We developed a nonlinear optical infrared microscope exploiting a thermally induced refractive index change in the mid-infrared regime and imaged a single biological cell with high spatial resolution that was not possible in conventional infrared microscopes. A refractive index change of a sample induced by infrared (~3.5 μm) absorption was probed by a visible (633 nm) laser beam. Thus the chemical specificity stems from the spectral absorbance of specimen and the spatial resolution from the short wavelength visible radiation. A reflecting objective (NA0.5) was used to focus the infrared and visible beams on the sample plane, and the sample was raster-scanned for 2-D imaging. The high resolution beyond the infrared diffraction limit was demonstrated by imaging fine grating lines made up of epoxy grooves (830 lines/mm). The probe wavelength dependence of the spatial resolution was investigated by imaging polystyrene beads. We found that the resolution was as small as 0.7 μm with 633 nm probe wavelength.
© 2011 OSA
Infrared (IR) microscopy in combination with spectroscopy has drawn much attention in various scientific fields ranging from material science to cellular biology as a useful chemical analysis method for microscopic scale specimens [1–3]. Due to the resonant absorption of mid-infrared radiation by specific molecular components in the specimen, it is possible to obtain highly selective molecular vibrational contrast images. Chemical composition of many diverse materials can be identified and spatially resolved by the infrared microscopy without any staining or labeling agents. In biomedical applications, infrared images have been utilized successfully to study pathological states of various kinds of tissues or cells and classify them as normal or diseased ones . Despite its indispensable role in biological studies, infrared microscopy has a fundamental limitation in its spatial resolution. Since the most infrared wavelengths useful in the IR microscopy resides in the mid-infrared (MIR) region between 3 and 11 μm, the spatial resolution is poor compared to visible light microscopy, limited to several micrometers at best in terms of Rayleigh criterion for the spatial resolution Δ = 0.61λ / NA, where λ and NA are the wavelength of the illumination and the numerical aperture (NA) of objective lens, respectively [5,6]. Thus, the poor spatial resolution makes it difficult to see the details of sub-cellular structures in the specimen. To overcome this problem, solid immersion lens (SIL) methods to increase the effective NA of the imaging system have been applied to IR microscopy to achieve good spatial resolution around 2 μm [7,8]. The near-field technique such as scanning near-field infrared microscopes have achieved high resolution down to below 30 nm [9,10]. But it could only provide the information near the cellular membrane surface. As a different approach, we have recently developed a nonlinear optical imaging technique, called nonlinear optical infrared (NLIR) microscopy , which exploits thermally induced refractive index change of the target objects we want to see. The technique has successfully demonstrated the chemical specificity by imaging CH and OH vibrational modes selectively. In this paper, we investigate the spatial resolution of the NLIR microscopy in a little more detail and apply it to a real biological sample to show the cellular imaging capability with unprecedented high spatial resolution in IR microscopy.
2. Operation principle of NLIR microscopy
The operating principle of NLIR microscopy is schematically illustrated in Fig. 1(a) . A long wavelength mid-infrared laser beam is resonantly absorbed by the sample to thermally induce the refractive index change around the focal region, and a near-infrared (NIR) or visible probe laser beam that is far off-resonant and collinear with the MIR beam deflects off the focal region where the refractive index profile is modified. Thus, the chemical specificity stems from the spectral absorbance of the sample and the spatial resolution from the short wavelength probe radiation. The thermally induced Kerr nonlinearity Δn is associated with the medium’s Kerr coefficient n 2 as [12,13]
3. Experimental setup
The optical layout for NLIR microscopy is shown in Fig. 1(b). A mode-locked Nd:vanadate laser (PicoTrain, High-Q Lasers) generates a 7 ps pulse train with 76 MHz repetition rate at 1064 nm. An intracavity doubled optical parametric oscillator (OPO: Levante, APE Berlin) is pumped by 90% output of the Nd:vanadate laser to generate a frequency-doubled signal beam at an NIR range from 776 nm to 820 nm with 6 ps pulse width. The OPO output and the remaining 10% of the Nd:vanadate laser power are used for difference frequency generation (DFG) in a 20 mm long periodically poled lithium niobate (PPLN) DFG crystal, which generates the MIR output of wavelengths from 2.87 μm to 3.57 μm by tuning the OPO wavelength. The MIR beam is modulated by an optical chopper and combined with a probe beam. To investigate the probe wavelength dependence of the NLIR microscopy, the probe beam is made switchable between the 1064 nm NIR beam and a continuous wave 633 nm HeNe laser. A Schwarzchild reflecting objective of 0.5 NA (Ealing) is used to focus the incident laser beams on the sample plane. The probe beam is then collimated by a conventional refracting objective (0.55 NA) and forwarded to a silicon photodiode. Since the small convex primary mirror of the reflecting objective induces a central obscuration, the collimated probe beam is annular. We put an aperture along the collimated probe beam before the detector. The aperture diameter is slightly bigger than the inner diameter of the annular beam. By optical chopping of the MIR beam, the annular beam is modulated in size and its transmission through the aperture is also modulated accordingly to generate an AC signal. The sample is mounted on an XY translation stage and raster-scanned by computer-controlled piezo actuation.
4. Results and discussion
To demonstrate that the NLIR microscopy exhibits high spatial resolution beyond the MIR diffraction limit, we imaged two grating samples, each surface of which is made up of fine epoxy grooves, 600 lines/mm, and 830 lines/mm. In this case, the measurements were carried out with MIR at 3.5 μm and visible probe at 633 nm wavelengths. For the sample preparation, a small amount of water was dripped on the grating surface to smear into all grooves, and a thin cover glass slip was put over the grating surface so that it closely adhered to the grating surface. The measured NLIR images of the grating samples are shown in Fig. 3(a) and (b) . Since some of CH vibrational modes of epoxy reside near 2850 cm-1 (~3.5 μm), whereas water does not have resonant absorption at this frequency region, strong signals are expected only from epoxy. The bright lines in the images represent epoxy ridges and the dark lines correspond to water inside the grooves. The images clearly show a strong contrast between epoxy and water even in the groove density 830 lines/mm (line spacing of 1.2 μm). It clearly demonstrates that our NLIR microscopy has higher spatial resolution beyond the MIR diffraction limit with chemical specificity. A conventional bright field image of the 830 lines/mm grating is presented in Fig. 3(c) for comparison. The image was taken with a refracting objective with NA0.74 (Olympus, UPlanSApo 20X). The intensity profiles in the direction perpendicular to the grating lines are plotted below the two 830 lines/mm grating images to show that the NLIR imaging with chemical specificity provides much higher contrast than the bright field imaging with index contrast. In the NLIR images, the actual grating lines appear to be curved owing to the mechanical nonlinearity of the piezo translation we implemented for raster scanning.
To investigate the dependence of the spatial resolution of NLIR microscopy on the probe wavelength, we obtained the NLIR images of polystyrene beads of 1 μm diameter at two different probe wavelengths, 633 nm and 1064 nm. The MIR wavelength was fixed at 3.3 μm for both images, which corresponds to aromatic CH vibrational modes of polystyrene. The results are shown in Fig. 2 , where the field of view is 25 μm x 25 μm with 100 x 100 pixels. We confirmed that those were NLIR images by blocking the MIR and the probe beams alternatively to observe the signal to disappear. Below the figures are the line intensity profiles measured across the bead indicated by arrows. The FWHMs of the intensity profiles are measured 1.2 μm and 1.8 μm for the probe wavelengths 633 nm and 1064 nm, respectively. It ascertains that both images are beyond the MIR diffraction limit ~3.4 μm estimated with the NA0.5 of the objective. And the results clearly show that the spatial resolution increases inversely with probe wavelength, implying a better resolution achievable for shorter probe wavelengths. Assuming the bead as a Gaussian emitter and using the Gaussian deconvolution of the intensity profile with it, we obtain an approximate value on the resolution as small as 0.7 μm with 633 nm probe wavelength. This is well above the value reported with the SIL techniques [7,8]. The ring pattern around each polystyrene bead was caused by the central obscuration of the reflecting objective.
Next, we demonstrate the capability of our NLIR microscopy in high spatial resolution cellular imaging. Pre-adipocytes 3T3-L1 were cultured and used for imaging of a single cell biological sample. In biomedical fields, the cell line has been widely used to study the obesity. Before NLIR imaging of the cell, the sample was imaged first by commercial Fourier-transform infrared (FTIR) microscope (Nicolet iN10, Thermo Scientific). Typically the FTIR images are obtained in such a way that the vibrational spectra over the whole MIR wavelength range are measured on the sample plane pixel by pixel and the spectral absorbance at a specific wavelength is then mapped to each sample position. The measured FTIR image of the pre-adipocyte corresponding to 2850 cm−1 is shown in Fig. 4(b) together with the corresponding bright field visible image (Fig. 4(a)), where the visible wavelength range from 450 nm to 650 nm is used for illumination and the field of view is 200 μm x 200 μm. The FTIR spectrum at a specific location marked with a red cross in the bright field image is also presented in Fig. 4(c). The image clearly shows high signal intensity around the nucleus, which results from high contents of intracellular lipid droplets that have a lot of CH vibrational modes. However, we notice that due to poor spatial resolution any distinctive individual lipid droplet cannot be seen. The spot size for the FTIR measurements was 10 μm, which is indicated as a blue square spot in Fig. 4(a). Although the FTIR microscope system has the smallest spot size of 5 μm, we could not apply that size because of the low signal-to-noise ratio. The globar, which is used as a thermal infrared light source in our commercial FTIR system, is known to be a very weak radiation source .
Finally we present the NLIR imaging result in Figs. 5(a) and 5(b). The field of view is 30 μm x 30 μm with 100 x 100 pixels, which is the maximum travel for each axis of our 2D translator. Due to the limited travel in our current setup, we could not cover the whole dimension of 3T3-L1 pre-adipocytes that extend as large as 100 μm typically. The MIR wavelengths used in the NLIR image measurements were 3.5 μm for Fig. 5(a) and 3.0 μm for Fig. 5(b), respectively, and the visible probe wavelength is 633 nm for both cases. The average incident powers were 30 μW for the probe beam and 5 μW for the MIR beam for both measurements. The modulation frequency of the MIR beam was 1 kHz and the pixel dwell time was 6 ms. Due to the resonant absorption of lipid molecules at 3.5 μm wavelength (or 2850 cm−1 wavenumber), we can clearly see many distinct lipid droplets in different size around the circular shape nucleus in Fig. 5(a), whereas the off-resonant NLIR image in Fig. 5(b) does not provide high contrast enough to clearly distinguish the lipid droplets from the surrounding matters. The smallest droplet which is indicated by an arrow in Fig. 5(a) is as small as 1.2 μm, which is hardly resolvable in conventional infrared microscopes. The lipid droplets in Fig. 5(a) appeared as an annular distribution with no detailed structures in Fig. 4(b). The two NLIR imaging results demonstrate that our NLIR microscopy can be a useful means of high spatial resolution cellular imaging with chemical specificity. A differential interference contrast (DIC) image of 3T3-L1 pre-adipocyte was measured for comparison with the NLIR images as shown in Fig. 5(c). Even though the lipid droplets are also visible in the DIC image, its contrast is not as pronounced as the NLIR image and no spectral information is contained in it. We now would like to comment on the cell viability after NLIR imaging. We did not expect a photo damage because of very low incident laser power less than 50 μW. In actual measurement, any signal decrease or morphological deformation on the cell like blebbing was not observed after some repeated imaging . We might conclude, from our observations, that there is no apparent cell damage while a cell viability test based on proliferation efficiency needs to be carried out for thorough verification.
A nonlinear optical infrared microscope exploiting thermally induced refractive index change in MIR regime was developed to demonstrate the imaging capability of a single biological cell with high spatial resolution beyond the infrared diffraction limit. The sample was illuminated by an MIR beam and probed by a visible laser beam. The probe wavelength dependence of the spatial resolution was investigated by imaging 1 μm diameter polystyrene beads. We found that the spatial resolution increased inversely with probe wavelength. In the cellular imaging experiment, the details of small lipid droplets in 3T3-L1 pre-adipocytes could be well resolved with our NLIR microscope. We are now in progress for improving the spatial resolution further by increasing the NA of the objective. Since the NA of commercially available reflecting objectives is not so high, we are implementing a counter propagating geometry of the MIR and the probe beams. High NA refracting objectives for the probe beam are easily obtained. Finally we wish to image proteins amide modes, crucial spectral features in biological studies.
This work was supported by the grant from the Bio-signal Analysis Technology Innovation Program of the Ministry of Education, Science, and Technology, Republic of Korea.
References and links
2. J. M. Chalmers, N. J. Everall, M. D. Schaeberle, I. W. Levin, E. N. Lewis, L. H. Kidder, J. Wilson, and R. Crocombe, “FT-IR imaging of polymers: an industrial appraisal,” Vib. Spectrosc. 30(1), 43–52 (2002). [CrossRef]
3. R. Mendelsohn, E. P. Paschalis, and A. L. Boskey, “Infrared Spectroscopy, Microscopy, and Microscopic Imaging of Mineralizing Tissues: Spectra-Structure Correlations from Human Iliac Crest Biopsies,” J. Biomed. Opt. 4(1), 14–21 (1999). [CrossRef]
5. E. Hecht, Optics (Addison-Wesley, New York, 2001). [PubMed]
7. C. A. Michaels, “Mid-infrared imaging with a solid immersion lens and broadband laser source,” Appl. Phys. Lett. 90(12), 121131 (2007). [CrossRef]
8. D. A. Fletcher, K. B. Crozier, C. F. Quate, G. S. Kino, K. E. Goodson, D. Simanovskii, and D. V. Palanker, “Near-field infrared imaging with a microfabricated solid immersion lens,” Appl. Phys. Lett. 77(14), 2109–2111 (2000). [CrossRef]
9. B. Knoll and F. Keilmann, “Near-field probing of vibrational absorption for chemical microscopy,” Nature 399(6732), 134–137 (1999). [CrossRef]
10. M. K. Hong, A. G. Jeung, N. V. Dokholyan, T. I. Smith, H. A. Schwettman, P. Huie, and S. Erramilli, “Imaging single living cells with a scanning near-field infrared microscope based on a free electron laser,” Nucl. Instr. and Meth. B 144(1-4), 246–255 (1998). [CrossRef]
11. E. S. Lee and J. Y. Lee, “Nonlinear optical infrared microscopy with chemical specificity,” Appl. Phys. Lett. 94(26), 261101 (2009). [CrossRef]
12. L. Pálfalvi, J. Hebling, G. Almási, Á. Péter, and K. Polgár, “Refractive index changes in Mg-doped LiNbO3 caused by photorefraction and thermal effects,” J. Opt. A, Pure Appl. Opt. 5(5), S280–S283 (2003). [CrossRef]
13. P. P. Banerjee, R. M. Misra, and M. Maghraoui, “Theoretical and experimental studies of propagation of beams through a finite sample of a cubically nonlinear material,” J. Opt. Soc. Am. B 8(5), 1072–1080 (1991). [CrossRef]