Article | Open

Giant five-photon absorption from multidimensional core-shell halide perovskite colloidal nanocrystals

  • Nature Communications 8, Article number: 15198 (2017)
  • doi:10.1038/ncomms15198
  • Download Citation
Published online:


Multiphoton absorption processes enable many technologically important applications, such as in vivo imaging, photodynamic therapy and optical limiting, and so on. Specifically, higher-order nonlinear absorption such as five-photon absorption offers significant advantages of greater spatial confinement, increased penetration depth, reduced autofluorescence, enhanced sensitivity and improved resolution over lower orders in bioimaging. Organic chromophores and conventional semiconductor nanocrystals are leaders in two-/three-photon absorption applications, but face considerable challenges from their small five-photon action cross-sections. Herein, we reveal that the family of halide perovskite colloidal nanocrystals transcend these constraints with highly efficient five-photon-excited upconversion fluorescence—unprecedented for semiconductor nanocrystals. Amazingly, their multidimensional type I (both conduction and valence band edges of core lie within bandgap of shell) core–shell (three-dimensional methylammonium lead bromide/two-dimensional octylammonium lead bromide) perovskite nanocrystals exhibit five-photon action cross-sections that are at least 9 orders larger than state-of-the-art specially designed organic molecules. Importantly, this family of halide perovskite nanocrystals may enable fresh approaches for next-generation multiphoton imaging applications.


Multiphoton absorption (MPA) involves the simultaneous absorption of multiple monochromatic infrared photons that excites an electron to a higher energy state, whereupon its subsequent relaxation yields the emission of a shorter wavelength photon. Such frequency-upconverted fluorescence possesses many attractive merits over linearly excited fluorescence (for example, high spatial confinement, long penetration depth and low biological damage)1,2,3,4,5,6 due to the longer infrared excitation wavelengths. Two-photon/three-photon absorption (2PA/3PA) have found widespread applications in three-dimensional (3D) biomedical imaging1,2,3,4,5,6, optical power limiting7, sensing8 and 3D optical data storage9 and so on. Specifically, higher-order nonlinear absorption (for example, 4PA or 5PA) are especially desirable for imaging10. For such applications, materials with large MPA cross-sections (σn) and high photoluminescence (PL) quantum yield (PLQY, η) are required, hence the parameter known as action cross-section (that is, η × σn). Suitable media include: conventional inorganic semiconductor nanocrystals (NCs)11,12,13, organic chromophores14,15,16, polymers17,18, metal complexes19 and biomolecules20. Among them, the former exhibit exceptional performances with relatively large two- and three-photon action cross-sections ((ησ2)max 104 GM and (ησ3)max 10−76 cm6s2 photon−2)10,11,12,13,14,15,16,21,22. Nonetheless, it is non-trivial to extend to 4PA/5PA in conventional semiconductor NCs (for example, II–VI) and organics because of their relatively smaller ησn≥4, thereby requiring large intensities that can easily cause material damage. Hence, reports on four-photon-excited upconversion emission are few and far between23,24. To the best of our knowledge, five-photon-excited upconversion fluorescence from semiconductor NCs has yet to be demonstrated. Such endeavour for organic chromophores necessitates careful design and synthesis (for example, (E)-3-(4-(2-(1-hexyl-4-methyl-1H-imidazol-5-yl) vinyl) pyridinium-1-yl) propyl sulfate)10.

Organic–inorganic halide perovskites, which demonstrated superior optoelectronic properties for photovoltaics and light emission25,26, have recently demonstrated promising nonlinear optical properties27,28,29,30,31. These include: strong second-harmonic generation in organic/inorganic germanium perovskite compounds27; 2PA and two-photon-excited fluorescence in CH3NH3PbBr3 (MAPbBr3) perovskite bulk crystal at 800 nm (ref. 28); 2PA/3PA in CH3NH3PbI3 (MAPbI3) perovskite bulk crystal at three discrete wavelengths29; and large ησ2106 GM (at 800 nm) in cubic colloidal CsPbBr3 NCs (2 orders larger than conventional semiconductor NCs30,31). Nonetheless, detailed understanding into these low-order nonlinear properties is still severely lacking (for example, the broad spectral dependence of 2PA/3PA cross-sections; the influence of the organic or inorganic A cation (that is, CH3NH3+ or Cs+) on the MPA). We also note an emerging controversy over the contrasting nonlinear optical behaviour for CsPbBr3 NCs where both saturable absorption32 and 2PA30,31 at 800 nm were reported (Supplementary Table 1). Critically, their higher-order nonlinear optical properties (that is, 3-, 4-, 5PA) are presently unknown.

One viable approach to enhancing ησn in quantum confined NCs is through the core–shell structure that permits/facilitates: effective passivation of non-radiative surface traps33,34; antenna-like effect35; photoinduced screening of the internal field34,36; and/or local field effect34,37. We recently realized highly luminescent (PLQY 92%) type I core–shell multidimensional perovskite MAPbBr3/(OA)2PbBr4 NCs comprising of a wider bandgap 2D-layered perovskite ((OA)2PbBr4) shell encapsulating a 3D MAPbBr3 NC core38. These novel halide perovskite NCs possesses the necessary criteria discussed above (as detailed in the Discussion section) that will open exciting new prospects for tuning the higher-order nonlinear optical effects (that is, 4PA/5PA).

Herein, we reveal that the family of colloidal halide perovskite (that is, CsPbBr3, MAPbBr3 and core–shell MAPbBr3/(OA)2PbBr4) NCs possesses ultralarge MPA properties, with five-photon action cross-sections (ησ510−136 cm10s4 photon−4) that are at least 9 orders larger than the record values of designer organic molecules (ησ510−145 cm10s4 photon−4) reported recently10. The MPA cross-sections (2-, 3-, 4- and 5PA) and their spectral dependences over the wide wavelength range of 675–2,300 nm are meticulously established. Importantly, multidimensional core–shell halide perovskite NCs with superior PL stability afford enhanced tunability of their MPA properties that would enable fresh approaches for multiphoton imaging applications.


Giant 5PA from perovskite colloidal NCs

These halide perovskite colloidal NC samples (in toluene) were synthesized using various solution-processed approaches. Briefly, CsPbBr3 NCs (of side9 nm) were fabricated using a modified one-step technique39, while both the core-only MAPbBr3 NCs (diameter8-9 nm) and the core–shell MAPbBr3/(OA)2PbBr4 NCs (diameter9–10 nm) were synthesized using the ligand-assisted reprecipitation method38—see Methods and Supplementary Information for more details of the synthesis and the transmission electron microscopy images. The core–shell MAPbBr3/(OA)2PbBr4 NCs possess a type-I conduction and valence band edge alignment between the core and shell—see schematic in Figure 1a, as validated by both photoelectron spectroscopy in air measurements and theoretical calculations (Supplementary Figures 1,2 and Supplementary Methods). The linear optical properties (that is, 1PA and one-photon-excited PL—Supplementary Figure 1) of these halide perovskite NCs show clear signatures of excitonic and quantum confinement effects, where the larger blueshift in CsPbBr3 NCs stems from its larger Bohr diameter (dB7 nm) compared to MAPbBr3 NCs (dB4 nm), in agreement with literature reports (see Supplementary Methods for more details). Their PLQY are84%,92% and55% for MAPbBr3, MAPbBr3/(OA)2PbBr4 and CsPbBr3 NCs, respectively (see Supplementary Methods for details of the PLQY characterization). Figure 1b shows the five-photon-excited (that is, at 2,100 nm wavelength) excitonic PL emission (centred around 520 nm) from these samples with excitation fluence range3.30–5.61 mJ cm−2 (inset) and their quintic excitation fluence dependence (inset). The 5PA excitation process via virtual energy levels is schematically illustrated in Figure 1c. Figure 1d shows the ultralarge ησ5 (and their corresponding σ5 values (inset)) obtained through employing the σ2 values at 800 nm measured by Z-scan40 (Supplementary Figures 3,4 and Supplementary Notes) as a standard and applying the calculation equations (see Supplementary Notes for details)41 for these halide perovskite colloidal NC samples over the infrared excitation wavelengths of 2,050–2,300 nm. Record ησ510−136 cm10s4 photon−4 values are achieved with MAPbBr3/(OA)2PbBr4 NCs>CsPbBr3 NCs>MAPbBr3 NCs (see Table 1), highlighting the significance of the (OA)2PbBr4 shell for enhancing the nonlinear optical absorption action cross-sections. All three samples follow a similar spectral dependence with ησ5 for MAPbBr3/(OA)2PbBr4 NCs6–8 times larger than MAPbBr3 NCs (see Table 1).

Figure 1: Type I core–shell multidimensional perovskite NCs and their giant five-photon action cross-sections (ησ5).
Figure 1

(a) Schematic illustrating the core–shell multidimensional perovskite NCs with 3D MAPbBr3 as core and 2D (OA)2PbBr4 as shell, and their type-I energy level alignment. (b) 5PPL spectra from core-only MAPbBr3 NCs (2.0 μM in toluene), core–shell MAPbBr3/(OA)2PbBr4 NCs (2.1 μM in toluene) and CsPbBr3 NCs (1.0 μM in toluene), with femtosecond laser excitation at 2,100 nm. Inset shows the quintic dependence on the excitation fluence of the spectrally integrated PL intensity. (c) Schematic illustrating the 5PPL process in perovskite NCs. (d) Five-photon action cross-section (ησ5) spectra of the MAPbBr3, MAPbBr3/(OA)2PbBr4 and CsPbBr3 NCs. Error bars indicate experimental uncertainty of ±15%.

Table 1: Multiphoton action cross-sections ησ n of lead bromide perovskite NCs over their respective wavelength range*.

The 2PA cross-sections (σ2 values at 800 nm) of the colloidal NCs (measured using the open-aperture Z-scan technique) were employed as a standard for multiphoton excited PL (MEPL) measurements at different wavelengths (see ‘Methods’ section and Supplementary Notes for more details). Our measured σ2 values (from Z-scan) for CsPbBr3 NCs at 800 nm agrees well with literature reports30,31. The consistency of the ratio of the σ2 values (for MAPbBr3, CsPbBr3 and MAPbBr3/(OA)2PbBr4 NCs at 800 nm) obtained from Z-scan (Supplementary Figure 4) with the two-photon-excited PL measurements (Supplementary Figures 5, 6a and 7a) provides further validation of our approach. Moreover, the good agreement of the measured 3PA cross-sections with those acquired from open-aperture Z-scan measurements at 1,050 and 1,100 nm (see Supplementary Figures 6b,7b,8 and 9 and Supplementary Notes for more details) further confirms that the MPA cross-sections have been properly measured with the MEPL technique. Similar to Figure 1b, MEPL spectra under 2-, 3- and 4-photon excitation for the NCs at 800, 1,200 and 1,600 nm with excitation fluence ranging from0.33 to 1.16,0.66 to 1.82 and 1.16 to 2.31 mJ cm−2 (that is, higher upconversion PL intensities in the lower-order multiphoton processes even at lower excitation fluence), respectively, are shown in Supplementary Figures 6a–c, and illustrated schematically in Supplementary Figure 6d. Supplementary Figures 6a–c insets clearly show the nearly quadratic, cubic and quartic dependences of the spectrally integrated PL intensity on excitation fluence for the 2-, 3- and 4-photon processes with excitation fluence ranging from 0.33 to 1.16,0.66 to 1.82,1.16 to 2.31 mJ cm−2, respectively. The comparison between the normalized one-photon-excited PL spectra and MEPL spectra of the NCs are displayed in Supplementary Figure 10. As illustrated in Supplementary Figure 10, the slight redshift of the MEPL spectra with respect to one-photon counterpart has been well reported in traditional semiconductor NCs42 and can be ascribed to the reabsorption effect and size inhomogeneity43,44. Further excitation fluence-dependent MEPL measurements at wavelengths ranging from 675 to 2,300 nm were also performed. Photographs in Supplementary Figure 11 clearly demonstrate the frequency-upconverted PL from NCs when they were irradiated with infrared femtosecond laser pulses. Furthermore, the direct comparison between the five-photon excited upconversion PL from MAPbBr3/(OA)2PbBr4 NCs and that from R6G having the same concentration and under the same experimental conditions exemplify the superior 5PA properties of the MAPbBr3/(OA)2PbBr4 NCs (see Supplementary Figure 12 and Supplementary Notes for more details).

Multiphoton action cross-section spectra in perovskite NCs

Figure 2a shows a summary of the excitation wavelength dependence of the slopes (that is, orders of MPA processes) for these halide perovskite NCs spanning from 675 to 2,300 nm. The slopes of PL from the NCs are around 2 in the wavelength range of 675–1,000 nm, clearly indicating 2PA. As the excitation wavelength increases to 1,050–1,500 nm, the slopes increase to around 3—revealing a switch of the excitation mechanism to 3PA. 4PA processes (with slopes around 4) dominate as the excitation wavelength is further increased to the range of 1,550–2,000 nm. In the long wavelength range 2,050–2,300 nm, the slopes are around 5, indicating the dominance of 5PA processes. Due to the NCs’ size inhomogeneity, ησn at the wavelength boundaries contains an admixture of contributions from both the lower- and higher-order MPA process (that is, 2/3PA, 3/4PA and 4/5PA and so on). Instead of a sharp transition at the wavelength boundaries (of 1,050, 1,550 and 2,050 nm), the slope deviates from the integer value by20%. Figure 2b–d shows a direct comparison of the NCs’ MEPL brightness (that is, ησn) for lower orders, n=2, 3 and 4, over the wavelength range 675–2,000 nm. Similar to ησ5, their spectral dependences exhibit a general overall decreasing trend with increasing wavelength. Their corresponding σn values obtained through applying the calculation equations (see Supplementary Notes for details) and employing the σ2 values at 800 nm measured by Z-scan as a standard are given in Figure 2b–d insets. In all these cases, the core–shell MAPbBr3/(OA)2PbBr4 NCs possess6–8 times larger ησn values than the core-only MAPbBr3 NCs. The ησn spectra suggest that the growth of the 2D (OA)2PbBr4 shell over MAPbBr3 NCs only enhances their MPA and PLQY, while hardly changing their spectral response.

Figure 2: Excitation fluence dependence of multiphoton-excited upconversion PL(slopes) and action cross-sections as a function of excitation wavelengths.
Figure 2

(a) Slopes n plotted as a function of laser excitation wavelength (photon energy), where n is defined as the excitation fluence dependence of the MPPL signal that is proportional to (excitation fluence)n. (b) Two-photon action cross-section (ησ2) spectra of the MAPbBr3, MAPbBr3/(OA)2PbBr4 and CsPbBr3 NCs in the wavelength range 675–1,000 nm. (c) Three-photon action cross-section (ησ3) spectra of the perovskite NCs in the range 1,050–1,500 nm. (d) Four-photon action cross-section (ησ4) spectra of the perovskite NCs in the range 1,550–2,000 nm. Insets in (bd) show the corresponding spectral dependence of MPA cross-sections (σn) of the perovskite NCs. Error bars indicate the experimental uncertainty of ±15%.

Table 1 shows the elucidated ησn for MAPbBr3, MAPbBr3/(OA)2PbBr4 and CsPbBr3 NCs. Comparatively, the ησ2 values of these lead bromide perovskite NCs are 1–2 orders larger than that of large size CdSe/CdS dot in rod heterostructures (39 or 180 nm length CdS nanorod—Supplementary Table 2)35,45, and are 2–3 orders larger than traditional inorganic semiconductor NCs and organic chromophores (Supplementary Table 2). For ησ3 values, the NCs are 1–2 orders larger than the strongly (or overly) excited ZnSe/ZnS core–shell NCs46 (where three-photon-excited intraband absorption has a significant contribution47) and the large CdSe/CdS dot-in-rod heterostructure (39 nm length CdS nanorod)48 (Supplementary Table 3).

Our lead bromide perovskite NCs exhibit ησ3 values 2–3 orders of magnitude higher than the best performing conventional inorganic semiconductor NCs (that are not overly excited and with size ≤10 nm) and organic molecules (see Supplementary Table 3). For the higher-order ησ4 values, our NCs are 3–5 orders larger than the best reported results of organic chromophores16,23,24 (see Supplementary Table 4). However, there are only limited research efforts devoted to studying the four-photon process in conventional inorganic semiconductor NCs. Amazingly, the ησ5 values of our NCs are >9 orders larger than specially designed organic molecules10 (ησ510−145 cm10s4 photon−4) (see Supplementary Table 5), which is a record for semiconductor NCs. Although the ‘Luttinger-Kohn’ and ‘Pidgeon-Brown’ models within the k·p approach have been successfully applied to conventional metal-chalcogenide NCs49,50,51 to model their 2PA/3PA spectral dependences and estimate their σ2/σ3, such approach cannot be directly applied to lead bromide perovskite NCs. This is because these methods are only suitable for the two-photon transitions from s-type valence band to p-type conduction band as in the former, unlike the all p-type valence and conduction bands in the latter. The theoretical study on the MPA properties of the lead bromide perovskite NCs will be the focus of a future work.


The family of lead bromide perovskite NCs possesses much higher ησn values than those of traditional inorganic semiconductor NCs and organic chromophores (Table 1 and Supplementary Tables 2–5), highlighting their potential for nonlinear optics and bioimaging applications, such as optical limiting, 3D microscopy for deep tissue imaging and sensing. Apart from having high PLQY (that is, η84%, 92% and 55% for MAPbBr3, MAPbBr3/(OA)2PbBr4 and CsPbBr3 NCs, respectively), these lead bromide perovskite NCs also possess giant σn values, which stems mainly from the intrinsic strong MPA of lead bromide perovskites. Among our perovskite NCs, a combination of effects like the relatively stronger quantum confinement in CsPbBr3 NCs (dB=7 nm) and/or the influence from Cs+ cation give rise to their larger MPA cross-sections than MAPbBr3 NCs (weaker confinement). By growing a (OA)2PbBr4 shell over the MAPbBr3 core, the PLQY can be enhanced to92% and the photostability can be improved, as demonstrated here and in ref. 38. Most importantly, the presence of the (OA)2PbBr4 shell enhances the σn by almost an order compared to the core-only MAPbBr3 NCs across all wavelengths from 675 to 2,300 nm. Given that both our MAPbBr3 and CsPbBr3 NCs are in the weak confinement regime (dB4 nm<<8–9 nm diameter for MAPbBr3 and dB7 nm<9 nm side for the cubic CsPbBr3), it is possible to enhance these σn values further through even smaller strongly confined MAPbBr3 NCs, their core–shell counterparts (that is, NCs’ diameter <4 nm) and CsPbBr3 NCs (that is, NCs’ side <7 nm). However, it is presently extremely challenging to synthesize small MAPbBr3 (refs 52, 53) and CsPbBr3 NCs54,55 with high crystalline quality, low surface defects, uniform size distribution and relatively high reaction yield, which are needed to conduct detailed investigations of the quantum confinement effect on the MPA of lead bromide perovskite NCs. The all-inorganic CsPbBr3 NCs were prepared via the hot-injection method39 at temperatures between 140–180 °C. The NC size is determined by the reaction temperature, where larger NCs are obtained at higher temperature. However, below 140 °C, nanoplatelets (not NCs) are formed instead54,55. Moreover, at the current stage, despite MAPbBr3 NCs with small sizes can be obtained52,53, various problems associated with high density surface defects, low sample crystalline quality, low reaction yield and non-uniform size distribution need to be further addressed52,53. Presently, there have been no reports on the synthesis of small-sized core–shell MAPbBr3/(OA)2PbBr4 NCs. Therefore, significant advances in perovskite NCs synthesis are needed before this question on the detailed dependence of MPA on the quantum confinement effect could be answered.

Previous studies on conventional inorganic semiconductor NCs have shown that the 2PA and 3PA cross-sections can be enhanced by an outer shell covering through: surface passivation effects33,34; antenna-like effect35; photoinduced screening of the internal field34,36; and/or local field effects34,37. To establish the origins for the σn enhancement of the core–shell MAPbBr3/(OA)2PbBr4 NCs and to gain more insights into the dynamics of excited photocarriers, time-resolved PL measurements using single (400 nm) and multiphoton (for example, 2P (800 nm), 3P (1,200 nm), 4P (1,600 nm) and 5P (2,100 nm)) excitation were performed. Figure 3a–c shows the one-photon- and multiphoton-excited time-resolved PL decay lifetimes for the MAPbBr3, MAPbBr3/(OA)2PbBr4 and CsPbBr3 NCs, respectively. For the respective MAPbBr3 and CsPbBr3 NCs, their PL decay curves are almost invariant for one- and multiphoton excitation (Figure 3a,c), indicating excitation via virtual states to the same lowest excited state in these perovskite NC systems.

Figure 3: Comparison between the PL decay traces of the perovskite NCs under various multi-photon excitations.
Figure 3

(a) One- and multiphoton-excited PL decay curves in MAPbBr3 NCs. (b) Longer one- and multiphoton-excited PL decay lifetimes are obtained for MAPbBr3/(OA)2PbBr4 NCs than for MAPbBr3 NCs. (c) Shorter PL decay lifetimes in CsPbBr3 NCs under one- and multiphoton excitation.

The increased PL lifetimes (from one- and multiphoton excitation—Figure 3b) together with the enhanced PLQY and stability38 in MAPbBr3/(OA)2PbBr4 NCs compared to the core-only MAPbBr3 NCs points to the effective surface passivation provided by the (OA)2PbBr4 shell that reduces the nonradiative surface traps, thus increasing the multiphoton transition probability33,34. Furthermore, the longer one- and multiphoton excited PL lifetime, the good spectral overlap between the shell emission and the core excitation (as revealed in Supplementary Figure 13), as well as the intimate proximity between the core and shell strongly suggest the presence of non-radiative Förster-type energy transfer from the shell to the core through an antenna-like effect. This is another plausible origin for the enhanced σn in the core–shell MAPbBr3/(OA)2PbBr4 NCs19,35. The 2PA property of the (OA)2PbBr4 shell revealed by open-aperture Z-scan measurements in Supplementary Figure 14 further supports this interpretation. The non-radiative energy transfer from the shell to core enhances the PL from the MAPbBr3 NCs core, resulting in an overall increase of the PL lifetime19,35. The longer multiphoton excited PL lifetime compared to the one-photon excited PL lifetime in MAPbBr3/(OA)2PbBr4 NCs (Figure 3b) indicates a larger contribution from this non-radiative energy transfer channel to the resulting core PL, suggesting a more effective non-radiative energy transfer under nonlinear optical excitation19,35.

Next, the relatively large lattice mismatch between the core MAPbBr3 (ref. 56) and shell (OA)2PbBr4 (ref. 57) could also induce intrinsic piezoelectric polarization charges at the core–shell interface that result in an internal electric field. This indicates that the photoinduced screening of the internal field by the photoexcited electron–hole pairs could be another possible origin for the enhanced σn in MAPbBr3/(OA)2PbBr4 NCs34,36. Furthermore, the dielectric confinement (local field effect) resulting from the relatively large difference between the dielectric constants of the core MAPbBr3 (ref. 58) and shell (OA)2PbBr4 (ref. 59) could be another factor34,37. Therefore, it is likely that an interplay of various factors arising from the effective surface passivation, the photoinduced screening of the internal field, the nonradiative energy transfer from the shell to the core through antenna-like effect, the local field effect or its combination gives rise to the giant MPA cross-sections of these core–shell multidimensional halide perovskites NCs.

The outstanding higher-order nonlinear optical properties of the lead bromide perovskite colloidal NCs (particularly the highly efficient five-photon excited upconversion fluorescence in the multidimensional core–shell perovskite NCs (for example, MAPbBr3/(OA)2PbBr4)) indicate their great potential for developing next-generation highly efficient, sensitive multiphoton imaging applications with unmatched imaging depth and resolution. Although the intrinsic low stability of halide perovskite NCs in polar solvents (such as water) and their cytoxicity from the Pb2+ ion could be an issue for potential applications in multiphoton bioimaging, such technical challenges could be circumvented through the following three possible encapsulation approaches:

The first is encapsulation with a SiOx/SiO2 inert shell/matrix. Utilizing SiOx/SiO2 either as an additional shell or as a medium to embed NCs will help to enhance the stability as well as reduce the cytoxicity. Such approach has been successfully applied to traditional inorganic semiconductor NCs60,61,62,63,64 for efficient bioimaging applications. Very recently, encapsulating lead bromide NCs into a SiOx spherical matrix (150 and 470 nm in diameter) have been realized by Huang et al.65 This provides a crucial proof-of-concept on the viability of the approach, although more work is required to achieve finer control of SiOx/SiO2 encapsulation for each perovskite NC.

The second is encapsulation with ligands. Through utilizing 3-aminopropyltriethoxysilane (APTES) as a branched capping ligand to replace the conventional straight-chained ligands, Luo et al.66 achieved enhanced stability in the APTES-coated MAPbBr3 NCs. Such APTES-coated MAPbBr3 NCs show an increased stability in 2-propanol. Furthermore, the available silane group allows for additional sol–gel reactions, which may further increase the stability of the NCs.

The third is encapsulation with solid lipid structures. More recently, Gomez et al.67 reported the application of solid lipid structures to encapsulate CsPbBr3 NCs that were stable up to 2 months in water, albeit at a low PLQY (around 11%) likely due to initial water degradation. Further work is needed to circumvent the initial degradation and realize water-stable NCs at high PLQY.

These examples clearly show that the potential stability and cytotoxicity issues of halide perovskite NCs in aqueous media for multiphoton imaging applications could, in fact, be overcome through a judicious choice of the encapsulation approach and careful optimization. Through tuning the emission wavelengths to the infrared (for example, using CsSnI3 NCs), these halide perovskite colloidal NCs could enable even deeper imaging for deep-tissue bioimaging, where both penetration of the incident photons and extraction of the emitting photons are essential.

In retrospect, our findings reveal that the family of halide perovskite colloidal NCs possesses outstanding higher-order nonlinear optical properties (for example, highly efficient five-photon-excited upconversion fluorescence), which is unprecedented for semiconductor NCs. Particularly, overcoating the 3D perovskite nanocrystals (for example, MAPbBr3) with a 2D perovskite shell (for example, core/(OA)2PbBr4 shell) to form multidimensional core–shell perovskite nanocrystals offers amazing five photon action cross-sections that are at least 9 orders larger (ησ510−136 cm10s4 photon−4) than state-of-the-art specially designed organic molecules. Our work aptly demonstrates that this new family of perovskite nanocrystals is a promising class of nonlinear optical materials for developing next-generation multiphoton imaging applications with unmatched imaging depth, sensitivity and resolution.


Sample preparation

The methylammonium lead bromide perovskite (MAPbBr3) nanocrystals were synthesized using the recently developed ligand-assisted reprecipitation strategy38. A mixture of 0.16 mmol methylammonium bromide (MABr), 0.2 mmol lead bromide (PbBr2) was first dissolved in 5 ml dimethylformamide (DMF) in a glass vial. Then, 50 μl oleylamine (OAm) and 0.5 ml oleic acid (OAc) were added into the above DMF solution to make the precursor solution. 250 microlitres of the as-prepared precursor solution was then swiftly injected into 5 ml of toluene kept at 60 °C in a round bottom glass flask, and stirred vigorously for 5 min. The MAPbBr3 nanocrystals were formed as confirmed by the change of the solution colour into green. The mixture was transferred into a centrifuge tube to be centrifuged at a relative centrifugal force of about 6738g (Model: Eppendorf 5804R, rotor FA-45-6-30, rotor radius 12.3 cm). For 10 min to purify the nanocrystals. The supernatant containing the nanocrystals was collected for further investigations and the precipitate was discarded. More details can be found in ref. 38.

To fabricate the core–shell multidimensional perovskite nanocrystals MAPbBr3/(OA)2PbBr4, octylammonium bromide was first synthesized utilizing octylamine and Hydrobromic acid (HBr). Then, methylammonium bromide and octylammonium bromide with molar ratio of 8:2 were added to the precursor solution. The core–shell multidimensional perovskite NCs MAPbBr3/(OA)2PbBr4 were synthesized following the same synthetic procedures elaborated above. The core–shell NCs were formed through the co-precipitation of the mixed methylammonium and octylammonium bromide precursors. More details can be found in ref. 38.

In addition, the colloidal CsPbBr3 nanocrystals with cubic shape and edge length9 nm were synthesized following the reported modified one-step technique39. The steps are as follows: preparation of Cs-oleate precursor: 814 mg (2.5 mmol) of caesium carbonate (Cs2CO3) was mixed with 2.5 ml OAc and 40 ml 1-octadecene (ODE) in a 100 ml three-neck round bottom glass flask. At first, the mixed solution was dried by a vacuum pump for 1 h at 120 °C. Then, the reaction mixture was heated under nitrogen environment for another 1 h at 150 °C until all Cs2CO3 reacted with OA and becomes transparent. Cs-oleate precursor was preheated to 100 °C before injection. Synthesis of cubic caesium lead bromide NCs: 69 mg (0.188 mmol) of PbBr2 was mixed in 5 ml ODE, 0.5 ml OAc and 0.5 ml oleylamine in a three-necked flask. The flask was dried under vacuum for 1 h at 120 °C and degassed with nitrogen gas for another 1 h at same temperature. Cs-oleate precursor of 0.4 ml was injected into the main reaction solution at 170 °C via syringe, yielding a bright green colloid NCs. The reaction was quenched with a ice bath for 30 s after the injection of the Cs-oleate precursor. The colloidal CsPbBr3 NCs was transferred to centrifuge tube with some acetonitrile. The mixture was centrifuged at relative centrifugal force of about 7587 × g for 10 min and the precipitation was redispersed in toluene for further experiments.

Sample characterization

Structural characterization of the as-synthesized nanocrystals was performed using both transmission electron microscopy (Jeol JEM-2010) and the X-ray diffraction (XRD Bruker D8 Advance). The concentrations of perovskite nanocrystals in toluene solution were determined by the mass of lead, which were measured by the inductively coupled plasma optical emission spectrometry (ICP-OES Optima 8000; Perkin-Elmer).

1PA spectra of the perovskite nanocrystals in solution phase (toluene) were measured using a Shimadzu UV1800 Ultraviolet –Visible spectrophotometer. A Shimadzu RF-5301pc spectrofluorophotometer was employed to record the one-photon-excited PL spectra of the nanocrystals in toluene. Measurements of the absolute PLQYs of the perovskite nanocrystals were performed using an Ocean-optics USB4000 spectrometer with a BaSO4-coated integrating sphere excited by laser beam at 400 nm.

Multiphoton-excited PL and time-resolved PL measurements

For multiphoton-excited frequency-upconverted PL measurements41, a femtosecond amplified-pulsed laser system was used as the excitation source. The excitation laser pulses (50 fs, 1 kHz, 250–2,600 nm) were generated by an optical parametric amplifier (OperA-Solo, Coherent) pumped by a regenerative amplified femtosecond Ti:Sapphire laser system (50 fs, 800 nm, 1 kHz; Libra, Coherent). The Coherent Libra regenerative amplifier was seeded by a femtosecond Ti:Sapphire oscillator (50 fs, 80 MHz, Vitesse, Coherent). The temporal, spectral and spatial profiles of the applied excitation laser source at wavelengths 675–2,300 nm follow Gaussian distribution based on the specifications in the manufacturer’s data sheets, and the pulse widths at different wavelengths are in the range of 50–60 fs. Moreover, the Gaussian distribution of the temporal, spectral profiles of the applied excitation laser source at wavelengths 675–2,300 nm was experimentally verified by the characterizations with a single-shot autocorrelator (High-Resolution Single-Shot Autocorrelator, Coherent) (see Supplementary Table 6 and Supplementary Methods), a frequency-resolved optical gating (Swamp Optics, UPM-8-50) (see Supplementary Figure 15 and Supplementary Methods), visible monochromator (Acton, Spectra Pro 2750i) coupled with CCD (Princeton Instruments, Pixis 100B) and infrared monochromator (Acton, Spectra Pro 2300i) coupled with liquid-nitrogen-cooled InGaAs infrared detector (Princeton Instruments, 7490–0001) (see Supplementary Figure 16 and Supplementary Methods). Additionally, the pulse widths at different wavelengths were measured to be in the range of (50.6±5.1)–(62.9±6.3) fs, highly consistent with the specifications in the data sheet (see Supplementary Figure 16 and Supplementary Methods). Moreover, through applying knife-edge scans along both the x- and y-directions of the cross-sectional planes of the laser beams, the 2D spatial profiles of the laser beams at wavelengths 675–2,300 nm were validated to follow a Gaussian distribution (see Supplementary Figure 17 and Supplementary Methods). The optical parametric amplifier output was filtered by a suitable filter (long-pass filter cut at 750 nm for wavelength >800 nm, notch filter at 800 nm for wavelengths between 675 and 750 nm) to stop light at undesired wavelengths. A circular lens with focal length of 20 cm was applied to focus the laser pulses onto the perovskite nanocrystals in toluene solution contained in a 2-mm-thick quartz cuvette. The NC samples were placed 23.5 cm away from the lens (3.5 cm away from the focal point) to avoid the high excitation peak intensity on the samples and to have larger excitation area (thus larger frequency-upconverted PL signal). Two continuously variable neutral density filters were employed to control the incident energy of the laser pulses. The frequency-upconverted PL from the nanocrystals was collected at a backscattering angle of 150° utilizing a pair of lenses into an optical fibre which was coupled into a spectrometer (Acton, Spectra Pro 2750i) and detected by a charge-coupled device (Princeton Instruments, Pixis 100B). A short-pass filter cut at 650 nm was placed in front of the fibre to remove the scattered light at the excitation laser frequencies. For the time-resolved PL measurements, the one-photon- and multiphoton-excited PL emission were acquired by a Optronis Optoscope streak camera system with an ultimate temporal resolution of10 ps.

Open-aperture Z-scan measurements

For the Z-scan measurements40 at 800 nm, 1050, nm and 1100, nm excitations, a beam splitter was employed to divide the incident laser beam into two parts. The first part served as the reference and was directed into a reference power detector (DR, RkP 465, Laser Probe). The other part functioned as the signal beam and was focus by a circular lens with 20 cm focus length onto a 1-mm-thick quartz cuvette filled with the toluene solution of perovskite nanocrystals. The transmitted signal laser beam through the nanocrystals was detected by a signal power detector (DS, RkP 465, Laser Probe). Both the reference and signal detectors were coupled with a powermeter (RM6600, National Instrument), whose readings were exported by a Labview USB data acquisition (DAQ, National Instrument) to a computer. The sample was controlled by a linear motorized stage to travel back and forth along the propagation direction of the laser beam (z axis). The transmission of the signal beam through the sample was monitored while translating the sample through the focus, and the transmission was recorded as a function of the sample position (z). With the incident laser pulse energies kept at a constant level, the sample experiences various laser irradiance I(z) at different z-positions, giving rise to corresponding changes in transmission if the sample absorbs light nonlinearly. All optical measurements were performed at room temperature.

Data availability

The data that support the findings of this study are available from the corresponding author on reasonable request.

Additional information

How to cite this article: Chen, W. et al. Giant five-photon absorption from multidimensional core-shell halide perovskite colloidal nanocrystals. Nat. Commun. 8, 15198 doi: 10.1038/ncomms15198 (2017).

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.


  1. 1.

    , & Two-photon laser scanning fluorescence microscopy. Science 248, 73–76 (1990).

  2. 2.

    & Deep tissue two-photon microscopy. Nat. Meth. 2, 932–940 (2005).

  3. 3.

    et al. Noninvasive multiphoton fluorescence microscopy resolves retinol and retinal condensation productsin mouse eyes. Nat. Med. 16, 1444–1449 (2010).

  4. 4.

    et al. In vivo three-photon microscopy of subcortical structures within an intact mouse brain. Nat. Photon 7, 205–209 (2013).

  5. 5.

    et al. Quantum rod bioconjugates as targeted probes for confocal and two-photon fluorescence imaging of cancer cells. Nano Lett. 7, 761–765 (2007).

  6. 6.

    et al. Multiphoton absorption induced amplified spontaneous emission from biocatalyst-synthesized ZnO nanorods. Appl. Phys. Lett. 92, 233116 (2008).

  7. 7.

    et al. Two-photon absorption and broadband optical limiting with bis-donor stilbenes. Opt. Lett. 22, 1843–1845 (1997).

  8. 8.

    et al. Metal-ion sensing fluorophores with large two-photon absorption cross sections:  aza-crown ether substituted donor–acceptor–donor distyrylbenzenes. J. Am. Chem. Soc. 126, 9291–9306 (2004).

  9. 9.

    et al. Two-photon polymerization initiators for three-dimensional optical data storage and microfabrication. Nature 398, 51–54 (1999).

  10. 10.

    et al. Frequency-upconverted stimulated emission by simultaneous five-photon absorption. Nat. Photon 7, 234–239 (2013).

  11. 11.

    et al. Water-soluble quantum dots for multiphoton fluorescence imaging in vivo. Science 300, 1434–1436 (2003).

  12. 12.

    et al. Quantum dots for live cells, in vivo imaging, and diagnostics. Science 307, 538–544 (2005).

  13. 13.

    , , , & Quantum dots versus organic dyes as fluorescent labels. Nat. Meth. 5, 763–775 (2008).

  14. 14.

    et al. Design of organic molecules with large two-photon absorption cross sections. Science 281, 1653–1656 (1998).

  15. 15.

    , , & Observation of stimulated emission by direct three-photon excitation. Nature 415, 767–770 (2002).

  16. 16.

    , , & Multiphoton absorbing materials:  molecular designs, characterizations, and applications. Chem. Rev. 108, 1245–1330 (2008).

  17. 17.

    et al. Two-photon pumped lasing from a two-dimensional photonic bandgap structure with polymeric gain material. Adv. Mater. 14, 673–676 (2002).

  18. 18.

    et al. Nonlinear absorption spectrum in MEH-PPV/chloroform solution:  a competition between two-photon and saturated absorption processes. J. Phys. Chem. B 108, 5221–5224 (2004).

  19. 19.

    et al. Efficient energy transfer under two-photon excitation in a 3D, supramolecular, Zn(II)-coordinated, self-assembled organic network. Adv. Opt. Mater. 2, 40–47 (2014).

  20. 20.

    et al. Two-photon absorption investigation in reduced and oxidized cytochrome c solutions. Chem. Phys. Lett. 390, 506–510 (2004).

  21. 21.

    , , , & Exceptionally strong multiphoton-excited blue photoluminescence and lasing from ladder-type oligo(p-phenylene)s. J. Am. Chem. Soc. 134, 7297–7300 (2012).

  22. 22.

    et al. Multiphoton harvesting metal-organic frameworks. Nat. Commun. 6, 7954 (2015).

  23. 23.

    , , , & Three- and four-photon absorption of a multiphoton absorbing fluorescent probe. Appl. Opt. 43, 5394–5398 (2004).

  24. 24.

    , , & Three- and four-photon excitation of poly(2-methoxy-5-(2′-ethylhexyloxy)-1,4-phenylenevinylene) (MEH-PPV). Adv. Mater. 19, 2653–2656 (2007).

  25. 25.

    et al. Long-range balanced electron- and hole-transport lengths in organic–inorganic CH3NH3PbI3. Science 342, 344–347 (2013).

  26. 26.

    & Metal-halide perovskites for photovoltaic and light-emitting devices. Nat. Nanotechnol. 10, 391–402 (2015).

  27. 27.

    et al. Hybrid germanium iodide perovskite semiconductors: active lone pairs, structural distortions, direct and indirect energy gaps, and strong nonlinear optical properties. J. Am. Chem. Soc. 137, 6804–6819 (2015).

  28. 28.

    et al. Two-photon absorption in organometallic bromide perovskites. ACS Nano 9, 9340–9346 (2015).

  29. 29.

    et al. Amplified spontaneous emission from organic–inorganic hybrid lead iodide perovskite single crystals under direct multiphoton excitation. Adv. Opt. Mater. 4, 1053–1059 (2016).

  30. 30.

    et al. Nonlinear absorption and low-threshold multiphoton pumped stimulated emission from all-inorganic perovskite nanocrystals. Nano Lett. 16, 448–453 (2016).

  31. 31.

    et al. Two-photon-pumped perovskite semiconductor nanocrystal lasers. J. Am. Chem. Soc. 138, 3761–3768 (2016).

  32. 32.

    et al. Nonlinear optical properties of colloidal CH3NH3PbBr3 and CsPbBr3 quantum dots: a comparison study using Z-scan technique. Adv. Opt. Mater. 4, 1732–1737 (2016).

  33. 33.

    , , , & Surface-related two-photon absorption and refraction of CdSe quantum dots. Appl. Phys. Lett. 99, 021908 (2011).

  34. 34.

    , , , & Effect of shell thickness on two-photon absorption and refraction of colloidal CdSe/CdS core/shell nanocrystals. Appl. Phys. Lett. 99, 231903 (2011).

  35. 35.

    et al. Enhanced tunability of the multiphoton absorption cross-section in seeded CdSe/CdS nanorod heterostructures. Appl. Phys. Lett. 97, 061112 (2010).

  36. 36.

    et al. Intrinsic optical nonlinearity in colloidal seeded grown CdSe/CdS nanostructures: photoinduced screening of the internal electric field. Phys. Rev. B 78, 195313 (2008).

  37. 37.

    , , , & Size dependence of nonlinear optical absorption and refraction of Mn-doped ZnSe nanocrystals. Appl. Phys. Lett. 91, 201103 (2007).

  38. 38.

    et al. Highly stable, luminescent core-shell type methylammonium-octylammonium lead bromide layered perovskite nanoparticles. Chem. Commun. 52, 7118–7121 (2016).

  39. 39.

    et al. Superior optical properties of perovskite nanocrystals as single photon emitters. ACS Nano 9, 12410–12416 (2015).

  40. 40.

    , , , & Sensitive measurement of optical nonlinearities using a single beam. IEEE J. Quantum Elect. 26, 760–769 (1990).

  41. 41.

    & Measurement of two-photon excitation cross sections of molecular fluorophores with data from 690 to 1,050 nm. J. Opt. Soc. Am. B 13, 481–491 (1996).

  42. 42.

    et al. Multi-photon excitation properties of CdSe quantum dots solutions and optical limiting behavior in infrared range. Opt. Express 15, 12818–12833 (2007).

  43. 43.

    et al. Stimulated emission and lasing from CdSe/CdS/ZnS core-multi-shell quantum dots by simultaneous three-photon absorption. Adv. Mater. 26, 2954–2961 (2014).

  44. 44.

    et al. Wavelength dependence of optical nonlinearity of terpyridine-based Zn(II)-coordinated rigid linear polymers. Appl. Phys. Lett. 101, 213302 (2012).

  45. 45.

    et al. Ultralow-threshold two-photon pumped amplified spontaneous emission and lasing from seeded CdSe/CdS nanorod heterostructures. ACS Nano 6, 10835–10844 (2012).

  46. 46.

    , , & Three-photon absorption in ZnSe and ZnSe/ZnS quantum dots. Appl. Phys. Lett. 90, 133113 (2007).

  47. 47.

    , , & High efficiency and nearly cubic powerdependence of below-band-edgephotoluminescence in water-soluble, copperdoped ZnSe/ZnS quantum dots. Opt. Express 16, 5715–5720 (2008).

  48. 48.

    , , , & Three-photon absorption in seeded CdSe/CdS nanorod heterostructures. J. Phys. Chem. C 115, 17711–17716 (2011).

  49. 49.

    & Motion of electrons and holes in perturbed periodic fields. Phys. Rev. 97, 869–883 (1955).

  50. 50.

    & Interband magneto-absorption and Faraday rotation in InSb. Phys. Rev. 146, 575–583 (1966).

  51. 51.

    Physics of Optoelectronic Devices Wiley (1995).

  52. 52.

    , , , & Highly luminescent colloidal nanoplates of perovskite cesium lead halide and their oriented assemblies. J. Am. Chem. Soc. 137, 16008–16011 (2015).

  53. 53.

    et al. Solution synthesis approach to colloidal cesium lead halide perovskite nanoplatelets with monolayer-level thickness control. J. Am. Chem. Soc. 138, 1010–1016 (2016).

  54. 54.

    , , , & Pure white-light emitting ultrasmall organic-inorganic hybrid perovskite nanoclusters. Nanoscale 8, 17433–17439 (2016).

  55. 55.

    , , , & Control of emission color of high quantum yield CH3NH3PbBr3 perovskite quantum dots by precipitation temperature. Adv. Sci. 2, 1500194 (2015).

  56. 56.

    , & Colloidal organohalide perovskite nanoplatelets exhibiting quantum confinement. J. Phys. Chem. Lett. 6, 1911–1916 (2015).

  57. 57.

    , , & Blue-luminescent organic lead bromide perovskites: highly dispersible and photostable materials. J. Mater. Chem. A 3, 14039–14045 (2015).

  58. 58.

    et al. High-quality bulk hybrid perovskite single crystals within minutes by inverse temperature crystallization. Nat. Commun. 6, 7586 (2015).

  59. 59.

    , , , & Photostability of 2D organic-inorganic hybrid perovskites. Materials 7, 4789 (2014).

  60. 60.

    , & Robust, non-cytotoxic, silica-coated CdSe quantum dots with efficient photoluminescence. Adv. Mater. 17, 1620–1625 (2005).

  61. 61.

    et al. Synthesis and properties of biocompatible water-soluble silica-coated CdSe/ZnS semiconductor quantum dots. J. Phys. Chem. B 105, 8861–8871 (2001).

  62. 62.

    et al. Cellular effect of high doses of silica-coated quantum dot profiled with high throughput gene expression analysis and high content cellomics measurements. Nano Lett. 6, 800–808 (2006).

  63. 63.

    et al. Surfactant-assisted synthesis of water-soluble and biocompatible semiconductor quantum dot micelles. Nano Lett. 5, 645–648 (2005).

  64. 64.

    , & Highly luminescent and photostable quantum dot–silica monolith and its application to light-emitting diodes. ACS Nano 7, 1472–1477 (2013).

  65. 65.

    et al. Enhancing the stability of CH3NH3PbBr3 quantum dots by embedding in silica spheres derived from tetramethyl orthosilicate in ‘waterless’ toluene. J. Am. Chem. Soc. 138, 5749–5752 (2016).

  66. 66.

    et al. Organolead halide perovskite nanocrystals: branched capping ligands control crystal size and stability. Angew. Chem. Int. Ed. 55, 8864–8868 (2016).

  67. 67.

    , , & Color-stable water-dispersed cesium lead halide perovskite nanocrystals. Nanoscale 9, 631–636 (2017).

Download references


Financial support from Nanyang Technological University start-up Grants M4080514 and M4081293; the Ministry of Education Academic Research Fund Tier 1 Grants RG184/14 and RG101/15, and Tier 2 Grants MOE2013-T2-1-081, MOE2014-T2-1-044 and MOE2015-T2-2-015; the NTU-A*STAR Silicon Technologies Center of Excellence Program Grant 11235100003 and from the Singapore National Research Foundation through the Singapore–Berkeley Research Initiative for Sustainable Energy (SinBeRISE) CREATE Program and the Competitive Research Program NRF-CRP14-2014-03 is gratefully acknowledged.

Author information

Author notes

    • Guichuan Xing

    Present address: Institute of Applied Physics and Materials Engineering, Faculty of Science and Technology, University of Macau, Macao SAR 999078, China


  1. Division of Physics and Applied Physics, School of Physical and Mathematical Sciences, Nanyang Technological University (NTU), 21 Nanyang Link, SPMS-PAP 03-05, Singapore 637371, Singapore

    • Weiqiang Chen
    • , Guichuan Xing
    • , Qiang Xu
    •  & Tze Chien Sum
  2. Energy Research Institute @ NTU (ERI@N), Research Techno Plaza, X-Frontier Block, Level 5, 50 Nanyang Drive, Singapore 637553, Singapore

    • Saikat Bhaumik
    • , Sjoerd A. Veldhuis
    • , Michael Grätzel
    • , Subodh Mhaisalkar
    •  & Nripan Mathews
  3. Laboratory of Photonics and Interfaces, Department of Chemistry and Chemical Engineering, Swiss Federal Institute of Technology, Station 6, Lausanne 1015, Switzerland

    • Michael Grätzel
  4. School of Materials Science and Engineering, NTU, Nanyang Avenue, Singapore 639798, Singapore

    • Subodh Mhaisalkar
    •  & Nripan Mathews


  1. Search for Weiqiang Chen in:

  2. Search for Saikat Bhaumik in:

  3. Search for Sjoerd A. Veldhuis in:

  4. Search for Guichuan Xing in:

  5. Search for Qiang Xu in:

  6. Search for Michael Grätzel in:

  7. Search for Subodh Mhaisalkar in:

  8. Search for Nripan Mathews in:

  9. Search for Tze Chien Sum in:


T.C.S., W.C. and N.M. conceived the idea for the manuscript and designed the experiments. W.C. conducted the spectroscopic characterization. S.B. and S.A.V. prepared the samples and performed sample characterization. Q.X. performed the DFT studies. W.C., M.G., S.M., N.M. and T.C.S. analysed the data and wrote the manuscript. All authors discussed the results and commented on the manuscript at all stages. T.C.S. and N.M. led the project.

Competing interests

The authors declare no competing financial interests.

Corresponding authors

Correspondence to Nripan Mathews or Tze Chien Sum.

Supplementary information

PDF files

  1. 1.

    Supplementary Information

    Supplementary Figures, Supplementary Notes and Supplementary References

  2. 2.

    Peer Review File


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Creative Commons BYThis 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 article’s 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