The diffraction limit in conventional microscopy arises from the size of the spot that a light beam can be focused to with normal lens elements. At the focal point, the beam forms a symmetric pattern of concentric rings known as the Airy disk pattern. The dimensions of the Airy disk pattern were first described in detail by Ernst Abbe in 1873. From Abbe, the distance, d, from the highest intensity point located in the middle of the center spot to the first node in intensity is given by
d = 0.61(lo/nsinq) (1)
where lo is the vacuum wavelength, n is the refractive index of the medium in which the light travels, and q is the light convergence angle for the focusing element. The denominator in eq 1, nsinq, is also known as the numerical aperture (NA) for the objective and can be as high 1.3 - 1.4 for modern objectives working in high-index mediums such as water or oil. With the intensity profile of the focused light beam described, the question of resolution arises. How close can two objects reside and still be resolvable by an ideal optical system using a lens?
The accepted practice is to invoke the Rayleigh criterion which states that two objects are resolvable if they are separated by a distance at least equal to the distance given in eq 1. As stated earlier, numerical apertures of 1.3-1.4 are now obtainable with high-quality objective lenses; therefore, eq 1 is usually simplified to d = l/2. The maximal resolution is then approximately equal to half the wavelength of the radiation used, which for visible light applications results in a spatial resolution of 250-300 nm. The arbitrary nature of the definition, however, only provides guidance for determining the optical resolution, and in actual practice, the experimental situation is often quite different. For instance, with excellent signal-to-noise and/or an accurate description of the instrumental response function, smaller distances can, in theory, be determined. More often, however, the experimental conditions and aberrations in the optical components conspire to lower the resolution and prevent attainment of the theoretical limit. In common practice, therefore, the practical limitation on spatial resolution is often worse than the diffraction-limitedl/2.
These spatial resolution limitations have been well known for some time and, not surprisingly, led many to begin exploring alternative ways of achieving higher resolution optical measurements. Early in the 20th century, Synge published a series of visionary papers in which he proposed a new type of optical microscope designed to circumvent the limitations imposed by the diffraction limit. This remarkable collection of papers details the foundations upon which the modern day NSOM is based.
Near-Field Scanning Optical Microscopy (NSOM)
Although technically challenging to implement, Synge's idea to beat the diffraction limit was eloquently simple. Shown schematically in Fig. 1, Synge proposed forming a microscopic aperture with dimensions much smaller than the optical wavelength in an opaque screen. By illuminating the backside of the screen with a high-intensity light source, light passing through the aperture would be confined by the dimensions of the hole. Once positioned in close proximity to the sample surface, the light emerging from the aperture could be used to image a specimen before it had time to diffract out and degrade the resolution. Included in Synge's original proposal was a keen awareness of the technical difficulties that would have to be overcome to construct such a microscope. The difficulties in aperture formation, illumination, and sample manipulation were all recognized as hurdles that would have to be overcome.
The experimental feasibility of high-resolution imaging using a subwavelength aperture was first demonstrated by Ash and Nicholls in 1972 using microwave radiation. With 3 cm microwaves passing through a small aperture, periodic features in a metal grating sample were measured with l/60 spatial resolution. These exciting results illustrated the feasibility of Synge's idea and renewed interest in carrying out similar experiments using visible radiation. However, the much shorter wavelengths associated with visible light imposed technological difficulties in aperture formation and positioning that required another decade to overcome. It was not until the mid 1980s that Pohl's laboratory at IBM Zürich first reported sub-diffraction-limited optical measurements using the ideas outlined by Synge nearly half a associated with sub-diffraction-limited optical imaging and initiated the developmental activity which has resulted in the modern day NSOM instrument. While still not completely routine, high-resolution optical measurements with NSOM are beginning to address important question in a variety of samples. The sensitivity of the technique has been amply demonstrated through single molecule measurements while the high spatial resolution has revealed previously undiscovered features in a variety of samples.
For biological samples, the few NSOM investigations reported in the literature portray a technique posed to make a significant impact once issues involved in imaging soft and often dynamic samples are resolved. While NSOM measurements on fixed cells both dry and under buffered conditions have been reported, the extension to unfixed cells has proven problematic. This mainly results from the forces imparted to the sample during imaging that tend to damage soft and fragile samples. Once this hurdle is overcome, however, the potential for measuring the structure and dynamics of living cells at the nanometer level offers exciting possibilities.
The heart of any near-field microscope lies in the quality of the aperture used to deliver the nanometric spot of light. Early NSOM tip designs included etched quartz crystals and pulled micropipettes, but these tips generally suffered from low throughput and poor reproducibility. By far, the most successful NSOM tip design to date was introduced by Betzig et. al. at AT&T Bell Labs and makes use of a tapered fiber optic waveguide coated with a reflective metal coating as shown in Fig. 2.
Fiber optic NSOM tips are fabricated by heating and pulling a single mode optical fiber down to a fine point in a commercial micropipette puller. By controlling the heating and pulling parameters, highly reproducible tapers and tip diameters can be fabricated. In general, throughput for a given aperture size decreases as the aspect ratio increases. As the diameter of the fiber in the taper region is reduced beyond the mode-field cut-off of the waveguide, light escapes from the sides of the tip which prevents the formation of a well-defined aperture. To form an aperture, the sides of the probe must be coated with an opaque metal to confine the light. For visible radiation, aluminum has the smallest skin depth (~13 nm at 509 nm) and, therefore, requires the least amount of coating. This, however, is complicated by the propensity for grain formation in aluminum which decreases the reflectivity of the film. To block light from escaping, approximately 50-100 nm of aluminum is coated around the sides of the taper region. A schematic view of an aluminum-coated optical fiber tip is shown in Fig. 2 along with a high resolution SEM image of a typical NSOM tip fabricated in our laboratory.
The power output and throughput efficiency of these probes is highly dependent on the particular parameters of the tip. Typically, however, tips with 80-100 nm diameter apertures can deliver tens of nanowatts of light with hundreds of microwatts coupled into the fiber. Although restrictive for some applications, nanowatts of output power is still relatively high considering the small size of the aperture. The power density exiting the near-field tip is on the order of 100 W/cm2 which has proven adequate for many applications including single molecule detection. The inefficiency of the tips results from the transition from propagating to evanescent waves as the diameter in the taper region decreases beyond cutoff. The light lost is either reflected back up the fiber from the taper region or is lost by absorption into the aluminum coating. The heating from the absorption of light by the aluminum coating ultimately limits the output power available from a tip. Local heating near the end of the probe can damage the aluminum coating thermally or as a result of stresses introduced through the differential thermal expansions of the glass and aluminum. In either case, the tip aperture is usually damaged in the process, rendering it unsuitable for high-resolution imaging.
To obtain high-resolution optical images with NSOM, the tip must be positioned and held within nanometers of the sample surface during scanning. Various feedback mechanisms based on electron tunneling, photon tunneling, impedance, and reflection measurements have been introduced to accomplish this precise positioning. One of the most widely adopted is the shear-force technique introduced independently by two groups in 1992.
In the shear-force method, the NSOM tip is dithered laterally at one of its mechanical resonances. The amplitude of the vibration is kept low, usually <10 nm, to avoid diminishing the resolution in the optical image. As the tip approaches the sample surface, shear forces acting between the tip and the sample dampen the amplitude of the tip vibration. This drop in amplitude normally occurs over a range of tens of nanometers from the sample surface. The amplitude can be monitored by several methods and used to generate a feedback signal to control the tip-sample gap during imaging. This not only keeps the tip in close proximity of the sample to obtain high resolution in the optical measurement (Fig. 2), but also results in a force mapping of the sample surface much like AFM. Therefore with NSOM two images are collected simultaneously – a high resolution fluorescence image and a force image which maps the topography of the sample.
NSOM Microscope Design
The final near-field microscope design can take on several forms depending on the particular needs of the research. The microscope design utilized for NSOM measurements in our laboratory is shown in Fig. 3. In this design, the NSOM is built atop an inverted fluorescence microscope. This is particularly well suited for biological applications in which the normal imaging modes of the inverted microscope are still required to locate and study the sample before performing the higher resolution NSOM experiments.
As shown in Figure 3, laser light is passed through a band-pass filter to remove unwanted colors followed by a combination of half-wave and quarter-wave plates to control the polarization of the light. This light is then coupled into a single mode optical fiber, the end of which is fabricated into the NSOM tip. In the arrangement shown, the near-field tip is mounted in a z-piezo tube which adjusts the tip-sample gap during scanning. The transparent sample is mounted in a separate x - y piezo stage which scans the sample under the tip during the experiment. Light exiting the NSOM tip excites fluorescence in the sample which is collected from below using a high numerical aperture microscope objective. Residual laser excitation light is removed using filters, and the remaining fluorescence signal is imaged onto a high quantum efficiency avalanche photodiode detector operating in single photon counting mode. The software and electronics necessary to scan the piezos and record images are similar to that used in atomic force microscopy.
Despite minotip used as the excitation source and/or collection element and a mechanism for positioning and holding the NSOM tip within nanometers of the sample surface during scanning. It is these that we propose to dramatically alter and thus open the possibility of conducting NSOM measurements on viable biological specimens. Before we discuss this, however, it is
Single Molecule Detection
Single molecule detection and spectroscopy using both near-field and far-field techniques have generated much interest in the physical and biological sciences. These experiments avoid complications from ensemble averaging inherent in bulk measurements and allow for an unprecedented view into the individual species which encompass an entire population.
Single molecule NSOM measurements also provide one of the most convincing demonstrations of resolution. Because an individual molecule is so much smaller than an NSOM tip, single molecule fluorescence images provide an excellent diagnostic of the actual NSOM aperture. For example, Fig. 4 shows a single molecule fluorescence image taken in our laboratory using NSOM. The sample consisted of the fluorescent membrane probe DiIC18 dispersed in a lipid monolayer of DPPC. Each bright spot in Fig. 4 represents the fluorescence from a single dye molecule dispersed in the DPPC matrix. This is evidenced by the coverage which agrees with calculations, the observation of a well-defined transition dipole moment, quantized photobleaching, and the blinking behavior that is ubiquitous in single molecule studies. For example, the molecule in the bottom of the image is seen to blink off for a scan line (horizontal direction) before returning to the emissive state for the rest of the scan. A linecut through the top molecule is shown above the image. The full-width-half-maximum of the linecut is approximately 28 nm which provides a good measure of the particular NSOM aperture used in carrying out these experiments and the resolution attainable with that tip. It should be stressed that this spatial resolution is almost and order of magnitude better than that possible with state-of-the-art confocal microscopy under similar conditions. One drawback, however, is in reproducibility. As seen in Fig. 2, structures near the end of the fiber tip due to the aluminum coating are often complicated. These grains produce several undesirable effects. The structures physically restrict the approach of the tip to the sample, thus limiting the attainable resolution and lowering the intensity seen at the sample. They also reduce the symmetry of the probe aperture and complicate the polarization properties of the light delivered with the tip. This has led several groups, including ours, to explore new ways in which the precise geometry at the end of the NSOM tip can be controlled to provide enhanced performance.
Dunn, R. C., “Near-Field Scanning Optical Microscopy” invited review article for Chemical Reviews, 99, 2891-2927 (1999).
Shiku, H. and Dunn, R. C., “Near-Field Scanning Optical Microscopy”, invited A-page article for Anal. Chem., 71, 23A-29A (1999).
Lee, M. A., Talley, C. E., Vickery, S. A., Krogmeier, J. R., Hollars, C. W., Shiku, H., Dunn, R. C., “Progress Towards Imaging Biological Samples With NSOM”, SPIE Proceedings in Laser Techniques for Condensed-Phase and Biological Systems, 3607, 60-66 (1999).
Talley, Chad E., Cooksey, Greg, and Dunn, Robert C., “High Resolution Fluorescence Imaging With Cantilevered Near-Field Optic Probes”, Applied Physics Letters, 69, 3809-3811 (1996).