Exciton localization in solution-processed organolead trihalide perovskites

Abstract

Organolead trihalide perovskites have attracted great attention due to the stunning advances in both photovoltaic and light-emitting devices. However, the photophysical properties, especially the recombination dynamics of photogenerated carriers, of this class of materials are controversial. Here we report that under an excitation level close to the working regime of solar cells, the recombination of photogenerated carriers in solution-processed methylammonium-lead-halide films is dominated by excitons weakly localized in band tail states. This scenario is evidenced by experiments of spectral-dependent luminescence decay, excitation density-dependent luminescence and frequency-dependent terahertz photoconductivity. The exciton localization effect is found to be general for several solution-processed hybrid perovskite films prepared by different methods. Our results provide insights into the charge transport and recombination mechanism in perovskite films and help to unravel their potential for high-performance optoelectronic devices.

ARTICLE

Received 18 Mar 2015 | Accepted 1 Feb 2016 | Published 21 Mar 2016

Haiping He1, Qianqian Yu1, Hui Li1, Jing Li1, Junjie Si1, Yizheng Jin2, Nana Wang3, Jianpu Wang3, Jingwen He4, Xinke Wang4, Yan Zhang4 & Zhizhen Ye1

Organolead trihalide perovskites have attracted great attention due to the stunning advances in both photovoltaic and light-emitting devices. However, the photophysical properties, especially the recombination dynamics of photogenerated carriers, of this class of materials are controversial. Here we report that under an excitation level close to the working regime of solar cells, the recombination of photogenerated carriers in solution-processed methylammoniumleadhalide lms is dominated by excitons weakly localized in band tail states. This scenario is evidenced by experiments of spectral-dependent luminescence decay, excitation density-dependent luminescence and frequency-dependent terahertz photo-conductivity. The exciton localization effect is found to be general for several solution-processed hybrid perovskite lms prepared by different methods. Our results provide insights into the charge transport and recombination mechanism in perovskite lms and help to unravel their potential for high-performance optoelectronic devices.

DOI: 10.1038/ncomms10896 OPEN

Exciton localization in solution-processed organolead trihalide perovskites

1 State Key Laboratory of Silicon Materials, School of Materials Science and Engineering, Zhejiang University, Hangzhou 310027, China. 2 Center for Chemistry of High-Performance and Novel Materials and State Key Laboratory of Silicon Materials, Department of Chemistry, Zhejiang University, Hangzhou 310027, China. 3 Key Laboratory of Flexible Electronics (KLOFE) & Institute of Advanced Materials (IAM), Jiangsu National Synergetic Innovation Center for Advanced Materials (SICAM), Nanjing Tech University (NanjingTech), 30 South Puzhu Road, Nanjing 211816, China. 4 Department of Physics, Capital Normal University, Beijing Key Lab for Metamaterials and Devices, and Key Laboratory of Terahertz Optoelectronics, Ministry of Education, Beijing 100048, China. Correspondence and requests for materials should be addressed to H.H. (email: mailto:[email protected]

Web End [email protected] ) or to Y.J. (email: mailto:[email protected]

Web End [email protected] ) or toZ.Y. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:10896 | DOI: 10.1038/ncomms10896 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10896

Recently, organolead trihalide perovskites have been utilized in low-temperature solution-processed photovoltaics16 and light-emitting devices710. Certied power conver

sion efciency approaching 20.1% has been realized6. The impressive photovoltaic performance is believed to originate from the long-distance and balanced diffusion of charge carriers1112. Remarkably, the solution-processed perovskite lms also exhibit superior luminescence properties. Optically pumped lasing with low thresholds and tunable wavelengths7, and bright light-emitting diodes910 have been demonstrated.

Despite these remarkable advances, knowledge of the photo-physical properties of the perovskites is still lacking. One of the key questions concerns the recombination dynamics of photo-generated charges: whether exciton or free carrier (FC) is the dominant recombination channel in organolead trihalide perovskites. The answer will help to interpret the seemingly counterintuitive facts that organolead trihalide perovskites can act both as extraordinary photovoltaic materials and superior gain mediums for lasing13. In general, photovoltaic materials require efcient separation of photocarriers, and lasing materials require high recombination rates. The reported exciton binding energy of the perovskites14,15 is comparable to the thermal energy at room temperature (RT), which arouses arguments that in such a case excited states will tend to dissociate into FCs rather than recombine radiatively.

Several groups have used photoluminescence (PL) to study the competition between exciton and FC in organolead trihalide perovskites. In a few steady-state PL studies, the RT PL was attributed to exciton recombination,1618 but the conclusions lacked solid evidence. Recently, several groups8,14,19,20 attributed the RT PL to FC recombination. For example, DInnocenzo et al.14 argued that excitons generated by low-density excitation are almost fully ionized at RT when the exciton binding energy is moderately larger than the RT thermal energy. The band lling effect8 and quadratic dependence of the PL intensity on the excitation intensity19, the two characteristic features of FC recombination, were observed at relatively high excitation levels. We note that the observation of FC recombination at

relatively high excitation levels is not surprising because the reduced exciton binding energy originated from the screening effect of FCs, a phenomenon that has been well established in many semiconductors21,22.

Here we show that under an excitation level close to the working regime of solar cells, the radiative recombination of photogenerated carriers in solution-processed CH3NH3PbX3 perovskites is dominated by excitons localized in band tail states. The excitonic nature of the emission is evidenced by the excellent power-law dependence of the PL intensity on the excitation intensity expected for bound excitons, and is supported by the PL lineshape analysis. The localization effect is supported by the spectral dependence of the PL lifetime and frequency-dependent THz photoconductivity results. We also show that the exciton localization effect is general in several solution-process perovskite lms.

ResultsEvidence for exciton localization in CH3NH3PbBr3 lms. We use solution-processed CH3NH3PbBr3 lms for the PL studies. The lms show good crystalline and optical quality (Supplementary Fig. 1). To avoid degradation induced by air exposure, all samples were prepared in a nitrogen-lled glove box, coated with a polymethyl methacrylate (PMMA) layer and were measured in vacuum (10 1 Pa). The CH3NH3PbBr3 lm shows emission at 2.35 eV, in agreement with the reported values23,24.

Near-band-edge emission in semiconductors may have several origins, including exciton recombination, FC recombination (also known as band-to-band transition), free-to-bound recombination and donoracceptor pair recombination. To determine which process is dominant in our samples, we measured the PL spectra under various excitation densities close to or lower than the photovoltaic working regime (B5 1014 cm 3) (refs 14,25).

The PL lineshapes in Fig. 1a are almost identical in the whole range of excitation intensity. The PL intensity shows excellent power-law dependence on the excitation power, with a power-law exponent of 1.179. In direct bandgap semiconductors and

a b

Photocarrier density (cm3)

1012 1013 1014 1015

MAPbBr3

Iex decrease

PL intensity (a.u.)

105

104

103

102

101

100

Normalized integrated PL intensity

K = 1.179

102 101 100 101

Excitation fluence (nJ cm2)

2.1 2.2 2.3 2.4 2.5 2.6 2.7

Photon energy (eV)

100

101

102

103

104

Figure 1 | Excitation density-dependent PL of solution-processed CH3NH3PbBr3 lms. (a) Steady-state PL spectra recorded with excitation density from0.01 to 4 nJ cm 2. All spectra are measured in vacuum at RT. In all spectra, the peak energy (indicated by the dashed line), lineshape and linewidth are identical within the experimental error. (b) Logarithm plot of the integrated PL intensity versus excitation density. The data show a power-law dependence with k 1.179.

2 NATURE COMMUNICATIONS | 7:10896 | DOI: 10.1038/ncomms10896 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10896 ARTICLE

under non-resonant excitation conditions, the integrated PL intensity (IPL) is a power-law function of the excitation density26,

IPL Ikex 1 with k 2 for FC recombination, 1oko2 for recombination of

excitons (including free excitons and bound excitons) and ko1 for free-to-bound recombination and donoracceptor pair. The model was further rened by Shibata et al27, who provided an analytical formula to conrm that 1oko2, even for free excitons.

The physics behind the process is the photo-neutralization of the donors/acceptors, which are present in all semiconductors and result in competitive recombination channels. Our k value agrees well with those reported for excitons in semiconductors26,27.

Figure 2a shows the RT PL decay curves monitored at different excitation energies. On the low-energy side, the PL lifetime is almost constant for all emission energies. On the high-energy side, the PL lifetime decreases with increasing emission energy. The PL decay curves can be well tted by the thermalized stretching exponential line shape28,29

I t

I1exp

" #

t t2

b

; 2

where ti is the decay time and Ii is the weight factor of each decay channel. A typical tting result is plotted in Fig. 2b (for all tting results, see Supplementary Fig. 2). The tting curves do not match normal mono-exponential or simple stretched-exponential decay (Supplementary Fig. 2c). Stretched-exponential decay is regarded as evidence of the exciton localization, in which the parameter b is related to the dimensionality of the localizing centres. The former exponential term in equation (2) represents the relaxation of free or extended states towards localized states, whereas the latter stretched-exponential term accounts for the communication between the localized states. We found that all the decay curves can be well tted with a constant b of0.430.03. Both lifetimes t2 and t1 show clear spectral dependence (Fig. 2c; Supplementary Fig. 2d). It markedly decreases on the high-energy side, while remaining constant on the low-energy side.

The spectral dependence of t2 can be described by a well-established model30 for excitons localized in the tail states

t E

tLE

1 expE EmeE0

; 3

where tLE is the lifetime of localized excitons, Eme can be regarded as the mobility edge and E0 is a characteristic energy of the

density of band tail states, which can be a measure of the localization energy. The best t of the data gives tLE 61 ns,

Eme 2.419 eV and E0 41 meV. The localization energy,

41 meV, is higher than the RT thermal energy, which is consistent with localized excitons being observed even at RT.

We provide evidence to show that the PL in our perovskite lms is not due to FC recombination but to a localization effect. The rst evidence comes from the PL spectra under high excitation. As shown in Supplementary Fig. 3, increasing the excitation density leads to the occurrence and increase of a higher energy tail at B2.41 eV. The result suggests that the main peak at 2.32 eV does not originate from FC recombination because FC emission has the highest energy among all radiative recombination. With increasing excitation density, the 2.32 eV peak is still dominant over the 2.41 eV FC emission and is almost unshifted. The unshifted luminescence agrees with the reported31 feature of localized excitons, which means that the many-particle effect of these localized excitons is weak. The second evidence comes from the frequency-dependent THz measurements (the experimental details can be found in Supplementary methods). Supplementary Figure 4 plots the photoconductivity induced by 400-nm pump pulses. The real part of the induced photoconductivity Ds1(o) decreases with decreasing frequency, and the imaginary part Ds2(o) has a

negative value at low frequency. The results do not support the Drude model for FCs32, which predicts increasing Ds1(o) with decreasing frequency and positive Ds2(o). The real part increasing with frequency and the negative imaginary part are typical signatures of carrier localization33,34. The low conductivity at low energy is consistent with the insulating nature of the charge-neutral excitons32. Such features build up immediately after excitation (Dt 3 ps) and are maintained

within the entire timescale (Dt 300 ps).

The above results indicate that the PL decay is dominated by recombination of localized excitons rather than FCs at RT. We also conducted excitation density-dependent PL and spectral-dependent lifetime measurements on CH3NH3PbBr3 lms at 237 K. These experiments were designed to check whether the assignment of localized excitons is valid for the perovskite lms at low temperatures, which have higher quantum efciency than that of the lms at RT. The temperature of 237 K was chosen based on the cubic-to-tetragonal phase transition35 of CH3NH3PbBr3. As shown in Fig. 3, the relative internal quantum efciency (IQE) at 237 K is estimated to be B80% based on the temperature-dependent PL intensity (Supplementary Note 1), a method that has been widely used in inorganic compound

t t1

I2exp

a b c

2.520 eV2.470 eV2.422 eV2.375 eV2.331 eV2.288 eV2.246 eV2.206 eV

T=300 K

103

102

101

100

0 500 1,000 1,500 2,000Delay time (ns)

PL intensity (a.u.)

100

101

102

103

0 500 1,000 1,500 2,000Delay time (ns)

Normalized PL intensity

1.2

1.0

0.8

0.6

0.4

0.2

0

2.1 2.2 2.3 2.4 2.5 2.6 Photon energy (eV)

100

80

60

40

20

0

[afii9848]1 = 8.1 ns

[afii9848]2 = 55.2 ns

[afii9826] = 0.456

Intensity (counts)

Lifetime (ns)

Figure 2 | Spectral-dependent PL decay of solution-processed CH3NH3PbBr3 lms. (a) PL decay curves monitored at various emission energies. The lifetime decreases markedly on the high-energy side of the emission. (b) Typical tting of a decay curve by the thermalized stretching exponential model described by equation (2). (c) The lifetime of localized excitons t2 (circles) as a function of emission energy. The data are tted with equation (3) (magenta line), with the lifetime of localized exciton tLE 60.5 ns, mobility edge of 2.419 eV and the localization energy E0 40.9 meV.

NATURE COMMUNICATIONS | 7:10896 | DOI: 10.1038/ncomms10896 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10896

Integrated PL intensity (a.u.)

1

Cubic Tetragonal

0.1

3 4 5 6 7 8 9 10

1,000/T (K1)

Figure 3 | PL thermal quenching behaviour of solution-processed CH3NH3PbBr3 lm. Integrated PL intensity as a function of the reciprocal temperature under low excitation (photocarrier density

B3.7 1014 cm 3). There is a cubic (RT) to tetragonal phase transition at

236 K, and tetragonal-to-tetragonal phase transition at 155 K. The magenta solid curve represents the tting result for the cubic data (Supplementary Note 1). The magenta dashed curve is the extrapolation result. The tetragonal data (open square) are plotted for comparison.

semiconductors36. The results (Supplementary Fig. 5) show that the k value and PL decay features at 237 K are similar to those at RT, indicating that localized excitons still dominate the PL under the condition of high quantum yield.

Recently, there have been arguments that at RT excitons cannot exist in perovskites because the dielectric constant of perovskite materials is very large3740, which results in dielectric screening and consequent dissociation of excitons. However, our evidence for exciton localization suggests that the role of the dielectric screening effect might be more complicated than expected. In fact, excitons do exist in materials with large dielectric constants. In ferroelectric oxides, such as SrTiO3 and

FeBiO3, it is accepted that self-trapped excitons41 and charge transfer vibronic excitons may exist42. The mechanism for these excitons can be quite different from the normal one. For example, charge transfer vibronic excitons is correlated electronic and hole polarons induced by the JahnTeller type lattice distortion, which are localized and can be quite stable at RT42. In view of these facts, we suggest that excitons can also exist in trihalide perovskites, despite the possible large dielectric constant under illumination. As additional experimental evidence, the absorption of exciton resonance is observed in the optical absorption spectra of CH3NH3PbBr3 lm (Supplementary Fig. 1b). Recent analysis of the absorption spectra suggested that bound exciton states also exist in CH3NH3PbI3 lms15.

The properties of perovskite lms may depend on the preparation methods and/or processing conditions. To test whether the exciton localization is a general mechanism, we prepared CH3NH3PbBr3 lms by the two-step method and measured the excitation-dependent PL and spectral-dependent lifetime (Supplementary Fig. 6). The samples also exhibit features similar to that shown in Fig. 2 and Fig. 3, indicating the localized exciton nature.

Exciton localization in CH3NH3PbI(Cl)3 lms. The emissions from solution-processed CH3NH3PbI3 and CH3NH3PbI3 xClx

lms are also dominated by localized excitons. The power-law dependence of the PL intensity on the excitation density reveals k values of B1.5 (Fig. 4a), and the PL decay shows spectral dependence (Supplementary Fig. 7) similar to that observed in CH3NH3PbBr3. The k values are larger than the value obtained in

CH3NH3PbBr3. According to the theory developed by Schmidt et al.26 and Shibata et al.27, such a difference mainly represents the different material properties such as the probabilities of radiative recombination and competitive nonradiative recombination. For example, the decrease of crystal perfection is expected to increase the k value. Moreover, the contribution of FC recombination may also lead to a change of k value. The PL lineshape analysis (Supplementary Fig. 8; Supplementary Note 2) indicates a small fraction (B9.5%) of FC recombination in the emission of CH3NH3PbI3 xClx lm, which is reasonable because

the reported exciton binding energy43 of CH3NH3PbI3 and CH3NH3PbI3 xClx is lower than CH3NH3PbBr3, so the thermal

dissociation of excitons is easier. The coexistence of exciton and FC recombination in the PL spectra has been observed in other materials44 with exciton binding energy comparable to the RT thermal energy. Excitation-dependent PL at low temperature reveals smaller k values (Supplementary Fig. 9), which can be interpreted by reduced thermal dissociation of excitons at low temperature.

It is of interest to determine whether the conclusion of localized excitons in perovskite materials is valid for the same material in a photovoltaic device structure. We construct such a structure using CH3NH3PbI3 xClx, as shown in the inset of

Fig. 4a. The PL intensity still shows power-law dependence on the excitation density, with a k value of 1.547, very close to the result of the bare lm. Moreover, the PL decay spectra (Supplementary Fig. 10) show dependence on the emission energy and the lifetime can be well described by equation (3) with E0 B17 meV, as seen in Fig. 4b. The results indicate that the PL of CH3NH3PbI3 xClx

lm in a typical photovoltaic structure is also dominated by the localized excitons rather than FC recombination.

DiscussionThe physical picture of the recombination of localized excitons is illustrated in Fig. 5. The density of localized states is approximated by an exponential tail with the form of Bexp ( E/E0). The excitons can be either partly localized (one carrier

is localized with another carrier bound to it by Coulomb attraction) or wholly localized21. With increasing energy, the localized excitons may transit to the extended exciton states (approaching free excitons) at the transition region known as the mobility edge. Under low excitation, most of the photocarriers occupy the tail states. The picture can also be understood in the space coordinates as shown in Fig. 5b. In this case, the tail states are represented by the local potential minima in the conduction and/or valence bands. The photogenerated carriers transfer to these potential minima to form localized excitons, which have much longer lifetime than free excitons due to the transfer between localized states28,29. The long lifetime of the localized excitons accounts for the observed long PL lifetime. This phenomenon is also observed in inorganic semiconductors. For example, localized exciton lifetime as long as 65 ns at RT has been reported29 in InGaN.

Tail states are very common in semiconductors and can be induced by doping, compositional changes and structural deformation21,45. Although solution-processed perovskites are materials with reasonable crystal quality, structure imperfections are inevitable. For example, the large rotational freedom of the polar CH3NH3 cation can produce structural disorder37,46.

Unintentional/intentional doping is possible. The weak bonding between lead and halogens may also produce local disorder, especially in the surface and crystal boundary region. Recent studies47 revealed that the grain boundaries exhibited faster nonradiative decay. Other results48 suggest that perovskites with larger grains exhibit better photovoltaic performance. Given these

4 NATURE COMMUNICATIONS | 7:10896 | DOI: 10.1038/ncomms10896 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10896 ARTICLE

a b

MAPbl3x CIx device

MAPbl3x CIx

MAPbl3

k = 1.547

k = 1.569

k = 1.513 PCBM Perovskite

PEDOT:PSS

ITO

100

101

102

103

104

102 101 100 101

Excitation fluence (nJ cm2)

Normalized integrated PL intensity

PL intensity (a.u.)

25

20

15

10

5

0

Lifetime (ns)

1.50 1.55 1.60 1.65 1.70 1.75

Photon energy (eV)

Figure 4 | Steady-state PL spectra and transient PL decay of CH3NH3PbI3 and CH3NH3PbI3 xClx. (a) Logarithm plot of the integrated PL intensity versus

excitation density for CH3NH3PbI3 and CH3NH3PbI3 xClx lms. Insert is a typical photovoltaic structure with a CH3NH3PbI3 xClx lm sandwiched

between two charge-transporting interlayers. The data show good power-law dependence with k values of 1.569, 1.513 and 1.547. (b) PL lifetime (solid

circles) of ITO/PEDOT:PSS/ CH3NH3PbI3 xClx/PCBM photovoltaic structure as a function of the emission energy. The data are tted with equation (3)

(magenta line), with the lifetime of localized exciton tLE 20.5 ns, mobility edge 1.689 eV and localization energy E0 17.3 meV. The PL lifetime is greatly

reduced due to the quenching effects of the adjacent layers.

Energy

Extended states ~(E Eme)1/2

Mobility edgeEmeLocalized states ~ exp(E Eme) / E0

Density of states (a.u.) PL intensity (a.u.) Space coordinate

a b

Relaxation

Localized exciton

EC

EV

Above gap

excitation

Figure 5 | Physical picture of exciton localization. (a) The density of the extended and localized states, and emission in perovskites (schematic). The density of the localized states is approximated by an exponential tail with the form of Bexp( E/E0). The localized and extended states are divided

by the mobility edge. Under low excitation, the excitons mainly occupy the localized states. Under high excitation, the localized states can be lled and the photocarriers also occupy the extended states, leading to emissions of free carriers. (b) Schematic drawing of exciton localization in space coordinates. With the presence of structural disorder, the tails of the localized states form local potential uctuation in the energy bands. These potential minima can localize electrons and holes to form localized excitons. The carriers can transfer between the local potential minima, leading to long PL lifetime.

facts, it is reasonable to suggest the existence of tail states in solution-processed perovskite lms.

The energy and intensity of localized state-related emission depend strongly on the nature (for example, energy level) and density of the localized states. In the case of strong localization, such as deep-level centres, the emission energy may be

substantially reduced. The localized states may emit weakly provided their density is low. In the present case, strong near-band-edge emission can be realized, indicating that excitons are weakly localized in the band tail states. The weak localization is evidenced by the relatively small localization energy reected by the E0 values in equation (3) and by the small Stokes shift of the luminescence (Supplementary Fig. 1b). The localization effect can even be benecial to light-emitting devices. Localized exciton may contribute to the optical gain because the localized states can be easily lled, provided their density is not too high31,49,50. We performed temperature-dependent PL measurements under moderate excitation to check whether nonradiative channels become dominant when the excitation density increases. Supplementary Fig. 11 shows that the relative IQE of moderate excitation is much higher than that of low excitation, in agreement with the result reported by Deschler et al8. This result implies that other decay channels such as Auger recombination do not become dominant, which will be of great benet to low threshold lasing. We conducted PL experiments at high excitation (photocarrier concentration up to B1019 cm 3).

Amplied spontaneous emission was observed on the low-energy side (at 2.26 eV) of the localized exciton emission with a threshold of B300 mJ cm 2 (Supplementary Fig. 3b). It is noteworthy that

FC emission (B2.41 eV) emerges with increasing excitation density. These results suggest that the amplied spontaneous emission is likely from localized excitons, which indicates that the optical gain can come from the lled localized states, and it is possible to achieve a low threshold by controlling the density of localized states. Moreover, exciton localization may also increase the luminescence efciency because the oscillator strength of the transitions in the localized states is greatly enhanced49,51.

We emphasize that the generation of localized excitons at excitation levels close to the working regime of solar cells does not conict the fact that the hybrid perovskites are superior photovoltaic materials. The localized excitons can diffuse by a thermally activated multiple trappingescaping process52,53. This process leads to an exponential dependence of the diffusion

NATURE COMMUNICATIONS | 7:10896 | DOI: 10.1038/ncomms10896 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10896

Fabrication of the photovoltaic device structure. The structure of ITO/PEDOT:PSS/perovskite/PCBM was fabricated on patterned ITO-coated glass substrates (sheet resistance: 15 O sq 1). The substrates were cleaned sequentially in acetone, ethanol, deionized water and ethanol for 10 min each, followed by oxygen plasma treatment for 15 min. A poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) layer was spin-coated onto the substrates at 4,000 r.p.m. for 60 s and then annealed in air at 140 C. The sample was transferred into a glove box. Next, the CH3NH3PbI3 xClx precursor solution was spin-coated at 3,000 r.p.m. for 45 s,

followed by annealing on a hot plate at 100 C for B60 min. The [6,6]-phenyl-C61-butyric acid methyl ester (PCBM) layers were deposited from a 30 mg ml 1 chlorobenzene solution at 2,000 r.p.m. for 45 s. To avoid the degradation induced by air exposure, the devices were packaged with glass.

PL measurements. The steady-state and time-resolved PL measurements were performed on a FLS920 uorescence spectrophotometer (Edinburgh Instruments). To minimize the effect of air exposure, all the measurements were taken in vacuum with pressure of o0.01 torr. Pulsed laser diodes (EPL405 and EPL635) with tunable repeating frequency of 20 kHz to 20 MHz were used as the excitation source. For CH3NH3PbI3 and CH3NH3PbI3 xClx lms, EPL635 with a wavelength

of 638.8 nm and pulse width of 86.4 ps was used. For CH3NH3PbBr3, we used EPL405 with a wavelength of 404.2 nm and pulse width of 58.6 ps. The excitation uence for both wavelengths was B4 nJ cm 2. For excitation density-dependent measurements at the low level, the lasers operated at 20 MHz and the light uence was tuned by a neutral attenuator. For moderate excitation experiments, a 355 nm frequency-tripled Nd:YAG laser (FTSS 355-50, CryLaS GmbH) with a pulse width of 1 ns and a repetition rate of 100 Hz was used. For lifetime measurements by time-correlated single-photon counting, the lasers operated at 200 kHz. The temperature-dependent measurements were performed with a closed-cycle helium cryostat.

Calculation of the photocarrier density. The corresponding photocarrier density can be calculated as rexc

light uence density of a single pulse/(photon energy optical penetration

depth) 4 nJ cm 2/(3.07 eV 1.6 10 19 J 220 nm)B3.7 1014 cm 3

Here the optical penetration depth of CH3NH3PbBr3 is taken as B220 nm (ref. 55). In this calculation, we assume a constant excitation because the effective excited volume is remarkably larger than the directly excited one due to the carrier diffusion during the long carrier lifetime21. For CH3NH3PbI3 and

CH3NH3PbI3 xClx, the photon energy of the excitation laser was 1.94 eV, and the

optical penetration depth was 250 nm (ref. 1); hence, the photocarrier density was B5.2 1014 cm 3.

References

1. Kojima, A., Teshima, K., Shirai, Y. & Miyasaka, T. Organometal halide perovskites as visible-light sensitizers for photovoltaic cells. J. Am. Chem. Soc. 131, 60506051 (2009).

2. Lee, M. M., Teuscher, J., Miyasaka, T., Murakami, T. N. & Snaith, H. J. Efcient hybrid solar cells based on meso-superstructured organometal halide perovskites. Science 338, 643647 (2012).

3. Burschka, J. et al. Sequential deposition as a route to high-performance perovskite-sensitized solar cells. Nature 499, 316319 (2013).

4. Jeon, N. J. et al. Compositional engineering of perovskite materials for high-performance solar cells. Nature 517, 476480 (2015).

5. Bai, S. et al. High-performance planar heterojunction perovskite solar cells: preserving long charge carrier diffusion lengths and interfacial engineering. Nano Res. 7, 17491758 (2014).

6. Research Cell Efciency Records. Available at http://www.nrel.gov/ncpv/

Web End =http://www.nrel.gov/ncpv/ (2015).

7. Xing, G. et al. Low-temperature solution-processed wavelength-tunable perovskites for lasing. Nat. Mater. 13, 476480 (2014).

8. Deschler, F. et al. High photoluminescence efciency and optically pumped lasing in solution-processed mixed halide perovskite semiconductors. J. Phys. Chem. Lett. 5, 14211426 (2014).

9. Tan, Z. K. et al. Bright light-emitting diodes based on organometal halide perovskite. Nat. Nanotechnol 9, 687692 (2014).

10. Wang, J. P. et al. Interfacial control toward efcient and low-voltage perovskite light-emitting diodes. Adv. Mater. 27, 23112316 (2015).

11. Stranks, S. D. et al. Electron-hole diffusion lengths exceeding 1 micrometer in an organometal trihalide perovskite absorber. Science 342, 341344 (2013).

12. Xing, G. C. et al. Long-range balanced electron and hole-transport lengths in organic-inorganic CH3NH3PbI3. Science 342, 344347 (2013).

13. Laquai, F. All-round perovskites. Nat. Mater. 13, 429430 (2014).14. DInnocenzo, V. et al. Excitons versus free charges in organo-lead tri-halide perovskites. Nat. Commun. 5, 35863591 (2014).

15. Saba, M. et al. Correlated electronhole plasma in organometal perovskites. Nat. Commun. 5, 50495058 (2014).

coefcient on the exciton energy (Eexc), as well as on the

temperature53

D D0exp

Eexc EmekBT ; 4

where D0 is the diffusion coefcient of the extended excitons. Taking the energetic distance Eexc Eme B70 meV (as extracted

from Figs 2c,4b), we estimate the RT diffusion coefcient of localized excitons D as 0.067D0, only approximately one order of magnitude lower than the free exciton. If we invoke the reported11,12, D value of 0.011 0.054 cm2 s 1 for the

perovskites, a D0 of B0.16 0.80 cm2 s 1 is obtained. This

value is comparable to the reported values53,54 for CdS1 xSex

(0.3 cm2 s 1) and In1 xGaxN (0.51.1 cm2 s 1), two alloy

semiconductors known as the typical examples for the study of localized excitons. In combination with the very long lifetime, a long carrier diffusion length in perovskites is expected in the scenario of exciton localization.

In summary, we have presented solid evidence that the RT PL in organolead trihalide perovskites is dominated by weakly localized excitons. The evidence includes: (i) the excellent power-law dependence of the PL intensity on the excitation intensity with 1oko2, (ii) the localization effect indicated by the spectral dependence of the PL lifetime and frequency-dependent THz photoconductivity and (iii) the coexistence of exciton and FC recombination under high excitation. Exciton localization is suggested as the origin of the long PL lifetime in this class of materials. We nd that the localization effect is general in solution-process perovskite lms due to the presence of crystal imperfections. The localization of excitons strongly inuences the transport and recombination properties of perovskite materials. The dominance of the localized exciton in the recombination channels as well as its higher IQE under moderate excitation, strongly suggests that it is possible to utilize these benets to realize low threshold lasing in perovskites, as has been demonstrated in IIIV and IIVI semiconductors and devices. The elaborate tailoring of the localization effect in perovskites is thus highly attractive in designing future high-performance optoelectronic devices.

Methods

Synthesis of perovskite lms. All the indium-tin oxide (ITO)-coated glass substrates were cleaned sequentially in deionized water, ethanol, acetone and oxygen plasma before spin-coating. The perovskite CH3NH3PbI3 xClx was pre

pared according to the reported procedure11. Methylamine iodide was prepared by reacting 33 wt % methylamine in ethanol (Sigma-Aldrich), with 57 wt % hydroiodic acid in water (Sigma-Aldrich), at RT. Hydroiodic acid was added dropwise while stirring. After drying at 100 C, the resultant white powder was dried overnight in a vacuum oven and was recrystallized from ethanol before use. To form the CH3NH3PbI3 xClx precursor solution, methylammonium iodide and lead (II)

chloride (Sigma-Aldrich) were dissolved in anhydrous N,N-dimethylformamide (DMF) in a 3:1 molar ratio of methylamine iodide to PbCl2, with nal concentrations 0.88 M lead chloride and 2.64 M methylammonium iodide. The precursor was ltered through a 220-nm polytetrauoroethylene (PTFE) lter head, then spin-coated at 3,000 r.p.m. for 30 s on ITO-coated glass; nally, it was annealed at 95 C for B10 min. CH3NH3PbI3 was prepared via the sequential deposition route3. A PbI2 (Sigma-Aldrich) solution in DMF (462 mg ml 1) was spin-coated on glass substrate and then kept at 70 C. After drying, the lms were dipped in a solution of CH3NH3I in 2-propanol (10 mg ml 1) for B60 s and rinsed with 2-propanol, and then spin-coated to form uniform CH3NH3PbI3 thin lms. For CH3NH3PbBr3 preparation,24 CH3NH3Br was rst prepared by mixing methylamine with hydrobromic acid (48% in water; CAUTION: exothermic reaction) in 1:1 molar ratio in a 100-ml ask under continuous stirring at 0 C for 2 h. CH3NH3Br was then crystallized by removing the solvent in an evaporator, washing three times in diethyl ether for 30 min and ltering the precipitate. The material, in the form of white crystals, was then dried in vacuum at 60 C for 24 h and was then kept in a dark, dry environment until further use. A 20-wt % solution of CH3NH3PbBr3 was prepared by mixing PbBr2 and CH3NH3Br in a 1:3 molar ratio in DMF. The precursor was spin-coated at 4,000 r.p.m. for 30 s on ITO-coated glass and was annealed at 60 C for B10 min.

6 NATURE COMMUNICATIONS | 7:10896 | DOI: 10.1038/ncomms10896 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10896 ARTICLE

16. Sun, S. et al. The origin of high efciency in low-temperature solution-processable bilayer organometal halide hybrid solar cells. Energy Environ. Sci. 7, 399407 (2014).

17. Savenije, T. J. et al. Thermally activated exciton dissociation and recombination control the organometal halide perovskite carrier dynamics. J. Phys. Chem. Lett. 5, 21892194 (2014).

18. Zhang, W. et al. Enhancement of perovskite-based solar cells employing core-shell metal nanoparticles. Nano Lett. 13, 45054510 (2013).

19. Yamada, Y., Nakamura, T., Endo, M., Wakamiya, A. & Kanemitsu, Y. Photocarrier recombination dynamics in perovskite CH3NH3PbI3 for solar cell applications. J. Am. Chem. Soc. 136, 1161011613 (2014).

20. Lin, Q. Q., Armin, A., Nagiri, R. C. R., Burn, P. L. & Meredith, P. Electro-optics of perovskite solar cells. Nat. Photon. 9, 106112 (2015).

21. Klingshirn, C. F. Semiconductor Optics (Springer-Verlag, 1997).22. Cingolani, R. et al. Radiative recombination processes in wide-band-gap IIVI quantum wells: the interplay between excitons and free carriers. J. Opt. Soc. Am. B 13, 12681277 (1996).

23. 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, 15841589 (2014).

24. Edri, E., Kirmayer, S., Cahen, D. & Hodes, G. High open-circuit voltage solar cells based on organic-inorganic lead bromide perovskite. J. Phys. Chem. Lett. 4, 897902 (2013).

25. Stranks, S. D. et al. Recombination kinetics in organic-inorganic perovskites: excitons, free charge, and subgap states. Phys. Rev. Appl. 2, 034007 (2014).

26. Schmidt, T., Lischka, K. & Zulehner, W. Excitation-power dependence of the near-band-edge photoluminescence of semiconductors. Phys. Rev. B 45, 89898994 (1992).

27. Shibata, H. et al. Excitation-power dependence of free exciton photoluminescence of semiconductors. Jpn J. Appl. Phys. 44, 61136114

2005:

28. Sun, Y. J. et al. Nonpolar InxGa1-xN/GaN (1-100) multiple quantum wells grown on g-LiAlO2 (100) by plasma-assisted molecular-beam epitaxy. Phys. Rev. B 67, 041306 (2003).

29. Onuma, T. et al. Localized exciton dynamics in nonpolar (11-20) InxGa1 xN

multiple quantum wells grown on GaN templates prepared by lateral epitaxial overgrowth. Appl. Phys. Lett. 86, 151918 (2005).30. Gourdon, C. & Lavallard, P. Exciton transfer between localized states in CdS1xSex alloys. Phys.Stat. Sol. (b) 153, 641652 (1989).

31. Majumder, F. A., Shevel, S., Lyssenko, V. G., Swoboda, H. E. & Klingshirn, C. Luminescence and gain spectroscopy of disordered CdS1-xSex under high

excitation. Z. Phys. B 66, 409418 (1987).32. Kaindl, R. A., Carnahan, M. A., Hagele, D., Lovenich, R. & Chemla, D. S. Ultrafast terahertz probes of transient conducting and insulating phases in an electron-hole gas. Nature 423, 734738 (2003).

33. La-o-vorakiat, C. et al. Elucidating the role of disorder and free-carrier recombination kinetics in CH3NH3PbI3 perovskite lms. Nat. Commun. 6, 79037909 (2015).

34. Nemec, H., Kuzel, P. & Sundstrm, V. Charge transport in nanostructured materials for solar energy conversion studied by time-resolved terahertz spectroscopy. J. Photochem. Photobiol. A. 215, 123139 (2010).

35. Mashiyama, H., Kurihara, Y. & Azetsu, T. Disordered cubic perovskite structure of CH3NH3PbX3 (X Cl, Br, I). J. Korean Phys. Soc. 32, S156S158

(1998).36. Okamoto, K. et al. Surface-plasmon-enhanced light emitters based on InGaN quantum wells. Nat. Mater. 3, 601605 (2004).

37. Juarez-Perez, E. J. et al. Photoinduced giant dielectric constant in lead halide perovskite solar cells. J. Phys. Chem. Lett. 5, 23902394 (2014).

38. Frost, J. M. et al. Atomistic origins of high-performance in hybrid halide perovskite solar cells. Nano Lett. 14, 25842590 (2014).

39. Menendez-Proupin, E., Palacios, P., Wahnon, P. & Conesa, J. C. Self-consistent relativistic band structure of the CH3NH3PbI3 perovskite. Phys. Rev. B 90, 045207 (2014).

40. Even, J., Pedesseau, L. & Katan, C. Analysis of multivalley and multibandgap absorption and enhancement of free carriers related to exciton screening in hybrid perovskites. J. Phys. Chem. C 118, 1156611572 (2014).

41. Kan, D. et al. Blue-light emission at room temperature from Ar -irradiated SrTiO3. Nat. Mater. 4, 816819 (2005).

42. Vikhnin, V. S., Eglitis, R. I., Kapphan, S. E., Borstel, G. & Kotomin, E. A. Polaronic-type excitons in ferroelectric oxides: microscopic calculations and experimental manifestation. Phys. Rev. B. 65, 104304 (2002).

43. Tanaka, K. et al. Comparative study on the excitons in lead-halide-based perovskite type crystals CH3NH3PbBr3 CH3NH3PbI3. Solid State Commun.

127, 619623 (2003).44. Wang, H., Wong, K. S. & Wong, G. K. L. in Proceedings of SPIE 3624, 1324 (International Society for Optical Engineering, San Jose, CA, USA, 1999).

45. Permogorov, S. & Reznitsky, A. Effect of disorder on the optical-spectra of wide-gap II-VI semiconductor solid-solutions. J. Lumin. 52, 201223 (1992).

46. Wasylishen, R. E., Knop, O. & Macdonald, J. B. Solid State Commun. 56, 581 (1985).

47. deQuilettes, D. W. et al. Impact of microstructure on local carrier lifetime in perovskite solar cells. Science 348, 683686 (2015).

48. Nie, W. Y. et al. High-efciency solution-processed perovskite solar cells with millimeter-scale grains. Science 347, 522525 (2015).

49. Satake, A. et al. Localized exciton and its stimulated emission in surface mode from single-layer InxGa1-xN. Phys. Rev. B 57, R2041R2044 (1998).

50. Chen, R. et al. Exciton localization and optical properties improvement in nanocrystal-embedded ZnO core-shell nanowires. Nano Lett. 13, 734739 (2013).

51. ODonnell, K. P., Martin, R. W. & Middleton, P. G. Origin of luminescence from InGaN diodes. Phys. Rev. Lett. 82, 237240 (1999).

52. Chichibu, S. F. et al. Recombination dynamics of localized excitons in cubic InxGa1 xN/GaN multiple quantum wells grown by radio frequency molecular

beam epitaxy on 3CSiC substrate. J. Vac. Sci. Technol. B 21, 18561862 (2003).53. Schwab, H., Pantke, K. H., Hvam, J. M. & Klingshirn, C. Measurements of exciton diffusion by degenerate four-wave mixing in CdS1-xSex. Phys. Rev. B 46,

75287532 (1992).54. Okamoto, K. et al. in Proceedings of SPIE (International Society for Optical Engineering) 4278, 157150 (Bellingham, WA, USA, 2001).

55. Kumawat, N. K., Dey, A., Narasimhan, K. L. & Kabra, D. Near infrared to visible electroluminescent diodes based on organometallic halide perovskites: structural and optical investigation. ACS Photon. 2, 349354 (2015).

Acknowledgements

This work was supported by the Natural Science Foundation of China (nos 51372223, 91333203, 51522209, 11474164, 61405091, 11474249 and 91433204), the Program for Innovative Research Team in University of Ministry of Education of China (no. IRT13037), the National Basic Research Program of China- Fundamental Studies of Perovskite Solar Cells (2015CB932200), the Natural Science Foundation of Jiangsu Province, China (BK20131413, BK20140952), the National 973 Program of China (2015CB654901), the Synergetic Innovation Center for Organic Electronics and Information Displays and the Fundamental Research Funds for the Central Universities (nos. 2014FZA4008 and 2015FZA3005). We thank Mr Yunzhou Deng (Zhejiang University) for his help in the numerical tting of the PL spectra.

Author contributions

H.H. and Z.Y. supervised the study. H.H., Q.Y., H.L. and J.L. contributed to the PL measurements and analysis. Q.Y., J.S. and N.W. contributed to the synthesis and characterization of the materials. J.H., X.W. and Y.Z. contributed to the THz measurements and analysis. Y.J. and J.W. provided input to the data analysis and discussed the results. H.H. and Y.J. wrote the manuscript. All authors assisted in manuscript preparation.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions

Web End =reprintsandpermissions /

How to cite this article: He, H. et al. Exciton localization in solution-processed organolead trihalide perovskites. Nat. Commun. 7:10896 doi: 10.1038/ncomms10896 (2016).

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 articles 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/

Web End =http://creativecommons.org/licenses/by/4.0/

NATURE COMMUNICATIONS | 7:10896 | DOI: 10.1038/ncomms10896 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7