Here, we present a protocol to obtain luminescent hyperspectral imaging data and to analyze optical anisotropy features of lanthanide-based single crystals using a Hyperspectral Imaging System.
In this work, we describe a protocol for a novel application of hyperspectral imaging (HSI) in the analysis of luminescent lanthanide (Ln3+)-based molecular single crystals. As representative example, we chose a single crystal of the heterodinuclear Ln-based complex [TbEu(bpm)(tfaa)6] (bpm=2,2’-bipyrimidine, tfaa– =1,1,1-trifluoroacetylacetonate) exhibiting bright visible emission under UV excitation. HSI is an emerging technique that combines 2-dimensional spatial imaging of a luminescent structure with spectral information from each pixel of the obtained image. Specifically, HSI on single crystals of the [Tb-Eu] complex provided local spectral information unveiling variation of the luminescence intensity at different points along the studied crystals. These changes were attributed to the optical anisotropy present in the crystal, which results from the different molecular packing of Ln3+ ions in each one of the directions of the crystal structure. The HSI herein described is an example of the suitability of such technique for spectro-spatial investigations of molecular materials. Yet, importantly, this protocol can be easily extended for other types of luminescent materials (such as micron-sized molecular crystals, inorganic microparticles, nanoparticles in biological tissues, or labelled cells, among others), opening many possibilities for deeper investigation of structure-property relationships. Ultimately, such investigations will provide knowledge to be leveraged into the engineering of advanced materials for a wide range of applications from bioimaging to technological applications, such as waveguides or optoelectronic devices.
Hyperspectral Imaging (HSI) is a technique that generates a spatial map where each x-y coordinate contains a spectral information that could be based on any kind of spectroscopy, namely photoluminescence, absorption and scattering spectroscopies1,2,3. As a result, a 3-dimensional set of data (also called “hyperspectral cube”) is obtained, where the x-y coordinates are the spatial axes and the z coordinate is the spectral information from the analyzed sample. Therefore, the hyperspectral cube contains both spatial and spectral information, providing a more detailed spectroscopic investigation of the sample than traditional spectroscopy. While HSI has been known for years in the field of remote sensing (e.g., geology, food industries4), it recently emerged as an innovative technique for the characterization of nanomaterials2,5 or probes for biomedical applications3,6,7,8. Generally speaking, it is not limited to the UV/visible/near-infrared (NIR) domain, but can also be extended using other radiation sources, such as X-rays – for instance in order to characterize elemental distribution in different materials9 – or Terahertz radiation, where HSI was used to perform thermal sensing in biological tissues8. Further, photoluminescence mapping has been combined with Raman mapping to probe the optical properties of monolayer MoS210. Yet, amongst the reported applications of optical HSI, there are still only a few examples on HSI of lanthanide-based materials11,12,13,14,15,16,17. For instance, we can cite: detection of cancer in tissues6, analysis of the light penetration depth in biological tissues7, multiplexed biological imaging3, analysis of multicomponent energy transfer in hybrid systems11, and investigation of aggregation-induced changes in spectroscopic properties of upconverting nanoparticles12. Clearly, the attractiveness of HSI arises from its suitability for generating knowledge about environment-specific luminescence, providing simultaneous spatial and spectral information about the probe.
Taking advantage of this powerful technique we herein describe a protocol to investigate the optical anisotropy of the heterodinuclear Tb3+-Eu3+ single crystal [TbEu(bpm)(tfaa)6] (Figure 1a)13. The optical anisotropy observed resulted from the different molecular packing of the Ln3+ ions in the different crystallographic directions (Figure 1b), resulting in some crystal faces showing brighter, others showing dimmer photoluminescence. It was suggested that the increased luminescence intensity at specific faces of the crystal was correlated with more efficient energy transfer along those crystallographic directions where the Ln3+···Ln3+ ion distances were the shortest13.
Motivated by these results, we propose the establishment of a detailed methodology to analyze optical anisotropy through HSI, opening the path for better understanding of ion-ion energy transfer processes and tunable luminescent properties stemming from specific molecular arrangement18,19. These structure-properties relationships have been recognized as important aspects for innovative optical materials design including, but not limited to waveguide systems and opto-magnetic storage devices at nano and microscale – addressing the demand for more efficient and miniaturized optic systems20.
CAUTION: It is recommended to use safety goggles specific for the excitation wavelength being used at all times when operating the imager.
1. Configuration of the hyperspectral microscope
NOTE: An overview of the hyperspectral imaging system is given in Figure 2a, with the main components of the imager being described. The imaging system can be used for the detection of the visible or the near-infrared (NIR) emission from a sample. Depending on which detection is desired (visible or NIR), the light goes through two different light paths (Figure 2e). A combination of different beam turning cubes and dichroic filter cubes (optical cubes) must be positioned at specific positions in the instrument to select the respective path.
2. Hyperspectral imaging of a [TbEu(bpm)(tfaa)6] single crystal
3. Hyperspectral data analysis
To illustrate the configuration of the hyperspectral microscope for the data acquisition on a Ln-based, molecular single crystal (i.e., [TbEu(bpm)(tfaa)6], Figure 1a), Figure 2 shows an overview of the system as well as the right placement of the optical cubes in the setup. Figure 3 shows a screen shot of the PHySpec software containing the menus used during the HSI acquisition. Figure 4 and Figure 5 show the microscope stage in greater detail, including the placement of the glass slide containing the sample to be analyzed. The selected UV illumination was turned on to show the visible red luminescence of the crystal (Figure 4a and 1 in Figure 5). Figure 6a shows a bright field image of the crystal recorded after adjusting the sample in the proper focus. The needle-like morphology of the crystal can be clearly seen. Figure 6b,c show the image of the same crystal under UV excitation with either full view (Figure 6b) or locally confined (Figure 6c) illumination. Under the wide UV illumination, the differences of emission brightness from the different faces of the crystal are immediately visible. The confined illumination can be used as an option, mainly to investigate any effects of energy or light transfer in the crystal, which may trigger waveguide-like behavior. In this case, a strong emission is detected in a point not directly under excitation. This suggests that efficient energy migration takes place through the crystal13 (5 and 6 in Figure 7).
From the acquired hyperspectral cube, it is further possible to obtain the spectral distribution in form of an image representing a specific wavelength, the intensity profile of a specific emission wavelength, as well as the emission spectra at any pixel or area of the acquired hyperspectral cube. As an example, the emission spectra given in Figure 7 (panel 4) show the most characteristic emission bands of the Eu3+ ion: the band observed at 590 nm is assigned to the magnetic dipole (MD) 5D0→7F1 transition of Eu3+, while the emission peaks in the region from 610 to 630 nm stem from the hypersensitive forced electric dipole (ED) 5D0→7F2 Eu3+ transition. The ratio between the integrated intensity of these two transitions is well known to be an excellent probe of the chemical environment around the Ln3+ ion in the structure of the single crystal21: the lower the symmetry around the Ln3+ ion, the larger is the ED/MD ratio. This allows to draw conclusions about the symmetry character of the chemical environment of the Ln3+ ion. Moreover, the Stark splitting of the 5D0→7F2 transition can also be correlated with the symmetry around the Ln3+ in its crystallographic environment – the lower the symmetry, the higher is the number of Stark sub-levels. In case of the needle-like polymorph crystallized in the low symmetric triclinic crystal system, the 5D0→7F2 transition splits into four sub-peaks (spectra shown in Figure 7, panel 4). Such analysis is particularly appealing when comparing the optical properties of several polymorphs of a luminescent crystal. We previously demonstrated that the information about the chemical environment deduced from the optical analysis correlated well with the molecular crystal structure obtained by single crystal X-ray analysis13. Moreover, the spectral profile along the different crystal faces shown in Figure 7 (panel 3), indicating brighter emission at the tip and side faces, can also be correlated with the Ln3+···Ln3+ ion distances in the three spatial directions (Figure 1b): the denser Ln3+ packing along the axes perpendicular to the tip and side faces, respectively, favor ion-ion energy transfer. Hence, emission enhancement is observed at the respective faces, thus, optical anisotropy.
Overall, the various options of data analysis, shown in Figure 7 and Figure 8, constitute the most important features of the combined spectroscopic and spatial information, which is possible to be explored by HSI analysis of luminescent samples.
Figure 1: Molecular structure and crystallographic arrangement. (a) Structure of the heterodinuclear Ln-based complex [TbEu(bpm)(tfaa)6], where Ln1 and Ln2 are Tb3+ and Eu3+ ions. Disordered groups and hydrogen atoms are omitted for clarity. Color code: Eu: dark cyan; C: grey; O: red; N: blue; F: lime green. (b) Representation of the molecular packing in the crystal: (i) top view and (ii) tip view of the needle-like single crystal structure with selected intermolecular and intramolecular Ln···Ln distances (tfaa subunits and hydrogen atoms are omitted for clarity). (iii) Crystal packing arrangement of the [TbEu(bmp)(tfaa)6] dimers (hydrogen atoms are omitted for clarity). (iv) Diagram of the crystal growth faces of the dimer revealing the shortest Ln···Ln distances in the (0 1 0) and (2 -1 1) crystallographic directions. The figure has been modified from reference 13. Please click here to view a larger version of this figure.
Figure 2: Overview of the Hyperspectral Imaging System. Shown is the configuration required for luminescence mapping at the visible spectral region using UV excitation. (a) General view of the system, where 1 is the microscope stage, 2 is the section containing the optical configuration, and 3 is the spectrometer with the visible and NIR detectors. (b) Open view of the optical set-up close to the microscope stage (right side of a) showing the optical configuration for the experiment: optical cube position 4 remains empty and the confocal microscope cube is placed in position 5 in order to route the light through the visible path, the visible cube is placed in position 6 in order to direct the visible light to the detection path, and the confocal pinhole cube is placed in position 7 in order to route light to the visible detection path. (c) Open view of the optical set-up closer to the detectors (left side of a), showing position 8, where the confocal spectrometer cube is placed to reflect light to the spectrometer and visible camera. The inset 9 shows the screw to adjust the opening width of the spectrometer slit. (d) View of the microscope stage, computer and broadband lamp (used for UV excitation) controller. In the inset, the broadband lamp controller is shown with more detail: 10 is the on/off button, 11 is the knob to control the intensity of the lamp, and 12 is the shutter button. (e) Scheme showing the visible/NIR optical path from the microscope stage to the detectors, including the optical cube positions from 4 için 8. Please click here to view a larger version of this figure.
Figure 3: Screenshot of the PHySpec software showing the menu with the parameters to be adjusted for the HSI. 1 allows to insert the scale bar in the color camera image; 2 and 3 allow to control exposure time and gain value of the color camera, respectively; the proper objective lens must be selected in 4; 5 allows the selection of the aperture of the pinhole ; 6 (Diverter) and 7 (Filter) allow to choose the detector and grating, respectively; the exposure time for the visible detector is set in 8. Please click here to view a larger version of this figure.
Figure 4: General view of the microscope stage. (a) Placement of the glass slide containing the sample at the stage, with the UV illumination ON showing the sample’s red luminescence (small red dot in the center of the glass slide). (b) View of the microscope stage with the white light illumination condenser on top. (c) Stage controller showing the joystick that controls the movement of the stage in the directions indicated by orange and yellow arrows (also shown in (a). (d) Detailed view of the focus button, which moves the stage in the directions indicated by the red arrow (also shown in (b)). Please click here to view a larger version of this figure.
Figure 5: The components of the microscope stage. 1 microscope stage with the sample on the glass slide placed on the sample stage on top of the objective lenses; 2 wheels to adjust the focus (large wheel) and to direct the captured emission (small wheel) either only to the detector (L), partially to the detector and partially to the camera (R), or solely to the binocular lenses (eye); 3 excitation/emission filters wheel used to choose the excitation wavelength range. The detail on the right shows the filter cube holding the UV filter and long-pass filter used in this experiment; 4 in the top/bottom are show the knobs to move the excitation beam through the sample, while in between, the circular field aperture control; 5 objective lenses; 6 ON/OFF button of the white light illumination; 7 knob to adjust the brightness of the white light lamp. Please click here to view a larger version of this figure.
Figure 6: Optical microscopy images of the analyzed single crystal. These images were obtained under (a) white light illumination, (b) full-view UV illumination, using the excitation circular aperture completely open, and (c) locally confined UV illumination (marked by the white circle), using a closer excitation circular aperture. Please click here to view a larger version of this figure.
Figure 7: Screenshot of the PHySpec software showing the hyperspectral cube data analysis process. Diverse spectral analysis methods can be applied on the acquired hyperspectral cube: 1 shows the wavelength which was chosen for the spectral image distribution shown in 2; 3 shows the 613.26 nm horizontal (7) and vertical (8) intensity profiles; 4 shows the emission spectra extracted from the targets 5 and 6 as well as from the area highlighted in 9. Please click here to view a larger version of this figure.
Figure 8: Alternative application of HSI probing the synergy between upconverting nanoparticles and lanthanide complexes. This example shows the hyperspectral analysis of a hybrid system comprised of molecular crystals ([Tb2(bpm)(tfaa)6]) combined with upconverting nanoparticles (NaGdF4:Tm3+,Yb3+). (a) Photomicrographs under white and UV light illumination along with the region of interest (ROI) used for hyperspectral imaging under 980 nm light irradiation. (b) Tm3+ and indirect Tb3+ emissions monitored over an area of 20 x 20 µm2. (c) Variation of the absolute intensity of the emission bands fluctuated throughout the hybrid system indicating some variability in the total amount of material distributed over the surface. (d) The constancy of the ratio between the integrated emission of the complex vs. Tm3+: 1G4→3H6 (squares) and Tm3+: 1G4→3F4 (circles) confirmed the simultaneous presence of the two moieties throughout the hybrid system and the homogeneous interaction between them. Scale bars are 20 µm in the photomicrographs and 5 µm in ROIs and spectral maps. Photomicrographs are presented in real colors. The Figure has been modified from reference 11. Please click here to view a larger version of this figure.
The hyperspectral imaging protocol here described provides a straightforward approach that allows to obtain spectroscopic information at precise locations of the sample. Using the described setup, the spatial resolution (x and y mapping) can reach down to 0.5 µm while the spectral resolution can be of 0.2 nm for the mapping at the visible range and 0.6 nm for the NIR range.
In order to conduct hyperspectral mapping on a single crystal, sample preparation follows an easy procedure: the crystal can simply be placed on a glass microscopy slide, covered by a cover glass as needed. Focusing the sample using the proper objective lenses and the bright field image at the Color Camera set-up in the software is a very important step during the pre-analysis stage to obtain best resolved hyperspectral images. Typically, higher emission intensities are obtained when the sample is well focused. Once this is done, the choice of the parameters of the analysis such as the x and y counts and the step size will dictate the field of view and spatial sampling, respectively of the obtained hyperspectral cube. However, the objective´s numerical aperture and the excitation/emission wavelengths dictate the real volume of the sample that is probed at each acquisition point. For example, for the objective used in this study, with numerical aperture (NA) of 0.4, and using the excitation in the UV spectral range (390 nm), the focused laser spot has a size of approximately 0.6 x 0.6 µm in the x and y direction. The size of the laser spot was calculated using the website (https://www.microscopyu.com/tutorials/imageformation-airyna, accessed on Sep. 26, 2019). If the chosen step size is larger than the spatial sampling, one may actually be sampling an area smaller than the one given by the step size. If the sample is homogeneous, undersampling may not be an issue. Yet, if spatial variations in the sample are important to detect, optimal sampling is obtained with a step size set at half the size of the laser beam on the sample. The chosen exposure time and the intensity of the UV selected illumination will control the intensity of the obtained spectra. Such parameters vary from sample to sample, depending on the emission intensity and the sensitivity towards the UV excitation.
At this point, the critical steps within the protocol can be listed as: the alignment of the optical system, correct placement of the optical cubes, regulation of the circular aperture of the excitation and detector slit openings, choice of the pinhole, focus of the sample in the color camera using the proper objective lenses, the intensity of the broadband lamp and proper choice of the long pass filter cube (allowing UV excitation), as well as choice of the proper step size and exposure time as mentioned above. Lastly, the room lights must be off during all time of the hyperspectral cube acquisition.
In case of poor signal detection either with the color camera or the spectrometer, the troubleshooting of the technique should include checking carefully each one of the critical steps given above, before starting the hyperspectral cube acquisition. The configuration of the output signal at the microscope stage is also important. The three possible configurations are: eye (100 % of the output signal is sent to the microscope binocular mount), L (100 % of the output signal is sent to the detectors) and R (80 % of the output signal is sent to the detectors and 20 % to the microscope binocular mount). During the hyperspectral cube acquisition, the R or L configuration must be used. If all the above listed parameters are properly chosen, high resolution spatial and spectral information of the sample can be obtained.
Some possible modifications of the technique herein described can be exemplified by the hyperspectral imaging of other systems, such as luminescent microparticles14 or optical hybrid systems comprised of molecular crystals combined with upconverting nanoparticles (Figure 8)11. In these examples, the NIR laser diode (980 nm) was used as the excitation source, replacing UV excitation, while detecting the generated visible emission. In the latter example of the hybrid system, HSI revealed the homogeneity of hybrid films that combine upconverting nanoparticles (NaGdF4:Tm3+,Yb3+) and [Tb2(bpm)(tfaa)6] crystals into a multiwavelength-responsive isotropic system, exhibiting energy transfer between materials and molecules (Figure 8)11. Furthermore, by using the InGaAs detector of the system, detection of emissions in the NIR spectral region (1000 to 1700 nm) becomes possible. This is of particular interest when seeking the investigation of NIR-based optical probes for biomedical applications3. In this case, the optical cubes configuration of the system (Figure 2e) has to be set for the NIR path. In case of NIR excitation-NIR emission, one of the limitations of the hyperspectral technique here described becomes evident: as being wavelength-dependent, the spectral resolution in the NIR region is lower than that for the visible detection, i.e. approximately 0.6 nm (vs. 0.2 nm). Moreover, for sub-1 µm features, in e.g. smaller molecular crystals, nanoparticles or hybrid systems, the spatial resolution, which is dictated by the system configuration (objectives used and excitation/emission wavelengths), becomes another potential limitation.
Finally, (irrespective of the chosen HSI configuration and wavelength regime) the data manipulation can be done either with the instrument’s software or, as shown for the case of the spectral profiles, can be exported to be analyzed in other software packages such as Origin® or Microsoft Excel. In our example, the optical anisotropy of the crystal was also promptly revealed at the color camera image, namely by the strong intensity variation along the different crystal faces. Moreover, under the wide UV excitation, different emission intensities are obtained depending on which face is analyzed (Figure 7). The possibility to obtain the emission intensity profile in different points of interest at the crystal (targets in Figure 7) further allows to study variation in emission intensity and, if present, also in spectral shape. The block-like polymorph of the [TbEu(bpm)(tfaa)6] constitutes an example where two crystal faces exhibited equally high emission intensities and a lower emission intensity for the third one13. Expending on this, in case of a system without any anisotropy, the emission intensity would be the same for all crystal faces.
Complementary methods to probe optical anisotropy are for instance related to the presence of polarized emission from a sample. These include polarization memory or spectroscopic ellipsometry. The first consists in the correlation between the polarization state of the light emitted by the material with the polarization state of the incident excitation light,22,23 while the latter measures the change in the polarization state of the light after being reflected obliquely by a thin sample film24. However, an advantage of using hyperspectral imaging as a tool for probing optical anisotropy, as shown here, comes with the fact that the presence of polarization is not a requirement for sample analysis. Moreover, the sample preparation does not require the fabrication of thin films, neither a very careful orientation of the crystal with respect to the incident and collected light. These aspects make the technique of HSI potentially more widely applicable. Moreover, the anisotropic features are promptly visualized upon image acquisition, and data analysis is straightforward (as exemplified above).
Considering a broader scope, the significance of the HSI technique can be attributed to its unique characteristic to corelate the optical signal with environment-dependent features. For instance, such connection is essential to improve the understanding of nano-bio interactions3,15,16 in the growing field of nanomedicine or even to understand structure-properties relationship in materials science10,11,12,13. As such, future potential applications of the herein described technique can be named as, but not being limited to: analysis of biological samples in vitro, ex vivo and in vivo performing the mapping of molecules of biological interest4,6, adaptation of the microscope stage to study environment-specific opto-electronic properties (e.g. samples embedded in electrical circuits), optical temperature sensing8,15 (by adding a temperature controller to the microscope stage) or gas sensing (by adapting a gas chamber to the microscope stage). HSI has further been demonstrated as suitable for fluorescence excitation techniques instead of fluorescence emission25. A good example of the use of this particular hyperspectral technique adaptation is the detection of cancer cells in biological tissues6. Consequently, the protocol here described has the potential to be largely extended to the study of spectroscopic features of many different types of luminescent structures.
The authors have nothing to disclose.
The authors thank Mr. Dylan Errulat and Prof. Muralee Murugesu from the Department of Chemistry and Biomolecular Sciences of the University of Ottawa for the provision of [TbEu(bpm)(tfaa)6] single crystals. E.M.R, N.R., and E.H. gratefully acknowledge the financial support provided by the University of Ottawa, the Canadian Foundation for Innovation (CFI), and the Natural Sciences and Engineering Research Council Canada (NSERC).
Microscope glass slides | FisherBrand | 12-550-15 | Glass slides used for sample preparation |
Visible and Near Infrared Hyperspectral Confocal Imager | PhotonETC | Microscope used for the analysis, builted according to the user needs, therefore it is no catalog number |