wafer-level encapsulation and sealing of ... - Semantic Scholar
Wafer-Level Encapsulation and Sealing of Electrostatic HARPSS Transducers Siavash Pourkamali* and Farrokh Ayazi School of Electrical and Computer Engineering, Georgia Institute of Technology Atlanta, GA 30332, USA Email: [email protected], [email protected] * Currently with Department of Electrical and Computrer Engineering, University of Denver, Denver, CO 80208
actuators in silicon substrate [8-14]. The unique feature of the HARPSS fabrication process is its capability to integrate nano-scale self-aligned lateral capacitive transduction gaps with thick structures made of bulk silicon and polysilicon. Several types of high resolution vibrating gyroscopes [8-10] as well as high frequency high-Q silicon resonators for frequency referencing [11-14] have been demonstrated using different versions of the HARPSS fabrication process. Figure 1 shows the SEM view of a typical SiBAR (Silicon Bulk Acoustic Wave Resonator) with resonance frequency of ~100MHz fabricated through the HARPSS-on-SOI process [12-14] as well as its width extensional resonant mode shape. The resonating body is made of SCS (Single Crystal Silicon) and its electrodes are made of polysilicon.
Abstract- This paper reports on a thin-film wafer-level encapsulation technique for packaging and CMOS integration of MEMS sensors and actuators fabricated through the HARPSS process. This approach takes advantage of the stationary parts of the micromechanical device itself for encapsulation of its sensitive moving parts, and therefore can be performed without addition of extensive processing steps. Encapsulated high frequency capacitive silicon resonators are demonstrated using this technique. Reliability and performance tests conducted on the encapsulated resonators reveal a high level of hermeticity and reliability with minimal interference with device operation. This technique can be applied to a wide variety MEMS sensors and actuators. Silicon transducers encapsulated using this approach can run through the regular IC fabrication and/or packaging processes.
Vp
I. INTRODUCTION Packaging is one of the most critical and costly steps in microelectronic manufacturing. Packaging of movable MEMS devices is specially challenging [1] because the package has to provide complete isolation and protection for the extremely sensitive micromechanical structures without being in physical contact with them and interfering with their operation. Furthermore, for some of the MEMS devices, such as resonant sensors, operation in vacuum is required or preferred. Therefore, low cost batch-fabrication techniques that can provide a suspended impermeable seal on top of the micromechanical structures are of great interest. Several types of such techniques have been reported that are mostly based on wafer bonding [2-4] or removal of a sacrificial layer from the top of the devices after deposition of a sealing layer [5-7]. In general, packaging techniques based on wafer bonding add significant complexity and cost to the process and therefore thin-film sacrificial layer based techniques are preferred. Among interesting examples of such techniques are using thermally-decomposable polymers as sacrificial layer [6] and epitaxial growth of silicon on top of silicon structures patterned on SOI substrates for sealing and CMOS integration [7].
vin vibration direction (a)
(b)
Figure 1. a) SEM view of a low impedance SiBAR fabricated using the HARPSS-on-SOI process, b) Mode shape for its fundamental width extentional mode.
It was demonstrated in [12-14] that for the single crystal silicon HARPSS-on-SOI resonators, polysilicon electrodes and their interconnects, can bridge over the SCS structures without physically contacting them. In this work, the suspended overlapping polysilicon electrodes of silicon bulk acoustic wave resonators (SiBAR) are extended all the way on top of the resonators providing a suspended polysilicon cap covering the resonating body of the resonators. Complete encapsulation is then achieved by deposition of a non-conformal sealing layer to close small release openings in the polysilicon cap without affecting the underlying movable structures.
HARPSS (High Aspect Ratio Poly- and Single-crystal Silicon) fabrication process is a 3-D silicon bulk micromachining technology with superior capabilities in embedding high-performance electrostatic sensors and
1-4244-1262-5/07/$25.00 ©2007 IEEE
vout
49
IEEE SENSORS 2007 Conference
and oxide layers used for encapsulation of MEMS devices are fully CMOS-compatible and stable at high temperatures, the encapsulated devices can run through a regular CMOS fabrication and packaging process, providing MEMS products with on-chip integrated electronics without any changes or significant cost added to the regular IC fabrication process.
II. FABRICATION PROCESS The resonators in this work are fabricated through an advanced version of the HARPSS-on-SOI process [14] without any modifications or extra steps added to the fabrication process and only by making slight adjustments to the resonator layout. Changes in the layout include extension of the polysilicon electrodes all the way on top of the resonators while leaving a narrow opening (~1-2 µm wide) in the polysilicon cap to electrically isolates the electrodes and allow HF release of the underlying structures. In addition, as opposed to the previously demonstrated devices, to allow further lithography steps, the surrounding SOI device layer was kept on the substrate while separating the resonators from it by narrow trenches (~4-8µm wide) etched down to the SOI buried oxide layer around the entire structure. Figure 2 shows the SEM view of a fabricated SiBAR with extended electrodes.
Isolation and Release Trench
a) Released resonator with extended electrodes on top
b) Deposit thick PECVD oxide (~5-10µm)
Polysilicon Cap c) Polish (etch back) PECVD oxide
d) Deposit and pattern LPCVD or metal film for improved sealing
Figure 3. Schematic diagram of the in-situ wafer level vacuum encapsulation fabrication process steps. Figure 2. Released SiBAR with extended poly electrodes covering the SCS resonator body.
The encapsulation process steps are shown in Fig. 3. After completion of the HARPSS SiBAR process with extended electrode, a thick non-conformal layer of PECVD oxide is deposited that gradually narrows down and closes all the openings in the polysilicon cap as well as the surrounding isolation trenches. Figure 4 shows the SEM view of similar resonators after step by step deposition of different thicknesses of PECVD oxide showing how the opening in the polysilicon cap starts narrowing down on the top by the deposited layer. Finally, after deposition of ~10µm of PECVD oxide, all the trenches are completely closed and covered by the thick layer. Since PECVD layers usually suffer from excessive defects and pinholes in their structure, the PECVD oxide layer proposed and used in this work may not provide a strong and impenetrable layer with enough hermeticity. Therefore, as shown in Fig. 3c and 3d, one can add a dense sealing layer such as a LPCVD or metal film on top of the PECVD layer. This is done preferably by polishing the PECVD layer back to the surface of the poly layer while keeping the devices sealed to avoid deposition of the LPCVD or metal film on top of the devices.
2µm PECVD Oxide
8µm PECVD Oxide
10µm PECVD Oxide Figure 4. SEM view of resonators after deposition of different thicknesses of PECDV oxide. Complete sealing of all the openings occurs after deposition of 10µm of oxide.
Figure 5 shows the SEM view of a number of encapsulated resonators with different resonant frequencies batchfabricated on an SOI substrate. As presented in the next section, the thick PECVD oxide layer used in this work turned out to provide a good enough seal with excellent resistance against gas and liquid penetration inside the cavity, and successfully passed the reliability and endurance tests available to the authors. Therefore, the resonators in Fig. 5 are sealed and isolated from the outside only by the
Finally, openings need to be etched in the sealing layer to allow access to the resonator pads for electrical connections (Fig. 5). The encapsulated resonators using this technique can then run through the regular packaging steps used in the IC industry providing a reliable package without significant cost added to the encapsulated MEMS devices compared to the regular IC products. Furthermore, since the polysilicon 50
The difference between the Q factor of SiBARs in air and vacuum in this frequency range is typically larger than the observed value in Fig. 6 and a Q of at least ~40,000-60,000 in vacuum is expected from a 103MHz resonator with Q of 25,000 in air [12-14]. The Q value being lower than expected could be a result of undesired deposition of very small amounts of PECVD oxide on the resonator body during the encapsulation process. Slight reduction (~420ppm) in the resonance frequency of the resonator after encapsulation also confirms the same assumption. This can be reduced or totally avoided by increasing the polysilicon cap thickness (currently 4µm) and/or decreasing its opening width.
thick PECVD oxide and the remaining photoresist used for patterning the oxide.
To further investigate the resistance of the seal against air pressure, comparatively long term tests were performed on the devices. The encapsulated resonators were first measured in air and then placed under sub-milli-Torr vacuum for a period of 60 hours. The resonant frequency and quality factor of the resonator was observed and recorded at intervals of a few hours. The resonators were then transferred back to the atmosphere and kept under measurement for another 50 hours. Figure 7 shows measured resonance peaks of an encapsulated 134MHz SiBAR in atmosphere, under mTorr level vacuum and back to atmosphere. Resonator Q was constant during the test period and independent of the pressure of its surrounding environment. Slight changes in the measured values of quality factor are in the order of measurement uncertainties and/or test set-up effects, and are much smaller than the environmental-pressure-induced Q variations [11-14], which further indicates the impermeability of the PECVD oxide seal.
Figure 5. SEM view of SiBARs after encapsulation by the thick PECVD oxide layer and patterning the sealing layer for electrical access to the resonator pads.
III. ENDURANCE AND RELIABILITY TEST Several tests were performed on a number of the encapsulated resonators to investigate the strength and reliability of the sealing layers against penetration of liquid or gas molecules in and out of the resonator cavity as well as the effects of the encapsulation process on resonator performance. To observe the effect of the encapsulation process on the resonator characteristics, some of the devices were tested in air before and after encapsulation. Figure 6 shows the measured frequency response of a 103MHz SiBAR before and after encapsulation. As demonstrated in Fig. 6, the resonator Q increases slightly (~22%) after encapsulation compared to its measured value in air prior to its encapsulation; this shows existence of some level of vacuum inside the sealed cavity. Since the sealing of the cavity takes place inside the PECVD chamber under a 200-300mTorr pressure, it is expected to maintain the same level of vacuum inside the cavity after encapsulation in case of a perfect and impermeable seal. Therefore, it can be concluded that the deposited PECVD layer has a good resistance against penetration of air molecules into the cavity.
Vp = 10V
f = 134.392 MHz f = 134.393 MHz Q = 18,500 Q = 17,500 Rm = 37.1kΩ Rm = 37.5kΩ In Air After Encapsulation
In Air Before Encapsulation
Vp = 15V f = 103.853 MHz Q = 25,300
In Air After Encapsulation
After 60hrs in submTorr Vacuum
W = 30µm L = 300µm t = 20µm g = 125nm
f = 134.392 MHz Q = 18,700 Rm = 37.1kΩ Back in Atmosphere after 50hrs
Figure 7. Measurement results for an encapsulated resonator. Showing no difference in the measured quality factor when transferring the device from vacuum to atmosphere and vice versa.
After leaving the encapsulated resonator sample in the lab with no protection for over two weeks, to determine the resistance of the seal against liquids, the sample was dispensed in a used paper cup filled with city water and was left there for over two days. The sample was then taken out, and after blowing the remaining liquid off its surface, the same resonator was tested again and a similar resonance peak with the same resonant frequency and the same quality factor was observed. This indicates the hermeticity of the oxide seal against water and the contaminants in it. Figure 8
Vp = 10V f = 103.809 MHz Q = 30,900
Figure 6. Measured resonance peaks for a 40µm wide, 300µm long, 20µm thick SiBAR before and after PECVD oxide encapsulation.
51
step and therefore minimal added cost. This fabrication technique can be used for encapsulation of several types of HARPSS based sensors and actuators as well as other types of MEMS devices. The sealed devices can then run through the regular IC packaging process. Furthermore, the polysilicon and oxide films used as the cap and sealing layer in this technique are fully CMOS compatible and can tolerate high temperatures in a CMOS fabrication sequence. Therefore, similar sealing technique can be used to fabricate encapsulated MEMS devices and then run the wafers in a regular CMOS fabrication and packaging process to implement CMOS integrated MEMS sensor and actuators in a very convenient and low cost manner.
shows the measured resonance peaks of the encapsulated resonator before and after dispensing in city water. Slight change in the resonant frequency of the resonator before and after the test is in the order of temperature induced frequency changes and can not be due to penetration of water into the cavity. before dispensing
After 2 days under city water
f = 134.378 MHz Q = 18,100
f = 134.384 MHz Q = 18,000
REFERENCES [1]
Figure 8. Measured response of the encapsulated 134MHz resonator before and after dispensing the sample in city water for over two days.
[2]
Finally, the sample was placed in a temperature controlled oven, and the oven temperature was swept from room temperature (~25C°) to 75°C. The resonance frequency of the resonator was measured and recorded at 10°C intervals. Figure 9 depicts the resulting temperature induced frequency shift graph showing similar linear trend with the same slope for the encapsulated resonator as that of regular unsealed SCS HARPSS resonators [11-14]. This certifies that the thick PECVD oxide sealing layers do not induce significant stress on the enclosed structures.
[3]
[4] [5]
[6]
Frequency (MHz)
134.45
[7]
134.4
[8]
134.35
-28.4ppm/°C
134.3
[9]
134.25 134.2
[10]
134.15 20
40
60
80 [11]
Temperature (C) Figure 9. Measured temperature drift of the 134MHz encapsulated resonator which is similar to that of unpackaged SCS resonators.
[12] [13]
IV. CONCLUSIONS A low cost batch fabrication process for wafer-level vacuum sealing of high frequency HARPSS capacitive resonators was demonstrated. This fabrication technique uses stationary parts of the device itself for forming a protective cap on top of the micromechanical movable structures and requires addition of only one extra lithography
[14]
52
K. Najafi, “Micro packaging technologies for integrated microsystems: applications to MEMS and MOEMS,” Proc. SPIE MOEMS/MEMS, Jan 2003. B. Ziaei, J. V. Arx, M. Nardin, K. Najafi, “A hermetic glass silicon micro package with high-density on-chip feedthroughs for sensors and Actuators,” J. Microelectromechanical. Syst. vol. 5, pp.166-179, 1996. Y. Mei, G.R. Lahiji, K. Najafi, “A robust gold-silicon eutectic wafer bonding technology for vacuum packaging,” in Tech. Dig. Solid-State Sensor and Actuator Workshop, Hilton Head Island, June 2002. L. Lin, “MEMS Post-Packaging by Localized Heating and Bonding,” IEEE Trans on Advanced Packaging, vol. 23, pp. 608-616, 2000. B. H. Stark, M. Dokmechi, T.J. Harspter, K. Najafi, “A wafer-level vacuum packaging technology utilizing electroplated nickel,” in ASME Congress, MEMS and Nanotechnology Symposium, New Orleans, November, 2002. P. Monajemi, P. Joseph, P. Kohl, and F. Ayazi, "A low cost waferlevel MEMS packaging technology," Proc. IEEE Micro Electro Mechanical Systems Conference (MEMS'05), Miami, FL, Jan. 2005,pp. 634-637. R. N. Candler, et al., “Single wafer encapsulation of MEMS devices,” Trans. on Advanced Packaging, vol. 26, no.3, pp.227-232, 2003. F. Ayazi and K. Najafi, "A HARPSS polysilicon vibrating ring gyroscope," IEEE Journal of Microelectromechanical Systems, Vol. 10, June 2001, pp. 169-179. M. Zaman, A. Sharma, B. Amini, and F. Ayazi, "The resonating star gyroscope," Proc. IEEE Micro Electro Mechanical Systems Conference (MEMS'05), Miami, FL, Jan. 2005, pp. 355-358. H. Johari and F. Ayazi, "High frequency capacitive disk gyroscopes in (100) and (111) silicon,” Tech. Dig. 20th IEEE International Conference on Micro Electormechanical Systems Conference 2007, Kobe, Japan, Jan. 2007. S. Pourkamali, A. Hashimura, R. Abdolvand, G. Ho, A. Erbil, and F. Ayazi, "High-Q single crystal silicon HARPSS capacitive beam resonators with sub-micron transduction gaps," IEEE Journal of Microelectromechanical Systems, Vol. 12, No. 4, August 2003, pp. 487-496. S. Pourkamali, G. K. Ho and F. Ayazi, “Vertical capacitive SiBARs,” MEMS’05, pp. 211-214. S. Pourkamali and F. Ayazi, “High frequency low impedance silicon BAR structures,” Proceedings, Hilton Head 2006, solid-state sensor, actuator and Microsystems workshop, pp. 284-287. Siavash Pourkamali, “High frequency capacitive single crystal silicon resonators and coupled resonator systems,” PhD Thesis, Georgia Institute of Technology, August 2006.
actuators in silicon substrate [8-14]. The unique feature of the HARPSS fabrication process is its capability to integrate nano-scale self-aligned lateral capacitive transduction gaps with thick structures made of bulk silicon and polysilicon. Several types of high resolution vibrating gyroscopes [8-10] as well as high frequency high-Q silicon resonators for frequency referencing [11-14] have been demonstrated using different versions of the HARPSS fabrication process. Figure 1 shows the SEM view of a typical SiBAR (Silicon Bulk Acoustic Wave Resonator) with resonance frequency of ~100MHz fabricated through the HARPSS-on-SOI process [12-14] as well as its width extensional resonant mode shape. The resonating body is made of SCS (Single Crystal Silicon) and its electrodes are made of polysilicon.
Abstract- This paper reports on a thin-film wafer-level encapsulation technique for packaging and CMOS integration of MEMS sensors and actuators fabricated through the HARPSS process. This approach takes advantage of the stationary parts of the micromechanical device itself for encapsulation of its sensitive moving parts, and therefore can be performed without addition of extensive processing steps. Encapsulated high frequency capacitive silicon resonators are demonstrated using this technique. Reliability and performance tests conducted on the encapsulated resonators reveal a high level of hermeticity and reliability with minimal interference with device operation. This technique can be applied to a wide variety MEMS sensors and actuators. Silicon transducers encapsulated using this approach can run through the regular IC fabrication and/or packaging processes.
Vp
I. INTRODUCTION Packaging is one of the most critical and costly steps in microelectronic manufacturing. Packaging of movable MEMS devices is specially challenging [1] because the package has to provide complete isolation and protection for the extremely sensitive micromechanical structures without being in physical contact with them and interfering with their operation. Furthermore, for some of the MEMS devices, such as resonant sensors, operation in vacuum is required or preferred. Therefore, low cost batch-fabrication techniques that can provide a suspended impermeable seal on top of the micromechanical structures are of great interest. Several types of such techniques have been reported that are mostly based on wafer bonding [2-4] or removal of a sacrificial layer from the top of the devices after deposition of a sealing layer [5-7]. In general, packaging techniques based on wafer bonding add significant complexity and cost to the process and therefore thin-film sacrificial layer based techniques are preferred. Among interesting examples of such techniques are using thermally-decomposable polymers as sacrificial layer [6] and epitaxial growth of silicon on top of silicon structures patterned on SOI substrates for sealing and CMOS integration [7].
vin vibration direction (a)
(b)
Figure 1. a) SEM view of a low impedance SiBAR fabricated using the HARPSS-on-SOI process, b) Mode shape for its fundamental width extentional mode.
It was demonstrated in [12-14] that for the single crystal silicon HARPSS-on-SOI resonators, polysilicon electrodes and their interconnects, can bridge over the SCS structures without physically contacting them. In this work, the suspended overlapping polysilicon electrodes of silicon bulk acoustic wave resonators (SiBAR) are extended all the way on top of the resonators providing a suspended polysilicon cap covering the resonating body of the resonators. Complete encapsulation is then achieved by deposition of a non-conformal sealing layer to close small release openings in the polysilicon cap without affecting the underlying movable structures.
HARPSS (High Aspect Ratio Poly- and Single-crystal Silicon) fabrication process is a 3-D silicon bulk micromachining technology with superior capabilities in embedding high-performance electrostatic sensors and
1-4244-1262-5/07/$25.00 ©2007 IEEE
vout
49
IEEE SENSORS 2007 Conference
and oxide layers used for encapsulation of MEMS devices are fully CMOS-compatible and stable at high temperatures, the encapsulated devices can run through a regular CMOS fabrication and packaging process, providing MEMS products with on-chip integrated electronics without any changes or significant cost added to the regular IC fabrication process.
II. FABRICATION PROCESS The resonators in this work are fabricated through an advanced version of the HARPSS-on-SOI process [14] without any modifications or extra steps added to the fabrication process and only by making slight adjustments to the resonator layout. Changes in the layout include extension of the polysilicon electrodes all the way on top of the resonators while leaving a narrow opening (~1-2 µm wide) in the polysilicon cap to electrically isolates the electrodes and allow HF release of the underlying structures. In addition, as opposed to the previously demonstrated devices, to allow further lithography steps, the surrounding SOI device layer was kept on the substrate while separating the resonators from it by narrow trenches (~4-8µm wide) etched down to the SOI buried oxide layer around the entire structure. Figure 2 shows the SEM view of a fabricated SiBAR with extended electrodes.
Isolation and Release Trench
a) Released resonator with extended electrodes on top
b) Deposit thick PECVD oxide (~5-10µm)
Polysilicon Cap c) Polish (etch back) PECVD oxide
d) Deposit and pattern LPCVD or metal film for improved sealing
Figure 3. Schematic diagram of the in-situ wafer level vacuum encapsulation fabrication process steps. Figure 2. Released SiBAR with extended poly electrodes covering the SCS resonator body.
The encapsulation process steps are shown in Fig. 3. After completion of the HARPSS SiBAR process with extended electrode, a thick non-conformal layer of PECVD oxide is deposited that gradually narrows down and closes all the openings in the polysilicon cap as well as the surrounding isolation trenches. Figure 4 shows the SEM view of similar resonators after step by step deposition of different thicknesses of PECVD oxide showing how the opening in the polysilicon cap starts narrowing down on the top by the deposited layer. Finally, after deposition of ~10µm of PECVD oxide, all the trenches are completely closed and covered by the thick layer. Since PECVD layers usually suffer from excessive defects and pinholes in their structure, the PECVD oxide layer proposed and used in this work may not provide a strong and impenetrable layer with enough hermeticity. Therefore, as shown in Fig. 3c and 3d, one can add a dense sealing layer such as a LPCVD or metal film on top of the PECVD layer. This is done preferably by polishing the PECVD layer back to the surface of the poly layer while keeping the devices sealed to avoid deposition of the LPCVD or metal film on top of the devices.
2µm PECVD Oxide
8µm PECVD Oxide
10µm PECVD Oxide Figure 4. SEM view of resonators after deposition of different thicknesses of PECDV oxide. Complete sealing of all the openings occurs after deposition of 10µm of oxide.
Figure 5 shows the SEM view of a number of encapsulated resonators with different resonant frequencies batchfabricated on an SOI substrate. As presented in the next section, the thick PECVD oxide layer used in this work turned out to provide a good enough seal with excellent resistance against gas and liquid penetration inside the cavity, and successfully passed the reliability and endurance tests available to the authors. Therefore, the resonators in Fig. 5 are sealed and isolated from the outside only by the
Finally, openings need to be etched in the sealing layer to allow access to the resonator pads for electrical connections (Fig. 5). The encapsulated resonators using this technique can then run through the regular packaging steps used in the IC industry providing a reliable package without significant cost added to the encapsulated MEMS devices compared to the regular IC products. Furthermore, since the polysilicon 50
The difference between the Q factor of SiBARs in air and vacuum in this frequency range is typically larger than the observed value in Fig. 6 and a Q of at least ~40,000-60,000 in vacuum is expected from a 103MHz resonator with Q of 25,000 in air [12-14]. The Q value being lower than expected could be a result of undesired deposition of very small amounts of PECVD oxide on the resonator body during the encapsulation process. Slight reduction (~420ppm) in the resonance frequency of the resonator after encapsulation also confirms the same assumption. This can be reduced or totally avoided by increasing the polysilicon cap thickness (currently 4µm) and/or decreasing its opening width.
thick PECVD oxide and the remaining photoresist used for patterning the oxide.
To further investigate the resistance of the seal against air pressure, comparatively long term tests were performed on the devices. The encapsulated resonators were first measured in air and then placed under sub-milli-Torr vacuum for a period of 60 hours. The resonant frequency and quality factor of the resonator was observed and recorded at intervals of a few hours. The resonators were then transferred back to the atmosphere and kept under measurement for another 50 hours. Figure 7 shows measured resonance peaks of an encapsulated 134MHz SiBAR in atmosphere, under mTorr level vacuum and back to atmosphere. Resonator Q was constant during the test period and independent of the pressure of its surrounding environment. Slight changes in the measured values of quality factor are in the order of measurement uncertainties and/or test set-up effects, and are much smaller than the environmental-pressure-induced Q variations [11-14], which further indicates the impermeability of the PECVD oxide seal.
Figure 5. SEM view of SiBARs after encapsulation by the thick PECVD oxide layer and patterning the sealing layer for electrical access to the resonator pads.
III. ENDURANCE AND RELIABILITY TEST Several tests were performed on a number of the encapsulated resonators to investigate the strength and reliability of the sealing layers against penetration of liquid or gas molecules in and out of the resonator cavity as well as the effects of the encapsulation process on resonator performance. To observe the effect of the encapsulation process on the resonator characteristics, some of the devices were tested in air before and after encapsulation. Figure 6 shows the measured frequency response of a 103MHz SiBAR before and after encapsulation. As demonstrated in Fig. 6, the resonator Q increases slightly (~22%) after encapsulation compared to its measured value in air prior to its encapsulation; this shows existence of some level of vacuum inside the sealed cavity. Since the sealing of the cavity takes place inside the PECVD chamber under a 200-300mTorr pressure, it is expected to maintain the same level of vacuum inside the cavity after encapsulation in case of a perfect and impermeable seal. Therefore, it can be concluded that the deposited PECVD layer has a good resistance against penetration of air molecules into the cavity.
Vp = 10V
f = 134.392 MHz f = 134.393 MHz Q = 18,500 Q = 17,500 Rm = 37.1kΩ Rm = 37.5kΩ In Air After Encapsulation
In Air Before Encapsulation
Vp = 15V f = 103.853 MHz Q = 25,300
In Air After Encapsulation
After 60hrs in submTorr Vacuum
W = 30µm L = 300µm t = 20µm g = 125nm
f = 134.392 MHz Q = 18,700 Rm = 37.1kΩ Back in Atmosphere after 50hrs
Figure 7. Measurement results for an encapsulated resonator. Showing no difference in the measured quality factor when transferring the device from vacuum to atmosphere and vice versa.
After leaving the encapsulated resonator sample in the lab with no protection for over two weeks, to determine the resistance of the seal against liquids, the sample was dispensed in a used paper cup filled with city water and was left there for over two days. The sample was then taken out, and after blowing the remaining liquid off its surface, the same resonator was tested again and a similar resonance peak with the same resonant frequency and the same quality factor was observed. This indicates the hermeticity of the oxide seal against water and the contaminants in it. Figure 8
Vp = 10V f = 103.809 MHz Q = 30,900
Figure 6. Measured resonance peaks for a 40µm wide, 300µm long, 20µm thick SiBAR before and after PECVD oxide encapsulation.
51
step and therefore minimal added cost. This fabrication technique can be used for encapsulation of several types of HARPSS based sensors and actuators as well as other types of MEMS devices. The sealed devices can then run through the regular IC packaging process. Furthermore, the polysilicon and oxide films used as the cap and sealing layer in this technique are fully CMOS compatible and can tolerate high temperatures in a CMOS fabrication sequence. Therefore, similar sealing technique can be used to fabricate encapsulated MEMS devices and then run the wafers in a regular CMOS fabrication and packaging process to implement CMOS integrated MEMS sensor and actuators in a very convenient and low cost manner.
shows the measured resonance peaks of the encapsulated resonator before and after dispensing in city water. Slight change in the resonant frequency of the resonator before and after the test is in the order of temperature induced frequency changes and can not be due to penetration of water into the cavity. before dispensing
After 2 days under city water
f = 134.378 MHz Q = 18,100
f = 134.384 MHz Q = 18,000
REFERENCES [1]
Figure 8. Measured response of the encapsulated 134MHz resonator before and after dispensing the sample in city water for over two days.
[2]
Finally, the sample was placed in a temperature controlled oven, and the oven temperature was swept from room temperature (~25C°) to 75°C. The resonance frequency of the resonator was measured and recorded at 10°C intervals. Figure 9 depicts the resulting temperature induced frequency shift graph showing similar linear trend with the same slope for the encapsulated resonator as that of regular unsealed SCS HARPSS resonators [11-14]. This certifies that the thick PECVD oxide sealing layers do not induce significant stress on the enclosed structures.
[3]
[4] [5]
[6]
Frequency (MHz)
134.45
[7]
134.4
[8]
134.35
-28.4ppm/°C
134.3
[9]
134.25 134.2
[10]
134.15 20
40
60
80 [11]
Temperature (C) Figure 9. Measured temperature drift of the 134MHz encapsulated resonator which is similar to that of unpackaged SCS resonators.
[12] [13]
IV. CONCLUSIONS A low cost batch fabrication process for wafer-level vacuum sealing of high frequency HARPSS capacitive resonators was demonstrated. This fabrication technique uses stationary parts of the device itself for forming a protective cap on top of the micromechanical movable structures and requires addition of only one extra lithography
[14]
52
K. Najafi, “Micro packaging technologies for integrated microsystems: applications to MEMS and MOEMS,” Proc. SPIE MOEMS/MEMS, Jan 2003. B. Ziaei, J. V. Arx, M. Nardin, K. Najafi, “A hermetic glass silicon micro package with high-density on-chip feedthroughs for sensors and Actuators,” J. Microelectromechanical. Syst. vol. 5, pp.166-179, 1996. Y. Mei, G.R. Lahiji, K. Najafi, “A robust gold-silicon eutectic wafer bonding technology for vacuum packaging,” in Tech. Dig. Solid-State Sensor and Actuator Workshop, Hilton Head Island, June 2002. L. Lin, “MEMS Post-Packaging by Localized Heating and Bonding,” IEEE Trans on Advanced Packaging, vol. 23, pp. 608-616, 2000. B. H. Stark, M. Dokmechi, T.J. Harspter, K. Najafi, “A wafer-level vacuum packaging technology utilizing electroplated nickel,” in ASME Congress, MEMS and Nanotechnology Symposium, New Orleans, November, 2002. P. Monajemi, P. Joseph, P. Kohl, and F. Ayazi, "A low cost waferlevel MEMS packaging technology," Proc. IEEE Micro Electro Mechanical Systems Conference (MEMS'05), Miami, FL, Jan. 2005,pp. 634-637. R. N. Candler, et al., “Single wafer encapsulation of MEMS devices,” Trans. on Advanced Packaging, vol. 26, no.3, pp.227-232, 2003. F. Ayazi and K. Najafi, "A HARPSS polysilicon vibrating ring gyroscope," IEEE Journal of Microelectromechanical Systems, Vol. 10, June 2001, pp. 169-179. M. Zaman, A. Sharma, B. Amini, and F. Ayazi, "The resonating star gyroscope," Proc. IEEE Micro Electro Mechanical Systems Conference (MEMS'05), Miami, FL, Jan. 2005, pp. 355-358. H. Johari and F. Ayazi, "High frequency capacitive disk gyroscopes in (100) and (111) silicon,” Tech. Dig. 20th IEEE International Conference on Micro Electormechanical Systems Conference 2007, Kobe, Japan, Jan. 2007. S. Pourkamali, A. Hashimura, R. Abdolvand, G. Ho, A. Erbil, and F. Ayazi, "High-Q single crystal silicon HARPSS capacitive beam resonators with sub-micron transduction gaps," IEEE Journal of Microelectromechanical Systems, Vol. 12, No. 4, August 2003, pp. 487-496. S. Pourkamali, G. K. Ho and F. Ayazi, “Vertical capacitive SiBARs,” MEMS’05, pp. 211-214. S. Pourkamali and F. Ayazi, “High frequency low impedance silicon BAR structures,” Proceedings, Hilton Head 2006, solid-state sensor, actuator and Microsystems workshop, pp. 284-287. Siavash Pourkamali, “High frequency capacitive single crystal silicon resonators and coupled resonator systems,” PhD Thesis, Georgia Institute of Technology, August 2006.
