidov in:
Search for X. Y. Zhu in:
Search for V. Podzorov in:
Contributions
V.P. designed the research project and supervised the experiment. Y.C. and H.T.Y. performed device fabrication and measurements. Y.C., X.W., R.H., X.Y.Z and A.Z. grew perovskite thin films and single crystals, Y.N.G., Y.I.R. and K.S.T. performed theoretical calculations, Y.C., Y.N.G. and V.P. wrote the paper. All authors discussed the results.
Competing interests
The authors declare no competing financial interests.
Corresponding author
Correspondence to V. Podzorov.
Supplementary information
PDF files
 1.
Supplementary Information
Supplementary Figures 14, Supplementary Notes 13 and Supplementary References
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
About this article
Further reading

Critical assessment of charge mobility extraction in FETs
Nature Materials (2017)

Understanding the physical properties of hybrid perovskites for photovoltaic applications
Nature Reviews Materials (2017)

Piezoelectric scattering limited mobility of hybrid organicinorganic perovskites CH3NH3PbI3
Scientific Reports (2017)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.