We report on a novel chromatic confocal microscope system using supercontinuum white light generated from a photonic crystal fiber. The chromatic aberration of a pair of singlet lenses is employed to focus the different spectral components of the supercontinuum at different depth levels. An effective depth scanning range of 7 μm is demonstrated. The corresponding depth resolution is measured to be less than 1 μm (FWHM).
©2004 Optical Society of America
Confocal microscopy [1–3] is a powerful technique for visualization of 3D structures due to its unique ability to optically section a thick sample. It has found a wide range of applications in engineering and life sciences [2, 3]. A conventional confocal microscope requires a point to point 3D scanning and therefore has a relatively slow speed. Various schemes have been proposed to address this problem. For example, the Nipkow disk [4, 5] improves the scanning speed dramatically. Another potential technique is the chromatic confocal microscopy [6–12], which takes advantage of the chromatic aberration of the optical system. Since different spectral component of the light source is focused at different depth levels, the depth scanning can be effectively realized by analyzing the spectrum of the light reflected from the sample. In an early paper by Molesini et al. , a chromatic profilometer was proposed. In Refs. [7–9], the chromatic aberration of the microscope objective lens was employed in the design of a chromatic confocal microscope. Recently, a research team at UCSD developed a system based on diffractive Fresnel lens [10–12]. They also demonstrated a non-translational confocal microscope by using a DMD (Texas Instrument) in conjunction with the Fresnel lens. While system performance depends on the method to get controllable chromatic aberration, which prior research efforts primarily focused on, it can also be limited by the broadband light source. Among the often used, Xenon lamps have low spatial coherence and therefore low illumination efficiency. Tunable lasers have speckle noise. Diodes do not have enough bandwidth and hence have only limited scanning range. Recently, supercontinuum light with huge bandwidth was observed when a short pulse is coupled into a highly nonlinear photonic crystal fiber [13, 14]. In this paper we describe a chromatic confocal microscope system using the supercontinuum light. The chromatic aberration of a pair of singlet lenses is used to obtain an effective depth scanning range of about 7 μm. The high spatial and low temporal coherence of the supercontinuum light result in high illumination efficiency and no speckle noise, which can improve the signal to noise ratio and potentially the scanning speed.
Figure 1 shows the schematic diagram of our experimental setup. Femtosecond pulses from a mode locked Ti:Sapphire laser (KM Labs) is coupled to a 7 cm photonic crystal fiber (Crystal Fibre NL-2.0-770, http://www.crystal-fibre.com) to generate the supercontiuum white light. A typical spectrum measured by an Ando optical spectrum analyzer (AQ6315E) is shown in Fig. 2. Although the supercontinuum has a very broad bandwidth (more than 1000nm), we only used the visible part in our experiment mainly because the alignment is easier and the sensitivity of the CCD camera is higher. Recently a spectrally much flatter (< 6dB over a wavelength range of about 1000nm) supercontinuum white light system is also available (http://www.blazephotonics.com). The collimated supercontinuum light is then focused by a singlet lens with nominal focal length F1 = 25.4mm. The focal length F of a singlet lens is given by the following ,
where R1 and R2 are the radius of the two spherical surfaces, and n(λ) is the refractive index at wavelength λ of the material from which the lens is made. It is straight forward by differentiating Eq. (1) to show that the change of the focal length due to chromatic dispersion of the material is as follows,
Therefore, different spectral components of the supercontinuum are continuously focused to slightly different position with red spectrum at the farther end (longer focal length) and the blue spectrum at the nearer end (shorter focal length). This line of focused light is subsequently imaged onto the sample by an imaging system which consists of another singlet lens with nominal focal length F2 = 19mm and a 100X achromatic oil immersion objective lens (Edmund Optics) with N.A. = 1.25 and effective focal length F3 = 2.9mm. The function of the imaging system is two-fold. Firstly, it can achieve the necessary lateral resolution. Secondly, the additional singlet lens can further increase the chromatic aberration of the system, and consequently, the depth scanning range. The relative position δz, at which different spectral component of the supercontinuum is focused, is given by the following,
where noil is the refractive index of the immersion oil. Light reflected from various depth position of the sample is collimated, reflected by a beam splitter, and focused by a 20X objective lens. After going through a pin-hole with a nominal diameter of 1 μm and re-collimated, it is dispersed by a grating (Edmund Optics, 1200 grooves per mm) and collimated by a lens. Finally, its spectrum is captured by a 14 bit CCD camera (Apogee 32ME). In our experiments, only the visible part of the supercontinuum spectrum was used. If we look at a particular spectral component, the system behaves just like a conventional confocal microscope. Only the light reflected from a particular location in the sample, i.e., the location where this spectral component is focused, can make through the pin hole and therefore be detected. Since each spectral component of the supercontinuum deals with a different depth position of the sample, we can obtain the depth information of the sample in parallel by simply recording the spectrum of the reflected and pin-hole filtered light. Therefore, the 3D structure information of the sample can be obtained by scanning only in the lateral (transverse) directions.
In order to characterize the performance of our system quantitatively, we first used a mirror as the test sample. By moving the mirror in the axial (longitudinal) direction (away or towards the lens), we captured a sequence of spectrum. The result is shown in Fig. 3, in which each row of the image is a spectrum of the reflected light with the corresponding mirror position specified by the vertical reading. The horizontal axis is the un-calibrated wavelength measured by the pixel number on the CCD camera. Longer wavelength corresponds to larger pixel number. Since the red spectrum is focused at the farther end while the blue spectrum is focused at the nearer end to the lens, as the mirror moves away from the lens the blue spectrum will first appear, and then the green spectrum, and finally the red spectrum. Due to the non-uniform spectral distribution of the supercontinuum, there is a portion of the spectrum in the Fig. 3 that remains relatively weak throughout the scanning process. Each column of Fig. 3 represents the depth response curve at a specific wavelength. Figure 4 shows three typical normalized curves at short, medium, and long wavelength. The depth resolution is defined as the full width half maximum (FWHM) of these curves and is found to be between 0.6 μm and 0.75 μm. Longer wavelength has slightly worse resolution. Figure 5 shows for each wavelength the corresponding mirror (depth) position at which maximal reflected light intensity is obtained. This is essentially a mapping between the wavelength and the depth position. The relationship is quite linear and can be fitted by a straight line. The slope of the line is about 8.5 nm/pixel. The depth scanning range is roughly 7 μm as we can tell from the Fig. 5.
We used this apparatus to image a field effect transistor (FET) on a micro circuit chip. The chip was scanned in the lateral (transverse) directions only with a step size of 1 μm. The depth information can be obtained by recording the spectrum of the reflected light as described in detail previously. The images at different depth position are shown in Fig. 6. The image size is 40 μm by 100 μm. For the purpose of demonstration, the chip was tilted slightly such that the left side of the FET is slightly higher than the right side (the 100X objective lens was mounted vertically during the experiment, and therefore higher means nearer to the lens). If we look at the short wavelength (shorter focal length, i.e., nearer to the lens), the left side is on focus while the right side is dark as shown in Fig. 6(a). By increasing the wavelength (pixel number), i.e., the depth, gradually the middle part and then the right side appears. If we keep on increasing, the whole transistor becomes out of focus and therefore disappears as shown in Fig. 6(f). The equivalent depth range from Fig. 6(a) to Fig. 6(b) is about 1 μm and that between Fig. 6(b) and Fig. 6(f) is also about 1 μm. This clearly shows the optical sectioning ability of our chromatic confocal system.
In summary, we developed a reflection type chromatic confocal microscope with an effective depth scanning range of 7 μm and depth resolution less than 1 μm. A photonic crystal fiber is incorporated into the system to generate supercontinuum light. Major advantages of our system include 1) simple system design, a large depth scanning range with just a pair of off-the-shelf singlet lenses and 2) high illumination efficiency, no speckle noise, and therefore high signal-to-noise ratio and potentially allowing higher scanning speed in the lateral directions. As an example, if we focus light from a 100 W Xenon Lamp to a 1 μm2 spot the illumination efficiency is about 10-6 assuming an effective emission area of 1 mm2 (see, for example, specs from http://www.oriel.com), which corresponds to an illumination power of 100μw.On the other hand, supercontinuum light with a power of 100mW can have an illumination efficiency of nearly 100%. Therefore, the signal-to-noise ratio can be potentially enhanced by 1000 times.
The authors thank Chun Zhan for providing the micro circuit chip. Z. L. acknowledges the department of electrical engineering and college of engineering at Penn State for generously providing the start-up fund, by which this work is supported.
References and links
1. M. Minsky, “microscopy apparatus,” U.S. Patent 3 013467 (19 Dec. 1962).
2. T. Wilson, Confocal Microscopy (Academic Press, London, 1990).
3. T. R. Corle and G.S. Kino, Confocal Scanning Optical Microscopy and Related Imaging Systems (Academic Press, London, 1996).
4. M. Petran and M. Hadravsky, “Tandem-scanning reflection-light microscope,” J. Opt. Soc. Am. 58, 661–664 (1968). [CrossRef]
6. G. Molesini, G. Pedrini, P. Poggi, and F. Quercioli, “Focus-wavelength encoded optical profilimeter,” Opt. Commun. 49, 229–233 (1984). [CrossRef]
7. M.A. Browne, O. Akinyemi, and A. Boyde, “Confocal surface profiling utilizing chromatic aberration,” Scanning 14, 145–153 (1992). [CrossRef]
9. M. Maly and A. boyde, “Real-time stereoscopic confocal reflection microscopy using objective lenses with linear longitudinal chromatic dispersion,” Scanning 16, 187–192 (1994).
10. S.L. Dobson, P. Sun, and Y. Fainman, “Diffractive lenses for chromatic confocal imaging,” Appl. Opt. 34, 4744–4748 (1997). [CrossRef]
11. P. C. Lin, P. Sun, L. Zhu, and Y. Fainman, “Single-shot depth-section imaging through chromatic slit-scan confocal microscopy,” Appl. Opt. 37, 6764–6770 (1998). [CrossRef]
12. S. Cha, P. C. Lin, L. Zhu, P. Sun, and Y. Fainman, “Nontranslational three-dimensional profilometry by chromatic confocal microscopy with dynamically configurable micromirror scanning,” Appl. Opt. 39, 2605–2613 (2000). [CrossRef]
14. J.K Ranka, R.S. Windeler, and A.J. Stentz, “Visible continuum generation in air-silica microstructure optical fibers with anomalous dispersion at 800 nm,” Opt. Lett. 25, 25–27 (2000). [CrossRef]
15. See, for example, M. Born and E. Wolf, Principles of Optics (Cambridge University Press, Seventh edition, 1999), Chap. 4 or E. Hecht, Optics (Addison Wesley, Fourth Edition, 2002), Chap. 5.