Materials Today
Volume 28, September 2019, Pages 31-39
Journal home page for Materials Today

Research
Electric-field effect on photoluminescence of lead-halide perovskites

https://doi.org/10.1016/j.mattod.2019.01.003Get rights and content

Abstract

With the development of novel semiconductors for optoelectronic applications, new device functionalities utilizing unique characteristics of emerging materials can be particularly appealing. Here, we demonstrate a reversible control of photoluminescence (PL) emission from lead-halide perovskites achieved in perovskite electric-double-layer transistors. PL in several prototypical lead-halide perovskite compounds is shown to be reversibly tuned by a small gate voltage in the range ±1.2 V applied to the ionic-liquid gel on the perovskite surface, with the intensity modulation that can reach one to two orders of magnitude. This effect may be mediated by a reversible migration of oxygen ions affecting the crystal region near the interface with the ion gel. The resulting passivation (or activation) of non-radiative recombination centers (traps) by oxygen ions would then modulate the population of mobile photogenerated electrons and holes that give rise to PL, which is thus tuned with an electric “knob” (the gate) in these devices.

Introduction

Lead-halide perovskites of a general composition APbX3, where A is a molecular or atomic cation (such as, for instance, methylammonium (MA+), formamidinium (FA+), or Cs+) and X is a halide anion (Br, I or Cl), exhibit a range of attractive physical properties important for solar cell applications [1], including unusually long charge carrier lifetimes and diffusion lengths, modest intrinsic carrier mobilities in the range 1–60 cm2 V−1 s−1, and low bimolecular recombination rates [2], [3], [4], [5], [6]. Furthermore, it was observed recently that hybrid perovskite solar cells prepared and measured in air showed better photo-voltaic conversion efficiencies, higher photoluminescence intensities, and longer carrier lifetimes compared to similar measurements in inert atmosphere or vacuum [7], [8], [9], [10], [11].

Here, we fabricated electric double-layer transistor (EDLT) devices [12], [13], based on single crystals of several prototypical lead-halide perovskites, including CsPbBr3 (CPB), MAPbBr3 (MAPB), and FAPbBr3 (FAPB), and observed that the photoluminescence (PL) intensity of these crystals can be modulated reversibly with a gate voltage applied to the ionic liquid (or gel) in contact with the surface of perovskite crystals. EDLTs utilizing ionic liquids or semi-solid ion gels have been developed as an effective tool to control electronic and magnetic properties of organic and inorganic materials via generating a very strong local transverse electric field at the interface, where a Helmholtz double layer is formed. This technique is powerful for the exploration of the fundamental properties of new materials, where high carrier densities can be generated without the need for chemical doping [12], [13], [14].

In this work, perovskite EDLT-like devices are fabricated with the facile “cut and stick” transparent ion gels [15]. It is found that the PL intensity, IPL, in these devices can be reversibly tuned by up to 1–2 orders of magnitude with a low gate voltage, VG, in the range of −1.2 V ≤ VG ≤ 1.2 V. The devices we report here do not function as conventional field-effect transistors (FETs) or electric EDLTs, as we did not observe any appreciable modulation of the source–drain current under the action of the gate voltage. Instead, the modulated observable in our EDLTs is the PL emitted by the crystal photoexcited with laser beam incident on the crystal through the transparent ion gel. X-ray photoelectron spectroscopy (XPS) shows that all the studied perovskite crystals feature a halide deficiency (both at the surface and in the bulk), as well as the presence of some oxygen but primarily at the surface of the crystals kept in air (below we refer to such surfaces as “as-grown”, as opposed to surfaces freshly exposed by cleaving the crystals). Prompted by the transport, photoluminescence, and XPS measurements, we propose that this effect may be mediated by a reversible electrochemical passivation of electron traps associated with the halide vacancies present in these crystals. Such traps can be passivated (at VG < 0) or reactivated (at VG > 0) via the interaction with negatively charged oxygen species (for instance, O2, OH) migrating within the surface region of the crystal under the transverse gate electric field generated at the interface between the crystal and the ion gel in EDLT devices. A reduced density of electron traps at VG < 0 leads to a lower probability of non-radiative recombination (trapping) and a higher population of mobile photogenerated carriers, thus enhancing the PL from the radiative electron–hole recombination. An increased density of traps at VG > 0, on the other hand, would lead to a smaller population of mobile carriers and the suppressed PL. A reversible modulation of the overall PL intensity is thus achieved by controlling the non-radiative recombination rate in the surface region with VG. We emphasize that although PL is a common focus in the extensive perovskite research, this work presents the first demonstration of a reversible tuning of PL with an “electric knob” (the gate voltage) in these materials.

The structure of our ion-gel gated perovskite transistors (or, simply, perovskite EDLTs) is shown in Fig. 1a and b. Source and drain contacts were deposited on the relatively large single crystals (1–5 mm in size) by sputtering Au through a contact shadow mask, and a 10-μm-thick ion-gel film was placed on the channel area, overlapping with the source and drain contacts. The ion gel consists of an inert polymer matrix infused with molecular cations (1-ethyl-3-methylimidazolium, or [EMI]+) and anions (bis(trifluoromethylsufonyl) amide, or [TFSA]) [14], [15]. To allow optical photoexcitation and PL collection from the channel area, the Au gate electrode was deposited on the side of the device. In steady-state PL measurements, a blue (λ = 465 nm) continuous-wave (cw) excitation laser beam with a relatively low power in the range 0.85–85 mW·cm−2 was incident upon the channel area through the transparent ion gel, and the emitted PL was collected back to a spectrometer through a microscope objective. Fig. 1b shows a schematic cross-section of our EDLTs.

The effect of the electric field generated by the ion-gel gate on the photoluminescence of inorganic CsPbBr3 perovskite crystals is shown in Fig. 1c and d. The PL was excited at a fixed incident photon flux F = 5 × 1015 cm−2 s−1 in high vacuum (∼10−5 Torr), while the gate voltage VG was set to discrete values varied between −1.2 and 1.2 V. Fig. 1c shows PL spectra collected at different VG. Red, blue, and green curves correspond to the spectra at VG = −1.2, 0 and 1.2 V, respectively. The intensity of the 540-nm PL band of the CPB perovskite, corresponding to the band-gap emission, is strongly modulated by VG. Fig. 1d shows the PL intensity emitted by the device as a function of gate voltage, IPL(VG), for VG varied in a counter-clockwise close loop. To prevent performance degradation caused by possible chemical interaction between the ion gel and the perovskites (or metal contacts), we have limited the range of gate voltage to |VG| ≤ 1.2 V. Repeated measurements in multiple loops show that the IPL(VG) data are stable and reproducible, without any noticeable degradation, as long as VG is kept within this range (Supplementary Information, Fig. S1). Each data point in Fig. 1d is collected after the electrolytic displacement current ceases, and the PL is allowed to stabilize for 1–2 min at every new value of VG. Note that for a better comparison of the PL at different VG, all PL intensities in this work are normalized by dividing the spectra by the maximum intensity of PL band at the highest positive VG. IPL increases with negative and decreases with positive VG. As a convenient parameter characterizing these devices, we use an on/off ratio defined as the ratio of photoluminescence intensity at VG = −1.2 V to that at 1.2 V. Given the above normalization procedure, this on/off ratio is simply given by the value shown in our PL plots at VG = −1.2 V. For instance, for the device in Fig. 1d, the on/off ratio is 50, meaning that the PL emission intensity of this device can be tuned by a factor of 50 within the indicated range of VG.

It is worth noting that these EDLT devices do not function as conventional (electric) FETs or EDLTs. We were not able to observe any appreciable modulation of the source-drain current with the gate voltage in any of the EDLT devices based on various perovskite single crystals. Although under a laser excitation an extremely small effect of VG on the source-drain photocurrent, ISD, could be seen (Supplementary Information, Fig. S2), the modulation magnitude is not significantly greater than the ion-gel displacement current. Furthermore, the effect of VG on ISD in the dark is not at all discernible. Nevertheless, it is important to keep in mind that the absence of an electric field effect does not contradict the observation of a VG-modulated photoluminescence, because these phenomena rely on different mechanisms. For instance, as described below, the gate voltage control of PL does not require carrier injection from contacts and their accumulation at the interface. Thus our focus in this work is not achieving a FET action, but to investigate the observed novel effect of PL gating.

To investigate the universality of the observed ionic-liquid gating effect on the PL, we have expanded our study to several types of perovskites. Fig. 2 shows IPL(VG) measurements in several EDLTs based on different lead-halide perovskite single crystals, including CPB, FAPB, MAPB, and MAPI. It is evident that most of the perovskites exhibit a clear electric field effect on photoluminescence, although with different on/off ratios. The polarity of the effect is, however, the same in all the studied compounds: the PL is enhanced at negative and suppressed at positive VG. We note that at this stage attempts to find a correlation between the magnitude of this effect and the carrier diffusion length LD reported for different perovskites are not worthwhile. Indeed, (a) the effect of PL modulation with VG shown in Fig. 2a and b depends on several factors simultaneously, including the thickness of individual crystals and device geometry; (b) the carrier diffusion length LD itself quite strongly depends on the purity and trap concentration of particular crystals and, in our experience, can vary substantially from batch to batch [5], [6].

We have also investigated the dependence of the VG-induced modulation of PL on the crystal thickness. As shown in Fig. 2a, the on/off ratio in IPL(VG) of CPB EDLTs drastically increases with a decreasing thickness, t, of the crystals. Fig. 2c is a set of PL on/off ratios, plotted as a function of the crystal thickness for the devices based on several compounds presented in Fig. 2a and b. It clearly shows that there is an overall correlation between the crystal thickness and the magnitude of photoluminescence modulation with VG. Note that the effect of the crystal thickness t has nothing to do with the electric field that one can assume by having a simple-minded parallel-plate capacitor model in mind, because there is no back electrode at the bottom of our samples. Instead, the electrodes are on top of the crystal and in contact with the ion gel. The same data are then plotted in Fig. 2d as a function of the dimensionless parameter quantifying the aspect ratio of the excited volume of the sample: the square-root of the photoexcited area of the crystal, A1/2, divided by the crystal thickness, t. For small devices, where the incident laser beam covers the entire channel area, A is defined by the physical dimensions of the channel as A ≡ L × W, where L and W are the channel length and width, while for big devices that are greater than the laser beam’s cross-section area a, the photoexcitation area is simply equal to the beam’s cross-section area, A ≡ a. Fig. 2d shows that the on/off ratio in IPL(VG) increases linearly with A1/2/t when plotted on a semi-log scale. The horizontal error bars of A1/2/t values in Fig. 2d were determined from the 10% accuracy of the measurements of the linear sample dimensions (L, W and t) via the error propagation formula. For FAPB device (gray solid square), since the illuminated area in this case corresponds to the laser beam cross-section, the characteristic error of the Gaussian beam diameter can be used, which is the difference between the full width at half maximum (FWHM) and the so-called 1/e2 width of the beam (approx. 70%). This error is then again used in the error propagation formula. Note that the intensity of the incident laser beam (in W·cm−2) is maintained the same in all these measurements of different devices in Fig. 2, indicating that the observed behavior is not a photoexcitation density effect.

The observed tendency of the PL gating to become stronger with a decreasing thickness of the crystals (and with an increasing illuminated area) is consistent with a surface origin of this effect. Indeed, given the long photocarrier diffusion lengths (of up to several hundred μm), which is much longer than the typical light penetration length of ∼80 nm in perovskites [5], [16], [17], [18], the carrier diffusion and the radiative recombination of mobile electrons and holes resulting in the photoluminescence in these materials are expected to occur essentially in the bulk of these crystals, while only the surface (or a very thin near-surface region) of the crystals is typically affected by the ion-gel gating. A greater relative effect of VG on IPL in the samples possessing a greater surface to volume ratio would therefore be reasonable. The observation of a significant dependence of the effect on the thickness of the crystals, when the latter is varied in the range t = 200–800 μm (as in Fig. 2a), is an indirect manifestation of the very long diffusion length of charge carriers, initially photogenerated near the surface of the crystal and then reaching deep into the crystal’s bulk over the length scale of several hundred μm.

We also note that the fact that samples of several different compounds from various batches collapse at the same (within the error bar) dependence in Fig. 2d suggests that PL on/off ratio primarily depends on the dimensionless parameter A1/2/t, but not on the impurity content of individual samples. However, this is only a very preliminary conclusion that requires further verification via studies of samples with systematically varied impurity level (trap concentration).

The residual gate “leakage” current, IG, known to occur in EDLTs even after the electrolyte is polarized and the electrolytic displacement current ceases, was monitored in all our measurements and did not exceed a low value of 10–50 nA. Nevertheless, in order to investigate if the residual IG at fixed VG had any effect on the photoluminescence emitted from the channels of our devices, we have measured EDLTs with the gate electrode in the two configurations shown in Fig. 3a. As mentioned earlier, to allow optical photoexcitation and PL collection, the gate electrode in our EDLTs is positioned on a side (as shown in Fig. 1a and b), instead of being deposited on the ion gel directly above the channel. In principle, even with such a “remote” gate, the electrolyte can be polarized by application of a gate voltage. However, the gating efficiency will depend on the particular device geometry: the distance between the gate and the channel bridged by the ion gel and the overall area of the physical contact between the gate and the ion gel will both affect the VG range necessary to achieve a certain degree of ionic polarization of the electrolyte. First, we have measured an EDLT with the same gate configuration as in the devices discussed so far: the corresponding data are shown in Fig. 3a with solid black symbols. After that, we have extended the gate electrode of the same device by depositing an additional layer of gold on top of the ion gel, while still keeping the gate off the channel area: the resultant measurements are shown in Fig. 3a with open red symbols. In the first measurement, IPL(VG) of CPB EDLT exhibits an on/off ratio of ∼50 for |VG| ≤ 1.2 V. In the second measurement, however, the on/off ratio is enhanced to 70, and it is achieved within a much narrower range of the gate voltage, |VG| ≤ 0.3 V. Because the poling efficiency of the ionic liquid is improved, a greater modulation of IPL within a narrower VG window is realized in the second type of structures. Even more importantly, the maximum value of the residual gate-channel “leakage” current flowing through the ion gel is drastically reduced to about 1 nA in the second type of structures due to the smaller range of VG. Therefore the extended-gate geometry improves the electrostatic coupling between the gate electrode and the mobile ionic species in the ion gel, thus allowing to achieve a similar degree of polarization of the ionic liquid at a lower VG. This in turn results in a lower residual parasitic gate leakage current. This indicates that the very small residual current IG, flowing through the electrolyte at fixed VG while the PL is measured, does not contribute in any significant way to the effect of PL modulation observed here. This test also suggests strategies for future optimization of the PL modulation within a smaller VG range.

We have carried out several control experiments to ensure that the observed gating effect on the PL of lead-halide perovskites is not an artifact. First, we have studied the same type of devices but built on the organic molecular crystals of rubrene using the same ion gel and the same type of electrodes. Rubrene has a clear photoluminescence in the visible range of spectrum with the main band centered at ∼600 nm [19], [20]. Our measurements indicate that there is absolutely no effect of the gate bias on the PL of rubrene-based EDLT (Supplementary Information, Fig. S3). In addition, we have monitored the optical transparency of the ion gel, as well as the overall appearance of perovskite EDLTs under a microscope, while in operation under various negative or positive VG biases, and were not able to detect any changes (Supplementary Information, Fig. S4). Given these control tests, as well as the effect of different gate geometries (Fig. 3a), the observed gating effect on the PL of the perovskites appears to be an intrinsic phenomenon related to microscopic processes occurring in the perovskite material near the interface with the ionic liquid.

It is also worth comparing the behavior of CPB EDLTs in different atmospheres. Fig. 3b shows that measurements in dry air (solid black triangles) result in an increased overall level of photoluminescence and a stronger IPL(VG) dependence, as compared to measurements of the same device in high vacuum (open red triangles). This behavior suggests that oxygen residing at the surface of the crystals and in the ion gel is involved in the passivation of centers of non-radiative recombination in perovskites near the interface, thus resulting in the observed here higher overall level of photoluminescence and a stronger modulation of IPL with VG in air.

We have recently reported [21] that the steady-state photoluminescence of (ungated) lead-halide perovskites exhibits two regimes in its dependence on the incident photoexcitation flux, IPL(F). Namely, this dependence is a power law, IPL = I0·Fα, with the power exponent α sequentially taking two distinct values α = 2 and 3/2 with increasing photoexcitation flux F. The incident flux at the transition point, FT, between these regimes was found to be dependent on the material and measurement conditions (for more detail, see Ref. [21]), while the power exponents 2 and 3/2 appear to be very robust (observed in all the studied hybrid and all-inorganic perovskites at different temperatures). Remarkably, here in our gated EDLT measurements, we now observe that not only the PL intensity is modified by VG both in the α = 2 and α = 3/2 regimes but also the transition point FT is clearly affected by the gate voltage. Indeed, as shown in Fig. 4, at VG > 0, the transition occurs at a higher excitation intensity (making the trap dominated α = 2 regime more extended), while at VG < 0, the transition shifts to a lower intensity (making the bimolecular recombination dominated α = 3/2 regime more extended). The simultaneous modulations of the PL intensity and the transition flux FT by EDLT gating can be indicative of a common mechanism involved. With gating likely affecting the crystal region close to the surface, one particularly looks for rationales related to the effect of surface.

It is then imperative to consider our observations within the framework of the model we discussed in Ref. [21] (see also Supplementary Information, sec. 5, where the detailed modeling of the spatial carrier distribution in the presence of trapping, e-h recombination and surface recombination is performed). In addition to the two usual channels of the decay of photocarrier concentration n in the bulk: namely, (1) trapping (or the monomolecular recombination) that occurs with the rate n/τtr, where tautr is a trap-limited carrier lifetime, and (2) bimolecular e-h recombination occurring with the rate γn2, where γ is the bimolecular e-h recombination coefficient, this model also takes into account the additional non-radiative recombination process taking place in the surface region of the crystal and characterized by the effective surface recombination velocity S. Given the fact that the carrier diffusion length in perovskites is much longer than the light penetration length, the “effective surface region” here extends to the depth of the carrier photogeneration region. The two regimes in the dependence of PL on the incident photon flux F are related to the two limits of the system’s behavior under a steady-state photoexcitation, separated by the characteristic crossover carrier density nγ1/γτtr. Below this characteristic carrier density (n < nγ), a monomolecular (trap-dominated) carrier recombination governs the system’s behavior. At stronger photoexcitations leading to the carrier densities above the characteristic density (n > nγ), a bimolecular e-h recombination dominates. The model in Ref. [21] shows that, while the low-flux scaling regime (exponent α = 2) arises irrespective of the magnitude of the surface recombination velocity S, the exponent α at higher fluxes does depend on S. Specifically, the observed regime with the power exponent α = 3/2 is realized only for a significant surface recombination, when the surface recombination velocity S is greater than the so-called carrier diffusion velocity ν: SνD/τtr1/2 (here, D and τtr are the diffusion coefficient and the carrier monomolecular decay time, respectively) With substantial surface recombination, the model yields the PL intensity at lower fluxes F behaving as:IPL0.5Dτtr1/2·FS2

At higher fluxes, however, the model derives the behavior:IPL2D3γ1/2·FS3/2

The transition point FT between the regimes of Eqs. (1), (2) is defined as a flux at which these PL intensities roughly match, that is:FT8S3γτtr

Eqs. (1), (2), (3) clearly show how the effective surface recombination velocity S affects PL intensities and the transition flux FT between the two regimes in PL characterized by different excitation power exponents. Smaller S lead to the increase in PL intensity and a shift of FT to lower fluxes, while stronger surface recombination would decrease PL and shift FT to the higher fluxes. This suggests that the effect of gating on the PL and FT that we observed here (Figure 1, Figure 2, Figure 3, Figure 4) originates from the influence of VG on S: a negative VG decreases the surface recombination velocity S, while a positive VG increases it. Moreover, there appears to be a correlation between the magnitudes of the variation of PL and FT with VG. Indeed, a smaller modulation of PL intensity in MAPB (Fig. 2b) is accompanied by a smaller modulation of its FT with VG (Fig. 4a), while a larger modulation of PL in CPB (Fig. 1d) is accompanied by a larger modulation of FT (Fig. 4b). How exactly the effective S is affected by VG in different samples would be an interesting topic for future studies. However, the experimental findings (Figure 1, Figure 2, Figure 3, Figure 4) and modeling (Eqs. (1), (2), (3)) presented in this work already suggest a picture of surface traps (or other non-radiative recombination centers) that are electrostatically passivated under a negative gate bias (and, correspondingly, reactivated under a positive gate bias) in perovskite EDLTs, leading to a reversible modulation of photoluminescence with a gate voltage in these devices.

To gain a better microscopic understanding of the observed effects, we have performed XPS studies of perovskite crystals that clearly show two important features. First, all the studied crystals are found to be slightly halide deficient, having a general composition APbX3-δ, where the average δ = 0.6 (Fig. 5a and Supplementary Fig. S5). This halide deficiency is observed both at as-grown surfaces of the crystals and in their bulk, as clearly shown by the comparison of XPS measurements of as-grown and cleaved crystals (Fig. 5a). Second, a noticeable concentration of oxygen is found at the surface of as-grown crystals, but not in the bulk (Fig. 5a). These results corroborate recent XPS work on similar systems [22]. In the correct stoichiometric composition, Pb in the perovskite structure, being a divalent element with 6s26p2 outer-shell configuration, carries a charge 2+, while A and halide sites carry charges 1+ and 1−, respectively (A+Pb2+X3). It was recently suggested that halide vacancies would create positively charged crystal defects, such as Pb2+, that can then act as electron traps [23]. Under a photoexcitation, these positively charged centers can be capturing photogenerated electrons (but not holes) and thus lead to an enhanced non-radiative recombination (trapping) of electrons. Recall that the photoluminescence in perovskites originates from a radiative recombination of mobile electrons and holes diffusing in the bulk, which is clearly suggested by the F2 scaling of PL observed at low-to-moderate photoexcitations [21]. Thus, trapping a significant fraction of one type of carriers (here electrons) would lead to a suppressed photoluminescence. This leads us to the mechanism schematically shown in Fig. 5b. Under a negative gate voltage VG < 0, the ionic liquid is polarized, with the [TFSA] anions driven toward the surface of the crystal, where they create a strong transverse electric field at the interface, which subsequently induces a migration of oxygen species from the surface to the bulk. These oxygen species likely originate from a physisorbed oxygen or dissociated water present at the surface of the crystal or dissolved in the ion gel (they are schematically shown in Fig. 5b as green circles labeled “O−”). The electrostatic interaction of these negative oxygen ions with the crystal defects schematically shown as “Pb+” would passivate the electron traps associated with these defects, leading to a recovery of the balance (equality) between the populations of mobile electrons and holes. At VG > 0, oxygen ions are driven back toward the surface, which reactivates the electron traps and weakens the photoluminescence.

It is to be understood, of course, that this picture does not imply that individual oxygen ions can freely traverse the crystal over appreciable distances under an applied gate bias; rather it is the spatial distribution of these species in the near-surface region that gets modified under the gate bias. We also need to emphasize here that, in addition to oxygen and lead (O and Pb+) species, other defects could also be present and contributing to the redistribution of the active traps in the sample under a gate bias. Phenomenologically, the cumulative effect of all the contributing species is absorbed in the VG-dependent effective surface recombination velocity parameter S. The likelihood of oxygen ions and Pb+ defects being particularly relevant is, however, supported by our control measurements in air vs vacuum (Fig. 3b) and XPS measurements (Fig. 5a).

It is also interesting to note that the proposed mechanism of the PL modulation with VG is also consistent with the asymmetry of IPL(VG) curves with respect to VG = 0 shown in Fig. 3b. Such an asymmetry is observed in most of our measurements and can thus have a fundamental origin. It is natural to assume that the concentration of oxygen species at the surface of the crystals is higher than in the bulk, and it gradually decreases with depth going away from the surface. This is consistent with our XPS measurements. In this situation, it would be easier to “push” O species away from the surface to the bulk (by applying VG < 0), rather than “pull” them back toward the surface (by applying VG > 0), where their concentration is already high. Such an asymmetry of the effect could thus be a direct consequence of the fact that the concentration of O species is usually higher at the surface.

In summary, we have observed an electric field effect on the photoluminescence in ionic-liquid gated lead-halide perovskites. We found that the modulation of the photoluminescence intensity in perovskite single crystals interfaced with a semi-solid ion gel can reach one to two orders of magnitude within a narrow gate voltage window, |VG| ≤ 1.2 V. The magnitude of the effect depends on: (a) the relative thickness of the crystal and the area of the channel subjected to illumination, (b) the presence of oxygen in the atmosphere during the measurements, and (c) the geometry of the gate electrode contacting the ion gel. These observations suggest that the reported VG-controlled modulation of photoluminescence can be further enhanced in future studies. Our measurements and modeling indicate that the mechanism of this effect can be related to a reversible electrostatic passivation of electron traps via ionic-liquid gating. The novel functionality of a reversible control of light emission from perovskites, achieved here with an “electric knob” (the gate voltage), could be beneficial for the development of future photonic applications based on solution-processed perovskite materials.

Section snippets

Acknowledgments

We thank Prakriti P. Joshi and Xiaoyang Zhu of the Department of Chemistry, Columbia University for providing some of the crystals used in this study. YNG is grateful for support through the Department of Energy, Office of Basic Energy Science (DOE/OBES) grant DE-SC0010697. CDF acknowledges partial support from the MRSEC program of the National Science Foundation under Grant Number DMR-1420013. VP acknowledges support from the CMP program of the National Science Foundation under Grant Number

Author contributions

HTY and VP designed the experiments and performed electric and PL measurements. SR and RAB performed XPS measurements and analysis. BT and CDF fabricated the ion gel. YNG and VP devised the theoretical model of the effect of surface recombination on PL intensity and transition flux FT. HTY and VP wrote the manuscript. All authors discussed the results.

Data availability

The data that support the findings of this study are available from the corresponding author on request.

References (23)

  • M.A. Green et al.

    The emergence of perovskite solar cells

    Nat. Photon.

    (2014)
  • T.C. Sum et al.

    Advancement in perovskite solar cells: photophysics behind the photovoltaics

    Energy Environ. Sci.

    (2014)
  • T.J. Savenije

    Thermally activated exciton dissociation and recombination control the carrier dynamics in organometal halide perovskite

    J. Phys. Chem. Lett.

    (2014)
  • S.D. Stranks

    Electron-hole diffusion lengths exceeding 1 micrometer in an organometal trihalide perovskite absorber

    Science

    (2013)
  • Y. Chen

    Extended carrier lifetimes and diffusion in hybrid perovskites revealed by Hall effect and photoconductivity measurements

    Nat. Commun.

    (2016)
  • H.T. Yi

    Intrinsic charge transport across phase transitions in hybrid organo-inorganic perovskites

    Adv. Mater.

    (2016)
  • G.E. Eperon

    The importance of moisture in hybrid lead halide perovskite thin film fabrication

    ACS Nano

    (2015)
  • J.F. Galisteo-Lopez

    Environmental effects on the photophysics of organic-inorganic halide perovskites

    J. Phys. Chem. Lett.

    (2015)
  • Y. Tian

    Enhanced organo-metal halide perovskite photoluminescence from nanosized defect-free crystallites and emitting sites

    J. Phys. Chem. Lett.

    (2015)
  • H.H. Fang

    Ultrahigh sensitivity of methylammonium lead tribromide perovskite single crystals to environmental gases

    Sci. Adv.

    (2016)
  • M. Lorenzon

    Role of nonradiative defects and environmental oxygen on exciton recombination processes in CsPbBr3 perovskite nanocrystals

    Nano Lett.

    (2017)
  • View full text