Optical Cantilever Beam Accelerometer

 The purpose of this research is to invesitgate, design, fabricate, and test a micromachined silicon cantilever beam accelerometer which incorporates a channel waveguide structure on the device which can function reliably in any environment. There have be en reports in the literature of the use of fiber optics and micromachined structures for use as acoustic sensors [J. W. B. Spillman and R. L. Gravel, Moving Fiber-Optic Hydrophone, Opt. Lett, vol. 5, pp. 30-31, 1980; S. Wu and H. J. Frankena, In tegrated Optical Sensors Using Micromechanical Bridges and Cantilevers, SPIE, vol. 1793, pp. 83-89, 1992] .

The following figures show conceptual drawings, both top and side views, of the accelerometer. Light coupled into the channel waveguide from an optical source travels down the beam. The light coupled across the gap is related to the amount of deflection of the beam tip.

Edge and Top Views of Cantilever Accelerometer.

The next figure shows the principle of operation for this device. This is an intensity-type sensor in which light is coupled into the waveguide film from either edge.

Principle of operation of the cantilever beam.

The light in the channel guides is confined in-plane by the refractive index profile of the fillms and guided around the surface by etched ridges in the film which have been patterned by photolithography. Beam deflection is determined optically by a decr ease in the light coupled between the two separated waveguides. Maximum coupling occurs when the two channel guides are aligned. Small displacements on the order of a few hundred angstroms should be detectable.

Standard photolithography and micromachining techniques are being used in the fabrication of this device[K. E. Petersen, Silicon as a Mechanical Material, Proc. IEEE, vol. 70, pp. 420-457, 1982]. Beginning with silicon [100] , double-side polished wafers that are Å 10 mils (254 microns) thick. The wafers are given a standard base clean, consisting of 150 ml of DI water, 25 ml of ammonium hydroxide (NH4OH), and 25 ml of hydrogen peroxide (H2O2). This solution is h eated to boiling and the wafers are soaked in this for 10 minutes. The wafers are then rinsed in DI water and then placed in concentrated hydroflouric acid (HF) for 5 minutes. The wafers are then rinsed in DI water again, and then blown dry with a nitro gen gun. The initial oxidation is done in a lab furnace at 1100 degrees C, in a wet ambient. The bubbler is cleaned with DI water and HF acid and then filled with DI water. This is brought to a boil, and held there ± 2 degrees C during the oxidation gr owth. The oxygen flow rate is set to 0.3 slpm. This oxidation is done for a period of 3 to 4 hours, which provides an oxide layer of 1 to 1.5 microns. This was measured with a Gaertner Dual Mode Automatic Ellipsometer.

Cantilever Beam Fabrication Sequence.

The thermal oxide is removed from one side of the wafers, and then the wafers are placed in wet etching solution of 70 % potassium hydroxide (KOH). This solution is maintained on a closed-loop hot plate at 40 degrees C ± 1 degrees C. The wafers are thin ned to approximately 5 mils (127 microns). This is done to decrease the amount of time that the wafers will spend in the KOH etchant during the beam formation stage, which will help to reduce the amount of inside corner rounding experienced by the beams due to the etching characteristics of the KOH solution. The etchant attacks higher order planes (122, 121, etc.) at a faster rate This affect can be compensated for by using corner compensation techniques [B. Puers and W. Sansen, Compensation Struc tures for Convex Corner Micromachining in Silicon, Sensors and Actuators, vol. A21-A23, pp. 1036-1041, 1990]. When the wafers have been thinned, the remaining oxide is stripped and then regrown.

An isolation layer of silicon dioxide is grown to prevent the guided wave from seeing the silicon substrate. The waveguide material, generally silicon oxynitride, is deposited using low pressure chemical vapor deposition (LPCVD). After the silicon oxyn itride is patterned with the waveguides and etched, a layer of aluminum is deposited which is to be used as the masking layer for the subsequent reactive ion etching. This layer of aluminum is patterned with the beam openings, and the etch is then done t hrough the layers. Once this is done, the aluminum is stripped and the wafer is cleaned and ready for testing.

Reactive ion etching (RIE) is done in a Technics Micro-RIE 85 Series Plasma System.

Schematic of plasma reactor.

The unit is a parallel plate set-up, with the wafer electrode driven by the radio frequency (rf) signal. In this configuration, the fields established when a plasma is struck are perpendicular to the electrode on which the wafers are placed and anisotrop ic etching can occur provided that the sheath electric potentials are sufficiently high to overcome the randomizing effects of gas scattering [R. H. Bruce, Anisotropy Control in Dry Etching, Solid State Technology, vol. pp. 64-68, 1981].

We have done both theoretical and computer modeling of the device, computing coupling efficiencies across the gap using a diffraction integral and with BPM-CAD software, and with two separate waveguide pieces simulating the cantilever.

Coupling Efficiency.

Overlap Integral Calculations.

We have found good agreement between the diffraction integral and the simulation data. With the BPM software, we modeled two cases: rounded edges which are formed with chemical etching, and sharp, vertical edges which are formed with RIE. We found good agreement with the BPM software as well, but only for the sharp profile case. This fact is shown in the next figure. These are two photos; one of a beam formed from chemical etching(L) and the other by simulating a sharp etch profile by using two piece s and aligning the waveguides(R).

Photographs of light coupling across a gap.

Currently, work is being done to form sharp beam edges with RIE. We are using CHF3 with O2 as our gas precursor and are trying to determine the system parameters (flow rates, power, etc.) which will provide the best edge quality. F abrication up to this point has been successful.

A new mask has been designed for fabrication. One new feature added to the mask was corner compensation structures. This is due to the fact that etching in KOH proceeds in silicon at a higher etch rate on higher order planes, such as the <100> plane as compared to the <111> plane. This creates etching at a 54.74 degree angle from the horizontal. Compensating structures have been designed that provides the KOH with a higher order etch plane, and are calculated and sized for a chos en etch depth. When the etching is finished, these structures allow a sharp corner to be formed.

Corner compensation structures.


If you would like more information on this work, please contact jboyd@ece.uc.edu or kburcham@indsci.com



Return to Optical Sensors

Return to OptoElectronics Laboratory