As early as 1970, Arthur Ashkin at Bell Laboratories realized that laser beams' radiation pressure could be used to manipulate dielectric particles . The earliest applications involved levitating particles against an external force such as gravity . The interplay of forces due to counterpropagating beams made it possible to localize particles in three dimensions in a process known as optical trapping. In the early 1980's, Ashkin and coworkers [3, 4, 5] discovered that a single well-focused laser beam could trap sufficiently large dielectric particles in three dimensions, and so was born the optical tweezer. Such single-beam optical gradient force traps can conveniently grab and move particles whose dimensions range from tens of nanometers to tens of microns.
Optical tweezers' ability to manipulate micron-scale particles precisely, nondestructively, and at a distance has found a variety of applications in the biological and physical sciences. Biological applications of optical tweezers have been reviewed by Svoboda and Block . This review focuses instead on their applications in colloid and interface science.
Figure 1 shows a typical optical tweezer setup. The same lens used to focus the laser beam into an optical tweezer can be used to image the particle being trapped. Consequently, optical tweezers are often built into conventional light microscopes. A noteworthy exception is the highly versatile optical tweezer recently made by terminating a tapered optical fiber with a hemispherical lens . The combination of tweezer manipulation, high-resolution imaging, and digital image analysis constitutes a new and powerful class of experimental techniques for probing the structure, dynamics, and interactions of soft condensed matter systems including colloidal suspensions, polymer dispersions, and bilayer membranes.
Despite more than a decade of intense activity, the agreement between theories of optical trapping and experimental force measurements has been considered generally unsatisfactory . Two heuristic lines of argument suggest how optical tweezers work and why quantitative descriptions are difficult.
The energy density stored in an electric field can be minimized by placing a particle with a high dielectric constant at the point of greatest intensity. A small dielectric particle thus is drawn to the waist of a converging laser beam. Provided the particle is smaller than the beam waist, the trapping force is proportional to the volume of dielectric in the beam and thus increases with increasing particle size. Particles small enough that this line of reasoning can be applied are said to be in the Rayleigh scattering regime and range in size from a few nanometers to a few hundred nanometers .
Dielectric particles of a few microns or larger act as lenses in the Mie or ray optics approximation. Each refracted photon imparts an impulse to the particle. The time-averaged vector sum of these impulses draws the particle toward the focal point. The angle of refraction and thus the momentum transfer decrease with increasing particle radius. The trapping force therefore falls off with increasing size for Mie particles.
Particles comparable in size to the wavelength of light scatter light in complicated patterns so that a theoretical description of the particle-light interaction bridging the Mie and Rayleigh domains is difficult. Many optical tweezer experiments operate in this regime, however, because such particles are large enough to image yet small enough to probe interesting microscopic and mesoscopic phenomena. The most advanced theoretical descriptions in this regime have now advanced beyond the paraxial approximation  and include contributions from the light's polarization state [10, 11]. These sophisticated descriptions already are being used to optimize optical tweezers' design .
Despite considerable progress on the theory of optical trapping, practical traps' characteristics still have to be calibrated [6, 13, 14, 15, 16]. Typically, the trapping force is estimated to be the minimum external force required to dislodge the particle. Stokes drag exerted by fluid flowing past the trapped particle has been used most often in this capacity. Practical difficulties arising from distortion of the hydrodynamic flow field by nearby walls  limit the accuracy of such determinations unless particular care is taken.