Photophysical properties and fluorescence lifetime imaging of exfoliated near-infrared fluorescent silicate nanosheets

The layered silicates Egyptian Blue (CaCuSi4O10, EB), Han Blue (BaCuSi4O10, HB) and Han Purple (BaCuSi2O6, HP) emit as bulk materials bright and stable fluorescence in the near-infrared (NIR), which is of high interest for (bio)photonics due to minimal scattering, absorption and phototoxicity in this spectral range. So far the optical properties of nanosheets (NS) of these silicates are poorly understood. Here, we exfoliate them into monodisperse nanosheets, report their physicochemical properties and use them for (bio)photonics. The approach uses ball milling followed by tip sonication and centrifugation steps to exfoliate the silicates into NS with lateral size and thickness down to ≈ 16–27 nm and 1–4 nm, respectively. They emit at ≈ 927 nm (EB-NS), 953 nm (HB-NS) and 924 nm (HP-NS), and single NS can be imaged in the NIR. The fluorescence lifetimes decrease from ≈ 30–100 μs (bulk) to 17 μs (EB-NS), 8 μs (HB-NS) and 7 μs (HP-NS), thus enabling lifetime-encoded multicolor imaging both on the microscopic and the macroscopic scale. Finally, remote imaging through tissue phantoms reveals the potential for bioimaging. In summary, we report a procedure to gain monodisperse NIR fluorescent silicate nanosheets, determine their size-dependent photophysical properties and showcase the potential for NIR photonics.


Figure S1. Colloidal stability of planetary ball (PB)/McCrone (MC) milled and centrifuged Egyptian Blue (EB), Han Blue (HB) and Han Purple (HP) a
Picture of EB, HB and HP dispersions in water obtained after milling and centrifugation step #1. In addition to the standard PB technique, MC milling was also tested on EB. b Pictures of the same samples taken 3 days later to check the colloidal stability: while EB supernatants were still stable, HB and especially HP had started to settle down.

Figure S2. Colloidal stability of PB+MC/MC+PB milled and centrifuged EB, HB and HP
Before choosing the planetary ball mill (PB) as best compromise for the milling step of EB, HB and HP, other techniques were tested. These included McCrone milling (MC) and mixed approaches (PB+MC, MC+PB). a Picture of EB, HB and HP dispersions in water obtained after mixed approaches of milling, followed by centrifugation step #1. For this dataset, MC+PB was performed only on EB. b Pictures of the same samples taken 1 day later to check the colloidal stability: while EB supernatants were still stable, HB and especially HP had started to settle down. Figure S3. Size distribution of planetary ball milled (PB) EB, HB and HP after centrifugation step #1 (i.e. prior to tip sonication) Milled and centrifuged samples measured via laser diffraction particle sizer (LDPS). For EB, HB and HP alike, the efficiency of the centrifugation step is shown by the cut-off seen at ≈ 1 μm. All samples display a trimodal distribution, which is less pronounced in the case of HP. N = 1 independent sample per category, n = 3 measurement runs. Figure S4. Effect of tip sonication (TS) on the size distribution of planetary ball milled (PB) EB Laser diffraction particle sizer (LDPS)-based analysis of only milled and milled + tip sonicated EB is presented. The former (milled) sample was centrifuged and its supernatant decanted before measurement, as described in the main manuscript text. The latter, on the other side, underwent a tip sonication step in addition to the milling one, and is then measured directly after exfoliation (as a whole, i.e. without centrifugation). When plotted in the form of histograms (number % vs. particle diameter), the LDPS dataset shows that the trimodal distribution is less pronounced after tip sonication, and its lowest extremes are shifted to lower diameter values, indicating a further size reduction. N = 1 independent sample per category, n = 3 measurement runs. Figure S5. Monitoring of size distribution of Han Purple nanosheets (HP-NS) during tip sonication Laser diffraction particle sizer (LDPS)-based analysis shows the effect of tip sonication on the size distribution of a HP sample over time (1 h -3 h -6 h). When plotted in the form of histograms (number % vs. particle diameter), the LDPS dataset shows that the trimodal distribution gets less and less pronounced in time, and its lowest extremes are shifted to lower diameter values, indicating a further size reduction. N = 1 independent sample per category, n = 3 measurement runs. Figure S6. Comparison of exfoliation efficiency of different milling techniques Histograms (number % vs. particle diameter) obtained from a laser diffraction particle sizer (LDPS) show the obtained size distributions of EB (a), HB (b) and HP (c). Before measurement, all samples were centrifuged and the supernatant decanted. The efficiency of the employed centrifugation step is proven by the very low amount of particles larger than ≈ 1 μm. For EB, while the population's extremes seem to be consistent with all tested techniques, the trimodal distribution is more evident in the PB case. HB displays a much clearer difference between the tested PB and PB+MC; here, the former approach yields smaller particles and, thus, results more optimal. Finally, the HP sample does not show a pronounced difference between the previous two techniques. N = 1 independent sample per category, n = 3 measurement runs. Figure S7. PB milled, tip sonicated and centrifuged EB-NS, HB-NS and HP-NS Self-taken photograph of the fully exfoliated samples, i.e. the final NS batches that underwent complete morphological and photophysical characterization.

Figure S8. Colloidal stability of EB-NS, HB-NS and HP-NS in acidic conditions a
Standoff NIR fluorescence images of NS samples in cuvettes at pH 5. Measurements were taken over 1 h to assess the effect of acidic pH on the stability, i.e. the fluorescence of a central region of interest within the cuvette. Control = addition of H2O (instead of buffer solutions). Scale bar = 1 cm. b Corresponding mean fluorescence intensity, normalized to the first frame. HBand HP-NS performed worse than EB-NS. Nevertheless, all NS displayed a signal decrease not lower than 15%, thus proving to be significantly stable in such conditions. Error bars = standard deviation, N = 2 independent samples. Figure S9. Colloidal stability of EB-NS, HB-NS and HP-NS at neutral pH a Stand-off NIR fluorescence images of NS samples in cuvettes at pH 7. Measurements were taken over 1 h to assess the effect of neutral pH on the stability, i.e. the fluorescence of a central region of interest within the cuvette. Control = addition of H2O (instead of buffer solutions). Scale bar = 1 cm. b Corresponding mean fluorescence intensity, normalized to the first frame. All NS displayed a neglectable signal decrease, thus confirming their stability in such conditions. Error bars = standard deviation, N = 2 independent samples.

Figure S10. Colloidal stability of EB-NS, HB-NS and HP-NS in ionic environment a
Standoff NIR fluorescence images of NS samples in cuvettes in the presence of sodium chloride (NaCl). Measurements were taken over 1 h to assess the effect of ions (9 g/L NaCl, a typical concentration in blood) on the stability, i.e. the fluorescence of a central region of interest within the cuvette. Control = addition of H2O (instead of NaCl solution). Scale bar = 1 cm. b Corresponding mean fluorescence intensity, normalized to the first frame. HB-and HP-NS performed slightly worse than EB-NS. Nevertheless, all NS displayed a signal decrease not lower than 15%, thus proving to be significantly stable in such conditions. Error bars = standard deviation, N = 2 independent samples.

Figure S11. Colloidal stability of EB-NS, HB-NS and HP-NS in buffer solution a
Stand-off NIR fluorescence images of NS samples in cuvettes in the presence of buffer (phosphatebuffered saline, PBS). Measurements were taken over 1 h to assess the effect of PBS (1x) on the stability, i.e. the fluorescence of a central region of interest within the cuvette. Control = addition of H2O (instead of buffer). Scale bar = 1 cm. b Corresponding mean fluorescence intensity, normalized to the first frame. EB-NS performed slightly worse than HBand HP-NS. Nevertheless, all NS displayed a signal decrease not lower than 15%, thus proving to be significantly stable in the presence of PBS. Error bars = standard deviation, N = 2 independent samples. Figure S13. Scanning Electron Microscopy (SEM) images of gold-coated small NS NS with sizes comparable/smaller than the optical resolution limit for optical microscopy (λ ≈ 500 nm, in our case) show a slightly improved contrast after the evaporation of gold on their surfaces. Scale bar = 500 nm. Figure S14. Morphology of nanosheets Scanning electron microscopy (SEM) and scanning transmission electron microscopy (STEM) were used to assess NS morphology. a-c EB-NS, HB-NS and HP-NS spin-coated on a graphite substrate. Scale bar = 5 µm. d-f Magnified images of the same samples to highlight the presence of particles close to/lower than the resolution limit of optical microscopy. In the lower-right window, clear 2D nanosheet structures could be observed in STEM mode. Scale bar = 500 nm.

Figure S15. Scanning Transmission Electron Microscopy (STEM) images of small EB-NS
a-b EB-NS with sizes slightly larger than Abbe's resolution limit (λ ≈ 500 nm) and displaying the lamellar morphologies typically yielded by fragmentation of thicker EB particles. c A thin EB-NS with lateral sizes in the range of the optical resolution limit. d Example of an EB-NS of diameter below 500 nm. Scale bar = 500 nm.   λemi, HP-NS ≈ 923.9 nm (FWHM ≈ 123.7 nm). A slight discrepancy compared to our 2D dataset is likely the result of data correction for the quantum efficiency of the detector and for the spectral irradiance of the xenon lamp, which was here not performed; additionally, spectra normalization and background subtraction might be responsible for further slight shifts, too.           show NIR fluorescence emitted from the NS under visible excitation (white light source with a 700 nm short-pass filter). The emission signal managed to reach the NIR camera through multiple 1 mm-thick chicken phantom tissues, which were positioned on top of the capillary glass. A background image was subtracted in each image. Scale bar = 10 mm. N = 1 sample.

Figure S30. Tissue phantom experiments of HB-NS A concentrated solution of HB-NS (≈ 5 g/L) was sucked into capillary tubes and imaged in a stand-off detection setup. Pictures
show NIR fluorescence emitted from the NS under visible excitation (white light source with a 700 nm short-pass filter). The emission signal managed to reach the NIR camera through multiple 1 mm-thick chicken phantom tissues, which were positioned on top of the capillary glass. A background image was subtracted in each image. Scale bar = 10 mm. N = 1 sample.

Figure S31. Tissue phantom experiments of HP-NS
A concentrated solution of HP-NS (≈ 5 g/L) was sucked into capillary tubes and imaged in a stand-off detection setup. Pictures show NIR fluorescence emitted from the NS under visible excitation (white light source with a 700 nm short-pass filter). The emission signal managed to reach the NIR camera through multiple 1 mm-thick chicken phantom tissues, which were positioned on top of the capillary glass. A background image was subtracted in each image. Scale bar = 10 mm. N = 1 sample.

Figure S32. Cytotoxicity tests with cell line 3T3 Different concentrations of EB-NS, HB-NS
and HP-NS were dispersed in cell medium and incubated with cell lines of type 3T3 for 1 h, 6 h and 24 h. NS-induced cytotoxicity was assessed via a cell proliferation assay whose absorbance at 490 nm is directly proportional to the number of living cells in the culture (expressed in the figure as % cell viability). Only at 0.1 mg/mL (0.05 mg/mL for HB-NS) cell viability decreased slightly. The biologically more relevant lower concentrations showed no effect on viability. Average values calculated from quadruplicates of each sample are plotted. Error bars = standard deviation. Figure S33. Cytotoxicity tests with cell line A549 Different concentrations of EB-NS, HB-NS and HP-NS were dispersed in cell medium and incubated with cell lines of type A549 for 1 h, 6 h and 24 h. NS-induced cytotoxicity was assessed via a cell proliferation assay whose absorbance at 490 nm is directly proportional to the number of living cells in the culture (expressed in the figure as % cell viability). Only at 0.1 mg/mL cell viability decreased slightly. The biologically more relevant lower concentrations showed no effect on viability. Average values calculated from quadruplicates of each sample are plotted. Error bars = standard deviation.

Figure S34. Cytotoxicity tests with cell line MDCK-II Different concentrations of EB-NS,
HB-NS and HP-NS were dispersed in cell medium and incubated with cell lines of type MDCK-II for 1 h, 6 h and 24 h. NS-induced cytotoxicity was assessed via a cell proliferation assay whose absorbance at 490 nm is directly proportional to the number of living cells in the culture (expressed in the figure as % cell viability). Only at 0.1 mg/mL cell viability decreased slightly. The biologically more relevant lower concentrations showed no effect on viability. Average values calculated from quadruplicates of each sample are plotted. Error bars = standard deviation.

Figure S35. Microscopic fluorescence lifetime imaging (FLIM) on a home-built setup
Schematic of the custom-built optical setup for FLIM imaging of NS samples using a frequency domain-based lifetime camera.

Table S1. Comparison of fluorescence lifetime values of EB-NS, HB-NS and HP-NS measured using different experimental techniques: confocal TCSPC, microscopic PCO.FLIM lifetime imaging and oxygen sensor
For TCSPC measurements, the mean value and standard deviation were calculated from at least three separate measurements performed on the same samples (N = 3-4 measurements). For FLIM imaging, the lifetime values of each pixel of the acquired image were histogrammed and fitted with the single Gaussian fit; the mean lifetime and standard deviation were taken from the fit (N = 1 measurement). For the modulated LED dataset, the error bar simply corresponds to the standard deviation of the acquisition (N = 1 measurement). 4.5 ± 0.7 14.14 ± 0.04

Absorption (Reflection) Spectra of Egyptian Blue (EB), Han Blue (HB) and Han Purple (HP)
Bulk and Exfoliated Powders: Bulk powders were analyzed with no prior processing. Aliquots Following the PB milling step, an aliquot was poured into a Nalgene ® centrifuge tube (Thermo Fisher Scientific) and water was added until an overall volume of 150 mL (dilution factor ≈ 3) was reached. Then, a first centrifugation step (Heraeus Multifuge X3R, Thermo Fisher Scientific) was performed: T = 20°C, 800 r.p.m. (150 × g), 9 min 41 s, 5 s acceleration ramp, 5 s deceleration ramp, 5 cycles. These parameters were calculated from the Stokes Equation (corrected for centrifugation [1] ) in order to remove particles of diameter d > 1 μm: Concerning the mixed approaches, instead, the starting material was represented by the settled parts of PB and MC centrifugations (i.e. the sediment, which had been collected and dispersed in demineralized water): according to the order these millings were executed in, the samples were named "PB+MC" or "MC+PB". The former ones were prepared by loading a total volume of 10 mL of sediment slurries into the MC mill for 30 min: after washing steps, a final volume of ≈ 45 mL was collected. For "MC+PB" samples, instead, MC sediment slurries were first centrifuged according to the following settings: T = 20°C, 3000 r.p.m. (2103 × g), 5 min, 5 s acceleration ramp, 5 s deceleration ramp, 1 cycle. Next, the so-obtained volume of sediment (≈ 8 mL) was placed in PB beakers and milled as described above: after washing steps, a final volume of ≈ 45 mL could be collected.
For MC and mixed milling techniques alike, samples were diluted and centrifuged to produce the supernatant slurries, analogously to how described for the (reference) PB approach. For all exfoliation routes, size distribution measurements of "whole" and "centrifuged" samples were carried out by means of LDPS before and after the centrifugation step.

Scanning Electron Microscopy (SEM) and Scanning Electron Transmission Microscopy
(STEM): For SEM imaging (Quattro S SEM, Thermo Fisher Scientific), highly-oriented pyrolytic graphite (HOPG, grade ZYB, Bruker) were used as substrates. HOPGs were plasma-treated (Zepto Diener Electronic GmbH +Co. KG, 1 min of O2 supply, 1 min of plasma process) in order to clean their surfaces and increase their hydrophilicity. Next, a typical NS sample was vortexed and bath sonicated for 10 min. 10 µL of undiluted sample were spin-coated with the same parameters employed for AFM measurements. For each NS sample, the so-prepared HOPG was either imaged at the SEM as it is, or a ≈ 4 nm-thick gold layer was evaporated (Baltec MED-020, Baltec) onto it to decrease surface charging and, thus, increase imaging contrast. The HOPG was then placed into the SEM chamber and imaged in the following conditions: high vacuum mode, voltage = 5.00 kV, spot size = 3.0, working distance = 10.0 mm, previously drop-casted and dried on a #1 glass coverslip. Acquisition settings were the following: laser set power = 500 mW, measured power out of the objective ≈ 180 mW,