An imaging Fourier transform spectrometer (IFTS) designed for fluorescence emission measurements is reported. The spectral range extension from NIR to visible of the system is realized by using a simple and low-cost optical beam-folding position-tracking technique. Spectral resolution as high as 9.78cm−1(0.4nm at 632.8nm) and maximum image resolution up to 300×300 pixels are proved by the system tests on its optical performances. Imaging fluorescence spectra acquisition of quantum dot clusters and single 200nm diameter fluorescent beads have demonstrated the system’s potential for high throughput imaging spectroscopic measurements of fluorescent biological and chemical samples.
© 2009 OSA
In the past fifteen years, spectral imaging (also imaging spectroscopy or chemical imaging) has achieved profound progress in many aspects of life science research such as fluorescence microscopy of biological live cells , Förster Resonance Energy Transfer (FRET) quantification enhancement , high throughput spectral discrimination of multicolor fluorophores  and spectral shifts measurements caused by sample microenvironment variations .
Generally, three popular categories of methods are often used to realize laboratory spectral imaging instrumentation for biological and chemical applications, including variable optical filter methods with discrete filter-wheel, liquid crystal tunable filter (LCTF), and acousto-optical tunable filter (AOTF); optical dispersive methods based on prisms or gratings e.g. scanning confocal microspectroscopy; and indirect time-scan methods e.g. imaging Fourier transform spectroscopy . Among all these approaches, IFTS stands out as a powerful and flexible method inheriting the traits of the Jacquinot (throughput), Fellgett (multiplex)  and Connes advantages  from Fourier transform spectroscopy (FTS). Benefitting from these advantages in spectral performances over other methodologies, IFTS has successfully achieved some unique biological and chemical applications in conjugation with fluorescence microscopy. One of such renowned examples is spectral karyotyping (SKY) , which can discern human chromosomes simultaneously and unequivocally based on fluorescence in situ hybridization (FISH). Fisher et al. also combined IFTS with fluorescence microscopy to realize a fast aerosol analysis on the polycyclic aromatic hydrocarbons contamination .
An FTS favors to work in infrared longer wavelengths according to its fundamental principle and Nyquist sampling theorem . However, many biological and chemical fluorescence emissions are weak and broadband (compared to atomic absorption spectra) in the visible to NIR spectral range. To make 2D spectroscopic measurements on these types of signals, an extension of working spectral band of an IFTS from its favorable infrared down to visible short wavelengths is of necessity. Although the development of FTS over the past forty years have found many ways to largely extend the spectral range down to even far-UV with very high spectral resolution, the sophisticated optics, electronics and motion control mechanics involved plus more stringent experimental conditions usually increase the cost and complexity of the whole system a lot . In addition, lack of sensitive focal plane array (FPA) detectors with high readout speed has also restrained the performances of IFTS for low light intensity measurements. To enhance the popularity of IFTS as a routine instrument for biology and chemistry laboratories, a simple, compact, rugged and low-cost device with its inherent spectral as well as spatial resolving performances being reasonably preserved is expected for specific biological and chemical applications.
In this paper, we report such a visible-NIR (~500nm-1100nm) IFTS system with the spectral extension problem being resolved by a simple optical beam-folding position-tracking technique  and the FPA sensor being selected as an Electron Multiplying CCD (EMCCD). The whole system is kept simple and compact with its optical performances and applicability proved to be powerful and practical for biological and chemical laboratories.
2. Principles and experiments
2.1 Instrument setup
The schematic setup of the IFTS system based on beam-folding position-tracking technique is shown in Fig. 1 .
It comprises two multipass interferometers with both of their movable arms fixed on the same piezoelectric translation (PZT) stage (P-629.10L, PI). On the left-hand side of Fig. 1 is a reference four-path interferometer working with a He-Ne laser at wavelength of 632.8nm for tracking the PZT stage position. On the right-hand side is a two-path interferometer for obtaining the interferogram images of the object under measurement. Its movable arm is beam-folded by a large aperture hollow retroreflector and the optical path difference (OPD) scan is achieved by linearly translating the hollow retroreflector along the optical axis. This arrangement prevents lateral displacement errors of the stage from displacing the beam thus optical alignment can be well kept during the scan. In addition, the equivalent maximum OPD of the moving arm is doubled, which in principle increases the spectral resolution of the IFTS by a factor of two. A pair of collimating microscope objective and tube lens at the respective input and output ports of the interferometer makes up an infinity-corrected microscope with its magnification equal to their focal lengths ratio. An EMCCD camera (Cascade 1K, Photometrics) residing at the image plane is used to capture the interferogram frames. For fluorescence emission measurements, a standard wide-field epi-fluorescence excitation configuration is established by inserting a fluorescence filter cube into the parallel optical path after the objective as enclosed in the dashed rectangle in Fig. 1.
The spectral range of the IFTS system is not only dependent on frame sampling interval but also on the spectral response characteristics of the optical components used. With our focus primarily on fluorescence imaging, we deliberately selected gold-coated surfaces of both the flat mirrors and the hollow retroreflector in the measurement interferometer, which have <60% reflectance in the short wavelengths below 500nm. Therefore the short excitation wavelengths can be further suppressed by these reflection surfaces after leaking through the long-pass filter in the filter cube; and on the other hand, the instrument’s lifetime can be prolonged because of gold inertness. Considering the above-mentioned parameters, the spectral range of the IFTS is estimated to be from 500nm to 1100nm (20000cm−1-9091cm−1) in the visible to NIR region.
With a maximum PZT stage translation distance of 1500μm, the theoretical maximum spectral resolution of the IFTS can reach to (~0.05nm in the visible) according to Rayleigh criterion. Thus the resolving power of the whole system can be estimated as:
For an IFTS, the OPDs are the same for all the points on the object symmetric to the optical axis with same incident angle θ and the pixel interferogram can be described as:11] indicate that for a practical IFTS, Eq. (4) can usually be simplified as12].
During the scan, interferogram frames will be continuously captured by the EMCCD at every clocking pulse generated at fringe zero-crossings of the reference interferometer output. Once a preset number of interferogram images are acquired (Eq. (5)), the spectra at every pixel (Eq. (6), 7)) can be obtained by taking Fourier transforms and spectral images at different wavenumber/wavelength within the response band can also be reconstructed for storage and display.
2.3 Test experiments
The optical performances of the IFTS system were first evaluated by measuring an Ø200μm multimode fiber end illuminated by a 632.8 nm He-Ne laser and by a combination of a red and a green LED, respectively. A ×10 standard microscope objective was used as the collimating lens and a 200mm focal length tube lens was used to project the fiber end image onto the EMCCD chip. The epi-fluorescence excitation unit was not incorporated temporally.
To test the practical applicability of the system, imaging spectroscopic measurements of NAC-capped quantum dot  clusters (peak at 600nm) and fluorescent beads (F8809, Invitrogen) of ~200 nm in size were performed respectively under the epi-fluorescence excitation configuration shown in Fig. 1. The quantum dot cluster sample was prepared by drying a small droplet of diluted solution onto a clean quartz slide. The fluorescent bead sample was prepared by immobilizing extremely diluted sample solution between quartz slides to ensure individual particles being observed within the objective field-of-view (FOV). Now the ×10 objective was replaced by a ×50 infinity-corrected microscope objective (M Plan Apo ×50, MITUTOYO) for higher imaging magnification and better beam collimation. The quantum dot clusters and fluorescent beads were excited by a 455nm LED and a 488nm argon laser, respectively. A long-pass filter in the filter cube blocks the short wavelengths below 500nm and permits all the longer wavelengths to pass into the interferometer.
3. Results and discussion
Figure 2 shows some example fiber end interferogram images, pixel spectra and spectral images as measurement results of the first test.
In Fig. 2(a), an arbitrary interferogram frame of the fiber end illuminated by He-Ne laser from a total number of 4k frames is shown. To increase the camera’s frame rate, a binning factor of 4 was applied to all the original 320×320 pixels frames to obtain a final image resolution of 80×80 pixels. Speckle patterns due to high coherence nature of the laser can be distinguished in the image. The maximal interferogram image resolution is restricted by the compromise between the slowest PZT stage scanning speed and the fastest camera frame rate. With a readout speed of 10MHz of the current EMCCD camera, a maximum interferogram image resolution of 300×300 pixels is obtained by measuring the fiber end illuminated with two low coherent red and green LEDs as shown in Fig. 2(b). Figure 2(c) and (d) displays the reconstructed spectral images of Fig. 2(b) at the peak wavelengths of the two LEDs in pseudo color, respectively. Compared to Fig. 2(a), edge defects, surface scratches and inhomogeneities of the fiber end can be clearly seen in Fig. 2 b), (c) and (d) with better image resolution.
To demonstrate the spectral performance of the IFTS system, the He-Ne laser spectrum at an arbitrarily coordinate in the bright region of Fig. 2(a) is shown in Fig. 2(e). It can be found that without further calibrations, the peak intensity position was measured to reside very accurately at wavenumber of 15802.4cm−1 (632.8 nm in wavelength). The wavenumber accuracy is benefited from the Connes advantage  of the reference interferometer, which will practically facilitate the instrument usage with long-term stability by avoiding tiresome spectral calibrations frequently needed in other competitive methods such as the rotating Sagnac interferometer based IFTS , although the latter could be made more compact. The spectral resolution at FWHM intensity is found to be 9.78cm−1 (~0.4nm at 632.8nm in wavelength), which is lower than the theoretical value shown in the parenthesis of Fig. 2(e) for comparison. The degradation in practical spectral resolution is mainly due to the fact that pixel interferograms of an IFTS require higher stringency on the maintaining of optical moving alignment. Small misalignment during the scan would induce light intensity crosstalk among neighboring pixels in the image plane. Even so the obtained resolution is still better than typical value (~1nm) of the commercially available high quality laser line interference filters (e.g. 10LF01-633, Newport) for Raman scattering measurements up to date.
The property that spectral resolution can be tunable by scanning different OPDs makes the IFTS versatile for different biological and chemical applications. For low spectral resolution cases, a smaller number of frames are needed with fixed sampling intervals. This is preferred as to shorten the acquisition time and reduce the data cube size, hence producing faster rate in data display. Faster data capture is essential in fluorescence experiments as biological dynamics may be monitored and photobleaching may be alleviated. As an example, the test on the fiber end illuminated by LEDs was performed in a faster scan rate of 1k frames in ~60s. The spectrum of one of the pixels in the bright region in Fig. 2(b) is shown in Fig. 2(f). Two peaks near 18596cm−1(538nm) and 15771cm−1(634nm) with FWHM linewidths of 307cm−1(13.6nm) and 815cm−1(23.5nm) can be identified, which are consistent with the specifications of the LEDs. It is noteworthy that the acquisition time can be further reduced by sacrificing the image resolution, since the camera frame rate is inversely proportional to the frame resolution. This can be done by either choosing a bigger binning factor with fixed FOV or selecting a smaller FOV with fixed binning settings respect to specific applications.
The practical usage of the instrument was tested by measuring quantum dot clusters and single fluorescence beads, respectively. In order to reduce the acquisition time and increase frame SNR, frame resolution was degraded to 100×100 pixels with a binning factor of 8. In both cases, 1k frames were captured in only 25s and example results are shown in Fig. 3 .
Figure 3(a) shows an arbitrary interferogram image as micrograph of one quantum dot cluster. The cluster’s shape and intensity distribution can be well recognized. With known EMCCD pixel size of 8×8μm, image magnification of ×50 and binning factor of 8, the irregular particle size is estimated to be about 128×100μm (h×v). The spectrum taken from the position where the red pixel pointed by the arrow in Fig. 3(a) manifests the quantum dot fluorescence peaks at 600nm with a linewidth of about 50nm. In the micrograph of individual fluorescent beads shown in Fig. 3(c), several particles inside the FOV can be recognized as bright pixels since their sizes (~200nm) are too small to be resolved by the IFTS with a spatial resolving power of 500nm. Figure 3(d) shows the fluorescence spectrum of a bead depicted in the red circle in Fig. 3(c). As can be seen, the spectrum exhibits good SNR even without signal averaging and the peak at 560 nm with a narrow linewidth of 33nm is consistent with the specification of the fluorescent beads. Since the FOVs in both cases are very small (~130×130μm) and the objective is infinity-corrected, rescaling in Eq. (7) is omitted and no obvious spectral redshift was observed.
The occurrence of fluorescence photobleaching can be noted in the pixel interferogram as slow decreasing DC component (data not shown). But it usually will not affect the final spectra because the slow variation should only change the low frequency components after Fourier transform. Although this is an advantage of Fourier transform, in cases where serious photobleaching is of concern, the excitation light can be pulsed and synchronized with the reference sampling signal as further action to reduce the effective excitation dosage besides reducing the acquisition time of the IFTS and optimizing the excitation intensity .
The results shown in Fig. 3 have successfully proved good spectral performances and high throughput measurement capability of the IFTS system. Furthermore, the current setup itself is actually an independent imaging spectrometer even without the epi-fluorescence excitation unit and its front end can be easily adapted for other advanced and popular experiment configurations such as total internal reflection fluorescence microscopy (TIRFM)  and dark-field illumination scattering microspectroscopy . The former can be an ideal barcode reader for high throughput spectral discrimination of quantum dot nano-barcodes in molecular imaging ; and the latter is also a potential candidate for high efficiency biological and chemical sensing that based on localized surface plasmon resonance (LSPR) spectroscopy of noble metal nanoparticles .
In summary, we have constructed an IFTS with working spectral range from 500nm to 1100nm. Its working spectral range is extended from NIR to visible by a simple, low-cost yet accurate optical beam-folding position-tracking technique. Tests on measuring a fiber end have proved the system could achieve a highest spectral resolution of 9.78cm−1, which is less than 1nm in its working spectral band, and also a maximum imaging resolution of 300×300 pixels. Experimental results on quantum dot clusters and fluorescent beads measurements further demonstrated the practicability of the IFTS system for fluorescence imaging spectroscopic applications. The visible-NIR IFTS system holds great potential as a practical and versatile platform to meet the trend that more and more experiments will be executed in high throughput screening manner with the development of microarray and lab-on-chip technologies in biology and chemistry.
We gratefully acknowledge Dr. Zhao D, Prof. Chan WH, Dr. Yeung KM and Dr. Li HW from the Department of Chemistry of Hong Kong Baptist University for the preparation of quantum dot clusters and fluorescent beads samples and the financial support from the FRG/07-08/II-53.
References and links
2. T. Zimmermann, J. Rietdorf, A. Girod, V. Georget, and R. Pepperkok, “Spectral imaging and linear un-mixing enables improved FRET efficiency with a novel GFP2-YFP FRET pair,” FEBS Lett. 531(2), 245–249 ( 2002). [CrossRef] [PubMed]
3. E. Schröck, S. du Manoir, T. Veldman, B. Schoell, J. Wienberg, M. A. Ferguson-Smith, Y. Ning, D. H. Ledbetter, I. Bar-Am, D. Soenksen, Y. Garini, and T. Ried, “Multicolor spectral karyotyping of human chromosomes,” Science 273(5274), 494–497 ( 1996). [CrossRef] [PubMed]
4. G. L. Liu, J. C. Doll, and L. P. Lee, “High-speed multispectral imaging of nanoplasmonic array,” Opt. Express 13(21), 8520–8525 ( 2005), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-13-21-8520. [CrossRef] [PubMed]
5. Y. Garini, I. T. Young, and G. McNamara, “Spectral imaging: Principles and applications,” Cytom, Part A 69A, 735–747 ( 2006). [CrossRef]
6. R. J. Bell, Introductory Fourier transform spectroscopy (Academic Press, New York, 1972).
7. S. Wartewig, IR and Raman spectroscopy: fundamental processing, Spectroscopic techniques. An interactive course. (Wiley-VCH, Weinheim, 2003).
10. X. Z. Wang, R. K. Y. Chan, and A. S. K. Cheng, “Near UV-near IR Fourier transform spectrometer using the beam-folding position-tracking method based on retroreflectors,” Rev. Sci. Instrum. 79(12), 123108–123106 ( 2008). [CrossRef] [PubMed]
11. N. J. E. Johnson, “Spectral imaging with the Michelson interferometer,” Proc. SPIE 226, 2–9 ( 1980).
12. E. H. Wishnow, R. E. Wurtz, S. Blais-Ouellette, K. H. Cook, D. J. Carr, I. T. Lewis, F. Grandmont, and C. W. Stubbs, “Visible Imaging Fourier Transform Spectrometer: Design and Calibration,” Proc. SPIE 4841, 1067–1077 ( 2003). [CrossRef]
13. D. Zhao, Z. He, W. H. Chan, and M. M. F. Choi, “Synthesis and Characterization of High-Quality Water-Soluble Near-Infrared-Emitting CdTe/CdS Quantum Dots Capped by N-Acetyl-l-cysteine Via Hydrothermal Method,” J. Phys. Chem. C 113(4), 1293–1300 ( 2008). [CrossRef]
15. D. Axelrod, “Total internal reflection fluorescence microscopy in cell biology,” in Biophotonics, Pt B (Academic Press Inc, San Diego, 2003).
16. A. Curry, W. L. Hwang, and A. Wax, “Epi-illumination through the microscope objective applied to darkfield imaging and microspectroscopy of nanoparticle interaction with cells in culture,” Opt. Express 14(14), 6535–6542 ( 2006), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-14-14-6535. [CrossRef] [PubMed]
17. J. Yang, S. R. Dave, and X. H. Gao, “Quantum dot nanobarcodes: Epitaxial assembly of nanoparticle-polymer complexes in homogeneous solution,” J. Am. Chem. Soc. 130(15), 5286–5292 ( 2008). [CrossRef] [PubMed]