Multidimensional Widefield Infrared-Encoded Spontaneous Emission Microscopy: Distinguishing Chromophores by Ultrashort Infrared Pulses

Photoluminescence (PL) imaging has broad applications in visualizing biological activities, detecting chemical species, and characterizing materials. However, the chemical information encoded in the PL images is often limited by the overlapping emission spectra of chromophores. Here, we report a PL microscopy based on the nonlinear interactions between mid-infrared and visible excitations on matters, which we termed MultiDimensional Widefield Infrared-encoded Spontaneous Emission (MD-WISE) microscopy. MD-WISE microscopy can distinguish chromophores that possess nearly identical emission spectra via conditions in a multidimensional space formed by three independent variables: the temporal delay between the infrared and the visible pulses (t), the wavelength of visible pulses (λvis), and the frequencies of the infrared pulses (ωIR). This method is enabled by two mechanisms: (1) modulating the optical absorption cross sections of molecular dyes by exciting specific vibrational functional groups and (2) reducing the PL quantum yield of semiconductor nanocrystals, which was achieved through strong field ionization of excitons. Importantly, MD-WISE microscopy operates under widefield imaging conditions with a field of view of tens of microns, other than the confocal configuration adopted by most nonlinear optical microscopies, which require focusing the optical beams tightly. By demonstrating the capacity of registering multidimensional information into PL images, MD-WISE microscopy has the potential of expanding the number of species and processes that can be simultaneously tracked in high-speed widefield imaging applications.


Figures S1 to S21
Tables S1 to S7 SI References

II. Additional data on rhodamine 6G (R6G).
(a).Solution phase measurements.Fig. S1 shows the infrared-pump-visible-probe transient absorption (TA) results of R6G dissolved in chloroform and in dimethyl sulfoxide-d6 (DMSO-d6) solutions.The solutions are sandwiched between two infrared-transparent CaF2 windows with 56 microns in thickness and optical density of ~0.2 at 540 nm.Clear features of excited-state absorption and ground-state bleaching are observed.The ultrafast kinetics of the visible-region absorbance change induced by the 1600 cm -1 infrared (IR) pump are sensitive to the solvent environment.In deuterated DMSO, the kinetics only exhibit a decaying pattern, whereas the decay kinetics in chloroform exhibit a more complex pattern.The difference in decay patterns might be attributed to change of Franck-Condon factors and intramolecular vibrational energy redistribution (IVR) of R6G in different solvents.
Fig. S1.The TA results of R6G in DMSO-d6 (A) and chloroform (B).The IR pump frequency is 1600 ± 30 cm -1 , and the probe is a broadband whitelight pulse.The upper panels show TA spectra at various time delays, and the lower panels show kinetics at several wavelengths that correspond the wavelength positions marked by dashed lines in upper panels.The R6G solution in chloroform is investigated by measuring its fluorescence intensity following the excitation of a 1600 cm -1 IR pulse and then a narrowband visible pulse (520 or 550 nm).The instrument and solution used are the same as the TA experiments above, but the broadband whitelight probe pulse in TA experiments is replaced with a narrowband visible pulse, and the detector receives fluorescence signals rather than the whitelight probe beam.The relative change of fluorescence intensity induced by the IR pulse is measured as 100% × (  −   )/  , where Ion or Ioff are the intensity when the IR beam is unblocked or blocked by a chopper.As shown in Fig. S2, the kinetics of relative fluorescence intensity change show similar patterns to the TA kinetics in Fig. S1B.The IR-induced fluorescence intensity change is attributed to the IR-induced change of electronic absorbance in the visible region.As discussed in the main text, at 550 nm, more absorption of visible photons leads to more fluorescence photons.In contrast, at 520 nm, ground state bleaching leads to reduction of fluorescence signals.
The results support the mechanism 1 of MD-WISE imaging of silica beads in the main text, but the change level of fluorescence intensity is small (only a few percentages).This is attributed to the relatively large thickness of the solutions (56 microns) generating defocused background fluorescence signals that is not modulated by the IR pulse.Following the excitation of an IR pulse and then a narrow band visible pulse, the fluorescence signals emitted are collected in the wavelength range of 585 ± 18 nm using a bandpass filter.The IR pump frequency is 1600 ± 30 cm -1 , and the visible excitation range is 550 ± 5 nm (A) or 520 ± 5 nm (B).

(b). MD-WISE images of stained silica beads.
In Fig. S3A, MD-WISE images that correspond to the 1600 cm -1 kinetic curve in main text Fig. 2E are displayed at a series of delay times.As described in the main text, the kinetic curves are measured by averaging the relative intensity change among all the detector pixels in a 2.5 µm by 2.5 µm box that centers around the microbead.At later delay times, vibrational relaxation and energy redistribution reduces the counts per pixel and the quality of the difference images.The highest quality of difference images is obtained using short delays such as 1 ps.In Fig. S3B, MD-WISE images that correspond to the 1720 cm -1 kinetic curve in main text Fig. 2E  displayed at a series of delay times.Besides the IR frequencies listed in the main text, we also performed MD-WISE imaging using 1650 ± 30 cm -1 IR excitation.The IR absorption peak at 1650 cm -1 is assigned to another stretch mode of the xanthene ring in R6G.As shown in Fig. S4, the ultrafast decay kinetics are faster than the kinetics measured at 1600 cm -1 and 1720 cm 1 .Since the R6G dye molecules have overlapping IR spectral features with the silica beads at ~1600 cm -1 , we performed MD-WISE experiments on an additional set of substrate materials to investigate whether the IR absorption and the properties of substrate materials have large impacts on the mechanism of ultrafast IR-induced emission intensity change.
One substrate tested is the surface-modified silica microbeads.The surface of silica microbeads contains silanol groups which can be used to anchor organic molecules with distinct vibrational features.Using an established protocol, 3-(azidopropyl)triethoxysilane (Gelest Inc., catalog no.SIA0777.0,structure embedded in Fig. S5A) is grafted onto the surface of silica microbeads (1).In Fig. S5A, the azide-modified silica microbeads (Silica-N3) show a distinct azide stretch mode at ~2100 cm -1 in addition to the absorption features of silica at 1600 cm -1 and 1800-2000 cm -1 .
Another substrate tested is PMMA microbeads (EpruiBiotech, Shanghai, catalog no.3-001-3).As shown in Fig. S5B, PMMA lacks vibrational features at ~1600 cm -1 , and thus has no overlap with the ring stretch mode of R6G at 1600 cm -1 .The ultrafast kinetics of IR-induced fluorescence intensity change of R6G on three types of substrates are shown in Fig. S6A.The IR frequency is tuned to 1600 ± 30 cm -1 for the xanthene ring stretch mode.The results show that, no matter the substrate IR absorption feature overlaps with the R6G feature or not, the kinetics are nearly identical for the three types of substrates.Thus, the IR-induced fluorescence intensity change measured for the silica beads shall be attributed to the excitation of molecular vibrational modes of the adsorbed dyes, rather than the IR absorption of silica.Furthermore, we test whether exciting the azide stretch mode of Silica-N3 can transfer energy to R6G modes and cause modulation on fluorescence intensity.As shown in Fig. S6B, there is no modulation on fluorescence intensity using IR frequency at 2100 cm -1 .This result agrees well with the flat kinetic traces in main text Fig. 2E using IR frequencies in the range of 1800-2000 cm -1 .In Fig. S7, MD-WISE images of QD-stained silica beads that correspond to the kinetic curve in main text Fig. 3D are displayed at a series of delay times.For QDs, the IR pulse only has effect on the photoluminescence of QDs when it arrives later than the visible pulse (negative delay times).

(b). Fluorescein
In Fig. S8, MD-WISE images of fluorescein-stained silica beads that correspond to the kinetic curve in main text Fig. 4D are displayed at a series of delay times.Only IR frequency of 1600 cm -1 that resonantly excites the molecule show effects on fluorescence emission intensity, while the difference images acquired using IR frequency of 2040 cm -1 are blank.For 1600 cm -1 , similar to the case of R6G, at later delay times, vibrational relaxation and energy redistribution reduces the counts per pixel and the quality of the difference images.

V. Quantum chemical vibrational analysis of R6G dye (a). Overview and methods
Gaussian 09, Revision D (2) was used to perform quantum chemical calculations for identifying and analyzing the vibrational modes, vibrational frequencies, and coupling between functional groups in the molecule.The density functional theory (3) calculations were carried out using RB3LYP/6-311++G(d,p) basis sets (4).The structure of R6G molecular cation was optimized in vacuo using a crystal structure as the initial guess (5,6).Then vibrational frequency analysis was performed.The optimized results of R6G cation in vacuum was then used as the initial structure for the vibrational frequency analysis of R6G cations with polarizable continuum models in three different implicit solvent environments: water, methanol, and acetone.The optimized structure of R6G cation in vacuum and the numbering of atoms are displayed as an example in Fig. S13.The atomic coordinates of R6G cations in different environments are listed in Table S1-S2.The frequency and intensity of all the modes above 1000 cm -1 are listed in Table S3.

(b). Calculated IR absorption spectra
The calculated vibrational IR spectra of R6G cations in different environments are shown in Fig. S14.The spectra are similar for R6G cations in water, methanol and acetone, which are redshifted from the spectrum in vacuum.The calculated spectra largely match the FTIR pattern of R6G shown in main text Fig. 2A, though the exact center positions of the peaks differ slightly from the experimental results.Three calculated xanthene ring modes (number 149, 150, 153 in Table S3) related to MD-WISE imaging are marked by the hexagon signs in Fig. S14.The stretch mode (number 155 in Table S3) of the ester group is marked by the triangle signs.Each of these four modes is analyzed with details below using water solvent as the example, since different solvents only yield negligible results.Vibrational modes with negligible IR absorption intensities or modes outside of the frequency ranges used in MD-WISE imaging are not subject to further analysis here.

(d). Analysis of mode 153 (xanthene ring mode)
The calculated vibrational mode 153 in water at ~1650 cm -1 is local to the xanthene ring.The atoms involved and the displacements of each atom are shown in Fig. S16 and Table S6.S6.Atomic displacements of each atom involved in mode 153 in the unit of angstroms.The XYZ vectors are displayed in Fig. S13.

(e). Analysis of mode 155 (ester group mode)
The calculated vibrational mode 155 in water at ~1720 cm -1 is largely a local mode of the ester group.However, calculation results also reveal coupling to the atoms in the xanthene ring.This coupling could be the origin of IR-induced fluorescence intensity change of R6G when the IR laser frequency is tuned to 1720 cm -1 as discussed in the main text.The atoms involved and the displacements of each atom are shown in Fig. S17 and Table S7.

VI. Determination of scale bars, field of view, and spatial resolution of the MD-WISE images (a). Determination of scale bars
The 3-micron silica spheres stained with fluorescent dyes (806765, Sigma Aldrich, see Section I. Materials and staining methods) were used as the calibration target for the MD-WISE microscope.The scale bars in all the MD-WISE fluorescent images were calculated based on the diameter of the images of the silica beads acquired under the same imaging conditions.The absolute size of the silica beads was in turn calibrated with a standard reticle (Thorlabs, R1L3S6PR) using an optical microscope equipped with a x100 objective under whitelight transillumination configuration.As shown below in Fig. S18, the diameter of the silica beads is 3 microns.For MD-WISE imaging experiments, the field of view is adjustable depending on the size of the objects that need to be imaged.For individual silica beads, we typically chose the field of view of 10~20 microns by setting the beam size of the IR and visible pulses to a value in this range.Such beam size is more than sufficient to image the 3-micron beads.To capture the fluorescence image of the entire field, we imaged a sample with crowded beads.As shown below in Fig. S19, the image of beads reports a field of view of ~15 microns.The variations of image intensity could be due to the inhomogeneous loading of dyes in the beads.For images of single cells, they were acquired with larger field of view, and the large size of the cells themselves (20-30 microns) nearly filled the entire field of view.First, we determined whether MD-WISE has a resolution as good as regular photoluminescence (PL) images in our setup.As shown by the linecuts of MD-WISE and PL images in Fig. S20, they lay on top of each other, with identical widths for both the 2-micron (left) and 3-micron (right) silica microspheres, respectively.This suggested that MD-WISE has as good resolution as regular PL imaging using either quantum dots or R6G dye as the chromophore.This makes sense, because the resolution of MD-WISE imaging is defined by the optical objective we used to resolve PL or fluorescence signals.We then determined the resolution of our setup by analyzing the point spread function in regular PL images of individual perovskite nanocrystals (size 50 nm, composition MAPbBr3) prepared according a literature method 7 .The nanocrystal serves as a bright PL point source of which the physical size is much smaller than the resolution limit.The FWHM of the gaussian fitting of the point spread function as shown in Fig. S21 yields a

Fig. S2 .
Fig. S2.The transient change of fluorescence intensity of R6G in chloroform.Following the excitation of an IR pulse and then a narrow band visible pulse, the fluorescence signals emitted are collected in the wavelength range of 585 ± 18 nm using a bandpass filter.The IR pump frequency is 1600 ± 30 cm -1 , and the visible excitation range is 550 ± 5 nm (A) or 520 ± 5 nm (B).

Fig. S3 .
Fig. S3.(A) MD-WISE images obtained using 1600 ± 30 cm -1 IR pulse and 550 ± 5 nm visible pulse, and (B) MD-WISE images obtained using 1720 ± 30 cm -1 IR pulse and 550 ± 5 nm visible pulse.The fluorescence signals are collected in the range of 585 ± 18 nm.Each series in (A) or (B) includes an IR-off image and difference images (IRon -IRoff) at different delay times.Scale bars are 1 micron in size.The color bars represent CCD counts per pixel.

Fig. S4 .
Fig. S4.(A) MD-WISE images obtained using 1650 ± 30 cm -1 IR pulse and 550 ± 5 nm visible pulse.The fluorescence signals are collected in the range of 585 ± 18 nm.The series include an IR-off image and difference images (IR on -IR off ) at different delay times.Scale bars are 1 micron in size.The color bars represent CCD counts per pixel.(B) Ultrafast kinetics of fluorescence intensity change at IR excitation of 1650 cm -1 .The relative change of counts per pixel is measured using the pixels within the 2.5-micron square box marked by dashed lines in (A).
Kinetics of IR-induced change of fluorescence intensity measured using other substrate materials: azide-functionalized silica and polymethyl methacrylate (PMMA).

Fig. S5 .
Fig. S5.Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) results of azide-modified silica microbeads (A) and PMMA microbeads (B).The azide stretch mode is marked by the ▲ symbol in panel A.

Fig. S6 .
Fig. S6.(A) Normalized ultrafast kinetics of fluorescence intensity change measured with IR excitation of 1600 ± 30 cm -1 and visible excitation of 550 ± 5 nm.The results show the three types of substrates stained with R6G have nearly identical kinetics.(B) Ultrafast kinetics of fluorescence intensity change of R6G in azide-modified silica microbeads measured with IR excitation of 2100 ± 30 cm -1 and visible excitation of 550 ± 5 nm.No IR-induced emission intensity change is observed.The fluorescence signals are collected in the range of 585 ± 18 nm.

Fig. S7 .
Fig. S7.MD-WISE images of QD stained beads obtained using 2100 ± 30 cm -1 IR pulse and 550 ± 5 nm visible pulse.The photoluminescence signals are collected in the range of 585 ± 18 nm.The series include an IR-off image and difference images (IRon -IRoff) at different delay times.Scale bars are 1 micron in size.The color bars represent CCD counts per pixel.

Fig. S9 .FITC
Fig. S9.(A) MD-WISE images of FITC-stained beads obtained using 1600 ± 30 cm -1 IR pulse and 520 ± 5 nm visible pulse, and (B) MD-WISE images of FITC-stained beads obtained using 2040 ± 30 cm -1 IR pulse and 520 ± 5 nm visible pulse.The fluorescence signals are within the range of 585 ± 18 nm.Each series in (A) or (B) include an IR-off image and difference images (IRon -IRoff) at different delay times.Scale bars are 1 micron in size.The color bars represent CCD counts per pixel.

Fig. S10 .
Fig. S10.Three sets of MD-WISE images of PI-stained cells obtained using 2100 ± 30 cm -1 IR pulse and 550 ± 5 nm visible pulse.The fluorescence signals are collected in the range of 585 ± 18 nm.The series include IR-off images (left column) and associated difference images (IRon -IRoff) at -10 ps (middle column) and 1 ps (right column).Scale bars are 5 microns in size.The color bars represent CCD counts per pixel.

Fig. S11 .
Fig. S11.Three sets of MD-WISE images of QD-stained cells obtained using 2100 ± 30 cm -1 IR pulse and 550 ± 5 nm visible pulse.The photoluminescence signals are collected in the range of 585 ± 18 nm.The series include IRoff images (left column) and associated difference images (IRon -IRoff) at -10 ps (middle column) and 1 ps (right column).Scale bars are 5 microns in size.The color bars represent CCD counts per pixel.

Fig
Fig. S13.(A) Optimized structure of R6G cation in vacuum, viewed from two different angles.(B) Numbering of atoms in the R6G cation.

Fig. S14 .
Fig. S14.IR spectra of R6G cations in various environments.(A) full IR spectrum in vacuum,(B-E) IR spectra in the range between 1475-1875 cm -1 for vacuum (B), water (C), methanol (D), and acetone (E).△ marks the peak of ester and ⬡ marks the peaks of xanthene rings, which largely match the peaks observed in FTIR experiments.

Fig. S18 .
Fig. S18.The reticle has 250 line pairs (one light line and one dark line) per millimeter.Each light or dark line has a width of 2 microns.We can determine the diameter of the silica beads is 3 microns.

Fig. S19 .
Fig. S19.Fluorescence image of 3-micron silica beads showing an entire field of view of ~15 microns.

Fig. S20 .
Fig. S20.Linecuts of MD-WISE (blue) and plain PL images (orange) acquired without IR modulation.The linecuts were performed across the images of individual silica spheres stained with either quantum dots or R6G dyes as shown in the main texts.

Fig. S21 .
Fig. S21.Gaussian fitting of the point spread function measured using a bright nanocrystal.The fitting equation and results are embedded in the upright corner.The FWHM of the gaussian is 2.355 * (c1/1.414)= 1.6 microns.microns

Table S1 .
Atomic coordinates of R6G cation in vacuum and water.

Table S2 .
Atomic coordinates of R6G cation in methanol and acetone.

Table S3 .
Calculated vibrational frequencies and infrared absorption intensities of R6G vibrational modes.Vibrational modes related to MD-WISE experiments are highlighted by yellow.

Table S4 .
Atomic displacements of each atom involved in mode 149 in the unit of angstroms.The XYZ vectors are displayed in Fig.S13.

Table S5 .
Atomic displacements of each atom involved in mode 150 in the unit of angstroms.The XYZ vectors are displayed in Fig.S13.

Table S7 .
Atomic displacements of each atom involved in mode 155 in the unit of angstroms.The XYZ vectors are displayed in Fig.S13.