BIPOLAR EFFECTS IN THE FABRICATION OF SILICON MEMBRANES BY THE ANODIC ETCH STOP

Experiments with p-n junctions show that the two-terminal anodic etch-stop configuration commonly used to fabricate silicon membranes actually stops several microns before the metallurgical junction is reached. The reason for this is that the p-region of the thinning membrane acts increasingly as the base of a bipolar transistor. Minority charge carriers injected at the etching surface produce a steadily falling junction impedance as the transistor action builds up. This allows the floating potential of the etching silicon to drift towards its passivation value, and the system current to rise to produce the current peak associated with the etch stop. The peak reflects changes in the etching geometry, and is not a feature of passivation. A four-terminal configuration is used to demonstrate transistor action becoming apparent 100-mu-m before the junction is reached, and increasing as etching proceeds. Precise control over the diaphragm thickness requires control over the potential of the etching silicon. When this is done, an etch stop is obtained at a position in close agreement with the location of the metallurgical junction.