This application claims the priority benefit of Italian Application for Patent No. 102023000013011 filed on Jun. 23, 2023, the content of which is hereby incorporated by reference in its entirety to the maximum extent allowable by law.
The present invention relates to a microelectromechanical sensor assembly and a process for manufacturing a microelectromechanical sensor assembly.
In order to improve the quality of voice communication, especially with the use of True Wireless Stereo (TWS) earphones, it has been proposed to exploit the conduction of sound through the cranial bones, in addition to conventional microphones. Bone conduction of vibrations is not affected in fact by the acoustic noise and by external disturbance that are superimposed on the voice of the speaker and that normally are picked up by microphones. The information that may in this way be obtained enables improvement of identification of the useful spectral content in the voice signals picked up by the microphones and, consequently, noise cancellation. The quality of the voice signal transmitted may thus be improved.
The systems currently available use an inertial device purposely dedicated to detection of the voice via bone conduction, for example an independent accelerometer or accelerometer an integrated in an Inertial Measurement Unit (IMU). Of course, the dedicated inertial device is optimized for the characteristics of the human voice, in particular as regards the bandwidth and the noise level. Further inertial sensors are normally used to implement functions that are increasingly often demanded by the market, such as activity recognition or user-interface functions.
Although the desired functions may effectively be made available, known solutions present limitations above all as regards the freedom of design. An independent inertial sensor imposes constraints on the overall dimensions, this possibly being at the expense of the battery and thus of the autonomy of the devices. Furthermore, the use of a number of independent components usually results in an increase in the production costs. On the other hand, it is problematic for the devices integrated in a same IMU to be optimized for functions that require very different levels of performance, and compromises may be reached that not satisfactory.
Consequently, there is a need in the art to provide a microelectromechanical sensor assembly and a process for manufacturing a microelectromechanical sensor assembly that will enable the limitations described above to be overcome or at least mitigated.
Embodiments herein relate to a microelectromechanical sensor assembly and a process for manufacturing a microelectromechanical sensor assembly.
In an embodiment, a microelectromechanical sensor assembly comprises: a semiconductor die having a sealed cavity; a microelectromechanical inertial sensor, having a sensing mass; and a piezoelectric vibration sensor, having a piezoelectric membrane; wherein the sensing mass and the piezoelectric membrane are housed in the sealed cavity.
In an embodiment, a process for manufacturing a microelectromechanical sensor assembly, comprises: forming a semiconductor die having a sealed cavity; forming a microelectromechanical inertial sensor, having a sensing mass; forming a piezoelectric vibration sensor, having a piezoelectric membrane; and housing the sensing mass and the piezoelectric membrane in the sealed cavity.
For a better understanding of the present invention, preferred embodiments are described, by way of non-limiting example, with reference to the annexed drawings, wherein:
With reference to
The processing circuit 2 uses detection signals SBC received from the piezoelectric vibration sensor 10 in order to improve the quality of the audio signal transmitted. The detection signals SBC are generated basically by the voice of the speaker and are without components of environmental noise, being transmitted through the cranial bones. For instance, the processing circuit 2 may use the contents of the detection signals SBC to identify the spectral components of the signal SMIC supplied by the microphone that effectively correspond to the voice, and to filter out the remaining spurious components.
With reference to
The substrate 12 is coated by a separator dielectric layer 18 and by a sacrificial layer 19, for example a silicon-oxide layer, which form an insulating body, where buried conductive paths 20 are embedded.
The supporting body 13 comprises a first portion 13a and a second portion 13b. The first portion 13a is joined to the substrate 12 via the sacrificial layer 19 and the separator layer 18 and laterally delimits a first volume 17a of the cavity 17. The second portion 13b of the supporting body 13 is joined to the first portion 13b on a side opposite to the sacrificial layer 19 and laterally delimits a second volume 17b of the cavity 17, communicating with the first volume 17a. The second portion 13b further functions as support for the piezoelectric vibration sensor 10 and as spacer between the microelectromechanical accelerometer 8 and the piezoelectric vibration sensor 10, which are stacked on top of one another.
The cap 15 is joined to the second portion 13b of the supporting body 13 by bonding structures 21. The cap 15 and the bonding structures 21 delimit a third volume 17c of the cavity 17.
The microelectromechanical accelerometer 8 comprises the sensing mass 8a capacitively coupled in a differential way to fixed electrodes 8b, 8c. See,
The sensing mass 8a is supported by the first portion 13a of the supporting body 13 in the first volume 17a of the cavity 17. In detail, the sensing mass 8a of the microelectromechanical accelerometer 8 is kept suspended at a distance from the substrate 12 by elastic connections known as flexures 22, configured to enable the sensing mass 8 to oscillate along a sensing axis. The first portion 13a of the supporting body 13 is coupled to a respective one of the buried conductive paths 20 for biasing the sensing mass 8a through the flexures 22. The fixed electrodes 8b, 8c are defined by respective plane semiconductor laminas perpendicular to the sensing axis and fixed to the separator dielectric layer 18 and to the substrate 12 through respective buried conductive paths 20.
The piezoelectric vibration sensor 10 comprises a piezoelectric membrane 23 and a routing structure 25, which are electrically insulated from the supporting body 13 by a dielectric body 24.
The piezoelectric membrane 23 is defined by a multilayer comprising, in order, a first electrode 23a, a piezoelectric lamina 23b, and a second electrode 23c and has a generally polygonal or circular shape. The piezoelectric lamina 23b may, for example, be of aluminum nitride (AlN), lead zirconate titanate (PZT), or sodium-potassium niobate (KNN). The piezoelectric membrane 23 has a peripheral edge anchored to the second portion 13b of the supporting body 13 through the dielectric body 24 and delimits at least partially the second volume 17b of the cavity 17 on a side opposite to the first volume 17a. Furthermore, the piezoelectric membrane 23 separates at least partially the second volume 17b and the third volume 17c of the cavity 17.
In the embodiment of
The routing structure 25 electrically couples the electrodes 23a, 23c of the piezoelectric membrane 23 to respective connectors 26 accessible from outside. In greater detail, the routing structure 25 comprises: a first surface conductive path 27, a first plug 28, and a respective one of the buried conductive paths 20 for connecting the first electrode 23a to the respective connector 26; and a second surface conductive path 30, a second plug 31, and a respective one of the buried conductive paths 20 for connecting the second electrode 23c to the respective connector 26. The plugs 28, 31 extend through the supporting body 13 and connect the first surface conductive path 27 and the second surface conductive path 30 to the respective buried conductive paths 20.
Further connectors 26 (just one of which is illustrated in
According to a different embodiment, illustrated in
Alternatively, as illustrated in
The sensor assembly described enables integration of a microelectromechanical inertial sensor and a piezoelectric vibration sensor in a single device. The two sensors may have characteristics very different from one another and thus be optimized separately to perform distinct functions. In particular, the inertial sensor may be used for user-interface functions, such as recognition of commands through movements and/or gestures, whereas the piezoelectric vibration sensor may serve as a voice accelerometer dedicated to voice detection via bone conduction in order to improve the quality of audio communication.
Further, it is possible to integrate other possible functions, according to the design preferences. For instance, the piezoelectric vibration sensor has a sensitivity and speed of response sufficient to implement effectively functions for waking-up the system from conditions of hibernation. Contrary to the inertial sensors, which require an albeit low power supply also in the hibernation state, the piezoelectric vibration sensor is in any case active and may supply signals in response to vibrations even in conditions of practically zero power consumption. Consequently, in the hibernation state, the overall power consumption is negligible, to the benefit of autonomy of the battery-powered devices and in any case in line with the general tendency to privilege energy saving. However, the promptness of response of the piezoelectric vibration sensor enables wake-up from conditions of hibernation in sufficiently short times to prevent any loss of useful data.
The piezoelectric vibration sensor is integrated in the sealed chamber itself of the inertial sensor. Therefore, on the one hand, the piezoelectric vibration sensor is acoustically isolated from outside and is not affected by environmental noise. On the other, the stacked arrangement of the inertial sensor and of the piezoelectric vibration sensor makes it possible to obtain a structure that is compact and far from cumbersome, which is much appreciated especially for the production of miniaturized devices.
Integration affords important advantages also from the standpoint of costs and management of the manufacturing processes. In general, both the production and the purchase of a single component instead of two distinct ones are economically more convenient. Further, the reduction in the number of components is an evident benefit for assembly because the number of operations required is reduced accordingly.
Also, the process for manufacturing of the sensor assemblies is suited to being integrated in manufacturing processes of microelectromechanical devices.
For instance,
As illustrated in
The first structural layer 50 is then planarized by Chemical-Mechanical Polishing (CMP) to a desired thickness, for example comprised between 10 μm and 60 μm.
A trench etch is then carried out to define the sensing mass 8a, the fixed electrodes 8b, 8c, and the flexures 22 in the first structural layer 50. In this step, first insulation trenches 51′ are further opened that delimit bottom portions 28′, 31′ of the plugs 28, 31.
After the trench etch, a first sacrificial layer 52, for example of silicon oxide, is deposited on the first structural layer 50 by Low-Pressure Chemical-Vapor Deposition (LPCVD) of silicon oxide so as to penetrate into the gaps between the sensing mass 8a, the fixed electrodes 8b, 8c, the flexures 22, and the first insulation trenches 51′, filling them partially or completely. The first sacrificial layer 52 is then patterned. In particular, the remaining portion of the first sacrificial layer 52 covers the region where the accelerometer 8 is formed.
Next (
The second structural layer 53 is then selectively etched with a second trench etch that stops on the first sacrificial layer 52. In the second structural layer 53 islands 53a are thus formed separated by trenches 56 and second insulation trenches 51″ are opened aligned to respective first insulation trenches 51′ that delimit top portions 28″, 31″ of the plugs 28, 31. The etched portion of the second structural layer 53 corresponds to the second volume 17b of the cavity 17 of
A second sacrificial layer 55 is then formed on the second structural layer 53 so as to fill the trenches 56 completely between the islands 53a.
Next (
As illustrated in
A piezoelectric multilayer 65 is then deposited (
Then (
Next, an etch is performed, for example using hydrofluoric acid to remove the sacrificial parts where exposed. In this step, the sacrificial layer 55 is removed both on the second structural layer 53 and under the piezoelectric membrane 23, through the through slits 23e. In particular, the walls 60 between the cavities 58 are eliminated, and the cavities 58 are joined in the second volume 17b of the cavity 17 of
Next, a cap wafer 70 is joined to the wafer 40 (
Finally, the cap wafer 70, the second structural layer 53, and the first structural layer 50 are etched to define the supporting body 13 and the connectors 26, and the wafer 40 is diced, thus obtaining the structure of
Finally, it is clear that modifications and variations may be made to the sensor assembly and to the process described herein, without thereby departing from the scope of the present invention, as defined in the annexed claims.
For instance, the inertial sensor integrated in the microelectromechanical sensor assembly may comprise one or more uniaxial or multiaxial accelerometers, one or more uniaxial or multiaxial gyroscopes, or a combination thereof.
The shape of the piezoelectric membrane may conveniently be chosen on the basis of the design preferences.
Number | Date | Country | Kind |
---|---|---|---|
102023000013011 | Jun 2023 | IT | national |