Constructing an Optimized Optical Tweezers

Optical tweezers utilize the radiation pressure exerted by a tightly-focused laser beam to trap microscopic particles in three dimensions. Such devices are particularly useful in biological applications such as cell manipulation and studying molecular motors. For particles sufficiently larger than the wavelength of light, the trap strength depends on the gradient of the light intensity distribution at the focus. Achieving axial trapping is more difficult than achieving transverse trapping because the axial gradient force has to overcome the scattering force that pushes the particle away from the focus.

In this project, we investigated experimentally how the trapping power of an optical tweezers depends on the intensity distribution of the light entering the microscope objective. We used both Gaussian beams of two widths and annular (optical vortex) beams of order ℓ = 1, 3, 5 and 7. Our tweezers was assembled on an optical breadboard. We used a surplus HeNe laser (P = 38 mW, λ = 632.8 nm) and a surplus Nikon inverted microscope with an inexpensive 50X NA = 0.85 air objective. (An inverted microscope has the advantage that gravity partially counteracts the scattering force, thus increasing trap stability.) We trapped yeast cells 4-6 μm in diameter suspended in a drop of tap water placed on a No. 0 coverslip; we also used 10 μm latex spheres but were not able to trap these. The image of the trapped cells was focused by the objective directly on to a CMOS camera sensor 230 mm away; this distance is larger than the optimum 160 mm due to space constraints. The overall magnification of the imaging system was determined to be 2700X. The trap beam was prepared by first expanding the 1.4 mm diameter laser beam with a two-lens telescope; a third lens brought this collimated light to a focus 230 mm from the objective. To create optical vortices a spiral phase plate (RPC Photonics) was inserted into the beam just before the telescope. The laser beam path also included a dichroic mirror and two turning mirrors. Intensity measurements along the laser beam path revealed that the overall transmission to the sample cell is only about 27%. We plan to replace the metallic turning mirrors and uncoated lenses with higher quality components to improve this number.

We measured transverse trapping forces by the drag force method. A 1 RPM motor was coupled to the translation stage mechanism through a pulley arrangement; the shape of the motor pulley was such that the speed of the stage motion gradually increased as the motor turned. The recorded images were analyzed to determine the velocity at which the particle "fell out" of the trap. Our strongest traps had final velocities of ~100 microns per second, which corresponds to a trapping force of ~4 pN. Quantifying the axial trapping force is more difficult but we did find that a 7.9 mm Gaussian beam (1/e² diameter)didn't consistently trap in the axial direction, while a 9.5 mm beam did. This finding agrees with the customary advice that the trap beam should "overfill" the objective, which in our case was 6.0 mm in diameter. Analysis of our results with the optical vortex beams is still in progress, but we have observed that higher ℓ values reduce the transverse trap strength, in agreement with findings of the Glasgow group (O'Neil and Padgett, 2001).

This work was supported by the Laser Teaching Center and the Simons Foundation. We also thank RPC Photonics Inc. (Rochester, NY) for providing the spiral phase plate.