While it is always possible to use a measured etch rate and a specified time to determine the depth of etched features, it is highly desirable to be able to use a well-defined structural feature to stop an etch. There is a rich array of possible etch-stop structures in use.
Electrochemical Etch Stop
- When a silicon wafer is biased with a sufficiently large anodic potential relative to the etchant, it tends to oxidize.
- This process is called electrochemical passivation, because if there is oxide on the surface, the oxide masks the etch.
This passivation step is a redox reaction and requires current. If the supply of that current is blocked with a reverse biased pn junction, the passivation cannot occur and the etching can proceed.
The use of this electrochemical property to control the thickness of an etched diaphragm is illustrated in Figure above.
- A p-type wafer has a diffused n-region at its upper surface. The depth of the junction can be accurately controlled by the combination of ion implantation and drive in.
- The backside of the wafer is covered with a masking layer that contains an opening.
- When this wafer is placed into the anisotropic etchant with an anodic bias applied as shown, the reverse biased p-n junction prevents the passivation current from flowing.
- As a result, the p-silicon remains unpassivated and hence etches.
- However, the n-silicon becomes passivated and is not etched.
- Furthermore, as anisotropic etching proceeds, when the junction is reached there is no longer anything blocking the flow of passivation current.
- The n-silicon quickly passivates, terminating the etch. The resulting diaphragm thickness is equal to the depth of the original junction, at least ideally.
- In practice, it is necessary to control the junction leakage and current supply paths to be certain, first, that there is no pathway that can provide enough current to passivate the p-region, and, second, that once the junction is reached, all features have enough current supply so that they passivate at the same point.
p+ Etch Stop
- If instead of using an n-type diffusion into a p-wafer, a heavily boron doped p+ layer is formed by ion implantation and diffusion, the anisotropic etch will terminate without requiring application of anodic bias.
- This greatly simplifies the tooling required to perform the etch.
- However, the etch-rate selectivity is not quite as high as for passivating oxides.
- Furthermore, the p+ diaphragm will have residual tensile stress that might affect the performance of any device, such as a pressure sensor, that depends on the mechanical properties of structure.
- Finally, it is not possible to diffuse piezoresistors into already-heavily-doped p+ silicon.
- So while the p+ etch stop offers some advantages and can be used with great effect in some processes, it also has some important limitations.
Dielectric Etch Stop
- If the n-layer of Figure above were replaced by a material that is not etched, for example, silicon nitride, then the result of the anisotropic etch is a silicon nitride diaphragm suspended over a hole through the silicon.
- LPCVD silicon nitride is very brittle, and has a very high residual tensile stress, making it prone to fracture.
- However, if the ratio of dichlorosilane to ammonia is increased during silicon nitride deposition, a silicon-rich nitride layer is deposited which has lower stress.
- Membranes of silicon rich nitride can be used as the basis for mechanical devices.
- Masks for X-ray lithography can also be made in this fashion, depositing gold on the membrane to create the masking regions.