Resonance-enhanced three-photon luminesce via lead halide perovskite metasurfaces for optical encoding

Lead halide perovskites have emerged as promising materials for photovoltaic and optoelectronic devices. However, their exceptional nonlinear properties have not been fully exploited in nanophotonics yet. Herein we fabricate methyl ammonium lead tri-bromide perovskite metasurfaces and explore their internal nonlinear processes. While both of third-order harmonic generation and three-photon luminescence are generated, the latter one is less affected by the material loss and has been significantly enhanced by a factor of 60. The corresponding simulation reveals that the improvement is caused by the resonant enhancement of incident laser. Interestingly, such kind of resonance-enhanced three-photon luminescence holds true for metasurfaces with a small period number of 4, enabling promising applications of perovskite metasurface in high-resolution nonlinear color nanoprinting and optical encoding. The encoded information ‘NANO’ is visible only when the incident laser is on-resonance. The off-resonance pumping and the single-photon excitation just produce a uniform dark or photoluminescence background.

There are a few changes that should, in my opinion, be made to significantly improve the manuscript: -The title of the manuscript is misleading. It is not clear that the resonances claimed are due to the presence of metasurfaces, which should be stressed more.
-Did the authors measure the eventual presence of an enhancement at the excitation wavelength of 1500nm for the un-structured (without metasurfaces) perovskite material? If so, the enhancement factors should be normalized to that values.
-Resonances due to the excitonic nature of perovskites nanocrystals have been reported by Manzi et al. in Nature Communications 9 (1), 1518. Although in this manuscript the excitation sources do not have enough energy to generate multiple excitons, the authors should consider the possibility of an enhancement (if present also without the metasurfaces) in line with what reported previously.
The manuscript by Fan et al. reports on the resonance selectivity in three-photon-induced photoluminescence (PL) from an engineered metasurface based on a hybrid halide perovskite, MAPbBr3. While the work is straightforward and rather interesting in terms of optical encoding/decoding, it should be published in a more specialized journal because of the lack of general interest. Indeed, it does not have any surprising result to warrant the publication in any of Nature's sister journals.
There is one minor mistake: In page 5, line 149, "In principle, both of 3PA and THG are cubic proportional to the amplitude of electromagnetic field." They are cubic in the "intensity" of the field, not the "amplitude".
The reviewer also suggests that the authors should describe the fundamental difference between 3PA and THG, when they resubmit the revision elsewhere: One is the fifth-order process, whereas the other is the third-order process.
They should also describe why the experimental results in Fig. 5(c) is seriously different from the simulation in Fig. 5(b). Although the current study is a concept study, the real enhancement factor is not impressive at the present level thank the reviewer for the suggestion of acceptance. Following the reviewer's comments, we have carefully revised our manuscript. All the replies have been added in the manuscript accordingly.
Comment-1. The authors wrote "Herein, we fabricate near-infrared MAPbBr 3 perovskite metasurfaces with a standard dry-etching technique and study their corresponding nonlinear processes for the first time." However, the nonlinearly excited three-photon PL in perovskite metasurface was investigated in Ref [16] as well. So, the authors should provide a more detailed comparison with this work.
Our Reply: We really appreciate the reviewer for pointing out this valuable reference. We agree with the reviewer that Ref.
[16] has also discussed the three-photon PL in perovskite metasurface. Despite similar three-photon responses have been observed in Ref. [16] and this work, their fundamental bases and possible applications are quite different. In Ref.
[16], the wavelength of pumping laser is 1050 nm and the bandedge of material is around 770 nm. The three-photon process is realized by the intrinsic material properties of perovskites, i.e. the resonance in photobleaching at 3.2 eV. The resonance of perovskite metasurface is designed for the modification of local density of states (LDOS) around the PL wavelength. In our research, the three-photon PL is generated by the field enhancement of the incident laser, which is generated by the resonant mode of perovskite metasurface. Compared with the intrinsic material properties, the nanostructure induced enhancement is dependent on the structural parameters and thus can be tuned to any designed wavelength, e.g. 1500 nm-1600 nm. This is the fundamental basis for the applications of this research. With the resonance enhancement of the pumping laser, we have demonstrated the applications of high resolution (8000 dpi) nonlinear display and the optical encryption. The encoded information "NANO" can only be viewed when the incident laser is on resonance.
The most valuable part of this comment is that Reviewer-1 has provided us a new research direction. Since both of the resonances of nanostructures (for incident laser) and the materials (for three-photon) can greatly enhance the three-photon PL process, the enhancement factor shall be more dramatic if these two effects are combined together in the perovskite metasurface. The importance of Ref. 16 has been highlighted in para-2, page-7 of the revised manuscript. The corresponding experiment will be done in near future. "We must note here that the nonlinear processes within the perovskite metasurfaces are intrinsically interesting for both of fundamental researches and practical applications. As pointed out by Ref.
[16], the intrinsic material properties such as the state in photobleaching region (3.2 eV) can also enhance the three-photon luminescence by tens of times. Similarly, the material enhancement can also be realized with the multi-exciton resonances [52]. Consequently, the combination of structural enhancement and material enhancement in perovskite metasurface can further improve the nonlinear responses." To avoid the unnecessary confusion, we have also removed the statement of "for the first time" in the revised manuscript.
Comment-2. I recommend to stress the novelty of the enhanced third-harmonics generation from the perovskite metasurface.
Our Reply: We thanks the reviewer for this constructive and deep-insight recommendation.
We fully agree with the reviewer that the THG process in perovskite metasurface is also attractive and worth to emphasize. The perovskite materials can provide the exceptional nonlinear properties and cost-effective fabrication process, whereas the metasurface is able to precisely control the wavefront of the THG waves. As a result, a large number of nonlinear devices can be realized in a higher efficiency and more cost-effective form. This will be essential for practical applications. Following the reviewer's suggestion, we have also stressed the novelty of the enhanced THG process from perovskite metasurface in para-2, page-7. "In addition, the enhanced THG processes in perovskite metasurface is also interesting. In contrast to the photoluminescence, the THG signals are coherent and their local phases can be precisely controlled by the nanoantennas. While the enhancement of THG signals are limited by the material absorption, this can be simply relieved by tailoring the field distribution and the radiation of the THG signals  The Purcell factor around the photoluminescence wavelength region has also been numerically studied. As shown in Fig. R2, the transmission spectrum at the visible spectrum indeed has a dip around the spontaneous emission wavelength range. However, as the perovskite metasurface is designed for the near infrared wavelength, the higher order resonances are quite weak and can only be barely seen from the background (1-2% difference). As a result, its influence on the spontaneous emission cannot be as significant as Ref.
[39]. Our experimental results are also consistent with the above numerical simulation. As shown in Fig. 4(b) in the main manuscript, there is no obvious additional enhancement around 525 nm even though the spectra are plotted in log scale. Therefore, we can confirm that the output coupling efficiency and the Purcell factor are not as significant as the resonant enhancement of the incident laser.
In the revised manuscript, we have added the above discussion in para-1, page-2. "Such kind Our Reply: We thanks the reviewer for this valuable suggestion. We agree with the reviewer that the upconversion is not very accurate and will lead to unnecessary confusion. To make the statement accurately and avoid confusion, we have replaced all the "upconversion" and "three-photon upconversion" into "three-photon luminescence". The latter one is more accurate to describe the nonlinear multiphoton absorption via virtual states.

Reply to Reviewer 2:
We thank the reviewer for the very careful review and the valuable suggestions for revision.
We also thank the reviewer for the recommendation for publishing on Nature Communications. Based on his/her suggestion, we have carefully revised the manuscript and all the comments have been addressed accordingly.
Comment-1. The title of the manuscript is misleading. It is not clear that the resonances claimed are due to the presence of metasurfaces, which should be stressed more.
Our Reply: We thank the reviewer for this valuable comment. In our research, the enhancement is caused by the resonant mode of the perovskite metasurface at the wavelength of incident laser. This can be seen from the dependence of maximal enhancement factor on the lattice sizes in Fig. R3 below. With the change of the nanostructures, the resonant wavelength and the enhancement gradually shift from ~ 1320 nm to 1590 nm. This is different from the enhancement of material resonance, which is usually fixed at particular wavelength.
We think that the "upconversion" in title is indeed confusing. It usually leads the general readers to anti-stokes luminescence excited via real states such as the rare-earth-element upconversion nanoparticles. To avoid the possible confusion, we have replaced the "upconversion" in the title and the whole manuscript with "three-photon luminescence". The title of our manuscript is also changed to "Resonance-enhanced Three-photon Luminesce via Lead Halide Perovskite Metasurfaces". The results in Fig. R3 has also been added in Supplementary Information Note 1.  in line with what reported previously.
Our Reply: We thank the Reviewer for the suggestion of this very important reference. We have added this reference as Ref.
[52] in the revised manuscript. The multiple exciton has the ability of enhancing the two-photon photoluminescence intensity by orders of magnitude. We agree with the reviewer that it is very important to consider this possibility. Following the reviewer's suggestion, we have also experimentally verified the wavelength dependence of two-photon luminescence from our perovskite film. All the results are shown in Fig. R5 below. With the decrease of pumping wavelength, there is a dramatic increase at around 720 nm. This is also caused by the excitonic resonance at ~ 3.4 eV, which can greatly enhance the multi-photon absorption process. Such kind of additional enhancement is quite interesting and worth to study in future. For our current research, as the enhancement factor is calculated by normalizing with the perovskite film, the material enhancement won't change the final experimental results.
Since both the pure material resonance from multiple exciton resonance and the structural resonance can enhance the nonlinear processes, the combination of two effects (designing the structural resonance for the excitation of multi-exciton resonance) can further improve the nonlinear signals by orders of magnitude. This is very important and we will focus on such studies in near future. In the revised manuscript, we have added the corresponding discussion in para-2, page-7. "Similarly, the material enhancement can also be realized with the multi-exciton resonances [52]. Consequently, the combination of structural enhancement and material enhancement in perovskite metasurface can further improve the nonlinear responses." The results in Fig. R5 has also been added into the Supplementary Information Note 5.

Figure R5:
The wavelength dependent two-photon luminescence. Due to the presence of an excitonic resonance at around 3.4 eV, the two-photon absorption is also significantly enhanced.

Reply to Reviewer 3:
We thank the reviewer for the careful review and valuable suggestion. In the revised manuscript, we have fully addressed all the comments accordingly. Based on his/her suggestions, the quality of this research has been significantly improved.
The main comment raised by reviewer-3 is the general interest of this work. We thank the reviewer for this deep insight comment. As the reviewer knows, lead halide perovskites have been incorporated with extraordinary success in photovoltaics. Such successes come from the fact that lead halide perovskite can provide similar or better power conversion efficiency (PCE) while significantly reduce the device cost. Soon after, with the improvements in quantum efficiency and light coupling efficiency, similar successes have also been achieved in lead halide perovskite based light emitting diodes (LEDs) and photodetectors. Now, the recent researches reveal that the nonlinearity of lead halide perovskite is comparable to or even superior than Si and III-V direct bandgap semiconductors again. In this context, it has been recognized that it is really the turn of perovskite nonlinear photonics.
With this background, we started to explore the lead halide perovskite based nonlinear

Comment-1.
There is one minor mistake: In page 5, line 149, "In principle, both of 3PA and THG are cubic proportional to the amplitude of electromagnetic field." They are cubic in the "intensity" of the field, not the "amplitude".
Our Reply: Thank the reviewer for this careful review. The reviewer is correct that 3PA and THG are cubic proportional the intensity instead of amplitude of the incident laser. In the revised manuscript, we have changed the sentence in para-1, page-5. "In principle, while three-photon absorption and THG are the fifth-order process and the third-order process, respectively, they are both cubic proportional to the intensity of electromagnetic field." Comment-2. The reviewer also suggests that the authors should describe the fundamental difference between 3PA and THG, when they resubmit the revision elsewhere: One is the fifth-order process, whereas the other is the third-order process.
Our Reply: We thanks the reviewer for this fundamental suggestion. The reviewer is correct that three-photon absorption is the fifth-order process, whereas the THG is the third order process. This can be directly seen from the equations.
For an isotropic medium in vacuum illuminated by fundamental electronic field amplitude ( ) with the same frequency ω, complex amplitudes of the nonlinear polarization can be written as ( ) (3 ) = ε χ ( ) ( ).
In case of three-photon absorption, the change of irradiance with depth into the sample as below ( ) = − ( ) ( ).
Here is the refractive index of an isotropic medium at the irradiance wavelength, c is the speed of light and is the permittivity of vacuum. Following the reviewer's suggestion, we have added the information in para-1, page-5 in the revised manuscript. "In principle, while three-photon absorption and THG are the fifth-order process and the third-order process, respectively, they are both cubic proportional to the intensity of electromagnetic field." Comment-3. They should also describe why the experimental results in Fig. 5(c) is seriously different from the simulation in Fig. 5(b). Although the current study is a concept study, the real enhancement factor is not impressive at the present level Our Reply: We thanks the reviewer for this careful review and valuable suggestion. Based on the reviewer's comment, we have carefully checked our experimental results and the numerical simulation. The difference comes from the capping layer on top of our sample. In our simulation, we directly simulated the perovskite nanostructures and got the enhancement factor. In experiment, considering that the perovskite is not stable, we have to use a cap layer (ZEP) to protect the nanostructures in room temperature and ambient environment. Compared with the air, the capping layer reduce the difference of refractive indices of perovskite nanostructure and the surrounding medium. The reduced refractive index difference changes the local field enhancement and thus reduces the enhancement factor. If the capping layer is considered, as the dashed lines shown in Figure R6c below, the experimental results match the numerical simulation very well if the capping layer is considered. In the revised manuscript, we have replaced the data in Fig. 5 and added the corresponding discussion in para-1, page-6. "Compared with Fig. 5b, here the enhancement factors are influenced by the capping layer, which is applied to protect the perovskite metasurface in experiment (see Supplementary Information Note 3)." The detail discussions have been added into the Supplementary Information Note 3.
We agree with the reviewer that the enhancement factor in Fig. 5 is not that impressive. This is mainly caused the reduction of periodicity of metasurface. For metasurface with infinite periodicity number, the enhancement factor can be as large as 150 as shown in Fig. 2(c).
However, in order to achieve a large number of practical applications such as optical encryption, the spatial resolution should be taken into account. By decreasing the periodicity number, the enhancement factor will reduce progressively. The smallest one is realized with only four strips (N = 4). But it is still clear enough to be distinguished from the perovskite film (background, see top row in Fig. R6d below) and meet the needs of practical applications.
The intensity of nonlinear signals can also be significantly enhanced if the other material properties are considered. For example, the multi-exciton resonance or state in photobleaching region can further improve the enhancement of three-photon luminescence.
We have also discussed this possibility in para-2, page-7 of the revised manuscript. The authors have carefully revised the most of critical parts of the manuscript and improved its overall quality. The results sound technically and provide full understanding of the processes behind the observed phenomena. Generally, I satisfied with all the answers except the first one, because there is still a serious problem related to the comparison with previous studies.
a) The authors incorrectly wrote that "As pointed in Ref.
[16], the intrinsic material properties such as the state in photobleaching region (3.2 eV) can enhance the three-photon luminescence by tens of times.". However, even the title of this paper "Multifold emission enhancement in nanoimprinted hybrid perovskite metasurfaces" clearly implies that the three-photon luminescence enhancement originates from optical resonances in a perovskite metasurface at wavelength 1050 nm rather than from any material effects. Moreover, in this paper, exactly the same enhancement factor (up to 70 times) for three-photon luminescence was observed owing to resonant modes excitation. Thus, the discussion on page 7 should be revised, or put somewhere before the results on optical encoding. b) In this regard, the modified title of the submitted manuscript ideologically resembles that in Ref. [16]. However, the manuscript contains very important and novel results on optical encoding. I suggest to stress this in title directly to avoid considerable overlapping with previous papers. For example, "Resonance-enhanced Three-photon Luminesce via Lead Halide Perovskite Metasurfaces for Optical Encoding", or something like that.
I believe that the paper deserves publication in Nature Communication after the addressing the abovementioned comments.
Reviewer #2 (Remarks to the Author): The authors have adequately replied to my original comments and provided satisfactory answers to the points I have raised. I suggest acceptance of this work in its current form.
thank the reviewer for the high evaluation. Following the reviewer's comments, we have carefully revised our manuscript and title. All the replies have been added in the manuscript accordingly.
Comment-1. The authors incorrectly wrote that "As pointed in Ref.
[16], the intrinsic material properties such as the state in photobleaching region (3.2 eV) can enhance the three-photon luminescence by tens of times.". However, even the title of this paper "Multifold emission enhancement in nanoimprinted hybrid perovskite metasurfaces" clearly implies that the three-photon luminescence enhancement originates from optical resonances in a perovskite metasurface at wavelength 1050 nm rather than from any material effects.
Moreover, in this paper, exactly the same enhancement factor (up to 70 times) for three-photon luminescence was observed owing to resonant modes excitation. Thus, the discussion on page 7 should be revised, or put somewhere before the results on optical encoding.
Our Reply: We really appreciate the reviewer for this suggestion. We delete the discussion about the properties effect and revise the manuscript in paragraph 2, page 7. "The experiment in Fig. 5 is only the proof of principle. The incident laser is designed for linear polarization with electric field perpendicular to the strip of perovskite metasurface. By exploiting the multiple design degrees of metasurfaces, more complicated incident laser can be designed, e.g.
incident laser with different angular momentum [50][51][52]. Then the encoded information can be safer. In addition, by partially designing information, the encoded information can also be well concealed even though the unauthorized person scans the incident wavelength (see