Establishing charge-transfer excitons in 2D perovskite heterostructures

Charge-transfer excitons (CTEs) immensely enrich property-tuning capabilities of semiconducting materials. However, such concept has been remaining as unexplored topic within halide perovskite structures. Here, we report that CTEs can be effectively formed in heterostructured 2D perovskites prepared by mixing PEA2PbI4:PEA2SnI4, functioning as host and guest components. Remarkably, a broad emission can be demonstrated with quick formation of 3 ps but prolonged lifetime of ~0.5 μs. This broad PL presents the hypothesis of CTEs, verified by the exclusion of lattice distortion and doping effects through demonstrating double-layered PEA2PbI4/PEA2SnI4 heterostructure when shearing-away PEA2SnI4 film onto the surface of PEA2PbI4 film by using hand-finger pressing method. The below-bandgap photocurrent indicates that CTEs are vital states formed at PEA2PbI4:PEA2SnI4 interfaces in 2D perovskite heterostructures. Electroluminescence shows that CTEs can be directly formed with electrically injected carriers in perovskite LEDs. Clearly, the CTEs presents a new mechanism to advance the multifunctionalities in 2D perovskites.

I nitiated by fast-developing thin-film photovoltaic functionalities [1][2][3][4][5] , organic-inorganic metal halide hybrid perovskites have emerged as interesting light-emitting materials with high photoluminescence (PL) quantum yields [6][7][8] , large color tuning capabilities [9][10][11] , domain-size controllable exciton binding energies 12,13 and bandgap engineering properties [14][15][16] . With these attractive optical characteristics, the hybrid perovskites have demonstrated high-performance light-emitting diodes (LEDs) 7,17-20 based on solution-processed thin-film formation procedures. Undoubtedly, the intrinsic excitons formed within band structures function as the primary excited states responsible for developing highperformance light-emitting properties in such hybrid perovskites. Essentially, tuning the intrinsic excitons through the formation probability, binding energy, and radiative/nonradiative recombination can determine the light-emitting properties, as exampled by mixing different halides [9][10][11] , introducing nanostructures 15,21,22 , and passivating grain boundary defects 18,23,24 . On the other hand, it is interesting to note that CTEs have been introduced as artificially engineered excitons with delocalized wavefunctions, namely, spatially extended states, and widely observed in low-dielectricconstant organic materials through donor-acceptor design, to develop high-efficiency optoelectronic functionalities [25][26][27][28][29] . It has been recognized as an intriguing nature that CTEs can be conveniently tuned on energy, polarization, and spin parameters by physically changing the formulation of donor-acceptor heterostructures 30,31 . This leads to unique tuning capabilities to control optic, electric, and magnetic functionalities by using CTEs [32][33][34][35] . In general, CTEs can be formed at heterostructures where charge transfer occurs between different local structures to form electron-hole pairs 36,37 . The heterostructures can be both chemically and physically prepared by using two molecular structures where different local energies and electron negativities exist 38,39 . In general, the heterostructures can be naturally formulated between donor-and acceptor-type structures, exampled as highly efficient light-emitting exciplexes [40][41][42] . On the other hand, the heterostructures can be also formed between two same-type molecular structures where local energy disorders are occurred, shown as light-emitting excimers in organic materials 43,44 . In the past, the CTEs have been extensively explored in organic heterostructures prepared by chemical synthesis, physical mixing, and mechanical formation of film interfaces in light-emitting 10,11,45 , lasing 46,47 , photovoltaic 38,39 , sensing 48 , and magneto-optical 49 applications. Recently, it has been shown that the perovskitepolymer bulk heterostructure, prepared by mixing quasi-2D/3D perovskite with wide-bandgap polymer, demonstrates a broad transient absorption (TA) signal, which is quickly appeared in 1 ps and slowly relaxed in nanoseconds 19 . This presents the possibility to form CTEs between perovskite and organic polymer, known as high-and low-dielectric materials. Essentially, the CTEs can be formed when a charge-transfer process is occurred with the consequence of forming electron-hole pairs between two adjacent structures with different electron negativities and local energies. It has also been observed that CTEs can be formed between 2D perovskite (BA 2 PbI 4 ) and organic molecule (1,4,5,8,9,11-hexaazatriphenylenehexacarbonitrile) when the charge transfer at the interface is enhanced by orbital overlap 50 . Clearly, these published results provide the necessary condition to explore the CTEs within perovskite heterostructures.
In this work, we initially utilized the chemical method to prepare 2D perovskite heterostructures [(PEA 2 PbI 4 ) x : (PEA 2 SnI 4 ) 1−x ] by alternatively selecting the host and guest components when mixing two precursor (Pb and Sn perovskites) solutions with distinctly high and low concentrations. The heterostructures with alternatively switched host and guest components are both demonstrated with similar broad emission spectra and prolonged lifetime (~microseconds). ] to generate a broad emission in 2D perovskites. Remarkably, the broad emission can be reproduced when using hand-finger pressing method to mechanically shearing away the PEA 2 SnI 4 film onto the PEA 2 PbI 4 film surface to uniquely prepare the double-layered heterostructures without lattice distortion and doping effects. This hand-finger pressing method confirms that the broad emission represents the signature for CTEs formed in 2D perovskite heterostructures. The pump-probe TA studies present that the CTEs are quickly formed and essentially become metastable states, as compared to surface-trapped and intrinsic excitons in 2D perovskites.

Results
CTEs in heterostructured 2D perovskite thin film. The 2D perovskite films were prepared by spin coating from mixed two precursor solutions (PEA 2 PbI 4 and PEA 2 SnI 4 ) and followed with thermal annealing at 100°C for 10 min. The two precursor solutions were initially mixed with the volume ratio of 99  Figure 1a shows that mixing 0.1% PEA 2 SnI 4 into 99.9% PEA 2 PbI 4 has little effect on the absorption characteristics in the heterostructures [(PEA 2 PbI 4 ) 0.999 : (PEA 2 SnI 4 ) 0.001 ] where the bandgap is governed by the host (PEA 2 PbI 4 ) component. Interestingly, a very broad PL spectrum is observed between 575 and 800 nm with the peak position located at 669 nm. When we further increase the guest Sn perovskite concentration to 1% or even 5%, broad emission can always be observed at same wavelength ( Supplementary Fig. 4). As a comparison, the PEA 2 PbI 4 and PEA 2 SnI 4 perovskites give rise to their intrinsic PL at 524 and 621 nm, shown as spectral shoulders on the broad emission spectrum. The broad PL demonstrates a very broad spectral width of 127 nm, while the intrinsic emission from the PEA 2 PbI 4 and PEA 2 SnI 4 components give rise to much narrower spectral widths  Fig. 1c. Here, the formation of CTEs requires that (a) the charge transfer is occurred from local host to guest structures and (b) the electron-hole pairs are formed at the interfaces in the 2D perovskite heterostructures.
To further understand the origin of broad emission occurring in the 2D perovskite heterostructures, we measured the PL intensitydependence and lifetime characteristics. We can see in Fig. 2a, b that the broad emission exhibits the intensity-dependence slope of 1.62, while the intrinsic emission from the host PEA 2 PbI 4 component gives rise to the intensity-dependence slope of 1.14. We should note that germinate process dominates the radiative recombination of the intrinsic excitons in host/guest 2D perovskites due to large exciton binding energy, which leads to the power-dependence slope of~1. On the other hand, when electrons and holes are paired to form electron-hole pairs, the pairing probability p is proportional to the product between the number (n) of electrons and the number (m) of holes: p ∝ m ⋅ n. In the situation where n = m, then p ∝ n 2 . Therefore, the non-germinate recombination can lead to the intensitydependence slope of 2 in light emission. Here, the broad emission of our heterostructured [(PEA 2 PbI 4 ) 0.95 :(PEA 2 SnI 4 ) 0.05 ] film demonstrates the power-dependence slope of 1.62, presenting the suggestion that the CTEs are responsible for the broad emission through nongerminate recombination. Furthermore, the broad emission peaked at 669 nm shows much prolonged lifetime of 0.5 μs relative to the short lifetime of 1 ns of the intrinsic emission peaked at 524 nm from the host PEA 2 PbI 4 component, as shown in Fig. 2c. Detailed fittings of PL lifetimes of pure PEA 2 PbI 4 and PEA 2 SnI 4 films are shown in Supplementary Fig. 6. In general, the lifetime of CTEs is mainly determined by the electron-hole recombination rate governed by Coulomb attractive force. As a result, CTEs often exhibit an extended lifetime due to longer electron-hole separation distance as compared to intrinsic excitons in semiconducting materials. Here, the much prolonged PL lifetime further verifies that the broad emission originates from CTEs in heterostructured 2D perovskites. We should note that CTEs can still demonstrate various lifetimes when the Coulomb attraction between electron and hole located on different energetic structures is changed by local dielectric backgrounds, widely observed in organic-organic 51-53 , organic-inorganic 35,54 , and inorganic-inorganic mixtures 34,55 .
CTEs in double-layered 2D perovskite heterostructures. We note that, in the mixed two precursor solutions with 99.9% host and 0.1% guest concentrations, the PEA 2 PbI 4 and PEA 2 Supplementary  Fig. 7), forming the double-layered [PEA 2 PbI 4 /PEA 2 SnI 4 ] interfaces, functioning as heterostructures clearly without doping effects and lattice deformation. Therefore, our hand-finger pressing method provides a unique experimental approach to clarify the origin of broad emission widely observed in hybrid metal halide perovskite heterostructures formed by mixing different precursor solutions or in nanocrystals formed with lattice strains. Here, it is very interesting to show that our broad emission can be reproduced with the peak position at 671 nm in the double-layered PEA 2 PbI 4 /PEA 2 SnI 4 interfaces prepared by our hand-finger pressing method (Fig. 3b), very similar to the broad emission observed by mixing two precursor solutions. Because our hand-finger pressing method can only form a low density of double-layered PEA 2 PbI 4 /PEA 2 SnI 4 heterostructures, the broad emission becomes an appreciable shoulder on the intrinsic emission peaked at 638 nm from the PEA 2 SnI 4 perovskite.
Clearly, the similar broad emission observed from our hand-finger pressing method as compared to the precursorsolution mixing method provides an unambiguous evidence to confirm that the broad emission indeed represents the CTEs formed at the interfaces between PEA 2 PbI 4 and PEA 2 SnI 4 domains in 2D perovskite heterostructures.
Dynamic analysis of CTE formation. To explore the dynamic behaviors of the CTEs, we performed pump-probe TA studies on the heterostructured 2D perovskite [(PEA 2 PbI 4 ) 0.95 : (PEA 2 SnI 4 ) 0.05 ] film (Fig. 4a). The TA characteristics show an interesting phenomenon where a broad signal is observed between 600 and 800 nm (bottom panel in Fig. 4a). Obviously, the broad TA signal is consistent with the broad emission from CTEs in the heterostructured 2D perovskite film. Contrarily, the pristine PEA 2 PbI 4 film does not show any broad TA signal between 600 and 800 nm, only leading to a photobleaching signal peaked at 514 nm related to intrinsic excitons, as shown in Fig. 4b. Clearly, the broad TA signal provides the dynamic information of CTEs formed in 2D perovskite heterostructures. Figure 4c  shown in Supplementary Fig. 8 that the PL of pure PEA 2 PbI 4 film contains a weak tail located around 540 nm, slightly below the emission (peaked at 520 nm) from intrinsic excitons with decreased film thickness (~20 nm), leading to an asymmetric spectrum. This weak PL tail at 540 nm coincides with the surfacetrapped excitons at 537.5 nm shown in TA (Fig. 4a). With increasing the film thickness to~150 nm, this weak tail becomes negligible, making the entire PL spectrum symmetric. We believe this weak tail on PL spectrum at thinner film provides a further indication to support the surface-trapped excitons. As a comparison, the intrinsic excitons shown as the band-to-band transition peaked at 513.8 nm are quickly developed into a light emission within~1 ps. Therefore, the dynamics of CTEs include the following three processes. First, both intrinsic and surfacetrapped excitons experience a quick charge transfer, leading to the formation of CTEs at the heterointerfaces between host and guest perovskite structures within 3 ps. Second, the CTEs formed at heterostructures gradually develop a broad light emission, showing a TA bleaching signal after 230 ps. Third, the light emission of CTEs exhibits an extended lifetime (~0.5 μs), indicating that the CTEs are metastable states in 2D perovskite heterostructured film. To further verify the formation of CTEs, we have also measured TA for the double-layered heterointerfaces prepared by our hand-finger pressing method ( Supplementary  Fig. 9). The double-layered heterointerfaces demonstrate a broad TA signal between 640 and 800 nm, in addition to the band-toband transitions (peaked at 520 and 590 nm) in Pb and Sn perovskites. This broad TA signal observed from the double-layered heterointerfaces prepared by our hand-finger pressing method provides a further evidence to confirm the formation of CTEs.
Heterostructured 2D perovskite photodetectors and LEDs. Now we use below-bandgap photoexcitation to show that the CTEs are vital states formed in 2D perovskite heterostructures. Figure 5a illustrates the photocurrents for pristine (PEA 2 PbI 4 , PEA 2 SnI 4 ) and heterostructured [(PEA 2 PbI 4 ) 0.95 :(PEA 2 SnI 4 ) 0.05 ] films with the device architecture of ITO/poly (3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS)/perovskite/ phenyl-C 61 -butyric acid methyl (PC 61 BM)/PEI/Ag under the photoexcitation of 640 nm. We should note that the 2D heterostructured [(PEA 2 PbI 4 ) 0.95 :(PEA 2 SnI 4 ) 0.05 ] and the pristine PEA 2 PbI 4 films have the same absorbance at the wavelength of 640 nm (selected as the below-bandgap photoexcitation) (Fig. 5b). Interestingly, the heterostructured [(PEA 2 PbI 4 ) 0.95 : (PEA 2 SnI 4 ) 0.05 ] film demonstrates a much enlarged ON-OFF ratio of 532 on the photocurrent when the 640 nm photoexcitation is applied and removed periodically. This below-bandgap photoexcitation-induced photocurrent directly indicates that the charge-transfer states are vital states formed in 2D perovskite heterostructures. In contrast, both pristine PEA 2 PbI 4 and PEA 2 SnI 4 films show significantly reduced ON-OFF ratios of 29.7 and 3.91 on photocurrent upon repeatedly applying and removing the 640 nm photoexcitation. Furthermore, we observed that the CTEs can be generated by electrically injected charge carriers in 2D perovskite heterostructures. Figure Fig. 5d indicate the typical LED behaviors from CTEs under electrical injection. It should be pointed out that the injection current is quite low: (<1 mA/cm 2 ) due to the lower conductance caused by insulating organic long-chain ligands in 2D perovskite film even when the applied bias reaches 13 V. Clearly, the CTEs allow the generation of broad EL in 2D perovskite heterostructures under electrical injection.

Discussion
In summary, we found that the CTEs can be formed at the interfaces between PEA 2 PbI 4 and PEA 2  Mechanically preparing heterostructures. The double-layer 2D perovskite heterostructures were prepared by using hand-finger pressing method to avoid lattice distortion and doping effect. Specifically, the PEA 2 SnI 4 film coated on glass substrate was held on the PEA 2 PbI 4 film with the face-to-face contact by using hand-finger pressing method. Then, the PEA 2 SnI 4 film was shearing on the surface of PEA 2 PbI 4 film. This leaves the PEA 2 SnI 4 film islands on the PEA 2 PbI 4 film, leading to double-layer heterointerfaces (PEA 2 SnI 4 /PEA 2 PbI 4 ). The PEA 2 SnI 4 film islands can be visible as pale-yellow color on the PEA 2 PbI 4 film. Therefore, our hand-finger pressing method can prepare the heterostructures without lattice distortion and doping effects to confirm the charge-transfer excitons responsible for broad emission observed in 2D perovskite heterostructures.
Characterizations and measurements. The steady-state and time-resolved PL characteristics were measured with Flouro Log III spectrometer. The absorption spectrum was measured with UV3600 (Shimadzu, Japan). TA spectra were collected by using a Helios Fire spectrometer (Ultrafast Systems LLC). The pump beam (346 nm, 5 µW) was generated through a harmonic generator (Ultrafast Systems LLC, third harmonic) pumped by a Pharos laser (Light Conversion, 1 kHz, 1030 nm, 290 fs). The probe beam was selected from the broad light emission generated by CaF 2 crystal excited by 1030 nm fs primary beam.

Data availability
The data that support the findings of this study are available from the corresponding author on reasonable request. ] photodetector (red) as compared to pure PEA 2 PbI 4 (black) and PEA 2 SnI 4 (blue) devices when exciting at 640 nm (1000 mW/cm 2 ). b Absorption spectra of two films. c EL spectrum containing broad CTE emission and host emission. d Current density-voltage-EL curve of doped device. The device structure is shown as the inset.