Defect in Digital Micromirror Device

Defect in Digital Micromirror Device

Material Name: Silicon
Record No.: 118
Primary Chemical Element in Material: Si
Sample Type: Wafer
Uses: Etching
Etchant Name: None
Etching Method: Dry etching
Etchant (Electrolyte) Composition: No data
Procedure (Condition): No data
Note: Figure 1 shows just such trapped carbon-based material that caused the final image to be less than acceptable. As light travels through the window, a shadow of the material is cast upon the surface of the mirrors. This shadow is also transmitted back to the window in reflection, and therefore is a problem in at least two occurrences. If the window is observed using a directional light source that is approaching the unit at an angle similar to that used in the actual application, a pair of dark or bright spots will be observed. The window must be removed mechanically before further analysis is possible within the device. This is accomplished using methods dependent upon the type package used. Grinding through seam welds is the most common method. Extreme care must be taken to prevent grinding into the operational cavity, as this will distribute metal particles freely throughout that cavity and cloud the analysis.

The superstructure required to build a DMD is a 3-D array of metal and hollow space. Fig. 2 shows this as seen from an extreme viewing angle.

Removal of the mirrors can be carried out by several methods, but the most common method is exquisite in its simplicity. Common transparent adhesive tape is applied carefully to the surface of the mirrors with a light amount of pressure, then removed in one motion. With the tape come the mirrors, breaking the hinges in the process. The view of the substructure surface is now unobstructed, and on the tape we removed are all of the mirrors and yokes available for inspection in an inverted position as in Figure 4. This image shows the entire area of mirrors removed using the tape, and an expanded view of a single yoke/mirror assembly. This is the view offered by inspecting the rear of the mirrors, usually with a low accelerating voltage in a Field Emission SEM.

This allows us to inspect the lower surfaces of both the yokes and mirrors. If reasonable care is taken, no damage of the substructure surface is observed. Removal of only the mirrors without damaging the yokes requires a bit more finesse, and also an isotropic dry etch. While this is still a global approach much like the transparent tape method, it will fold the mirrors and leave the yoke assembly available for inspection and analysis. This approach is seen in Figure 6. Similar results are obtainable using a laser-cutting tool for removing one or more mirrors. There is some variation in laser power between units, and therefore each setup must be worked with to obtain the optimum power and spot size for single pulse laser mirror removal.
Reference: Cary Davis,Wes Mahin, and Becky Holdford, Failure Analysis of the Digital Micromirror Device, ISTFA 2002, Proceedings of the 28th International Symposium for Testing and Failure Analysis, 3-7 November 2002, Phoenix Civic Center, Phoenix, Arizona, pp. 291-294.

Figure 1: Carbon material which has dried on the surface of the window.

Figure 2: Typical pixel as seen from extreme viewing angle.

Figure 4: Tape after mirror removal.

Figure 5: View of material trapped behind the mirror imaged using High Acceleration Imaging.

Figure 6: Mirror reduced using RIE isotropic etch.