Pushing the Limits of TEFM - Toward Imaging Biomolecular Networks

The cartoon to the right depicts a hypothetical protein network embedded within a membrane patch, where each shape/color tile represents a particular type of protein. To determine the arrangement of molecules within such a patch requires a combination of nanoscale spatial resolution (a typical protein size is ~10 nm as indicated by the circle), single-molecule sensitivity, chemical recognition, and compatibility with aqueous environments. Optical microscopy, particularly tip-enhanced microscopy, has all the required characteristics in principle, and we have already demonstrated sufficient resolution (see TEFM page) but there are a number of formidable challenges associated with achieving the desired level of performance.

Optimizing TEFM Sensitivity

As the density of molecules in the sample of interest increases, the image contrast decreases (see TEFM page). Since high molecule densities characterize many of the biological samples we are interested in studying, it is critical to find ways to increase near-field contrast as much as possible. The issue of decreasing image contrast led to predictions that only non-linear optical process (e.g., two-photon fluorescence) could be used to image high density samples with tip-enhanced microscopy. However, those predictions did not account for the possibility of improving image contrast (sensitivity) by modulating the near-field signal in some way and then demodulating the fluorescence signal to suppress the far-field background. Contrast vs. Amplitude We developed a modulation scheme whereby the AFM cantilever is resonantly driven such that the tip undergoes rapid vertical oscillations of amplitude 10 100 nm at a frequency of 50 - 300 kHz. These induce modulations in the fluorescence signal as the tip is brought into intermittent contact with the sample, which indeed improves image contrast markedly. The figure to the left shows the measured image contrast for a single quantum dot as a function of the peak-to-peak oscillation amplitude of a silicon AFM probe. The top curve corresponds to using a lock-in amplifier to demodulate the fluorescence signal, while the bottom curve is for no demodulation.

We developed a semi-analytical model to study the image contrast with and without demodulation. This model predicts that the minimum tip-induced enhancement factor needed to detect a single fluorophore in a dense ensemble is proportional to (N/I)1/2n, where I is the far-field laser intensity, N is the number of molecules (or particles/features/etc.) in the laser focus, and n is the order of the optical process (n=1 for 1-photon fluorescence, n=2 for two-photon fluorescence, etc.). This 1/2n scaling is advantageous compared to the unmodulated case, N1/n, and the dependence on I adds a new “knob” to improve detection sensitivity. The model also predicts an optimum oscillation amplitude that agrees very well with experimental measurements (using no free parameters), as shown in the figure above (dashed curve). Together, these results suggest that even linear processes should be sufficient to image dense samples, particularly at higher laser intensities, and provide positive insight into the application of TEFM to biological samples.

Phase sensitive photon counting

To improve image contrast even further, we also developed a novel single-photon counting technique, as depicted in the figure to the right. Briefly, the arrival time of each detected photon is correlated with the instantaneous phase (or equivalently height) of the tip oscillation motion. Among other benefits, this technique makes it possible to apply arbitrarily precise phase-sensitive filters and other analysis algorithms to the photon data, which can yield larger detection sensitivity compared to simple lock-in demodulation. We used this technique to image the orientation of single dye molecules, as well as the dye-labeled ends of the DNA oligomers, as shown on the TEFM page. More recently, we used it to study the three-dimensional nano-optical interaction between a tip and sample and the resonant energy transfer from a quantum dot to a carbon nanotube, as discussed on the Energy Transfer page.

With respect to the performance of different tip materials, we find that metal-coated probes generally yield weaker near-field fluorescence signals due to a competition between field enhancement and fluorescence quenching, compared to high-dielectric materials such as silicon. In metals, fluorescence quenching can totally overwhelm enhancement at small tip-sample separation distances, resulting in a net decrease in the fluorescence signal and thus reduced image contrast. This is especially important when using metal-coated AFM tips, for which there is no well-defined surface plasmon polariton. Specialized tips with highly-engineered shapes such as bowtie antennas or even metal nanospheres attached to the tip of glass fibers, can exhibit strong enough field enhancement via plasmon polaritons to overcome quenching. On the other hand, we have shown that commercial silicon tips do not exhibit strong quenching, and can yield larger net signal enhancement and thus larger image contrast. Out-of-the-box silicon tips also have the advantage of being sharper than both metal-coated tips and most specialized metallic probes, thus yielding finer spatial resolution.

Operation in an Aqueous Environment

TEFM in water

Imaging in water presents a number of challenges. First, AFM operation becomes more tricky in liquids due to strong damping of the cantilever oscillations. Furthermore, illuminating the tip with vertically polarized light as required for field enhancement is more difficult in liquids, particularly when illuminating from below through a glass coverslip (episcopic configuration) as is common for single-molecule imaging. Finally, soft samples in an aqueous environment can be easily perturbed by the oscillating motion of the tip. To date, we have used metal tips to achieve negative (quenching) contrast in water for fluorescent beads on glass, as shown in the figure to the left (scale bars = 200 nm). Although silicon tips have thus far not produced strong near-field contrast, we are designing a new illumination scheme whereby a laser beam mask, a pockels cell, and a high numerical aperture objective lens are used to generate a tightly focused evanescent field whose polarization can be quickly dithered between vertical and transverse. This will produce strong modulation of the enhanced field and associated fluorescence signal without large amplitude oscillations of the tip. Thus, this illumination scheme should increase near-field contrast for silicon tips in water while minimizing damage to soft samples.

Potential Biological Systems

In collaboration with Markus Babst’s group in the Biology Department and John Conboy's group in the Chemistry Department at the University of Utah, we are studying a ubiquitous membrane-associated protein network thought to extend over areas of greater than 100 nm in size. This machinery, called the ESCRT network, functions to concentrate membrane proteins marked for degradation, and then package them into topologically inverted vesicles for trafficking to the lysosome, where they are broken down into the constituent amino acid residues and recycled. Much is known about the biochemical signatures of the ESCRT network, but there is no understanding of its detailed molecular-scale architecture and the relationship between this structure and its function. We are working to reconstitute this network on planar-supported lipid bilayers, and then we will use TEFM to image the molecular architecture of the network using the improved illumination scheme described above.

Prof. Jordan Gerton | James Fletcher Building | Room 314 | 115 South 1400 East | Salt Lake City, UT | 84112
Office: +1-801-585-0068 | Lab: +1-801-581-5078 | Email: jgertonphysics.utah.edu