OPTICAL FABRICATION: Maskless patterning can create micro-optics

Exposure of borosilicate glass to intense laser radiation creates areas in the glass that can be selectively etched to fabricate micron-scale features. This technique may be used to create glass-based optical elements for various microtechnology applications. Rapid prototyping of diffractive optical elements and microchannels for performing ultrahigh-speed capillary electrophoresis separations are some of the applications envisioned by developers Jae Kyung and Nabil Lawandy of the Department of Physics and Division of Engineering at Brown University (Providence, RI).

The selective etching process is based on two-photon excitation of carriers in boron E`-center containing borosilicate glasses. Diffusion and recombination of carriers excited by the two-photon process in mid-bandgapstates of the glass result in regions in which electrons are trapped at certain sites. The researchers exposed a piece of optically polished borosilicate glass with a borate concentration of at least 10% by weight (such as SK5; Schott Glass, Duryea, PA) to focused radiation from a frequency-doubled Nd:YAG laser.

The 532-nm output was created with a KTP crystal, the laser was modelocked at 76 MHz, and the pulse train consisted of 120-ps pulses in a 125-ns Q-switched envelope that itself had 10 pulses. A half-waveplate, quarter-waveplate, and a beamsplitting polarizer controlled the amount of laser power that exposed the glass sample. Mounting the sample on an x,y stage allowed direct patterning with the fixed beam. The glass showed no visible signs of exposure to intense laser light (peak intensities ranged from 1 to 20 GW/cm²) until it was placed in the hydrofluoric acid etching solution.

The researchers affixed the glass to a rapidly rotating mount (3000 rpm) and ultrasonically agitated the solution during etching to eliminate surface build-up of insoluble byproducts thereby increasing the etching depth. Atomic force microscopy revealed that the technique could create structures with depths greater than 3 µm, while their previous efforts only etched about 250 nm into the glass. Although longer etching times did increase the depth, it also contributed to broadening of the features. A polishing step with a slurry of 0.3-µm alumina particles removed rough background surface features; this process only took 10 to 15 small circular passes and could easily be automated.

To test this technique, various-sized lines and holes have been selectively etched. A series of parallel lines were fabricated in glass by illuminating the laser light through a 43X objective and using the two-axis stage to translate the sample back and forth. After polishing, lines were 20 µm apart with a width of 6 µm and a depth of 300 nm; the background roughness was 10-18 nm. According to Lawandy, channels slightly bigger than these could be used in ultrahigh-speed capillary electrophoresis schemes.

Using a spot-generator phase mask, Kyung and Lawandy also created a 4 × 4 array of holes that had dimensions of 50 µm (OD) and a depth of 0.47 µm. The phase mask produced 16 Gaussian spots of equal intensity from a beam at the focal point of a Fourier transform lens. With an average laser power of 750 mW, each spot received about 47 mW. Depending on where the sample was placed relative to the focus, holes with diameters ranging from 14 to 50 µm could be made.

Lawandy believes the technique could be used in rapid prototyping of diffractive optics by direct writing of phase masks with feature resolution of 5 × 5 µm and heights of 500 nm using a computer-controlled x,y stage. An all-optical switch/microchip could be written in a single step into glass by the selectively etching and polishing process. Subsequently spin-coating with photoresist (with a higher refractive index than the glass) and then baking would allow fabrication of integrated optics in glass.

Because the encoding is a two-photon process, linear absorption is nearly negligible and the pattern exists throughout the thickness of glass. Therefore, multiple copies of the same pattern could be obtained by slicing the slab into thin pieces. Lawandy says, "This capability could be important for large-scale production of glass-based devices in which the parallel nature of the encoding process can drastically reduce the production time."