AmericanChemicalSociety.com

A: Particles near fluidic interfaces.

We now have a simplified recipe for generating bi-continuous emulsions (bijels) and methalated silica particles with a fluorescent core, for studies of particlestabilized bijels. While commercial microfluidic droplet generators make droplets in the size range from 20 - 200 microns in diameter, we successfully generated smaller aqueous, monodisperse droplets, in the 0.5 - 5 micron range where optical trapping is optimal, and separated them into a large, low-flow "collection zone."   We were able to intercalate DNA origami with fluorescent dye, and can image these single molecules inside our lab-on-a-chip microdroplet generators.  

B: Trapping “extreme” samples

Optical tweezers use functionalized micron-scale spheres as “handles” for manipulating the micro-world, commonly with an index of refraction of around 1.5, compared to an index of 1.33 for the surrounding medium. Particles with a lower index of refraction than the surrounding medium are stably trapped, but instead are repelled by light. Yet a significantly higher index of refraction leads to increased radiation pressure, which overwhelms the gradient force required for stable trapping in 3D. If the sphere is made of a semiconductor or metal such issues arise: at 1064nm, silicon has n = 3.6 and (bulk) gold has n = 0.26. 

Trapped noble metal nanoparticles hold promise for new forms of fluorescence microscopy: silver nanoparticles enhance fluorescence, and silver spheres with diameters down to 18 nm have been individually optically manipulated. Hence, silver nanoparticles are obvious choices as handles for experiments where a clear fluorescent signal is desirable. On the other hand, the surface plasmon resonance of gold nanoparticles peaks in the visible, and is broad enough to overlap most regions where common fluorophores emit light; so gold can be used to locally quench fluorescence, and these have been trapped with diameters down to 9 nm!

Despite the “rules of thumb” noted in the opening paragraph, in a standard, Gaussian, single-beam gradient trap, gold and silver nanoparticles are not repelled by the beam, but are, in fact, well rapped. One key issue is that in calculating the strength of interaction with optical fields, it is the relative polarizability of the particles, compared to the surrounding medium, that matters, and this appears in a Clausius-Mossetti (or Lorentz-Lorenz) form: namely, it is not a linear function of the index of refraction that we see, but something depending upon the index squared. Because square of the (complex!) index of refraction is equal to the dielectric constant, this suggests that, especially for metallic spheres, instead of thinking about the (real part of the) index of refraction (as is done for the class of materials in common use), it is appropriate to think in terms of the dielectric constant (because the imaginary part of the index of refraction plays an important role for this class of materials). While gold, at 1064nm, has a very low n = 0.26, the real part of the dielectric constant is -48.5, which is a fairly large magnitude. It is this large dielectric constant that results in the very large trapping efficiency of “low-index” gold nanoparticles. – Yet, even after careful considerations, the stability of these (particularly the larger gold nanoparticles) has not been accounted for. Our new work on gold nanoparticles has led to hypotheses that, in turn, have led us to design independent tests that are in preparation.

Semiconducting quantum dots have also been individually trapped. Their narrow emission bands combined with a very low degree of photo-bleaching make quantum dots superior as markers or donors in the study of, e.g., single molecule systems. And we have begun to trap semiconducting nanomembranes, as controllable platforms for many types of sensors or sources (as well as Janus particles of extreme aspect ratio); here, my simulations showed that, because of the relatively high index of refraction, conventional single-beam gradient traps should not be as effective as it would be to exploit radiation pressure, using weakly focused counter-propagating beam arrays.

C: Compensating for optical aberrations in multi-phase and turbid media Generally, increase in complexity of optical systems comes at a cost. For every element added to the optical path there is insertion loss (a reduction in the transmitted intensity). With added complexity, systems also become more susceptible to optical aberrations. Optical trapping and microscopy depend critically on how tight the focus can be made. The high numerical-aperture (NA) lenses typically used create the desired strong gradients in the optical fields, giving both high resolution and trapping strength, but sensitivity to optical aberrations is exponentially worse for higher NA systems. Given that many of our samples, themselves, lead to enhanced optical aberration, efforts to compensate for aberrations became a productive focus of our work.  Minimizing aberrations allows us to maximize trapping strength while minimizing laser power (in the specimen plane), which in turn minimizes heating. 

As light travels through an optically inhomogeneous medium, different rays are delayed by varying amounts, especially when high-NA objectives are used. To remedy this, some authors have applied non-deterministic algorithms to search the phase-correction space to optimize a certain metric such as fluorescence intensity. Those fail in the presence of phase changes with high spatial frequency since only the lower spatial frequencies are probed, due both to limitations of the deformable mirror arrays used and time constraints. Others sample the wavefront of back-scattered light from the focal region using a wavefront sensor, but the back-scattered beam has to traverse the optically inhomogeneous medium a second time in returning from the focal point, and thus does not accurately represent the state at the focal plane. Recently deterministic methods have been developed: Vellekoop et al. optimized light transport through tissue by changing the phase of each pixel of a high pixel count spatial light modulator. This led to the first demonstration of optical trapping through highly turbid media in Nature Photonics in 2010 (in non-turbid media the required laser powers for optical trapping was also reduced by an order of magnitude). We have assessed trade-offs associated with such techniques, adding a method of our own design, submitted for publication.