Pressure Sensor Description

Pressure Sensor Project Description

Introduction

The pressure sensor fabricated at the University of Louisville as part of ECE544: Introduction to Microfabrication/MEMS was originally designed as a part of an NSF-funded interdisciplinary course in microfabrication and MEMS.  The goal of the laboratory course and companion lecture course is to provide an introduction to microfabrication techniques used in the fabrication of IC and MEMS devices.

Basic Theory

This pressure sensor operates on the basis of measuring stresses produced by the deflection of pressure sensitive, thin monocrystalline silicon diaphragm.  Diaphragm stress is measured using piezoresistive elements along the edges of the deflecting diaphragm.  These piezoresistive elements are created through the diffusion of boron impurity atoms into regions of localized high stress on the diaphragm surface (along the edges). These boron impurity atoms become substitutional elements within the silicon crystal lattice. The boron impurity atoms, in conjunction with internal stress within the diaphragm, produce an anisotropic mobility of electrons in the lattice resulting in resistivity that depends on the current direction within the lattice and internal stresses within the surface of the diaphragm.

The internal stresses created in the crystal lattice create a shift in the band-gap energy, resulting in a change in resistivity. It should be noted that this is not how a typical metallic strain gauge works. A metallic strain gauge works by measuring a change in resistance caused by the actual deformed geometry of the resistive element. This change in the geometry of the resistive element is actually negligible in the model of the piezoresitive element because the distortion is insignificant for that effect to be quantified.

Basic Sensor Fabrication

The construction of the sensor is based on 8 main process steps.

  1. Preliminary cleaning and oxidation
  2. Producing the oxide windows for diffusion
  3. Diffusion of boron
  4. Creating diaphragm oxide windows
  5. Machining the diaphragm
  6. Formation of the contact windows for the piezoresistive elements
  7. Sputtering aluminum and back substrate bonding
  8. Formation of the metal traces from the piezoresistive elements to contact pads.

Packaging and testing follow the wafer fabrication process.

The wafer begins processing as a single double-side polished 4” wafer.  The wafer is first base cleaned to remove organic, inorganic, and metallic contamination. After cleaning, an oxide film that will act as a masking layer for several of the process steps is thermally grown on the wafer through a wet oxidation process at 1100C for 1.0 hrs.

The next step in the process is opening “windows” in the oxide for thermal diffusion. A photosensitive polymer, known as photoresist, is spun on the wafer.  The wafer is then aligned to a photolithographic mask is via the wafer flat  and is exposed to UV radiation.  This radiation changes the polymer’s cross-linked structure allowing the wafer’s photoresist to be developed in the image of the mask pattern.  The photoresist then acts as a mask to open windows in the oxide via submersion in a BOE (buffered oxide etch).  These "dog bone" openings will permit boron to diffuse selectively into the silicon substrate and form piezoresistive structures along the diaphragm.

The boron diffusion consists of a constant-source diffusion of 30 min at 1000C, followed by a BOE soak, followed by a 30 min limited-source diffusion/dry oxidation at 1000C, followed by a BOE soak, followed by a 1100C dry oxidation for 0.7 hrs.

The diaphragm is created on the bottom side of the wafer through an anisotropic bulk micromachining process until the diaphragm is approximately 20 microns thick. A process similar to the photolithography above is used to produce a large rectangular window in the oxide on the bottom of the wafer.  This window is aligned to the piezoresistive features on the wafer front using a front to back mask aligner. The window opened in the oxide allows for selective access to the substrate for the anisotropic etchant.

After the diaphgram is created, it is now necessary to place contacts to the piezoresistive elements.  Photolithography is used to create contact windows to the substrate through the oxide formed during the diffusion process. These windows are located within the ends of the piezoresistive "dog bone" structures and at several locations for "grounding" the substrate.

An aluminum alloy is then sputter deposited across the entire substrate.  Some deposited aluminum fills the contact window openings and allows for contact to be made to piezoresistive elements.  The rest of the aluminum on the substrate is then patterned to serve as interconnect between the piezoresistive elements and corresponding aluminum bonding pads on the substrate.  The aluminum alloy is RF sputtered onto the wafer surface at 250 Watts for 10 minutes.  Photolithography is then used to protect the aluminum in regions where the interconnects are required. An aluminum etchant is used to remove the unwanted aluminum and form the interconnect pattern.   This step is followed by a soak in a piranha dip for photoresist removal.

After this stage of processing test structures on the wafer are evaluated. If the sensor is built within acceptable parameters, the wafer is electrostatically bonded to a glass substrate to create a chamber for referencing the pressure. The wafer is then diced and working individual die are epoxy mounted into a Dual In-line Package (DIP), and wire bonded.

Final testing of the chip occurs in a pressure chamber. Evaluations of the actual sensitivity and linearity of the sensor are made.

The layout of the device can be downloaded in L-Edit. Pressure Sensor Mask.