All-optical control of lead halide perovskite microlasers

Lead halide perovskites based microlasers have recently shown their potential in nanophotonics. However, up to now, all of the perovskite microlasers are static and cannot be dynamically tuned in use. Herein, we demonstrate a robust mechanism to realize the all-optical control of perovskite microlasers. In lead halide perovskite microrods, deterministic mode switching takes place as the external excitation is increased: the onset of a new lasing mode switches off the initial one via a negative power slope, while the main laser characteristics are well kept. This mode switching is reversible with the excitation and has been explained via cross-gain saturation. The modal interaction induced mode switching does not rely on sophisticated cavity designs and is generic in a series of microlasers. The switching time is faster than 70 ps, extending perovskite microlasers to previously inaccessible areas, e.g., optical memory, flip-flop, and ultrafast switches etc.

think it should be better to show the Y-axis in a logarithmic scale. In addition, it would be interesting to include in the supporting information the integrated intensity as a function of the excitation fluency in a log-log plot. 7. Absorption curve in Figure 1 indicates strong scattering at long wavelengths. Are results affected by this scattering? What is the size of the spot in the absorption measurements? 8. How n and k are measured? 9. I do not understand the section VII included in the supporting information. This section is even longer than the main manuscript and it is only called in the conclusions. 10. I found misprints in line 70 "confim" by "confims", 109 mechanism by mechanisms, line 169 "sepctrum" by spectrum. 11. Reference 39 is missing. 12. Line 56. Include a reference about cross gain saturation observed with other materials, and explain what it is. 13. I do not understand why the inset of Figure 2d is referred in Line 142. Anyway, this inset is not commented within the Figure (lines 95-104). 14. Line 149. Are cross-section interaction coefficients normalized magnitudes? or are units missed?
Reviewer #2 (Remarks to the Author): Comments to the Authors In the manuscript entitled "All-optical control of lead halide perovskite microlasers", N. Zhang, et al. reported 70-ps ultrafast all-optical mode switching based on organic-inorganic lead halide perovskite (MAPbBr3) microlasers. By adjusting the excitation energy, mode switching with high extinction ratio (~15 dB) between two distinct modes has been demonstrated. The experimental observations have been well explained with a cross-gain saturation model, which is also applicable to other microlasers (e.g. polymer microdisk lasers in the supplemental information). All-optical control on perovskite microlasers is currently the subject of extensive research efforts by many groups. The mechanism in this manuscript is timely and important for the developments of perovskite-based photonics in near future. All the experimental observations and theoretical discussion are interesting, convincing, well organized and accessible to a broad audience. Therefore, I recommend the acceptance of the manuscript after some minor revisions.
1) The references in the introduction section should be carefully checked, e.g. though the introduction mainly focuses on MAPbX3, some works on all inorganic perovskites have been cited; on Line-41, Page-2, the references 28-33 are not about photonic crystals; the reference 39 is missing in the reference list. 2) In Line-38, Page-2, the record values on threshold and Q factor of perovskite microlasers cannot be found in the reference. The authors should very accurate information or citation. 3) In Figure 1, authors mainly discuss the crystalline quality and optical properties of perovskites, which are not directly relevant to the topic of the manuscript. Thus I suggest that authors add them into the supporting information. 4) The description on the excitation is not proper: pumping density should be changed to pumping energy, considering that joules are used in the manuscript to describe the excitation. 5) In Figure 2c, the intensity of mode 2 slightly drops when the excitation energy is above 0.8 µJ.
What is the cause of the drop? It seems that the Auger effect starts to play an important role. The author should discuss this and give a possible way to solve this effect. 6) Do the two competing modes in Figure 2b have the same polarization? The polarization shown in the inset of Figure 2d has not been described with detail information. 7) Typos and grammar errors should be checked through the whole manuscript, e.g. on Line-21, Page-1, "reversable" should be "reversible"; on Line-22&23, Page-1, there are two redundant hyphens. We thank the reviewer for the careful review and valuable suggestions to improve the quality of our research. According to the reviewer's comments, the manuscript has been carefully revised and its quality has been improved greatly.
Comment-1 First of all, both the experiment and final device are not completely clear for me. According to the sentence "The microrod was placed onto another microrod with one end suspended in air via the micro-manipulation" I guess that there are two microrods coupled by the end edges, but then only one microrod is described or showed in the figures. Authors should include a scheme of the experiment indicating clearly where the excitation and the collection are. Here it is also necessary to specify the size of the spot in excitation and collection. Are both the same, 20 μm? Are authors studying just one rod or an ensemble?
What is the concentration of rods per cm 2 .
Our Reply: We thank the reviewer for this careful review and valuable suggestions. We are sorry for these confusions. In our experiments, all the experimental phenomena were obtained from one microrod every time. No coupling effect has been taken into accounted.
The two-microrod scheme was initially selected following Ref. 19. One rod was placed onto the other one and had one end suspended in air (see Fig. R1). As the refractive indices of two perovskite microrods are exactly the same, the guiding modes in the longitude direction shall experience significant scattering loss and leakage at the joint position. Then the longitude Fabry-Perot modes are completely suppressed and only the transverse WGMs can have high Q factors. In our experiment, only the suspended end of the top microrod was excited to generate laser emission (see Fig. R2). This has been confirmed by the fluorescent microscope image in the inset of Fig. 2(c) in the main manuscript, where bright spots are only observed at the suspended end of the top microrod. According to the SEM image, the ends of two microrods are separated tens of wavelengths, the coupling between transverse modes in two microrods can thus be neglected. Figure R1: The SEM image of the two microrods system. The dashed With the progress of our experiment, we also realized that the material absorption of perovskite at the band edge was very large when the microrod was not optically pumped. In this sense, if only one end of a perovskite microrod was pumped, the longitude Fabry-Perot modes could also be suppressed by the material absorption at un-pumped area. In some experiments, we only utilized a single microrod with partial pumping (see schematic picture in Fig. R2) and obtained the switching effect, e.g. the experimental results such as Fig. 6 of the main manuscript.
To avoid the possible confusion, we have replaced the SEM image in Fig. 2(a) with Fig.   R1. The original tilt-view SEM is added as an inset in Fig. 2(a). We have also added the corresponding description in para-3, page-3. "The microrod was transferred to a clean substrate and placed onto another microrod with one end suspended in the air (see Fig. 2(a)) via micromanipulation. In the optical experiment, only the suspended part was excited by adjusting the relative position between microrod and pumping laser spot (with R ~ 20 µm). At a low pumping density, a broad photoluminescence peak appeared at 540 nm." and "Due to the strong scattering loss and leakage at the overlapping region, the longitude Fabry-Perot modes in microrods will be suppressed 19 . Meanwhile, since the ends of two modes are widely separated, the coupling between transverse WGMs in two rods are also negligible. In this Following the reviewer's suggestions, we have also added the experimental details in the main manuscript. The excitation beam was focused on the perovskite microrod through a 40× objective (NA=0.6) and spot radius is around 20 μm. The emitted light was collected through the same objective and the collection area was much larger than the pumped region. The corresponding descriptions on the excitation and collection scheme have been included in the Methods section. "The perovskite samples were mounted onto a three-dimensional translation stage under a home-made microscope and excited by a frequency doubled laser (400 nm, using a BBO crystal) from a regenerative amplifier (repetition rate1 kHz, pulse width 100 fs, seeded by MaiTai, Spectra Physics). The pump light was focused onto the top surface of the samples through a 40x objective lens and the radius of the beam size was adjusted to R~20 micron. The emitted lights were collected by the same objective lens and coupled to a CCD (Princeton Instruments, PIXIS UV enhanced CCD) coupled spectrometer (Acton SpectroPro s2700) via a multimode fiber. All the emission spectra were measured using a 150 g/mm 6 / 19 grating with 0.3-nm resolution. The fluorescent microscope images were recorded by a CCD camera behind a longpass filter." During the experiment, only one microrod was studied. From the SEM image below ( Figure R3a), hundreds of perovskite microrods can be seen in a small area about 100 μm× 100 μm. To eliminate the influence of other perovskite microrods, only one perovskite microrod was selected by an optical fiber tip and transferred to a clean wafer by micro-manipulation ( Figure R3b, Wang, K. et al. Formation of single-mode laser in transverse plane of perovskite microwire via micromanipulation. Opt. Lett. 41, 555-558 (2016)). This information has also been added in para-3, page-3. "The microrod was transferred to a clean substrate and placed onto another microrod with one end suspended in the air (see Fig. 2(a)) via micromanipulation."

Comment-2
The second paragraph of section 2 is a bit confusing. First, authors argue that they choose FP or WGM by exciting entirely or partially the microrod, but then they say that "In general, both modes are observed" What does it mean? Are Fabry Perot (FP) modes or Whispering Gallery Modes (WGM)? How do they observe the modes experimentally? Can authors show an experimental near field pattern of the modes?
After reading the description of Figure 4 I understood that they are working with 2 WGM modes with practically the same pattern. Then, are they working with only one microrod instead of a coupled cavity system? If it is necessary to change the excitation area to observe the WGM modes, I understand that spot size is changed. Does this change affect the measurements since excitation fluency would be different? Can authors specify which modes present the highest Q factor (TEx)?
Our Reply: We thank the reviewer for this careful review and valuable comments. We realized that we hadn't described the typical lasing actions in perovskite microrods clearly. In a microrod, two kinds of resonant modes with relatively high Q factors could be supported: Fabry-Perot (FP) modes along the axial direction of microrod; and WGMs formed by four-bounce total internal reflection in the transverse plane of the microrod (MAPbBr 3 perovskite microrods have rectangular or even square cross-sections). Typically, the FP lasers were generated by entirely pumping the microrod (see Zhu et al. Nat. Mater. 14, 636-642 (2015)). As mentioned above, once the microrod was partially excited, the WGM laser can be generated. Both kinds of lasing modes are in one-to-one correspondence with the passive cavity resonances.
In our case, only the WG mode lasing were observed by partially exciting one end of the perovskite microrod as shown in Figure R2. Lett. 41, 555-558 (2016)). Since the resonant cavity length of FP mode is much larger than that of WG mode, the mode spacing of FP mode is thus much smaller than that of WG mode, according to ∆λ= 2 / (λ is the resonance wavelength, n is the effective refractive index and L is the effective cavity length). Therefore, for the same perovskite microrod (length: 53.18 μm, width: 1.51 μm, thickness: 1.22 μm), FP mode lasing shows periodic multimode lasing peaks, whereas WG mode lasing shows a single lasing peak. Besides, FP modes are bouncing between two end-facets and thus show two bright spots at two ends of microrod (see inset in  (2016)).
As mentioned in the above, the two competing modes are generated from the same perovskite microrod. The other microrod was utilized to completely suppress the FP mode in the microrod under investigation. And no coupling between different perovskite microrods has been involved. To eliminate the misunderstanding, the description on the pumping scheme has been revised in para-3, page-3. "The microrod was transferred to a clean substrate and placed onto another microrod with one end suspended in the air (see Comment-3 I also do not understand the sentence "without carefully designing the fundamental mechanism such as bistability" I think here authors want to claim that they obtain a stable laser device through a straightforward fabrication or without complicated design of the cavity. However, in this section of the paper the switching results are not yet introduced, hence the sentence could be confusing for the reader.
Our Reply: We thank the reviewer for this careful review and valuable suggestion. We agree with the reviewer that the statement is quite confusing. Following the reviewer's suggestion, we have changed the sentence as follows (see para-2, page-3). "In general, lasing modes are in one-to-one correspondence with the passive cavity modes. Under a fixed excitation scheme (entirely or partially pumping), main laser characteristics in a perovskite microrod were typically preserved very well during the lasing experiments."

Comment-4
The sentence "Interestingly, there are also some novel perovskite microlasers that can provide some hints to control the perovskite microlasers" needs to be rewritten. Do authors mean that some microrods work and others does not work? They should indicate some percentage in that case.  (2016)). The other transverse modes have much lower Q factors. In addition, the criterion < < ( ) ( ) requires that two modes largely overlap one another to ensure strong cross-interactions. This is usually not easy to be fulfilled. We can simply illustrate this strict requirement with the microrod in Fig. 4 of the manuscript. As shown in Fig. 5(a), an additional high Q mode at 533.26 nm can also exist in the transverse plane of microrod. While its mode profile (inset in Fig. R5) is still very close the ones in Fig. 4(b), the interaction between mode-1 and mode-3 cannot fulfill the criterion. In other world, if mode-1 and mode-3 are excited in experiment, they will show a regular mode competition instead of mode switching (see Fig. R5(b)) although mode switching has been observed between mode-1 and mode-2 in the same microrod. Therefore, based on the strict requirement of the criterion and the random sizes of as-grown, mode switching is not ubiquitous in the our as-grown perovskite microlasers. It is worth to note that the mechanism for mode switching is robust and can be repeated. Following the reviewer's suggestion, we have rewritten the first sentence of para-3, page-3 in the revised manuscript. "Interestingly, there are also some novel perovskite microlasers that show different performances. Although the percentage of such lasers is quite low, they can still provide some hints for a new mechanism to control the perovskite microlasers." We have also added the comment of production rate in para-1, page-9. "This mechanism is not limited in MAPbBr 3 perovskites. It also works well in other materials such as polymer. Importantly, with the mature top-down fabrication technique of polymer, this mode switching phenomenon can be well reproduced in a series of polymer microcavities (see Section VII in Supplementary), indicating the low production rate of switchable perovskite laser can be eventually improved with the development of top-down fabrication techniques."

Comment-5
The switching is just based on an increase of the excitation fluency. Since this is in principle a simple activation mechanism, I understand that it should not be difficult to observe this behavior once the threshold of stimulated emission is overcome.  Fig. R4(a)). The other one is the WGM lasers that are generated in the transverse planes of microrods, see Ref. 19.
In case of Fabry-Perot lasers, below we will analytically prove that their cross-interaction coefficients , are smaller than their self-interaction coefficients , , which do not meet the requirement of interaction-induced mode switching given by Eq. (4) in the main text. To simplify our discussion, we treat them as transverse waves ( ) described by the one-dimensional (1D) Helmholtz equation: Here x is the axial coordinate of the microrod, is the refractive index assumed to be uniform inside the microrod between = − /2 and /2, and is the free-space wave number. By imposing the outgoing boundary condition at both = − /2 and /2, i.e., we can solve the modes inside the cavity, which are given by either ( ) = sin( ) or ( ) = cos( ).
( 1) In principle we need to solve for the mode-dependent wave number and the amplitude or simultaneously, but since the latter will be modified using the normalization condition we eliminate them here and focus on the solutions of . This is done by employing the continuity boundary condition for the ratio   intending to synthesize the nanorods. In such systems, the transverse sizes are usually too small to support multiple high Q modes (see Fig. R4(b)), eliminating the possibility of mode switching between two modes. The most important reason is still the criterion of Eq. (4). As shown in Fig. R5, this criterion can only be realized with particular modes. Without the fine design, even in the large microrods, the mode switching effect is still rare due to the uncontrollability and randomness in chemical synthesis. We need to test hundreds of large perovskite rods to get several useful samples. Therefore, without a strong motivation and an extremely long time to screen the perovskite microlasers, it is hard to find the mode switching effect occasionally in the as-grown samples.
Importantly, while the production rate is low in the as-grown samples, the obtained mechanism for mode switching is quite robust and can be well reproduced. Of course, this requires the well-controlled top-down fabrication process. To verify this reproducibility, we have extended the mechanism to polymer microdisks that can be well defined with nanofabrication techniques. As shown in the supplemental information, the mode switching can be simply re-produced in a large number of samples on the same wafer. Those results

/ 19
clearly show that the mode switching shall also be well reproduced in perovskite devices if the precise nanofabrication technique is further improved.
In the revised manuscript, we have added the comment in the conclusion part (para-1, page-9). "This mechanism is not limited in MAPbBr 3 perovskites. It also works well in other materials such as polymer. Importantly, with the mature top-down fabrication technique of polymer, this mode switching phenomenon can be well reproduced in a series of polymer microcavities (see Section VII in Supplementary), indicating the low production rate of switchable perovskite laser can be eventually improved with the development of top-down fabrication techniques." We heartfully thank the reviewer for the very careful review of the quality and threshold.
In our experiments, both of the Q factors and thresholds are very close to the previous reports.
The Q factors are a few thousand and the thresholds are a few μJ/cm 2 . The original values are directly read from the power meter and we forgot to include the reduction of optical setup and the ND filters. Following the reviewer comment, we have changed all the threshold to the pumping density in the revised manuscript.

Comment-6 Is this strategy developed previously with other materials?
Our Reply: We thank the reviewer for this careful review. This strategy has not been demonstrated in any materials before. Our experiment is the first time to illustrate this mechanism. This is also the reason that we extend the researches from perovskite laser to typical polymer laser. We want to show that the mechanism developed from our perovskite systems can also impact the conventional microlaser community.
Comment-7 I miss in the supporting information the spectra below and above the lasing threshold. Here I think it should be better to show the Y-axis in a logarithmic scale. In addition, it would be interesting to include in the supporting information the integrated intensity as a function of the excitation fluency in a log-log plot.
Our Reply: We thank the reviewer for this valuable suggestion. We have added these information in the Section IV of revised supplementary. "In this section, we will represent the detail of the lasing action. As shown in Fig. S5, we give the spectra below and above the lasing threshold, and the integrated intensity as a function of the excitation also summarized in Fig. S5." Figure R8: The spectra below, at, and above the lasing threshold (black, pink, and blue solid 15 / 19 line), and the integrated intensity as a function of the pumping density (orange circles). (Fig.   S5 in supplemental information) Figure 1 indicates strong scattering at long wavelengths. Are results affected by this scattering? What is the size of the spot in the absorption measurements?

Comment-8 Absorption curve in
Our Reply: We thank the reviewer for this careful review. In the absorption measurement, the diameter of light spot is 40 μm. The increase of absorption at longer wavelength should be caused by the scattering loss of multi-perovskite microrods. The description on the absorption measurement has been included under Fig. S3 of Supplementary. "Basically, the white light is collimated and then focused by a 20× objective lens onto the top surface of the sample, where the diameter of the spot is about 40 μm." Comment-9 How n and k are measured?
Our Reply: We thank the reviewer for this careful review. The n and k of perovskites are measured by Ellipsometer. In the revised manuscript (para-2, page-6), "In our calculations, the structural parameters followed the SEM image, whereas the refractive index (n) and light extinction coefficient (k) were measured by ellipsometer experimentally (see Fig. S7)." Comment-10 I do not understand the section VII included in the supporting information. This section is even longer than the main manuscript and it is only called in the conclusions.
Our Reply: We thank the reviewer for this careful review. Section VII in the supplementary demonstrated that the proposed strategy to achieve mode switching is applicable to other materials and systems. Meanwhile, we also try to emphasize the robustness of the obtained mechanism. While the production rate is low, it is mainly because that the synthesized samples have random sizes. Once the cavity size and shapes can be precisely controlled with the top-down nanofabrication technique, the obtained mechanism can be easily reproduced in many samples on the same wafer. We hope this information can impact the microlaser researches in other material systems.
Our Reply: We thank the reviewer for this careful review. We have checked through the manuscript carefully and corrected the typos, misspellings and grammatical errors.

Comment-12
Reference 39 is missing.  871-875 (2014).], and the sentence has been rewritten as the following: "In this research, we explore nonlinear modal interactions in perovskite microlasers and demonstrate their impacts on ultrafast mode switching, especially with cross gain saturation 46 ,47 where the intensity of one lasing mode reduces the available gain for all other modes in the same system." Comment-14 I do not understand why the inset of Figure 2d is referred in Line 142. Anyway, this inset is not commented within the Figure (lines 95-104).
Our Reply: We thank the reviewer for this valuable suggestion. In the revised manuscript, the description on the inset of Figure 2(d) has been added in para-2, page-4. "The inset of Fig.   2(d) shows the polarization of these two WG modes, which are both transverse electrically (TE) polarized with dominant electric field perpendicular to the light propagation direction in the cross-sectional plane."

Comment-15
Line 149. Are cross-section interaction coefficients normalized magnitudes? or are units missed?
Our Reply: We thank the reviewer for this careful review. We thank the reviewer for this excellent question. Here the field distributions u ( ) and ( ) are normalized and dimensionless, and hence the cross-interaction coefficients defined in Line 149 is also dimensionless. We have clarified this point by revising the sentence below Eq.
(2) to "…u ( ) is the normalized and dimensionless field distribution of mode-μ in the cavity." We also corrected a related typo in Eq. (S7) of the SI, which is now consistent with the rest of the manuscript.

/ 19
Reply to Reviewer #2: We would like to thank the reviewer for the valuable suggestions and recommendation for publication in Nature Communications after a minor revision. Based on the comments, we have carefully revised our manuscript and the manuscript quality has been significantly improved.
Comment-1. The references in the introduction section should be carefully checked, e.g. though the introduction mainly focuses on MAPbX3, some works on all inorganic perovskites have been cited; on Line-41, Page-2, the references 28-33 are not about photonic crystals; the reference 39 is missing in the reference list. Our Reply: We thank the reviewer for this careful review. The record threshold and Q factor of perovskite microlasers have been corrected in the revised manuscript in para-1, page-2.
"The threshold and quality (Q) factors of perovskite microlasers have been improved to 220 nJ/cm 2 17 and 1×10 4 12 , respectively." Comment-3. In Figure 1, authors mainly discuss the crystalline quality and optical properties of perovskites, which are not directly relevant to the topic of the manuscript. Thus I suggest that authors add them into the supporting information.
Our Reply: We thank the reviewer for this valuable suggestion. We agree with the reviewer that some detail analysis of structural information can be placed in the supplemental information. Following the reviewer's suggestion, we have compressed the contents of this part (para-2, page-3) in the revised manuscript. "The single crystal nature of synthesized perovskite microrods were determined by the following X-ray diffraction (XRD) spectrum and high-resolution transmission electron microscopy (HRTEM) investigation (see Fig. S2).
Figure 1(b) shows the absorption and photoluminescence spectra of MAPbBr3 perovskites microrods (see methods, Fig. S3 and Fig. S4). A clear bandedge can be seen at ~ 2.32 eV, consistent with the previous reports well 12 ." As shown in Fig. R9 below, the corresponding contents in Fig. 1 have also been modified.
The structural information of MAPbBr 3 perovskite microrod have been moved to Fig. S2 in the supporting information in Section I. "In the main text, in order to characterize the material properties of single crystal perovskite microrod, we use the high-resolution transimission electron microscopy (HRTEM) and the fast Fourier transform (FFT) pattern,

/ 19
associated with the XRD spectrum, clearly confirms the single crystal nature of the synthesized MAPbBr 3 microrods. Fig. S2(b) is the X-ray diffraction ( Comment-4. The description on the excitation is not proper: pumping density should be changed to pumping energy, considering that joules are used in the manuscript to describe the excitation.
Our Reply: We thank the reviewer for this valuable suggestion. In the revised manuscript, the descriptions on excitation fluence have been changed to pumping density with unit of μJ/cm 2 .
Comment-5. In Figure 2c, the intensity of mode 2 slightly drops when the excitation energy is above 0.8 µJ. What is the cause of the drop? It seems that the Auger effect starts to play an important role. The author should discuss this and give a possible way to solve this effect.
Our Reply: We thank the reviewer for this careful review and valuable comment. The reviewer is absolutely right that Auger recombination began to affect the output power at high pumping fluence. As a result of multiple-particle loss (Auger nonradiative recombination), the perovskite microlaser exhibits a flat or even negative power slope at high pumping density (Zhang, C. et al. Adv. Opt. Mater. 4, 2057-2062(2016 Adv. Opt. Mater. 4, 2057-2062(2016), the Auger effect can be significantly eliminated by simply covering a few layer graphene. In the revised manuscript, the discussion on the intensity drop at high pumping density in Fig. 2(c) has been added (para-2, page-4). "It is worth noting that the emission intensity of mode-2 shows a slight drop with further increase of pumping density. This caused by Auger recombination at high pumping fluence. This limitation can be reduced with additional technique such as covering the sample with few-layer graphene 48 ." Comment-6. Do the two competing modes in Figure 2b have the same polarization? The polarization shown in the inset of Figure 2d has not been described with detail information.
Our Reply: These two WG modes have the same polarization. In the revised manuscript, the description on the inset of Figure 2(d) has been added in para-2, page-4. "The inset of Fig. 2(d) shows the polarization of these two WG modes, which are both transverse electrically (TE) polarized with dominant electric field perpendicular to the light propagation direction in the cross-sectional plane." and the detail of the polarization is shown in Figure R11.
Our Reply: We thank the reviewer for this careful review. We have checked through the manuscript carefully and corrected the typos and grammar errors.