Near field scanning optical microscopy (NSOM) was developed in the mid 1980's as a means to break the diffraction limit on spatial resolution attainable with optical measurement.  Traditional NSOM utilizes a tapered, metal-coated optical fiber with a small aperture as either an excitation source or collection device.  The spatial resolution attainable with this technique is roughly defined by the size of the aperture itself.  Unfortunately, rough metallic coatings on glass fibers, coupled with the finite skin depth of metals at optical frequencies, place a limit on the spatial resolution attainable with fiber probes.  Typical spatial resolution for the fiber approach to NSOM is ~50-100 nm, though careful probe and sample preparation have yielded ~30 nm resolution.

     The Xie Group has long been interested in mapping out the spatial distribution of the proteins in photosynthetic membranes.  These proteins are approximately 10 nm in diameter, and electron microscopy indicates that they are closley packed in their native configuration.  The density of this arrangement precludes optical mapping with either far-field measurement or traditional aperture NSOM.

R.C. Dunn et al., J. Phys. Chem. 98, 3094 (1994).

     In the mid-1990's, a number of groups began working on what would prove to be the solution to the fundamental resolution limitations in aperture NSOM.  The idea was to spatially map an optical signal from the surface of a sample by scattering that signal with a solid probe.  The choice of probe ranged from tungsten STM probes to metallized AFM tips to silicon probes.  While optical image interpretation was complicated by topographic coupling, the spatial resolution in the apertureless technique was improved to ~10's of nm.

     In 1999 the Xie group began work on a type of apertureless NSOM that takes advantage of the phenomenon of field enhancement by noble metal probes.  With the appropriate combination of excitation polarization and probe geometry, a gold or silver probe is capable of strongly enhancing the magnitude of the incident illumination.  This strong, localized field can be used as an excitation source for a spectroscopic transition of interest, such as fluorescence or Raman scattering.  By locally exciting a spectroscopic transition in the sample and detecting the Stokes shifted light in the far field, we greatly simplify the interpretation of apertureless NSOM images. 

     We utilize the nonlinear response of the sample by exciting two-photon fluorescence in the sample.  This tip enhanced nonlinear optical microscopy (TENOM) amplifies the contrast.  The enhancements achievable with specifically fabricated gold probes are sufficient to yield near-field two photon fluorescence signal above the background generated in the diffraction-limited far-field region.  Furthermore, TENOM has improved spatial resolution to the sub-20 nm regime.  This high resolution and strong signal generation should be capable of, for example, mapping the spatial distribution of PSI and PSII in the thylakoid membrane.

     While TENOM has presented a solution to issues of high resolution optical imaging, it does have several intrinsic drawbacks.  These drawbacks include the quenching of non-energy-transferred fluorescence, the necessity to carefully tailor the probe geometry to yield high field enhancement, and the practical requirement of a nonlinear signal source.

Topographic (left) and optical (right) images of J-aggregates of PICI dye.  The optical signal
is two-photon excited fluorescence.  Fast energy transfer along the J-aggregate fiber and 
away from the tip reduces the quenching of the fluorescence by the tip.
 

E.J. Sanchez et al., Phys. Rev. Lett. 82, 4014 (1999).
L. Novotny et al., Ultramicroscopy 71, 21 (1998).

     We are currently exploring a variety of approaches to eliminate the few drawbacks of TENOM and to expand the capabilities of apertureless NSOM in general.  Our first concern has been to address the problem of fluorescence quenching by the metal probe.  To this end, we have applied ion- and electron-beam assited deposition (IBAD and EBAD, respectively) of dielectrics to our apertureless probes.  The growth of glass with these techniques allows us to build a spacer layer between the probe and the sample, reducing quenching while still maintaining high field enhancement at the sample.  The theoretical treatment of fluorescence quenching by probes of complex geometries is currently being explored through the use of finite difference time domain (FDTD) simulation (see movies for example)and Poynting vector integration.

     Due to the complex nature of light-metal interactions, electromagnetic simulation is vital in our TENOM and apertureless NSOM research.  The FDTD technique is a flexible, powerful method for fully three dimensional simulation of the response of metal probes to incident radiation.  The power of the FDTD technique lies largely in its conceptual simplicity.  A geometry of interest is defined by a number of cells, called Yee cells.  The complex permittivities of each cell are assigned.  After a geometry has been defined, an incident optical field is simulated, and the Maxwell's equations are integrated in a leapfrog fashion.  The result is a method that can calculate the resultant fields for any arbitrary geometry illuminated with optical sources.

     FDTD has been applied to the study of NSOM in the past.  Unfortunately, the large computational burden associated with this technique made three dimensional simulations quite difficult.  Two dimensional calculations (e.g. R.X. Bian et al., Phys. Rev. Lett. 75, 4772 (1995).) have generated insight into the aperture method of NSOM.  We have utilized a commercial FDTD package (RemCom's XFDTD) to simulate apertureless probes in three dimensions, working to adapt this microwave-focused package to the complex permittivities intrinsic to the optical regime behavior of real metals.  This commercial software, coupled with the recent tremendous improvement in computational power, has provided a general method for simulating complex geometries, such as groups of spheres.  (See movies for examples.)

     In order to expand apertureless NSOM to linear signal sources, intensity enhancements on the order of 10,000 are required in our experiments.  In order to develop a physical intuition for the nature of strong field enhancement, and to rationally design probes for high field enhancement, we have used FDTD to model apertureless NSOM.  Our simulations indicate that probes with finite size on the scale of the excitation light are preferable to the "quasi-infinite" tips we have used in the past.  Furthermore, by properly tuning the size of a right trigonal pyramid to the excitation light, we predict field enhancements sufficient for exciting linear signal sources and yielding signal above the far field background.  (See images, right.)  The simulated high field enhancement has been experimentally verified.

     We use a focused ion beam instrument to make apertureless NSOM probes.  This instrument (FEI Strata DB-235) allows us to mill materials with ~10 nm resolution, image with ~5 nm resolution, and deposit glass or platinum with ~20 nm resolution.  We are utilizing this powerful tool for nano-optics and near field microscopy applications, including surface enhanced Raman scattering (SERS) and chemical sensors.