High-resolution chemical microscopy with multiple bright synchrotron beams

Fourier-transform IR (FTIR) microspectroscopy delivers chemical information without perturbing the sample and without the need for stains, dyes, or labels. It is an important tool in many disciplines, ranging from nanomaterials,¹ pharmaceuticals,² volcanic rocks,³ forensic⁴ and art conservation,⁵ to biomedical samples.^6–8 Up to now,⁹ however, the optical resolution of available instruments was limited by fundamental trade-offs between acquisition time, signal-to-noise ratio (SNR), sample coverage, and spatial resolution.

In microscopy, both source and detector characteristics are fundamental factors that affect the achievable spatial resolution. Benchtop thermal IR sources, like light bulbs, emit their light in all directions, which makes it impossible to collimate their light efficiently. The result is significant photon loss and effectively fewer photons arriving at the detector. Conventional single-synchrotron beams, on the other hand, can be collimated very efficiently, like lasers, and therefore work very well with confocal-type IR microscopes equipped with a single-element detector. However, even though state-of-the-art for high spatial resolution for many years,¹⁰ these systems are very slow and unable to cover larger sample areas, due to their raster-scanning mechanism. To solve this problem, we have now developed, for the first time, an IR imaging system that combines the speed and coverage advantage of wide-field benchtop setups with the high spatial resolution offered by multiple beams of bright synchrotron light.⁹

Figure 1. Schematic of the 12-beam synchrotron Fourier transform IR imaging beamline (a) at the Synchrotron Radiation Center. M1–M4 are sets of toroidal, flat, parabolic, and flat mirrors that collect and rearrange the bending magnet radiation. (b) False-color screenshot of the 128×128pixel focal plane array (FPA) showing the 12 overlapping beams illuminating an area of ∼50×50μm²(scale bar 40μm). (c) Visible light photograph of the 4×3 beam matrix at the position indicated by the dashed box in (a). Scale bar ∼1:5cm.

Figure 2. Transmission IR images of groups 8 and 9 (bar/gap width range: 0:78 1:95μm) of a high-resolution US Air Force test target at 3000cm ¹(wavelength 3:33μm); both images were measured on the same Bruker Hyperion 3000 FPA microscope with 16 scans, but (a) was acquired using a conventional thermal source and a 15×objective, while (b) was measured with the new multibeam synchrotron source and a 74×objective. Pixel sizes are given in parentheses.

Figure 3. IR image showing the amide I distribution at 1654 cm ¹in cancerous prostate tissue with chronic inflammation, measured with a conventional benchtop instrument (PerkinElmer Spotlight) equipped with a linear array detector (a) and the new multibeam synchrotron system with an FPA (b). Pixel sizes are given in parentheses.

We use 12 bright but slightly defocused overlapping synchrotron beams, arranged in a 3 × 4 matrix, to illuminate a large sample area quasi-homogeneously.^{9, 11,12} The sample is then imaged onto a focal plane array (FPA) multi-element detector in a wide-field geometry, where no time-consuming raster-scanning is necessary to cover a sample area up to ∼50×50μm² in a single shot (see Figure 1). While single synchrotron beams have been used with FPA detectors on a few occasions by pioneering groups,^13–16 we have designed our system to effectively use multiple beams, providing a decisively more homogeneous sample illumination which allows us to acquire larger areas in only few minutes.

The use of multiple bright synchrotron beams from a dedicated bending magnet at the Synchrotron Radiation Center (Stoughton, WI) gives us enough photons to increase the microscope objective's magnification to 74×, decreasing the effective sample-plane pixel size to 0:54×0:54μm² while maintaining a very high SNR and therefore short acquisition times. The correct pixel size is crucial, because to achieve the highest optical resolution for a given wavelength, one needs not only a high numerical aperture objective (according to the Rayleigh criterion) and high SNR, but also pixels small enough to correctly sample fine specimen details (according to the Nyquist theorem). A detailed analysis^{9, 17} shows that, for the mid-IR wavelength range of 2.5–12:5μm, pixels should not be larger than 0.6–0.8μm to achieve full diffraction-limited spatial resolution. Smaller pixels (oversampling) will not increase the resolution further (empty magnification), whereas larger pixels (undersampling) will lead to unavoidable loss of resolution. We use a 74× objective with an effective pixel size of 0:54×0:54μm², which is ∼100 times smaller in area than conventional benchtop wide-field instruments, spatially oversampling images at all mid-IR wavelengths.

It is important to note that simply using a higher magnification objective (yielding a smaller pixel size) to obtain a correct sampling on a benchtop FPA instrument unavoidably results in exceedingly poor SNR and thus bad spectral quality, as fewer photons are collected per pixel area. For example, using a ∼100 times smaller pixel size would require a 10⁴-fold increase in scanning time, making this approach impractical. Thus, the effective use of a bright source like the synchrotron is essential.

The resulting increase in spatial resolution can clearly be seen in Figure 2 (see also Supplementary Figure 3 elsewhere⁹), which shows IR images of the same groups of a commercial 1951 US Air Force target. They have been acquired under identical conditions on the same microscope system, but one is measured with a benchtop thermal source and a standard 15× objective (pixel size 2:7×2:7μm²) in Figure 2(a), and one is measured with our multi-beam synchrotron system using the 74× objective with the correct spatial sampling in Figure 2(b). The resolution improvement is also obvious in Figure 3. It shows the amide distribution in the same cancerous prostate tissue section—see Figure 3(c) and (d) elsewhere⁹—measured with a standard benchtop instrument in Figure 3(a) and our new system in Figure 3(b).

The spatial resolution of conventional benchtop mid-IR systems and synchrotron-based confocal-type microscopes is typically limited to ∼5μm². We show that the combination of multiple bright synchrotron beams with a wide-field FPA detector can deliver high-resolution and high-SNR chemical images that are truly diffraction-limited at all mid-IR wavelengths within minutes. In the future we plan to use the subcellular resolution of this instrument to study living biological cells as a function of time.

We thank T. Kubala, S. Janowski, G. Rogers, M. Fisher, G. Vlasak, M. Thikim, R. Julian, M. Rak, and Z. El-Bayyari for their help at the beamline. This work was supported by the National Science Foundation (NSF) under awards CHE-0832298, CHE-0957849, DMR-0619759, and by the Research Growth Initiative of the University of Wisconsin-Milwaukee. Part of this work is based on research conducted at the Synchrotron Radiation Center, University of Wisconsin-Madison, WI, which is supported by the NSF under award DMR-0537588. The project described was also supported by award R01CA138882 from the National Institutes of Health.