T E C H N I C A L P A P E R
A new hybrid fabrication process for a high sensitivity MEMS
microphone
Paul C.-P. Chao•Chun-Yin Tsai •Chi-Wei Chiu• Che-Hung Tsai• Tse-Yi Tu
Received: 31 October 2012 / Accepted: 21 May 2013 / Published online: 13 June 2013 Ó Springer-Verlag Berlin Heidelberg 2013
Abstract A novel multi-layer stacking capacitive type microphone is designed in this study based on theoretical analysis and numerical simulations, while fabricated via two standard stable silicon-based MEMS processes— PolyMUMPs and SOIMUMPs. The adoption of two stan-dardized processes helps greatly to increase yield rate. The sensitivity of the microphone is first determined by an analytical model based on an equivalent circuit, which is followed by finite element (FEM) analyses on the capaci-tance value, static pull-in voltage and dynamic character-istics. Based on the developed analytical model, varied dimensions of the microphone are optimized and then the performance is validated by analytical simulations. In the next step, micro-fabrication of the microphone is accom-plished using two standard processes, PolyMUMPs and SOIMUMPs provided by MEMSCAP. Experiments are conducted to acquire the information of pull-in voltage for safe operation and frequency response in sensitivity for performance evaluation. In the static case, experimental results show a good agreement with the analytical results with 90 Mpa residual stress assumed. As for dynamic
validation, the frequency response is measured in an anechoic room adopting the exciting frequency as the audible range from 100–10 kHz. The measured sensitivity is as around 0.78 and 1.7 mV/Pa from 100 to 10 kHz, under the biases of 2 and 4.5 V, respectively. Within the audible frequency range, the proposed device maintains the loss as less as 2.7 dB (ref. to V/Pa), under 3 dB—the commonly acceptable drop within audible frequency range.
1 Introduction
One of the most successfully-developed and widely-used MEMS devices is the electrostatic condenser microphone. The main benefit of this device is its characteristic high sensitivity and small power consumption. Recent improvements in MEMS processing have resulted in the batch fabrication of low cost, high performance, minia-turized condenser microphones (Bergqvist and Rudolf
1990; Scheeper et al. 1992, 1994; Royer et al. 1983; Leinenbach et al.2010; Chen et al.2010; Chan et al.2011). In addition to fabrication, past studies have accomplished static (Chao et al. 2006) and dynamic (Chao et al. 2008) pull-in analysis based on certain MEMS device geometries. In these applications, this device offers specific advantages, such as miniaturization, over traditional ECMs (electret condenser microphone). These past studies are however all fabricated in the laboratory environment which may lead to uncertainties in the future mass production (Pedersen et al.
1998; Hsu et al.1998).
The current study therefore proposes a reliable hybrid fabrication process, which combines two standard and stable processes, PolyMUMPs (Koester et al. 2003) and SOIMUMPs (Koester et al. 2003), developed from the MEMSCAP company, aiming for designing a high
P. C.-P. Chao (&) C.-W. Chiu C.-H. Tsai T.-Y. Tu Department of Electrical Engineering,
National Chiao Tung University, Hsinchu 300, Taiwan
e-mail: [email protected] P. C.-P. Chao
Department of Imaging and Biophotonics, National Chiao Tung University, Tainan 711, Taiwan
C.-Y. Tsai
Department of Mechanical Engineering, National Chiao Tung University, Hsinchu 300, Taiwan
sensitivity condenser microphone. The individual pro-cesses, PolyMUMPs and SOIMUMPs, have already shown their stabilities for maintaining fabrication quality. This work then aims to adopt the two standard procedures sequentially into microphone structure design with few modified steps and specifications in the fabrication pro-cesses. The standardized process helps greatly to increase yield rate, which is still the major issue to deal with nowadays for MEMS microphone fabrication. Based on specifications, the analysis for predicting microphone sensitivity is first conducted for designing preliminary sizes of the microphone. Previous studies show that the model-ing usmodel-ing finite element method (FEM) is efficient for fast verification about the design (Bergqvist1993; Wang et al.
2004; Senturia 2000). Hence, the FEM analysis using commercial software, COMSOL and ANSYS, are also conducted to predict the deflection of membrane with realistic boundary conditions assumed, which are clam-pings at edges of the membrane. It is found that the both simulation results are close to each other and accurately predict the deflection of the vibrating membrane of the MEMS microphone.
With simulation and design completed, the MEMS microphone was then fabricated using the combined pro-cesses of PolyMUMPs and SOIMUMPs based on the proposed design concepts and compatible fabrication steps. The static characteristic, frequency response and maximum vibration amplitude of the fabricated MEMS microphone were measured. With the devices being fabricated, two approaches are adopted in experimental validation—one under static input voltage while another subjected to vibratory sound in wide frequency range. In the case with static input voltage applied to the mem-brane, the largest displacement of 1 lm over the entire membrane is measured by a laser Doppler vibrometer; furthermore, the pull-in phenomenon is present with acquisition of corresponding pull-in voltage and mem-brane deflection. In the second case with vibratory sound as the excitation, the frequency responses are measured with moderate levels of noise present. The measured sensitivity of the fabricated microphone which is under the biases of 2 and 4.5 V is around 0.78 and 1.7 mV/Pa from 100 to 10 kHz. Within this measured frequency range, the proposed device maintains the loss as less as 2.7 dB (ref. to V/Pa), under 3 dB–the commonly accept-able drop within audible frequency range.
This study is organized as follows. Section 2 presents the designed microphone structure. Section 3 performed finite element analysis and then predict pull-in and fre-quency response of the proposed MEMS microphone device. Section4states the designed fabrication process for the microphone, Sect. 5 presents experimental results. Finally, Sect.6 concludes this study.
2 Design of the MEMS microphone
The designed microphone structure is presented in Fig.1a. It is composed of a diaphragm as vibrating membrane and a metal backplate as the electrode. The diaphragm is pressure sensitive and deforms as a flexible plate in a parallel-plate capacitor. Figure1b shows the layout mask of the proposed MEMS device. Through a systematically simulation anal-ysis, the basic sensitivity formula (Bergqvist and Rudolf
1990) shows that the sensitivity of the basic square microphone is in a direct proportion with DC bias voltage, while in an inverse proportion with the air gap. The basic design goal herein is to increase the sensitivity of the microphones.
The mechanical behavior of the microphone can be regarded as the part of an equivalent circuit, which relates forces (potential) and velocities (flow) in the microphone structure, aiming to predict well the microphone perfor-mance. The prediction is used for better design of micro-phone. Previous studies (Hsu et al. 1998; Senturia 2000) presented a detailed description of this circuit, which includes radiation and compliance of the diaphragm, air-streaming resistance in the air gap and acoustic holes in back plate, and compliance of the back plate and back chamber. The mechanical components are given by the following relations: Diaphragm radiation: Rr ¼ q0a4x2 2pc ; Mr¼ 8q0a3 3pp :ffiffiffip ð1Þ (a) (b) Backplate electrode Si substract Polysilicon diaphragm Nitride insulation
Fig. 1 aStructure of the MEMS microphone, b fabrication processes for the microphone
Diaphragm compliance and mass inertia: Cmffi 32a2 p6ð2p2Dþ a2TÞ; Mm¼ p4q 2pð 2Dþ a2TÞ 64T : ð2Þ
The viscosity loss in the air gap and its compliance: Rg¼ 12ua2 nd3p a 2 a2 8 ln a 4 3 8 ; Ca¼ d q0c2a2a2: ð3Þ
The viscosity loss of back plate holes is approximated as (Hsu et al.1998):
Rh
8uha2
pnr4 : ð4Þ
In the above equations, a is the side length of the microphone, q0is the density of air, c is the speed of sound
in air, x is the angular frequency (2pf), q is tensile force per unit length, n is the hole density in the backplate, a is the surface fraction occupied by the holes, u is the air viscosity coefficient, d is the average air gap distance, and h is the back plate height and r is the radius of hole. The sensitivity of the microphone is then considered as the output voltage with the presence of the acoustical pressure loading, i.e., Se¼ V0 P ¼ Vba2 jxdZt ; ð5Þ
where P is the sound pressure; Vb is the bias voltage between two electrodes; Zt is the total equivalent imped-ance of the circuit; a is the width of the membrane; d is the gap distance of the microphone. The goal of our design is to maximize the sensitivity within audible frequency range, 10–10 kHz. Based on the analysis tool with Eq. (5), the microphone is designed of 1.1 lm thick and 2 9 2 mm2. As shown herein, the baseline design determined based on the lumped coupled eletro-mechanical-acoustic model presented in this section, while the structure and geometry of the device is optimized via FEM analysis in Sect.3, with confirmation of no pull-in occurrence in Sect.5.
3 FEM analysis
Two commercial softwares, COMSOL and ANSYS, are utilized for predicting the capacitance of the designed microphone and the static/dynamic behaviors of the pro-posed MEMS microphone device, respectively. The capacitance value is first predicted by COMSOL, which is accomplished using the 3D electrostatics application mode in the electromagnetic module. The potential on each capacitor plate must obey the applied boundary condition and is therefore constant. Between the plates, the electric field is almost uniform, but fringing effects are seen at the edges of the plates. Under the proposed design dimension,
the capacitance of the parallel plates, C, is calculated from the COMSOL, as shown in Fig.2. It shows the resulted capacitance is approximately 2.4 pF by summing all the analyzed value.
Dynamic FEM analysis is next performed via ANSYS to predict pull-in and frequency response of the device. Since the mechanical force and electrostatic force are coupled, the simulation by ANSYS is carried out by a direct matrix-coupled formulation in a 3-D transducer element, TRANS126. The element couples structural motion with electrostatic fields, which can be used in a ‘‘lumped’’ or ‘‘distributed’’ environment. The capacitive microphone is seen as a pressed transducer, with the diaphragm suspended above the fixed substrate. The diaphragm is meshed with four node Solid 45 element. The top and bottom electrodes are separated by air gap, which is modeled using distributed TRANS126 transducer element (Fig.3). All the geometry parameters must fulfill fabrication requirements, which is the major consideration of the geometrical design. The area of the diaphragm is 2 9 2 mm2, the material properties of the diaphragm are Young’s modulus E = 170 GPa, pois-son’s ratio m = 0.34 and the density q = 2,300 kg/m2. The thickness of the diaphragm and the air gap are 1.1 and 3.75 lm, respectively. Finally, the residual stress is set as 90 MPa for FEM simulations. Simulated static deflection of the microphone membrane via ANSYS is shown in Fig. 5b, where it is seen that the microphone can be safely operated without pull-in occurring, possibly due to high membrane residual stress of 90 MPa assumed. Note that this assumed residual stress of 90 Mpa will be validated later by experi-ment, as stated in Sect.5.
4 Fabrication
Figure4a shows the cross-section of the customized hybrid processes designed and employed by this study for
1 0 2 3 4 5 6 Min: -0.936 Max: 6.983 (pF) (m)
microphones based on the standardized processes of PolyMUMPs (Koester et al. 2003) and SOIMUMPs (Koester et al. 2003). As shown in Fig. 5, the standard PolyMUMPs process offers to deposit seven layers, con-sisting of four kinds of materials, on the substrate. As it creates each layer, it converges to three basic steps, including deposition, lithography and etching, to manu-facture the structures, devices, or systems of micrometer size. The material layers and the lithography masks are also listed in Fig.5. Attention has to be paid to the difference between the layer and the mask names. Using eight masks to modify different aspects of the wafer, one can create a lot of different designs on a chip. On the other hand, the process of SOIMUMPs is employed in the later steps of fabrication, which is used to do back-etching to form the membrane of microphones. The Overview of SOIMUMPs is shown in Fig.6. In this process of fabrication, the inductively coupled plasma sputtering (ICP) technique is used to back-etch silicon substrate (the thickness is about 525 lm) to the Nitride level, the membrane of micro-phones is accomplished.
Employing the PolyMUMPs process, the fabrication designed for building microphone membranes in this study is started by depositing four layers, including a nitride layer called NITRIDE, three polysilicon layers (POLY0-2), two sacrificial oxide layers (OXIDE1 and OXIDE2), and a gold metal layer named METAL. Hydrofluoric (HF) acid solu-tion is used to release the sacrificial layers of OXIDEs and to form the semi-rigid back plate fabricated by two poly-silicon layers, POLY1-2. The SOIMUMPs process is finally employed for the backside bulk micromachining to form the low-stress nitride thin film (Royer et al. 1983) which is a deformed diaphragm.
In this study, three modifications are made to achieve desired microphone performance. The first one is that the
(a)
(b)
The capacitance and the electrostatic force shares by the TRANS126 element
Fig. 3 a Schematic TRANS126 element b TRANS126 electro-mechanical transducer element
Fig. 4 aFabrication processes for the microphone, b a photo of the fabricated microphone, c a micro-photo on the fabricated microphone by a scanning electron microscopic (SEM)
thickness of layer OXIDE1 is replaced by 3 lm instead of standard 2 lm, in order to enlarge air gap, as shown in Fig.4a. Note that this gap enlargement leads to a signifi-cantly higher sensitivity based on the design Eq. (1), and also avoid sticking during etching release. The second modification is that, unlike the standard POLY1 (Chao et al.2006) which was doped, the layer of POLY1 is un-doped to impose insulation between two charged micro-plates which are the diaphragm and the backplate. The last modification is that the backside etching via the standard SOIMUMPs etches the device through the substrate but stops at the nitride layer. Figure4b shows a number of fabricated microphones in varied designed patterns, while Fig.4c display a micro-photo on top of a given fabricated microphone membrane where holes are seen clearly in a squared membrane patterned successfully by the RIE process.
5 Experimental results
Static and dynamic experiments are conducted, respec-tively, to verify previous analytical predictions and predict microphone performance. In the case with a static input voltage applied to the membrane, the deflection of the entire membrane is measured by a laser Doppler vibrom-eter, Polytec OFV-5000. Figure7a shows the experimental apparatus, while Fig.7b shows the comparison results among the measurements and the simulated results using the aforementioned analytical tool in Sect. 3. With a residual stress of 90 Mpa afore-assumed, the theoretical predictions agree well with experimental counterparts, thus this 90 MPa serves legitimately as the residual stress of the membrane. It is also seen from Fig.7b that the microphone can be safely operated without pull-in occurring up to bias of 16 V, due to high membrane residual stress of 90 MPa assumed. Since commercial microphones used in the mobile device that can only supply bias up to 5 V, the presented microphone can be safely operated in mobile phones without pull-in occurring.
In order to test the sensitivity of the microphone, the device was placed in the anechoic room as shown Fig.8. The interior of the room is covered with SONEX prospec polyurethane composite foam which provides a sound barrier to external noise and internal sound absorption. The speaker is driven with a HAS 4051 amplifier, which inputs driving signals from SRS785 dynamic signal analyzer. A reference calibrated microphone, the JEIC 1051 free-field microphone with preamplifier is used for measuring nomi-nal ambient pressure. The measured MEMS microphone is next connected to a B&K 2671 preamplifier with an internal impedance of 1.5 GX. The preamplifier is connected a bias voltage supply which provides an internal DC polarization voltage for operating the designed MEMS microphone. The output voltage from the preamplifier is recorded using an SRS785 dynamical signal analyzer. Based on the recorded response from the MEMS microphone, the microphone
Fig. 6 Overview of SOIMUMPs (Koester et al.2003)
HAS 4051 Amplifier
Laser Doppler Vibrometer, Polytec OFV-5000 Measured MEMS Microphone dSPACE AD/DA System Measured point (central of the membrane)
(a)
(b)
Fig. 7 aExperimental apparatus, b measured, simulated and analyt-ical results
sensitivity is obtained by subtracting the reference response from the device response plus the calibration output level.
According to the common specifications of the com-mercial microphones used in the mobile device, the oper-ation bias voltage is around 2–5 V. Therefore, two bias voltages, 2 and 4.5 V, are considered for testing the sen-sitivity. Figure9 shows the measured sensitivities of the microphone. Note that since the major design criterion for microphones is to request flatness before 10 kHz to render satisfactory fidelity, thus the measurement presented herein is up to 10 kHz. It is clearly seen from this figure that with a bias voltage of 2 V, the 2 mm-sized microphone renders the sensitivity between -58 and -64 mV/Pa from 100 to 10 kHz, the audible frequency range. With bias of 4.5 V, the microphone exhibits sensitivities between -58 and -65 mV/Pa, from 100 to 10 kHz. As compared to the presented MEMS microphones, the sensitivity performance in this study, particularly 1.7 mV/Pa under bias 4.5 V, is found slightly better than those devices presented in the initial stage of MEMS microphone development (Bergqvist and Rudolf 1990; Scheeper et al. 1992, 1994) [around 1 mV/Pa with bias 2 V and above, as shown in (Scheeper et al. 1992)], while compatible to some most recent developed devices (Lee et al. 2012; Ganjia and Majlis
2009). The devices developed by the current study is however inferior to those recently developed in academia (Pedersen et al. 1998; Hsu et al. 1998) and industry (Knowles Company2013), by which the sensitivity could reach 10 mV/Pa, with an integrated CMOS pre-amplifier and/or low residual stress in microphone membrane. Note that the current study aims mainly to showcase the
feasibility of a hybrid process combining two available stable PolyMUMP and SOIMUMP to increase yield rate. The sensitivities and signal-to-noise ratio (SNR) can be further increased in the future by a better readout circuitry and techniques of signal processing.
As to the bandwidth of the microphone, also shown in Fig.9, the presented microphone offers relatively linear response up to 10 kHz, the performance of which is com-patible to other already existing ones. In fact, under the measured audible range, the proposed device with bias 4.5 V maintains the loss as less as 2.7 dB (ref. to V/Pa), under 3 dB—the commonly acceptable drop before the upper limit of the audible frequency range, 10 kHz (Pedersen et al.1998). However, small levels of fluctuation in both responses in Fig. 9along the increasing frequency are present, which could cause distortion in output sound signal. This is due to the low fabrication quality by
Fig. 8 Experimental apparatus in anechoic room
Fig. 9 Frequency response of a 2.0 mm wide microphone a with a bias voltage of 4.5 V, the microphone exhibits sensitivity between -65 and -58 dB from 100 to 10 kHz, b with a bias voltage of 2 V, the 2 mm-wide microphone has a sensitivity between -58 and -64 dB
preliminary fabrication trails, which leads to nonlinear, asymmetric and non-uniform distributions of mechanical/ electrical properties over the membrane, such as membrane thickness and residual stress. It is possible to improve fabrication quality along the development of mass pro-duction. Note lastly on frequency response that the phase response is not presented and discussed herein. This is due to the fact that the major cause of phase delay for a microphone unit comes from electronic readout circuit, not the device membrane (Chiang et al.2012), since only the factor of material damping of the membrane in the device leads to phase delay, which is usually much less than that caused by readout circuitry.
6 Conclusion and future works
A single-chip-processed silicon condenser microphone has been designed and fabricated. The microphone with a thin membrane of 2 9 2 mm2in area and a thick backplate are fabricated. The analysis via finite-element (FEM) and equivalent circuit are successfully conducted to design and evaluate the microphone. The integration of standard PolyMUMPs and SOIMUMPs processes is proposed, where the sacrificial layer sizes are re-assigned to suc-cessfully develop a microphone structure with better sen-sitivity. Static membrane deflection are also simulated and measured by computer software and a laser Doppler vib-rometer, respectively. It is confirmed that no pull-in occurs for normal microphone operations.
The sensitivity of the fabricated microphone within audible range is measured. The measured sensitivity of the microphone is around 0.78 and 1.7 mV/Pa from 100 to 10 kHz, under the biases of 2 and 4.5 V, respectively. This sensitivity performance is compatible to some microphones presented in the past. The device also maintains the loss as less as 2.7 dB (ref. to V/Pa), under 3 dB—the commonly acceptable drop within the audible frequency range. To increase the sensitivity in the future, the residual stress of the membrane ought to be reduced by lowering deposition temperature of membrane material. Further efforts are to improve fabrication quality, for lessening negative effects of nonlinear, asymmetric and non-uniform distributions of mechanical/electrical properties over the membrane, and also to perform drop tests (Meng et al.2012) for fortify the device structure against accidental drops.
Acknowledgments The authors appreciate the support from National Science Council of R.O.C under the grant no. NSC 101-2623-E-009-006-D and 100-2221-E-009-091-. This work was also supported in part by the UST-UCSD International Center of Excellence in Advanced Bio-Engineering sponsored by the Taiwan National Science Council I-RiCE Program under Grant NSC-100-2911-I-009-101.
References
Bergqvist J (1993) Finite-element modeling and characterization of a silicon condenser microphone with a highly perforated back-plate. Sens Actuators 39:191–200
Bergqvist J, Rudolf F (1990) A new condenser microphone in silicon. Sens Actuators 21:123–125
Chan C-K, Lai W-C, Wu M, Wang M-Y, Fang W (2011) Design and implementation of a capacitive-type microphone with rigid diaphragm and flexible spring using the two poly silicon micromachining processes. IEEE Sens J 11:2365–2371 Chao PC-P, Chiu C-W, Tsai C-Y (2006) A novel method to predict the
pull-in voltage in a closed form for micro-plate actuated by a distributed electrostatic force. J Micromech Microeng 16:986–998 Chao PC-P, Chiu C-W, Liu T-H (2008) DC dynamic pull-in predictions for a generalized clamped–clamped micro-beam based on a continuous model and bifurcation analysis. J Micromech Microeng 18:14 Chen J-M, Kuo C-F, Wang D-B, Huang C-W, Sun S-C, Hsieh Y-S,
Zhon W-Z (2010) Low bias voltage and high sensitivity CMOS condenser microphone using combined stress relaxation design. The 5th International Conference on Microsystems Packaging Assembly and Circuits Technology, pp 1–3
Chiang C-T, Wang C-H, Wu C-Y (2012) A CMOS MEMS audio transducer implemented by silicon condenser microphone with analog front-end circuits of audio codec. IEEE Trans VLSI Syst J 20(9):1656–1667
Ganjia BA, Majlis BY (2009) Design and fabrication of a new MEMS capacitive microphone using a perforated aluminum diaphragm. Sens Actuat A 149:29–37
Hsu PC, Mastrangelo CH, Wise KD (1998) A High sensitivity polysilicon diaphragm condenser microphone. In: The 11th Annual Interna-tional Conference on MEMS, Heidelberg, pp 580–585
Knowles Company (2013) Datasheet of the SP-series MEMS Microphone,http://www.knowles.com/search/products/m_surface_ mount.jsp
Koester ID, Cowen A, Mahadevan R, Stonefield M, Hardy B (2003) PolyMUMPsTMDesign Handbook. Revision 10.0, MEMSCAP Koester ID, Cowen A, Mahadevan R, Stonefield M, Hardy B (2003)
SOIMUMPsTMDesign Handbook. Revision 3.0, MEMSCAP Lee J, Je CH, Yang WS, Kim Y-G, Cho M-H, Kim J (2012) Thin
MEMS microphone based on package-integrated fabrication process. Electron Lett 48(14):29–37
Leinenbach C, van Teeffelen K, Laermer F, Seidel H (2010) A new capacitive type MEMS microphone. The IEEE 23rd Interna-tional Conference on MEMS, pp 659–662
Meng J, Mattila T, Dasgupta A, Sillanpaa M, Jaakkola R, Andersson K, Hussa E (2012) Testing and multi-scale modeling of drop and impact loading of complex MEMS microphone assemblies. The 13th International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems (EuroSimE) vol 14. pp 1/8–8/8
Pedersen M, Olthuis W, Beergvld P (1998) High performance condenser microphone with fully integrated CMOS amplifier and DC–DC voltage converter. JMEMS 7:387–394
Royer M, Holmen JO, Wurm MA, Aadland OS, Glenn M (1983) ZnO on Si integrated acoustic sensor. Sens Actuators 4:357–362 Scheeper PR, van der Donk AGH, Olthuis W, Bergveld P (1992)
Fabrication of silicon condenser microphones using single wafer technology. JMEMS 1:147–154
Scheeper PR, van der Donk AGH, Olthuis W, Bergveld P (1994) A review of silicon microphones. Sens Actuators 44:1–11 Senturia SD (2000) Microsystems Design. Kluwer Academic
Pub-lishers, pp 397–428
Wang WJ, Lin RM, Zou QB (2004) Modeling and characterization of a silicon condenser microphone. JMEMS 14:403–409
