We propose a novel microscopy technique based on the four-wave mixing (FWM) process that is enhanced by two-photon electronic resonance induced by a pump pulse along with stimulated emission induced by a dump pulse. A Ti:sapphire laser and an optical parametric oscillator are used as light sources for the pump and dump pulses, respectively. We demonstrate that our proposed FWM technique can be used to obtain a one-dimensional image of ethanol-thinned Coumarin 120 solution sandwiched between a hole-slide glass and a cover slip, and a two-dimensional image of a leaf of Camellia sinensis.
© 2006 Optical Society of America
Two-photon excited fluorescence (TPEF) microscopy has been widely used for imaging structures and dynamic interactions in biological samples since its introduction because of its low invasiveness and three-dimensional sectioning ability [1,2]. However, live samples are affected by the dye and the staining procedures. The fluorescence proteins, such as the green fluorescent protein (GFP) that are widely used in fusion with the target proteins, also often cause artifacts. Nonlinear microscopic techniques based on second-harmonic generation (SHG) [3,4], third-harmonic generation (THG) [5,6] and coherent anti-Stokes Raman scattering (CARS) [7–9] can be performed without staining and recently have been applied to the imaging of biological objects. In this paper, we propose a new type of nonlinear optical microscopy that is based on the four-wave mixing (FWM) process enhanced by two-photon electronic resonance. The first comprehensive study of the FWM process was carried out by Maker and Terhune  and the Raman resonance including the nonresonant electronic contribution to the FWM process was observed by Levenson and Bloembergen . However, the two-photon electronic resonance in the FWM process was hardly investigated. This may be explained by the fact that the relaxation time of the FWM process under the two-photon electronic resonance condition and the pulse widths at that time were largely mismatched. This mismatch has been solved because a Ti:sapphire laser with a pulse width shorter than 10 fs is now commercially available. As far as we know, the imaging technique based on the FWM process enhanced by the two-photon electronic resonance has not been reported before, particularly under the condition of tight focusing. Since this technique uses, as the output signal, one of the two photons emitted in the parametric process, hereafter we call this new technique stimulated parametric emission (SPE) microscopy. A different but similar technique called seeded parametric FWM spectroscopy has recently been suggested . This technique is based on the two-photon resonant FWM process and requires a phase-matched seeder field formed using a low-numerical-aperture lens. This technique is suggested for spectroscopic purposes but not for imaging. A fast time gate has been realized by the resonant degenerate FWM (DWFM) process . This technique enables the selective detection of collimated ballistic light in a turbid medium. Also note that the signal in SPE microscopy suggested here has already been reported as a source of noise in CARS microscopy . Since SPE and CARS microscopies both use the FWM process, SPE microscopy has the same characteristics in the three-dimensional spatial resolution as CARS microscopy. The coherent nature of the CARS and SPE processes necessarily results in coherent addition of the emitted radiation. Thus the signal beam intensity is proportional to the square of the interacting molecular density and the emission is unidirectional (forward). However, it is important to note that the targets and characteristics of optical pulses in SPE microscopy are usually quite different from those in CARS microscopy and the involved transition moments are entirely different. SPE microscopy does not share the same features as CARS microscopy.
2. The TPEF, CARS and SPE processes
TPEF, CARS and SPE processes are third-order nonlinear processes. Figures 1(a), (b) and (c) show the energy diagrams for the TPEF, CARS and SPE processes, respectively. In the TPEF process, two incident photons are almost simultaneously absorbed to excite the electronic transition from a ground state to an excited electronic state. After thermal relaxation, fluorescence is emitted. The SPE and CARS processes belong to the same class of FWM process [10, 11]. We explain first the SPE process. The SPE process is enhanced by the two-photon electronic resonance of pump pulses whose angular frequencies are ω 1 and ω 2, respectively, and is followed by the stimulated emission of an angular frequency, ω 3, induced by a dump pulse (ω 3). The photon at the angular frequency of ω 4 = ω 1 + ω 2 - ω 3 is generated as one of the two photons when the stimulated emission is induced. We detect the photon at ω 4 as the SPE signal. By contrast, the CARS signal (ω 4) is enhanced when the frequency difference (ω 1 - ω 3) between the pump pulse (ω 1) and the dump pulse (ω 3) is tuned to be resonant with molecular vibration.
FWM intensity is proportional to the squared modulus of the induced third-order nonlinear polarization, P (3)=χ (3) E 1 E 2 E 3, where χ (3) is the third-order susceptibility, and E 1 and E 2 are the pump fields and E 3 is the dump field. The general expression for χ (3) was given by Lotem et al. [9, 14] and the real and imaginary parts are given by
where ω t and Ω are the angular frequency of the allowed two-photon electronic transition and the molecular vibrational frequency, respectively. 2Γt and 2ΓR are the full widths at half maximum (FWHMs) of the two-photon electronic transition and the Raman line, respectively. At and AR are constants representing the two-photon absorption and the Raman scattering cross sections, respectively. The first term in eqs (2) and (3) is the two-photon electronic resonant contribution (Fig. 1(c)). The second term is the vibrationally resonant contribution (Fig. 1(b)). The third term is the nonresonant contribution.
In CARS microscopy, picosecond pulses are normally used because the Raman linewidth 2ΓR in the imaginary part of χ (3) is typically on the order of 10 cm-1 . In SPE microscopy, however, several tens of fs pulses can be used since the bandwidth 2Γt in the imaginary part of χ (3) is often more than the order of several hundred cm-1.
Let us consider a rough indication of the bandwidths of the real part of χ (3). The frequency differences between the central resonant frequency (ω t and Ω) and the frequency at half maximum of the real part are (2 + √3)Γt and (2 + √3)ΓR for SPE and CARS microscopies, respectively. As a rough indication of bandwidth, we may use 2(2 + √3)Γ, since the real part is inversion symmetry with regard to the central resonant frequency. We may conclude that the bandwidths of the real and imaginary parts have the same order of widths. Provided that broadband pulses shorter than 100 fs are used, we can expect that the contribution from the SPE process to the signal intensity is rather dominant in the FWM process and the signal from the CARS process may be considered as the noise component in this case. On the other hand, because of the broad bandwidth of the electronic resonance and the pump pulses, we cannot expect a high spectral resolution in SPE signals. For the discrimination of materials and molecules we may detect several independent spectral bands of SPE signals and then rely on the signal processing techniques for further feature extraction.
With the tight focus of laser pulses, the phase-matching condition for the SPE process is largely relaxed because of the wide cone of wave vectors and the short interaction length . SPE microscopy has the same three-dimensional spatial resolution as CARS microscopy under the condition that the same optical pulses are used, because these techniques are based on the same FWM process and have the same output frequency. However, imaging targets in SPE microscopy are different from those in CARS microscopy.
Provided that the frequency difference ω 1 - ω 3 is far from the Raman line of the sample, SPE microscopy can map distributions of the nonlinear refractive index or two-photon absorption coefficient of a certain electronic-excitation level since the nonlinear refractive index and the two-photon absorption coefficient are proportional to the real and imaginary parts of χ (3), respectively. Noted that not only the fluorophores but also the nonfluorescent molecules can be observed when the molecule has an appreciable electronic resonance. By appropriately shifting the wavelength of the pump pulse from the resonant frequency to the off resonant wavelength, we may expect that the fluorophores can be observed for a long time without photobleaching.
3. Experimental setup
Figure 2 shows a schematic diagram of SPE microscopy system. A mode-locked Ti:sapphire laser operates at a wavelength of 810 nm (ω 1 = ω 2) and a repetition rate of 80 MHz. The Ti:sapphire laser beam reflected from the input surface of the LBO crystal in an optical parametric oscillator (OPO) was used as the light source of the pump pulses for the SPE process. The pulses from the OPO pumped by the Ti: sapphire laser were used as a light source of the dump pulses. The pump and dump pulses were overlapped in time using an optical delay line, and in space by adjusting the angle of a dichroic mirror. The pulses were focused into the sample using a 50× microscope objective (OB1) with a numerical aperture (NA) of 0.55. We observed the TPEF and THG signals along with the SPE signals. The SPE, TPEF and THG signals were collected with the second objective (OB2, NA: 0.5) in the forward direction. After passing through a BG-39 color glass filter, the composite signals were analyzed using a fiber-optic spectrometer.
4. Results and discussion
We first obtained the SPE signal from a dye (Coumarin 120, 4.6 mM) dissolved in ethanol and sandwiched between a hole-slide glass and a cover slip. The central wavelength and spectral width (FWHM) of the dump pulse were 1506 (ω 3) and 20 nm, respectively. The pulse energies of the pump and dump pulses were 600 and 625 pJ, respectively. Figures 3(a) and (b) show the typical emission spectra of the signals from the dye solution and cover slip, respectively. The signals in Figs. 3(a) and (b) that peak at 554 nm (ω 4 = 2ω 1 - ω 3) are the emission spectra of the SPE signals. The signal peak at 502 nm in Fig. 3(b) is the emission spectrum of the THG signal. Figure 3(c) shows a magnified view of Fig. 3(a). We found that a weak TPEF signal is included in Fig. 3(a). The emission spectrum of the TPEF signal ranges from 400 to 500 nm as shown in Fig. 3(c).
The SPE signal intensity second-order dependence on pump intensity and first-order dependence on dump intensity were investigated. The intensities of the SPE signals from the dye solution in relation to the pump and the dump intensities are plotted on a log-log scale in Fig. 4, the slopes are 1.97 and 0.99, respectively. Owing to the nonlinearity, SPE microscopy provides three-dimensional resolution. In TPEF microscopy, an increase in pump intensity results in an increase in signal intensity. However, photobleaching of the fluorophores is induced by the increase in signal intensity. In SPE microscopy, however, signal intensity can be increased by increasing dump pulse intensity without the additional enhancement of photobleaching.
We also confirmed that the signal is not due to the resonant CARS process. Figure 5 shows the dependence of signal intensity on dump wavelength. We found that signal intensity increases with dump wavelength since the photon flux increases at a constant intensity with increasing wavelength. Because the peak in the resonant CARS signal is not found, it is clear that the contribution from the SPE process to the FWM signal dominates. We can conclude that the signal at 554 nm is not the CARS but the SPE signal.
To observe the one-dimensional (1-D) distribution of the SPE signal, the sample was scanned with a step size of 1 μm in the depth (z) direction (See Fig. 2). The mean values in the ranges from 550 to 558 nm (SPE), from 498 to 505 nm (THG) and from 422 to 485 nm (TPEF), were plotted in Figs. 6(a), (b) and (c), respectively. The weak THG signal is observed near the interface between the cover slip and the dye solution, and the large one is observed near the interface between air and the cover slip in Fig. 6(b). The TPEF image as shown in Fig. 6(c) indicates the distribution of the dye solution. By comparison of the SPE image with the TPEF and THG images, we found that the strong SPE signals are obtained in the dye and the cover slip. We conclude that the SPE signal from the dye is enhanced by two-photon electronic resonance because we could obtain the TPEF and SPE signals simultaneously. We also found that the signal from the cover slip has the same signal level as the dye. We consider that the enhancement is due to the coherence effect, because the glass has a molecular density higher than the dye solution.
To confirm the optical sectioning ability of SPE microscopy, we obtained a 1-D signal intensity distribution along the depth (z) direction near the interface between air and the cover slip with a step size of 100 nm (Fig. 7(a)). We estimated the axial resolution by calculating the derivative of the distribution. Figure 7(b) shows the derivative. The FWHM was 3.9 μm.
It is important to evaluate the SPE signal intensity of water because live samples live in and contain water. We measured the dependence of the SPE signal intensity of distilled water on pump wavelength (Fig. 8). We found that the SPE signal intensity of distilled water is minimum near 800 nm and the SPE signal intensity increases with pump wavelength. The increase may be considered to be due to the one-photon electronic resonance enhancement. To reduce the SPE signal intensity of water, we need to use a pump wavelength of 800 nm.
The practical applicability of the SPE imaging technique to biological samples is demonstrated. We observed vascular bundles in the 10 μm cross section of a leaf of Camellia sinensis with pump and dump pulses of 800 and 1150 nm, respectively. Figures 9(a) and (b) show the cross-sectional SPE and transmission images obtained by scanning the sample with step sizes of 0.5 and 0.5 μm in the x, y directions, respectively (See Fig. 2). The SPE and transmission images are composed of the signals in the ranges from 600 to 625 nm and from 790 to 810 nm, respectively. The image size was 60×60 pixels. The pulse energies of the pump and dump pulses were 210 and 100 pJ, respectively. The exposure time for each pixel was 100 ms. Cells in vascular bundles are surrounded by the cell wall that is a rigid layer of cellulose, other polysaccharides, lignins and proteins. To confirm that the SPE signal is enhanced by the two-photon electronic resonance, we obtained a wide-field autofluorescence image (Fig. 9(c)) with the fluorescence microscope under the condition that the excitation wavelength was 436 nm (FWHM: 10 nm). These results suggest that SPE microscopy can detect the cell wall of Camellia sinensis. We consider that we will be able to detect specific components in the cell wall by using multi-band SPE signals in the future.
In summary, we proposed a novel nonlinear optical technique of SPE microscopy that enables us to image distributions of nonlinear refractive index or two-photon absorption coefficient. Provided that broadband pulses shorter than 100 fs are used, we can expect that the contribution from the SPE process to the signal intensity is rather dominant in the FWM process and the signal from the CARS process may be considered as the noise component in this case. With a tight focus of laser pulses, the phase-matching condition for the SPE process is largely relaxed. SPE microscopy has the same three-dimensional imaging characteristics as CARS microscopy. SPE microscopy permits a lower incident average power than CARS microscopy, since shorter laser pulses with a higher peak power can be used. The intensity of the SPE signal can be increased by increasing dump pulse intensity without the additional enhancement of photobleaching. To confirm the optical sectioning ability, the Coumarin 120 solution sandwiched between a hole-slide glass and a cover slip was imaged by SPE microscopy. The practical applicability of SPE imaging to biological samples was also demonstrated. We may identify materials by the spectral content of the SPE signal. The supercontinuum (SC) generated in a photonic crystal fiber has recently been used as a multi-spectral light source in nonlinear microscopy [16, 17] and we consider that the SC can be used as a light source for multiplex SPE microspectroscopy as in multiplex CARS microspectroscopy  and are now working along this research direction.
A part of this experiment was carried out in the Research Center for Superconductor Photonics, Osaka University. We thank H. Sumikura, T. Nagashima, M. Tani, and M. Hangyo for providing the ultrafast lasers and essential assistance. One of the authors (KI) gratefully acknowledges the valuable discussions with K. Ohkubo of Shimadzu Corporation before and after the experiment. This work was partly supported by a grant from the Cooperative Link of Unique Science and Technology for Economy Revitalization promoted by the Ministry of Education, Culture, Sports, Science and Technology, Japan.
References and links
3. Y. Guo, P. P. Ho, H. Savage, D. Harris, P. Sacks, S. Schantz, F. Liu, N. Zhadin, and R. R. Alfano, “Second-harmonic tomography of tissues,” Opt. Lett. 22, 1323–1325 (1997). [CrossRef]
4. Y. Guo, H. E. Savage, F. Liu, S. P. Schantz, P. P. Ho, and R. R. Alfano, “Subsurface tumor progression investigated by noninvasive optical second harmonic tomography,” Proc. Natl. Acad. Sci. USA 96, 10854–10856 (1999). [CrossRef]
5. Y. Barad, H. Eisenberg, M. Horowitz, and Y. Siberberg, “Nonlinear scanning laser microscopy by third harmonic generation,” Appl. Phys. Lett. 70, 922–924 (1997). [CrossRef]
8. A. Zumbusch, G. R. Holtom, and X. S. Xie, “Three-dimensional vibrational imaging by coherent anti-Stokes Raman scattering,” Phys. Rev. Lett. 82, 4142–4145 (1999). [CrossRef]
9. J. X. Cheng and X. S. Xie, “Coherent anti-Stokes Raman scattering microscopy: Instrumentation, theory, and applications,” J. Phys. Chem. B 108, 827–840 (2004). [CrossRef]
10. P. D. Maker and R. W. Terhune, “Study of optical effects due to an induced polarization third order in the electric field strength,” Phys. Rev. 137, A801–A819 (1965). [CrossRef]
11. M. D. Levenson and N. Bloembergen, “Dispersion of the nonlinear optical susceptibility tensor in centrosymmetric media,” Phys. Rev. B 10, 4447–4464 (1974). [CrossRef]
12. M. J. Fernee, P. E. Barker, A. E. W. Knight, and H. Rubinsztein-Dunlop, “Infrared seeded parametric four-wave mixing for sensitive detection of molecules,” Phys. Rev. Lett. 79, 2046–2049 (1997). [CrossRef]
14. H. Lotem, R. T. Lynch, J, and N. Bloembergen, “Interference between Raman resonances in four-wave difference mixing,” Phys. Rev. A 14, 1748–1755 (1976). [CrossRef]
15. J. Cheng, A. Volkmer, L. D. Book, and X. S. Xie, “An epi-detected coherent anti-Stokes Raman scattering (E-CARS) microscope with high spectral resolution and high sensitivity,” J. Phys. Chem. B 105, 1277–1280 (2001). [CrossRef]
16. H. Kano and H. Hamaguchi, “Vibrationally resonant imaging of a single living cell by supercontinuum-based multiplex coherent anti-Stokes Raman scattering microscopy,” Opt. Express 13, 1322–1327 (2005), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-13-4-1322. [CrossRef]
17. K. Isobe, W. Watanabe, S. Matsunaga, T. Higashi, K. Fukui, and K. Itoh, “Multispectral two-photon excited fluorescence microscopy using supercontinuum light source,” Jpn. J. Appl. Phys. 44, L167–L169 (2005). [CrossRef]