Schematic of the fabrication process flow for epi-seal encapsulation
SEM cross section of episeal encapsulated MEMS devices. (a) MEMS devices are embedded underneath a -thick polysilicon cap layer. Passivation oxide covers this polysilicon cap layer and aluminum is deposited on top of the oxide to form bond pads and electrical connections. (b) A post in the polysilicon cap layer, which provides an electrical via between the aluminum electrical trances on the surface and on the encapsulated device.
(a) The flexure mode beam type MEMS resonators used in this study as pressure sensors inside encapsulation. (b) The pressure vs quality factor calibration curve of the MEMS resonators. This curve was obtained by measuring vented three different MEMS resonators in a vacuum chamber with controlled pressure and calibrating to the hydrogen equivalent pressure. Using this calibration curve, the encapsulation pressure can be measured within 25% measurement error (which corresponds to 15% uncertainty in the quality factor measurement) in the range 5–1000 mTorr.
Data from our previous work (Ref. 2). To investigate the hermeticity of the epi-seal encapsulation, a hydrogen diffusion test was repeated at elevated temperature (300, 350, and ) using MEMS resonators. (a) Quality factor of measured resonators vs time in the nitrogen-filled furnace at different temperatures. (b) Calibrated cavity pressure vs time in the nitrogen-filled furnace at different temperatures.
Gas diffusion behavior through epi-seal. Hydrogen can diffuse in and out of the encapsulation depending on the partial pressures, but the larger nitrogen molecules cannot. The situations in the FGA furnace (a) and the nitrogen furnace (b) are illustrated.
Cavity pressure measurement at . All monitored devices exhibited a pressure increase. In contrast, after the nitrogen purge in the test chamber, the increase in pressure had noticeably slowed, and when the nitrogen supply ran out, pressure increase resumed. During this measurement, some of the electrical connections aged and loosened, resulting in a variance in quality factor measurement. These connections were replaced and fixed in the middle of the measurement. Pressure was calibrated by the data in Fig. 3.
Long-term intermediate temperature hermeticity test setup. Encapsulated resonators were placed in the test chamber, which controlled the environmental temperature. To provide nitrogen environment, the test chamber was sealed by tape and nitrogen gas was pumped into the chamber. The pressure of the nitrogen bottle outlet was set slightly above atmosphere pressure.
(a) A gas furnace used in this study. Test samples were inserted into the furnace while providing various gas environments as well as various test temperatures between 300 and . (b) Measured quality factor data and calibrated pressure of four encapsulated resonators before and after 20 h in a argon-filled furnace. The quality factor and the pressure of all four measured devices remained within the measurement error range, which indicates argon did not penetrate the epi-seal encapsulation.
Hypothetical model of the encapsulation cavity pressure in a nitrogen-filled furnace. Hydrogen already existing inside diffuses out as time goes by, while at the same time nitrogen in the furnace diffuses into the encapsulated cavity at a much slower rate. This nitrogen influx sets the lower limit of the reachable cavity pressure.
Measured quality factor and calibrated pressure of four encapsulated resonators after long-term stay in a nitrogen-filled furnace. After more than 100 h in the furnace, a decrease in quality factor of all resonators was observed. During about 200 h (between 114 and 313 h), the cavity pressure decreased about 90 mTorr from its peak value (calibrated by the data in Fig. 3).
Cross-sectional schematic of various cap layer designs. These variations were patterned after the epitaxial silicon deposition; thus, the underlying MEMS devices and cavity conditions are identical.
Measured quality factor data of various cap layer designs (shown in Fig. 11) for (a) cap thickness of , (b) cap thickness of , and (c) different silicon encapsulation layer thickness (30.7 and ) designs. These devices were placed in a nitrogen-filled furnace, but no difference in the quality factor change rate was detected.
Hypothetical leakage pathway in the epi-seal encapsulation is drawn in dotted lines. Instead of diffusing through the cap layer, gases may leak through the sacrificial oxide, electrical via trenches, and passivation oxide.
Microscopic top-view images of various cap layers prepared for the leakage path trace experiment. The MEMS devices and the cavity are hidden under the polysilicon cap layer.
(a) Measured quality factor data and (b) corresponding calibrated cavity pressure data of various cap layers (shown in Fig. 14). During the same time in a nitrogen-filled furnace, Cap6, which has extra trenches, showed the most dramatic change in quality factor and the cavity pressure. In Cap7 and Cap8, aluminum blocks the leakage pathways; thus, they exhibit a very slow change in quality factor and cavity pressure. Cavity pressure was determined from the quality factor vs pressure curve in Fig. 3.
Atmospheric compositions of dry air (Ref. 26). Molecule species are listed in the order of abundance. Encapsulation permeability (the capability to penetrate the encapsulation to exhibit measurable pressure change at below in tens of hours) through the epi-seal is shown in the last column.
Diffusion activation energy of various gas molecules in a silicon dioxide thin film (Refs. 27–33). The activation energy of neon is in the similar level of hydrogen and deuterium.
Article metrics loading...
Full text loading...