Large polarization-dependent exciton optical Stark effect in lead iodide perovskites

Abstract

A strong interaction of a semiconductor with a below-bandgap laser pulse causes a blue-shift of the bandgap transition energy, known as the optical Stark effect. The energy shift persists only during the pulse duration with an instantaneous response time. The optical Stark effect has practical relevance for applications, including quantum information processing and communication, and passively mode-locked femtosecond lasers. Here we demonstrate that solution-processable lead-halide perovskites exhibit a large optical Stark effect that is easily resolved at room temperature resulting from the sharp excitonic feature near the bandedge. We also demonstrate that a polarized pump pulse selectively shifts one spin state producing a spin splitting of the degenerate excitonic states. Such selective spin manipulation is an important prerequisite for spintronic applications. Our result implies that such hybrid semiconductors may have great potential for optoelectronic applications beyond photovoltaics.

ARTICLE

Received 3 Jun 2016 | Accepted 15 Jul 2016 | Published 31 Aug 2016

Ye Yang1, Mengjin Yang1, Kai Zhu1, Justin C. Johnson1, Joseph J. Berry1, Jao van de Lagemaat1

& Matthew C. Beard1

A strong interaction of a semiconductor with a below-bandgap laser pulse causes a blue-shift of the bandgap transition energy, known as the optical Stark effect. The energy shift persists only during the pulse duration with an instantaneous response time. The optical Stark effect has practical relevance for applications, including quantum information processing and communication, and passively mode-locked femtosecond lasers. Here we demonstrate that solution-processable lead-halide perovskites exhibit a large optical Stark effect that is easily resolved at room temperature resulting from the sharp excitonic feature near the bandedge. We also demonstrate that a polarized pump pulse selectively shifts one spin state producing a spin splitting of the degenerate excitonic states. Such selective spin manipulation is an important prerequisite for spintronic applications. Our result implies that such hybrid semiconductors may have great potential for optoelectronic applications beyond photovoltaics.

DOI: 10.1038/ncomms12613 OPEN

Large polarization-dependent exciton optical Stark effect in lead iodide perovskites

1 Chemistry and Nanoscience Center, National Renewable Energy Laboratory, Golden, Colorado 80401, USA. Correspondence and requests for materials

should be addressed to Y.Y. (email: mailto:[email protected]

Web End [email protected] ) or to M.C.B. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:12613 | DOI: 10.1038/ncomms12613 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12613

The optical Stark effect (OSE) results from a coherent interaction between excitonic states and a non-resonant pump photon eld, and is not associated with the

photo-generation of real carriers or excitons1. The OSE results in a shifting of the exciton transition energy to higher energy in a semiconductor by optical pumping with photon energies that are below the bandgap (non-resonant)2,3. Because of the coherent nature, the OSE has many promising applications, for example, in quantum information processing and communication, serving as a potential ultrafast optical switch3,4, source of quantum entanglement5,6 and a method for spin manipulation710. The OSE can also be utilized to produce ultra-short laser pulses serving as a passive mode-locking saturable absorber1113.

To serve as a good OSE system, semiconductor candidates require quantum states that are strongly coupled to photons to enable effective optical control and they should have narrow transition bandwidths that can exhibit a high-contrast spectral modulation. Quantum-conned semiconductors such as quantum wells2,14,15, quantum dots7,10,16,17, two-dimensional semiconductor monolayers1820 and carbon nanotubes21 have been intensively studied for OSE because of their strong excitonphoton interactions and discrete exciton transitions. Unfortunately, detecting OSE in bulk semiconductors requires extremely low temperatures to achieve a well-resolved exciton transition that is usually buried in continuum band transitions14,22. On the other hand, recent spectroscopic investigations of lead-halide perovskites have uncovered that the excitonic transitions in these semiconductors have the prerequisite conditions for an OSE system; narrow bandwidth and large transition oscillator strength2326.

Here we report the observation of polarization-selective OSE of bandedge excitonic states in methylammonium lead iodide (MAPbI3) perovskite lms at room temperature. The OSE is generated and detected by using transient absorption (TA) spectroscopy. Under similar pump intensity and detuning, the observed OSE-induced energy shift is nearly an order of magnitude larger than in previously reported quantum-conned systems2,20,21 and is similar to that observed in monolayer WSe2, a two-dimensional semiconductor19 that is gaining increasing interest in spintronic investigations. Within this context a broader set of applications beyond photovoltaic can be envisioned for these solution-processed hybrid semiconductors.

ResultsOptical Stark effect. We are able to selectively dress the excitonic states within MAPbI3 lms by using circularly polarized photons from a femtosecond laser source that is tuned below the exciton transition energy, so that ideally no excitons/carriers are created. A low-intensity circularly polarized broadband probe pulse, with photon energies that span the exciton transition energy, conrms that the dressed excitonic states are shifted towards higher energy, and the shift depends linearly on the pump intensity. The corresponding energy shift (dE), applying a perturbation of the excitonphoton interaction in a two-level system, is given by19,21,27

dE

m0X

2 F

h i2

. Optical transitions from the ground state to

the exciton states are allowed for Dmj 1. At large detuning

energies compared with the exciton-binding energy (DcRex) the

degenerate excitonic states can be treated as multiple independent two-level systems and equation (1) is still valid to predict the energy shift of each excitonic state29 (see Supplementary Note 1 for a derivation of equation (1)).

The polarization selectivity of the OSE is the basis for its experimental demonstration, and our experiment is described in Fig. 1. Excitonic states with mj 1 can only couple to circularly

polarized photons with spin magnetic quantum number (ms) of

1, represented by s , to preserve angular momentum.

Therefore, when a s photon couples to a mj 1 excitonic state

a new entangled state results with transition energy E0 dE. Since

this new entangled state also carries total angular momentum with mj 1, it can only be detected by another s photon that is tuned

to E0 dE. If, in contrast, a counter-polarized probe photon (s )

with ms of 1 is used, no difference in the energy level will be

detected since it cannot induce exciton transitions with mj 1. For

the same reason, a s photon cannot couple with an exciton state

of mj 1, leaving that transition unperturbed.

Linear absorption. The preparation of the MAPbI3 lms is described in the Methods section, and the lm is characterized by X-ray diffraction (Supplementary Fig. 1), showing a highly crystalline perovskite structure in the tetragonal phase. A relatively large in-plane grain size (around 12 mm) is indicated by the scanning electron microscope image (Supplementary Fig. 2). The linear absorption spectrum of the sample (Fig. 2, red circles) is determined by measuring transmission and reection (Supplementary Fig. 3). There is a sharp onset followed by a nearly constant absorption, and such behaviour can be attributed to Coulombic interactions between the electron and hole within

Probe

|1

|0 |0

|1

E0

mj= 1

ms= 1

[afii9846]

D 1

where m0X is the transition dipole moment between the ground and a non-degenerate exciton state, F

h i is the time-averaged

electric eld of excitation light and D is the detuning of the excitation energy from the exciton resonance (D E0 :o). From

equation (1), the energy shift, dE, is proportional to both the square of the transition dipole moment and the excitation intensity as F

h i22I0= e

p e0c , where I0 is light intensity, e is dielectric

constant of the sample, and e0 and c are the vacuum permittivity

and speed of light, respectively. Optical selection rules implicitly govern dE through the transition dipole moment.

Excitonic states are primarily built from conduction and valence bandedge states, and for lead-iodide perovskites both bands are twofold degenerate with a total angular momentum number of 1/2 (ref. 28). The two lowest bright exciton states are denoted by the azimuthal quantum number of the total angular momentum mj

Figure 1 | Energy diagram of the exciton energy shift due to OSE. The

diagram shows energy levels for the ground state, 0

j i, and the two bright

exciton states, 1

j i, labelled by the azimuthal quantum number of the total

angular momentum, mj. The exciton transition energy is E0. When a photon

with energy E0 D and with spin momentum s interacts with the mj 1

exciton, then an entangled state with energy E0 dE forms (shaded box).

The new state can be detected by a probe photon with the same spin, while

the mj 1 exciton is left unperturbed.

2 NATURE COMMUNICATIONS | 7:12613 | DOI: 10.1038/ncomms12613 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12613 ARTICLE

the Elliott model (Fig. 2, black dash line)2325. The absorption components of the excitonic and continuum transitions (green and blue dash lines, respectively) are decoupled from the spectral tting. The exciton resonance (E0) is centred at 1.631 eV, and the bandgap (Eg) of continuum transitions is found at 1.642 eV, which yields an exciton binding energy (Rex Eg E0) of 11 meV,

consistent with recently reported values25,3032. The full-width at half-maximum of the exciton transition is found to be 73 meV (tting parameters are tabulated in Supplementary Table 1).

Transient absorption. A high-intensity polarized monochromatic pump pulse (:op 1.55 eV) is used to couple to the

exciton states (thus the detuning, D, is 81 meV and therefore DcRex, justifying the use of equation (1)). The resulting change in absorption (DA) is captured by a weak broadband polarized probe pulse (:o 1.552.1 eV). Because, in our experiment, OSE

is created and probed by femtosecond pulses, the duration of the OSE response is determined by the pumpprobe correlation time (around 180 fs). There is a slight spectral overlap between the excitation pulse and the unperturbed exciton transition, resulting in a small population of excitons that quickly dissociate to form free carriers (o5 1016 cm 3). Since the density of real

carriers is too low to screen Coulomb interactions (Debye length is around 50 nm) the major effect is to interfere with the observation of the OSE through the reduction of the exciton transition strength via phase space lling. The carrier-induced TA signal lasts for over tens of nanoseconds until the carriers recombine30, however, due to the drastic difference in lifetime, DA resulting from OSE and from the presence of carriers can be isolated from one another using a simple spectral subtraction procedure (Supplementary Figs 4 and 5). To focus on the OSE, we only show the subtracted TA spectra (raw TA spectra are provided in Supplementary Fig. 4).

The OSE is rst studied under different congurations of pumpprobe polarization. The pump pulse intensity (I0) is

controlled to be the same regardless of polarization in this experiment. When the pump and probe are co-circularly polarized, denoted as (s ,s ), a prominent absorption change anti

symmetric to the exciton resonance centre (:o 1.631 eV)

appears at zero pumpprobe delay (Fig. 3a). The narrow temporal width (around 200 fs) of DA is consistent with the pumpprobe correlation time. On the contrary, when the pump and probe pulses are circularly polarized opposite to one another, denoted as (s ,s ), no OSE features are detected (Fig. 3b). The

above observations suggest that the s pump selectively induces

an energy shift of the exciton transitions with Dmj 1, and

the s probe can detect the modied exciton transitions while

the s probe cannot see them, consistent with angular

momentum conservation considerations discussed above. The pumpprobe congurations of (s ,s ) and (s ,s ) show the

same spectra as (s ,s ) and (s ,s ), respectively

(Supplementary Fig. 6).

The sample is also examined under two linearly polarized pumpprobe congurations, parallel (VV) and perpendicular (HV) pumpprobe polarization. The TA spectra show that similar DA signals due to OSE are observed under both congurations, and the magnitudes are around half of that measured under (s ,s ) conguration (Supplementary Fig. 7).

1.0

0.8

0.6

0.4

0.2

0.0

Absorption (OD)

1.5

1.6

1.7

1.8

1.9

Photon energy (eV)

Figure 2 | Absorption spectrum and Elliotts model of the MAPbI3 lms.

The absorption spectrum (red circles) near the bandgap was t by Elliotts

model (black dash line). The contribution from excitonic (green dash line)

and continuum band (blue dash line) transitions are also plotted.

1.00.80.60.40.20.0 0.2 0.4

a

b

1.00.80.60.40.20.0 0.2 0.4

1.5

[afii9846]+ pump

[afii9846]+ probe

1.0

[afii9846]+ pump

[afii9846] probe

1.5

0.5

1.0

0.0

A (mOD)

0.5

0.5

0.0

A (mOD)

1.0

0.5

Delay (ps)

1.5

Delay (ps)

1.0

1.55

1.60

1.65

1.70

1.75

1.55

1.60

1.65

1.70

1.75

Photon energy (eV)

Photon energy (eV)

1.5

Figure 3 | OSE-induced TA spectra of MAPbI3 lms. The TA spectra are measured by (a) co- (s ,s ) and (b) counter-circularly (s ,s ) polarized

pumpprobe congurations. The y axis and x axis represent the pumpprobe delay and probe photon energy, respectively. As indicated by the colour scale

bars, the blue and red represent the photon-induced absorption and bleach, respectively. The colour intensity reects signal magnitude.

NATURE COMMUNICATIONS | 7:12613 | DOI: 10.1038/ncomms12613 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12613

Because linear polarization consists of equal amounts of s and

s components, both (VV) and (HV) congurations can be

considered as a mixture of 50% (s,s8) and 50% (s,s). Therefore, the OSE magnitude detected by the (VV) and (HV) congurations are the same, and the magnitude is only 50% of that detected by (s ,s ).

DiscussionBecause we use TA to observe the OSE, the resulting DA spectra arise from the energy shift of the exciton transition, which are the difference spectra of shifted and original exciton absorption. Since dE is much smaller than E0, DA can be approximated by the following equation:

DA

1

2

@Aex

@ o dE 2 where :o is probe photon energy and Aex is the exciton absorption spectrum obtained from deconvolution of the linear absorption spectrum (Fig. 2). The pre-factor of 1/2 accounts for the selection rule of (s ,s ) congurations because only half

of the exciton states are coupled to the s pump and s probe

each. We also investigated the dependence of the OSE on I0 that represents the photon energy ux within the pulse duration. The DA recorded under different values of I0 are plotted in Fig. 4a (red circles), the blue traces are non-linear least squares best-ts of equation (2) to the data with dE as the only adjustable parameter. The model deviates slightly from the experimental spectra (Fig. 4a) on the higher energy side, which may arise from a pump-induced shift of the continuum band.

We nd a linear relationship between dE and pump intensity (Fig. 4b) as predicted by equation (1). Considering that D and F

h i2vary with different experimental conditions, m0X is the only

intrinsic parameter determining the energy shift. In our experiment, m0X is determined as 46 D, which is larger than the reported values of semiconductor quantum wells (about 6 D)2, carbon nanotubes (about 12 D)21 and WSe2 monolayers (about 7 D)20, and is comparable to that in WS2 monolayers (about 51 D)19. Equation (1) also implies that the OSE is strongly associated with the intrinsic properties of the material rather than sample preparation method, which is consistent with the observations of OSE in the samples made from different methods

(Supplementary Figs 8 and 9). In contrast, the TA signal caused by OSE depends on the width of the exciton peak (equation (2)), and thus the OSE-induced TA features may show different widths and amplitudes depending on the sample inhomogeneity. It is remarkable that such sharp excitonic features, necessary to observe the OSE, are present in a room-temperature solution-processed semiconductor.

In summary, a large OSE is observed in MAPbI3 at room temperature that is easily resolved. Our observation is consistent with a narrow excitonic transition that is optically active at the absorption bandedge within these inorganic/organic hybrid semiconductors. We also demonstrate that the OSE-induced absorption change can be controlled by the photon spin polarization, suggesting a method to control spin states which is a requisite for spin-based quantum computing. Realizing a functional spin-based qubit requires the spin coherence time (around 2 ps in this case) to be lengthened so that multiple independent operations8,10 can be performed before decoherence. The mechanisms limiting the dephasing time and to what extent these can be mitigated in the lead halide perovskite materials remain open questions. It is nonetheless clear from our demonstration of strong OSE in this solution-processed bulk semiconductor that additional experimental study to determine their potential for quantum information applications and femtosecond lasers is warranted.

Methods

Synthesis. The samples were prepared following published literature33.

A nonstoichiometric mixture of methylammonium iodide and lead iodide (molar ratio of MAI:PbI2 is 1.2:1) was dissolved in a solvent comprising a mixture of N-methyl-2-pyrrolidinone/g-butyrolactone (7:3, weight ratio) to form anB50 wt% precursor solution. The precursors were used for the depositionof perovskite lms on quartz substrates. Briey, 50 wt% precursor with 20% more organic salt was dispersed on top of substrate by spin-coating at 4,500 r.p.m. for 25 s, and as-formed wet lm was immediately transferred into a diethylether (Fisher Chemical) bath to form perovskite lm. The perovskite lm was further annealed at 150 C for 15 min with a Petri dish covered to remove excess organic salt.

Transient absorption spectroscopy. The femtosecond transient absorption spectrometer is based on a regeneratively amplied Ti:sapphire laser system.

The wavelength of the fundamental beam is at 800 nm, and the pulse repetition rate is 1 kHz. The fundamental beam is then split into two beams. One beam is attenuated by neutral density lter and chopped at a rate of 500 Hz, which is used as the pump. The other beam is also attenuated and focused into a sapphire crystal

a

b

Pump [afii9846]+

Pump [afii9846]+

4

2.0

1.5

1.0

0.5

0.0

2

A (mOD)

0

E (meV)

2

4

1.55

1.60

1.65

1.70

1.75

0.00

0.04

0.08

0.12

Photon energy (eV)

Pump pulse intensity (GW cm2)

Figure 4 | TA spectrum tting and pump intensity-dependent energy shift. (a) OSE-induced TA spectrum (red circles) for different pump intensities

recorded at time zero. The blue traces are best ts to equation (2). (b) Shift of the exciton transition energy for different pump pulse intensities. The dashed

line represents a linear t of the data points. The average pump intensities are also listed in Supplementary Table 2.

4 NATURE COMMUNICATIONS | 7:12613 | DOI: 10.1038/ncomms12613 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12613 ARTICLE

to generate the broadband probe (420830 nm). The probe is delayed in time with respect to the pump using a motorized translation stage mounted with a retroreecting mirror. The pump and probe are spatially overlapped into the sample, and the transmitted probe pulses are directed to the detectors. The polarization of the pump and probe is controlled by polarizers and waveplates.

The total pump intensity is determined by measuring the photon ux after a pinhole with radius of 400 mm at the sample position.

Data availability. The authors declare that all data supporting this work are contained in graphics displayed in the main text or in supplemental information. Data used to generate these graphics are available from the authors on request.

References

1. Scully, M. O. & Zubairy, M. S. Quantum Optics (Cambridge Univ. Press, 1997).2. Von Lehmen, A., Chemla, D. S., Heritage, J. P. & Zucker, J. E. Optical Stark effect on excitons in GaAs quantum wells. Opt. Lett. 11, 609611 (1986).3. Mysyrowicz, A. et al. Dressed excitons in a multiple-quantum-well structure: evidence for an optical Stark effect with femtosecond response time. Phys. Rev. Lett. 56, 27482751 (1986).

4. Vasa, P. et al. Optical Stark effects in j-aggregate-metal hybrid nanostructures exhibiting a strong exciton-surface-plasmon-polariton interaction. Phys. Rev. Lett. 114, 036802 (2015).

5. Muller, A., Fang, W., Lawall, J. & Solomon, G. S. Creating polarizationentangled photon pairs from a semiconductor quantum dot using the optical Stark effect. Phys. Rev. Lett. 103, 217402 (2009).

6. Nick Vamivakas, A., Zhao, Y., Lu, C.-Y. & Atature, M. Spin-resolved quantum-dot resonance uorescence. Nat. Phys. 5, 198202 (2009).

7. Berezovsky, J., Mikkelsen, M. H., Stoltz, N. G., Coldren, L. A. & Awschalom, D.D. Picosecond coherent optical manipulation of a single electron spin in a quantum dot. Science 320, 349352 (2008).8. Gupta, J. A., Knobel, R., Samarth, N. & Awschalom, D. D. Ultrafast manipulation of electron spin coherence. Science 292, 24582461 (2001).

9. Unold, T., Mueller, K., Lienau, C., Elsaesser, T. & Wieck, A. D. Optical Stark effect in a quantum dot: ultrafast control of single exciton polarizations. Phys. Rev. Lett. 92, 157401 (2004).

10. Zhang, J., Tang, Y., Lee, K. & Ouyang, M. Tailoring light-matter-spin interactions in colloidal hetero-nanostructures. Nature 466, 9195 (2010).

11. Wilcox, K. G. et al. Ultrafast optical Stark mode-locked semiconductor laser. Opt. Lett. 33, 27972799 (2008).

12. Quarterman, A. H. et al. A passively mode-locked external-cavity semiconductor laser emitting 60-fs pulses. Nat. Photon. 3, 729731 (2009).

13. Tsuda, S., Knox, W. H., Cundiff, S. T., Jan, W. Y. & Cunningham, J. E. Mode-locking ultrafast solid-state lasers with saturable Bragg reectors. IEEE J. Sel. Top. Quantum Electron. 2, 454464 (1996).

14. Joffre, M., Hulin, D., Migus, A. & Combescot, M. Laser-induced exciton splitting. Phys. Rev. Lett. 62, 7477 (1989).

15. Gansen, E. J., Jarasiunas, K. & Smirl, A. L. Femtosecond all-optical polarization switching based on the virtual excitation of spin-polarized excitons in quantum wells. Appl. Phys. Lett. 80, 971973 (2002).

16. Le Gall, C., Brunetti, A., Boukari, H. & Besombes, L. Optical Stark effect and dressed exciton states in a Mn-doped CdTe quantum dot. Phys. Rev. Lett. 107, 057401 (2011).

17. Press, D., Ladd, T. D., Zhang, B. & Yamamoto, Y. Complete quantum control of a single quantum dot spin using ultrafast optical pulses. Nature 456, 218221 (2008).

18. Xu, X., Yao, W., Xiao, D. & Heinz, T. F. Spin and pseudospins in layered transition metal dichalcogenides. Nat. Phys. 10, 343350 (2014).

19. Sie, E. J. et al. Valley-selective optical Stark effect in monolayer WS2. Nat. Mater. 14, 290294 (2015).

20. Kim, J. et al. Ultrafast generation of pseudo-magnetic eld for valley excitons in WSe2 monolayers. Science 346, 12051208 (2014).

21. Song, D. et al. Measurement of the optical Stark effect in semiconducting carbon nanotubes. Appl. Phys. A 96, 283287 (2009).

22. Frhlich, D., Nthe, A. & Reimann, K. Observation of the resonant optical Stark effect in a semiconductor. Phys. Rev. Lett. 55, 13351337 (1985).23. Saba, M. et al. Correlated electron-hole plasma in organometal perovskites. Nat. Commun. 5, 5049 (2014).

24. Sestu, N. et al. Absorption F-sum rule for the exciton binding energy in methylammonium lead halide perovskites. J. Phys. Chem. Lett. 6, 45664572 (2015).

25. Yang, Y. et al. Observation of a hot-phonon bottleneck in lead-iodide perovskites. Nat. Photon. 10, 5359 (2016).

26. Yang, Y. et al. Low surface recombination velocity in solution-grown CH3NH3PbBr3 perovskite single crystal. Nat. Commun. 6, 7961 (2015).

27. Combescot, M. Optical Stark effect of the exciton. II. Polarization effects and exciton splitting. Phys. Rev. B 41, 35173533 (1990).

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

127, 619623 (2003).29. Combescot, M. & Combescot, R. Optical Stark effect of the exciton: Biexcitonic origin of the shift. Phys. Rev. B 40, 37883801 (1989).

30. Yang, Y. et al. Comparison of recombination dynamics in CH3NH3PbBr3 and CH3NH3PbI3 perovskite lms: inuence of exciton binding energy. J. Phys. Chem. Lett. 46884692 (2015).

31. Miyata, A. et al. Direct measurement of the exciton binding energy and effective masses for charge carriers in organicinorganic tri-halide perovskites. Nat. Phys. 11, 582587 (2015).

32. 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).

33. Yang, M. et al. Square-centimeter solution-processed planar CH3NH3PbI3 perovskite solar cells with efciency exceeding 15%. Adv. Mater. 27, 63636370 (2015).

Acknowledgements

This work was supported by the Division of Chemical Sciences, Geosciences, and Biosciences, Ofce of Basic Energy Sciences of the US Department of Energy through the Solar Photochemistry programme under contract DE-AC36-08GO28308 to the National Renewable Energy Laboratory, Golden, Colorado. Perovskite lms were supplied from the Hybrid Perovskite Solar Cell program of the National Center for Photovoltaics funded by the US Department of Energy, Ofce of Energy Efciency and Renewable Energy, Solar Energy Technologies Ofce.

Author contributions

Y.Y. and M.C.B. conceived the original ideas and designed the experiment; Y.Y. carried out the experiment and analysed the data; M.Y. and K.Z. prepared and characterized the samples; Y.Y., J.C.J, J.J.B., J.v.d.L. and M.C.B. wrote the manuscript. All authors discussed the results and commented on the manuscript.

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: Yang, Y. et al. Large polarization-dependent exciton optical Stark effect in lead iodide perovskites. Nat. Commun. 7:12613doi: 10.1038/ncomms12613 (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/

r The Author(s) 2016

NATURE COMMUNICATIONS | 7:12613 | DOI: 10.1038/ncomms12613 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5