Exfoliated near infrared fluorescent CaCuSi4O10 nanosheets with ultra-high photostability and brightness for biological imaging

Imaging of complex (biological) samples in the near infrared (nIR) range of the spectrum is beneficial due to reduced light scattering, absorption, phototoxicity and autofluorescence. However, there are only few near infrared fluorescent materials known and suitable for biomedical applications. Here, we exfoliate the layered pigment CaCuSi4O10 (known as Egyptian Blue, EB) via facile tip sonication into nanosheets (EB-NS) with ultra-high nIR fluorescence stability and brightness. The size of EB-NS can be tailored by tip sonication to diameters < 20 nm and heights down to 1 nm. EB-NS fluoresce at 910 nm and the total fluorescence intensity scales with the number of Cu2+ ions that serve as luminescent centers. Furthermore, EB-NS display no bleaching and ultra-high brightness compared to other nIR fluorophores. The versatility of EB-NS is demonstrated by in vivo single-particle tracking and microrheology measurements in developing Drosophila embryos. Additionally, we show that EB-NS can be uptaken by plants and remotely detected in a low cost stand-off detection setup despite strong plant background fluorescence. In summary, EB-NS are a highly versatile, bright, photostable and biocompatible nIR fluorescent material that has the potential for a wide range of bioimaging applications both in animal and plant systems.


Introduction
Fluorescence imaging provides important insights into the structure, function and dynamics of biological samples 1,2 . Imaging in the near infrared (nIR) spectral range (800-1700 nm) promises higher tissue penetration, higher contrast and lower phototoxicity due to reduced nIR light scattering and absorption [3][4][5] . However, these approaches are limited by the scarcity of nIR fluorescent materials. nIR fluorescent organic dyes such as indocyanine green bleach and are therefore not suitable for long term imaging 6,7 . In contrast, nanomaterials provide beneficial photophysical properties such as ultra-high photostability that would enable tracking in living systems without time constrains. nIR fluorescent nanomaterials include InAs quantum dots, lanthanide doped nanoparticles or semiconducting single-walled carbon nanotubes (SWCNTs) [8][9][10][11][12] . For example, SWCNTs have been used as building blocks for nIR imaging and as fluorescent sensors that detect small signaling molecules, proteins or lipids [13][14][15][16][17][18][19] . They can be chemically tailored and have been used to reveal spatiotemporal release patterns of neurotransmitters from single cells 2,15,[20][21][22] . However, most nIR fluorescent nanomaterials often have low quantum yields, lack biocompatibility or are restricted to certain emission/excitation wavelengths. Therefore, there is a major need for novel nIR fluorescent and biocompatible nanomaterials for sophisticated applications such as long time single particle tracking in organisms or multiscale bioimaging such as stand-off detection in plants 23,24 .
One of the first colored pigments created by mankind is the calcium copper silicate called Egyptian Blue (CaCuSi4O10, EB), which has been synthesized and used as early as 2500 BC in Ancient Egypt 25 . Current ancient works of art decorated with EB have lost none of their vibrant color, a testimony to the remarkable chemical stability of this compound. Interestingly, bulk EB displays nIR fluorescence, which was only recently identified 26,27 and attributed to a 2 B2g-2 B1g electronic transition of the copper ion that ranges from 910 to 930 nm 26,28 . Bulk EB shows a remarkable high quantum yield of 10.5% for a nIR emitter compared to SWCNTs, quantum dots, metal nanoclusters and FDA-approved fluorophores like indocyanine green 5,28 . Recently, µm-sized monolayer sheets of EB were isolated by stirring in hot water for several days 29,30 . However, the remarkable properties of EB have not been explored for developing nIR luminescent nanomaterials for bioimaging applications. The layered structure of EB suggests that exfoliation procedures known from other 2D materials including graphene and transition metal chalcogenides could efficiently exfoliate it [31][32][33][34] . Such 2D materials have been shown to possess novel optoelectronic properties and are a rich playground for physics and chemistry 35 .
Herein, we use a facile tip sonication technique to exfoliate CaCuSi4O10 (EB) nanosheets (EB-NS) that allows one to control the nanomaterial size/thickness and retain the unique nIR fluorescent properties of macroscopic CaCuSi4O10. We report the photophysical properties of EB-NS and how nIR fluorescence scales with nanosheet size. Furthermore, we demonstrate the first use of this material for in vivo nIR microscopy and stand-off detection.

Results and Discussion
The size of a nanomaterial determines its properties and interactions with the environment. For fluorescence imaging in cells or whole organisms fluorophores should be as small as possible to not perturb the system. Exfoliation into 2D sheets is one step but it is also important to decrease the lateral size into the nanoscale. Therefore, we exfoliated EB-NS via tip sonication, which allowed for the controlled decrease in height and diameter with sonication time ( fig. 1a, fig. S1). In a next step, we explored how reducing the size and dimensionality of EB into the nanoscale regime affects its luminescence properties relative to macroscopic EB powder 27 . Fluorescence quantum yields of 1D materials such as SWCNTs have been shown to decrease with size, probably due to exciton diffusion and their collision with SWCNT ends 39 fig. 2c, fig. S4). EB-NS have a similar zeta potential (-22 mV) as spherical silica nanoparticles, which highlights that they can be dispersed and applied in aqueous solutions ( fig. S5) 40 .

Fig. 1: Exfoliation of Egyptian Blue (EB) into Egyptian Blue nanosheets (EB-NS
To estimate the number of luminescent centers in a single EB-NS, we performed single particle fluorescence saturation measurements of EB-NS using scanning confocal microscopy and pulsed laser excitation (20 MHz pulse rate). Only particles with sizes not exceeding 100 nm were studied The slightly lower saturation intensity value that was obtained for the process of diminishing excitation power could be caused by heating of the sample by the excitation light, which affects the fluorescence lifetime in bulk EB 28 . show that there are many luminescent Cu 2+ centers in one EB-NS, but because of the resolution limit it is not possible to assess the actual size. To provide an unambiguous answer, a correlative method that measures both size and fluorescence intensity of the same single nanosheet is required.

Fig. 3: nIR imaging of EB-NS. a,b Scanning electron microscopy (SEM) images of larger EB particles. c nIR image of a larger EB-NS. d,e nIR images of EB-NS at increasing magnification
show that single EB-NS can be resolved via nIR fluorescence microscopy down to the resolution limit of light microscopy. In e, EB-NS with a diameter below the resolution limit (≈ 500 nm) are shown. Excitation wavelength 561 nm.
To understand if and how nanomaterial fluorescence intensity is affected by size, we performed single particle tracking of EB-NS in a viscous glycerol solution. This approach allowed us to simultaneously quantify fluorescence intensity and Brownian motion as a measure of size of the same EB-NS. We used the maximum intensity during the whole trace as a measure of fluorescence intensity to account for out of focus movement or rotations. A size equivalent can be where η is the dynamic viscosity of the solvent, T the temperature and kB the Boltzmann constant.
As EB-NS are rather anisotropic with a large aspect ratio ( fig. 1, fig. 3, fig. S7), we assumed that the diffusion is dominated by the diameter and not the height similar to a spherical particle.
Approximations that correct the Stokes-Einstein equation for anisotropy have been used for example to analyze carbon nanotube diffusion and length, but there is no analytical solution available for 2D nanosheets 44 .  The dynamics of the nuclei and their associated centrosomes is determined by the cortical link and internuclear interactions mediated by microtubules 46 . It has been hypothesized that the viscoelastic properties and the molecular motors play a key role in this process 45 . To elucidate the micromechanical properties of the cellular matrix, we introduced EB-NS into Drosophila embryos and tracked single EB-NS via nIR fluorescence microscopy. Imaging in Drosophila is challenging due to autofluorescence in the visible region and its high sensitivity to phototoxicity, therefore nIR imaging would be very beneficial.  45 . These differences can be attributed to the size of EB-NS, which is much smaller than the typical micrometer-sized probes used for microrheology. EB-NS are likely to probe the mesh-size of the embryo's dense cytosol on the nanoscale without being easily trapped in between cellular filaments and other subcellular barriers. Therefore, EB-NS probe mechanical properties on a size scale that is not accessible by typical micrometer-sized beads used for microrheology 45 . Active processes and interactions inside the embryo (including movement of the nuclei, contractions and even flow) increase the movement of the nanoprobes beyond diffusion, which can be seen from the increasing slope in the MSD plots ( fig. 5c,d) 47  In the context of biological applications, toxicity is a major concern and for example a potential drawback of quantum dots that contain toxic elements and 2D materials in general 49,50 . To evaluate cytotoxicity, viability assays with EB-NS exposure to different cell lines (A549, 3T3 and MDCK-II) were performed. We observed no significant effects on the viability of these cell lines,  . 6b, bottom), EB-NS are significantly 2x brighter than ICG and 10x brighter than SWCNTs (fig. 6b, fig. S9). Interestingly, stand-off detection of the nIR emission of EB-NS is also possible without LED excitation: room or sun light alone is sufficient to generate detectable nIR fluorescence from EB-NS ( fig. 6e).

Fig. 6: Stand-off detection of EB-NS fluorescence in plants through low cost and widely available imaging devices. a Picture of a plant (Arabidopsis thaliana) placed in a low cost stand-off imaging system, which consisted of a LED, nIR filters and a Si-CMOS camera. b Visible (top) and nIR fluorescence (bottom) images of EB-NS (≈ 0.1 mg/mL) compared to other nIR nanomaterials and fluorophores at similar concentration (single-walled carbon nanotubes (SWCNTs), indocyanine green (ICG)). Water is used as negative control. c Visible (left) and nIR (right) images of an Arabidopsis plant, which was infused with EB-NS (frame 1), SWCNTs (frame 2) and buffer only (frame 3). d The nIR fluorescence spectrum of the leaf confirms the presence of EB-NS and its strong emission compared to the leaf background. Both results demonstrate the unprecedented brightness of EB-NS compared to state-of-the-art nIR nanomaterials (SWCNT)
and that this platform can be applied for stand-off detection using a low cost optical setup. e The

EB-NS emission can even be detected without LED excitation under room light conditions.
Buffered EB-NS were infused into the leaves of Arabidopsis thaliana, a well-established plant model system ( fig. 6c,d) 51 , using a standard method for nanomaterial delivery into plant leaves in vivo 51 . The same approach was used for the infiltration of SWCNTs in separate leaves in the same plant, which served as a visual comparison with EB-NS. As it can be observed from the nIR image in fig. 6c, plant leaves are characterized by a background autofluorescence in this wavelength region that hinders the visualization of SWCNTs fluorescence (frame 2) through a Si-based CMOS camera, whereas EB-NS can be detected with a strong fluorescence signal (frame 1). Similarly, fluorescence spectra of these leaves ( fig. 6d) show the strong EB-NS emission compared to the (autofluorescence) background. These stand-off detection experiments highlight that EB-NS can be delivered into living organisms with high background fluorescence such as plants and can be detected using low cost and widely available stand-off imaging devices.
Our results show that EB-NS have extremely useful properties for nIR bioimaging applications including ultra-high photostability, extreme brightness for nIR fluorophores and biocompatibility.
The remarkable properties of EB-NS are retained even when the dimensions of the nanosheets are reduced. Our data indicate that most if not all copper ions serve as luminescent centers that emit nIR light. Therefore, size reduction does not decrease the nIR fluorescence quantum yield.
It only affects the absorption cross section by changing the number of light absorbing Cu 2+ ions.
In contrast, the quantum yield of nIR fluorescent SWCNTs increases with length. As a consequence, shorter SWCNTs (< 100 nm) are much less bright than longer ones 39 . Therefore, SWCNTs cannot be shortened without disadvantages. Although it has been reported that for bioimaging applications the nIR-II (1000-1700 nm) windows further improves tissue penetration 5 , a major advantage of the EB-NS fluorescence in the nIR-I region (800-1000 nm) is the wide availability of cameras 53,54 . Instead of expensive (>40k €), liquid nitrogen-cooled low resolution InGaAs-based cameras, low cost (1k €) high-resolution Si-based cameras can be used for detection. In the past, (6,4)-SWCNTs with smaller diameter that emit in the nIR-I region at 870 nm have been isolated exactly for this purpose 55 . Even though the quantum yields of Si-based cameras rapidly decrease in the nIR (around 5% > 900 nm), they are still able to image even single EB-NS using a normal microscope or a low cost stand-off detection device. Furthermore, there exist (bulk) pigment homologues (Ba, Sr), which have a red-shifted emission spectrum up to 1000 nm. Nanosheets of those materials could therefore extend the spectral range further into the nIR. The capability of imaging EB-NS even with only room light excitation using Si-based cameras is particularly relevant for commercial applications beyond the well-equipped research laboratories or biomedical research, e.g. the engineering of smart plants for high-throughput phenotyping and precision agriculture. Another major advantage of EB-NS is their low toxicity, proven by the performed viability assays on cells ( fig. S8). With our study we add EB-NS to the library of 2D materials in general and especially exfoliated silicates and clays 56,57 . In the future, EB-NS and their physicochemical properties could be further investigated to find non-expected properties as in other 2D materials such as transition metal dichalcogenide nanosheets 58 .
Additionally, the surface chemistry could be tailored with biomolecules similar to silica nanoparticles and silicate coated core-shell nanoparticles 59,60 .

Conclusion
In summary, we developed a method to exfoliate the calcium copper silicate Egyptian Blue into nanosheets (EB-NS) and demonstrated that their nIR optical properties are retained at the nanoscale with copper ions as luminescent centers. We add a novel 2D material to the class of nIR fluorophores with ultra-high photostability, brightness and biocompatibility. EB-NS have a large potential for demanding biological applications such as in vivo imaging in complex biological matrices, including animal and plant model systems.

Exfoliation of Egyptian Blue (EB) into Egyptian Blue Nanosheets (EB-NS)
EB powder was purchased from Kremer Pigmente GmbH & Co. KG. EB-NS and prepared as follows. EB powder was ground with a ceramic pestle and mortar to mechanically reduce the crystal size until the powder was visibly brighter, a consequence of the dichroism of EB 25

Fluorescence Saturation Measurements
The main components of the employed setup were a laser source (Supercontinuum laser SC400-4-20, Fianium), a photodetector (single photon avalanche diode PDM series, MPD) and a 60x objective lens (Apo N, 60x/1.49 NA oil immersion, Olympus). 10 µL of the supernatant of a 6 hsonicated EB-NS sample (1:100 diluted in isopropanol) were spin-coated on a glass cover slide.
Spin-coating parameters were the same ones chosen for AFM measurements. Despite the polydispersity of EB-NS, scanning them using a confocal microscope through the diffraction limited focal spot of 1.49 NA objective allowed us to select only the smallest particles with sizes estimated to be not exceeding ≈ 100 nm. The spot size of larger EB-NS exceeded the dimensions of a diffraction limited focal spot, thus allowing us to distinguish them from smaller particles. After the dimmest particles could be detected on the sample surface, we acquired its fluorescence images at different excitation powers. The dependence between the brightness of the EB-NS and the excitation light power (measured directly before the objective lens) was studied. The excitation wavelength was 640 nm.

Fluorescence Polarization Measurements
The Here, a temperature of 293 K and a dynamic viscosity of glycerol of 1412 centipoises were assumed 62 . The Stokes-Einstein equation is only strictly valid for spherical particles. Due to the high anisotropy of EB-NS, we assumed that Brownian motion is dominated by the diameter of the nanosheets and not the much smaller height.

Scanning Electron Microscopy (SEM)
A 6 h-tip sonicated sample was observed under a LEO SUPRA 35 microscope (Zeiss) with an Inlens detector at 20 kV (secondary electrons). 10 μL of EB-NS suspension were drop-casted on a Si wafer. Both gold sputtering and evaporation (≈ 2 nm of gold layer) were tested on different samples, nevertheless the best imaging conditions were met when no gold deposition step was performed at all. Interestingly, EB was conductive enough to be seen at SEM without gold deposition.

In vivo Microrheology of Drosophila Embryos
Drosophila embryos expressing Histone2Av-GFP 63 with an age of 0-1 hour were collected and dechorionated with hypochlorite for 120 s, washed with water thoroughly, aligned on a piece of agar, transferred to a cover slip coated with glue and covered with halocarbon oil (Voltalef 10S, Lehmann & Voss) after a slight desiccation. An aliquote of suspension in water of a 6 h-tip sonicated EB-NS sample (filtered with a 0.20 μm syringe filter) was injected using Microinjector FemtoJet® (Eppendorf) on an inverted microscope: the injection volume is calculated according to literature 64 . After around 30 min of incubation, the sample could be moved to the previously mentioned nIR setup for co-localization experiments. EB-NS were excited with the 561 nm laser (300 mW), while a fluorescence lamp (X-Cite® 120Q, Excelitas Technologies) was used for GFP excitation. Both channels were observed through a 100x objective and recorded with a Zyla 5.5 sCMOS camera. An exposure time of 0.1 s was chosen and images on both channels were taken for 60 s (10 fps). A 2×2 pixel binning was performed to lower the size of acquired data and thus facilitate the following particle tracking analysis. As for the EB-NS tracking in glycerol, ImageJ with the Mosaic Particle Tracker was employed to analyze the acquired frames. Only the EB data was evaluated, since the nuclei did not display any significant motion during the acquisition. The following parameters were employed for analysis: radius = 5, cutoff = 0, per/abs = 0.1, link range = 100, displacement = 2, dynamics = Brownian. The diffusion coefficient was then evaluated as previously described. The radius of each EB-NS was estimated from 10 images of the movie. Finally, the Stokes-Einstein equation was used to estimate a corresponding dynamic viscosity.

ASSOCIATED CONTENT
Supporting Information. Supporting Information is available for this article.

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

CODE AVAILABILITY
The codes that support the findings of this study are available from the corresponding author upon reasonable request.

AUTHOR INFORMATION
Corresponding Author * Correspondence should be addressed to skruss@gwdg.de (Sebastian Kruss)

ACKNOWLEDGMENT
We thank the Volkswagen foundation and the life@nano cluster for funding (S.K.).
We thank Dr. Burkhard Schmidt for recording reflectance spectra, Elena Polo for initial exfoliation experiments, Nelli Teske and Jeremias Sibold for help with the SEM and Angela Rübeling for technical assistance. Furthermore, we thank Wentao Peng/Prof. Dr. Vana for assistance with zeta potential measurements. We thank the Steinem and Janshoff labs for discussions and support.
We are grateful to Dr. Ellen Hornung for providing us with Arabidopsis plants and related information. This work was funded by the National Science Foundation under Grant No. 1817363 to J.P.G.