Abstract
This paper presents radio frequency (RF) microelectromechanical system (MEMS) filters with extremely high bandwidth widening capability. The proposed filtering topologies include hybrid configurations consisting of piezoelectric MEMS resonators and surfacemounted lumped elements. The MEMS resonators set the center frequency and provide electromechanical coupling to construct the filters, while the lumpedelementbased matching networks help widen the bandwidth (BW) and enhance the outofband rejection. Aluminum nitride (AlN) S0 Lamb wave resonators are then applied to the proposed filtering topologies. AlN S0 first and secondorder wideband filters are studied and have shown prominent performance. Finally, the AlN S0 firstorder wideband filter is experimentally implemented and characterized. The demonstrated firstorder filter shows a large fractional bandwidth (FBW) of 5.6% (achieved with a resonator coupling of 0.94%) and a low insertion loss (IL) of 1.84 dB. The extracted bandwidth widening factor (BWF) is 6, which is approximately 12 times higher than those of the current ladder or lattice filtering topologies. This impressive bandwidth widening capability holds great potential for satisfying the stringent BW requirements of bands n77, n78, and n79 of 5G new radio (NR) and will overcome an outstanding technology hurdle in placing 5G NR into the marketplace.
Introduction
The proliferation of 5G communications has led to the allocation of wider bands for faster data transmission and processing^{1}. Radio frequency (RF) bandpass filters, which are used to select the desired signals while suppressing the unwanted signals, are the keys to define these frequency bands. Therefore, the realization of filters with a wide bandwidth (BW) is attracting tremendous attention. Filters based on various technologies, including microstrip, metal cavity, dielectric, lumped element, and acoustic or microelectromechanical systems (MEMS), have all been developed in recent years. Among these technologies, MEMS filters stand out for their compact size, low cost, and excellent filtering performance (especially the sharp rolloff). These advantages make MEMS filters the most promising solution to 5G filtering applications. Researchers have made extensive investigations of MEMS filtering technologies in three main directions: exploiting resonant modes with higher coupling (k_{t}^{2})^{2,3,4}, exploring strong piezoelectricity materials (e.g., AlScN^{5,6,7}, LiNbO_{3}^{1}), and hybridizing acoustic resonators to expand the bandwidth^{8,9,10}. For example, the shear horizontal (SH0) mode and firstorder asymmetric (A1) mode in lithium niobate (LiNbO_{3}) have been demonstrated with high k_{t}^{2} values of 50% and 22%, respectively^{11,12,13}. Acoustic resonators with these modes can be used to construct wideband filters^{14,15}. However, the severe powerhandling issue and temperature instability make the two modes still not promising enough for wideband or highfrequency applications. Nevertheless, AlN is the first choice in piezoelectric resonators^{16,17,18} and sensors^{19} since it has a high powerhandling capability and excellent thermal stability and offers a mature process that can be integrated with complementary metaloxidesemiconductor (CMOS) integrated circuits (ICs)^{20}. Scandium (Sc) has been studied for its doping with AlN to increase the piezoelectric coefficients that determine the couplings of the resonators^{21,22}. It has been reported that doped Sc_{0.12}Al_{0.88}N can increase the piezoelectric coefficient d_{33} by 50% and reduce the stiffness constant c_{33} by 10%^{23}. Thanks to this achievement, k_{t}^{2} is improved by 1.7 times. Nevertheless, the doped AlN always comes with the degradation of the quality factor, which directly affects the insertion loss (IL) of the filters. In addition to the two developments above, researchers have also made efforts to hybridize filters using quartz surface acoustic wave (SAW) resonators and lumped elements or microwave transmission lines^{8,9,10}. However, these hybridized filters suffer from small bandwidth (BW), poor rolloff, and low outofband rejection.
To solve the above issues, this paper proposes AlN MEMS filters with extremely high bandwidth widening capabilities. The wideband filters utilize AlN Lamb wave resonators in conjunction with surfacemounted lumped elements. Compared with conventional ladder and lattice topologies, the proposed filters show a much larger bandwidth widening capability.
This paper is organized as follows: the Method and Modeling section presents the models of the two filtering topologies: firstorder and secondorder wideband filters. Both filters are designed based on a general MEMS resonator. The Material and Device section illustrates the AlN MEMS firstorder and secondorder wideband filters. It begins with the introduction of the AlN S0 Lamb wave resonator, followed by the implementation of the two wideband filtering topologies. The Results and Discussion section shows the experimental realization of the AlN firstorder wideband filter. The filter fabrication, assembly, and characterization results are covered. The final section concludes this paper with some potential future endeavors.
Method and modeling
Firstorder wideband filter
The filter construction starts with the MEMS resonator. The electrical response of a MEMS resonator is often represented with the modified Butterworth–Van Dyke (MBVD) model, in which C_{0} is the static capacitance of the resonator and L_{m}, C_{m}, and R_{m} are the motional inductance, motional capacitance, and motional resistance, respectively. Extraction of the MBVD parameters can be found in many literature reports^{24,25,26,27,28,29}. The motional branch forms the mechanical resonance equivalently in the electrical domain via piezoelectricity. As an example, the measured response of a MEMS resonator is plotted and fitted in Fig. 1a. The demonstrated resonator has a typical resonant frequency f_{s} and an antiresonant frequency f_{p}. The quality factor and the piezoelectric coupling factor are two important parameters for the performance of the resonators, which can be related to the MBVD model parameters. Therefore, the MBVD model is extracted from the measured response and will be used as the basis for the following filter analysis.
As shown in Fig. 1d, the proposed firstorder filter is obtained from the combination of the topology in Fig. 1b, c, which consists of a MEMS resonator, a parallel inductor L_{0}, and a pair of threelumpedelement matching networks, which includes a series inductor L_{s}, a shunt inductor L_{p}, and a shunt capacitor C_{p}. When the MEMS resonator parallel with the inductor L_{0}, two symmetric transmission zeros f_{a1} and f_{a2} are generated, as indicated in Fig. 1b. The filter is centered at the same resonant frequency (f_{s}) as the MEMS resonator, and it features a large bandwidth and two deep transmission zeros (TZs).
The series resonance frequency f_{s} is the frequency where L_{m} and C_{m} cancel each other out, while the antiresonance frequency f_{p} arises when C_{0} and C_{m} collectively cancel out L_{m}. The parallel L_{0} is added to decouple the antiresonance and generate two symmetric TZs: f_{a1} and f_{a2}. The two TZs are determined by the MBVD model and L_{0}, as shown in Fig. 1b and computed as
where
If we set, ω_{s} = ω_{s0}, then
Then, Eq. (1) can be rewritten as
Since the piezoelectric coupling factor of the resonator device is proportional to C_{m}/C_{0}^{3} and the TZs determine the maximum bandwidth of the filter, Eq. (3) implies that a higher bandwidth widening capability can be obtained by using resonators with a larger piezoelectric coupling factor.
As demonstrated in Fig. 1c, the lumpedelement matching network consists of a series inductor L_{s}, a shunt inductor L_{p}, and a shunt capacitor C_{p}. The purpose of the matching networks is to widen the bandwidth of the filter. The threelumped elements resonate at f_{s} and are regulated by
Only when L_{s}, L_{p}, and C_{p} obey Eq. (4) can we obtain a functional matching network. If any two of three are determined, then the third one can be acquired by Eq. (4). Therefore, only two parameters are independent, and the left parameter is dependent. In later discussion, L_{s} and L_{p} will be chosen as independent variables and C_{p} as the dependent variable.
Finally, as shown in Fig. 1d, the proposed filter is formed by combining the two topologies in Fig. 1b, c. In this filtering topology, the MEMS resonator and L_{0} provide sharp rolloff, TZs, and set the center frequency of the filter, while the matching network, which in essence is an LC bandpass filter, offers the wide bandwidth for the filter. The filtering topology enjoys highperformance flexibility by tuning the lumped elements in the matching networks. As shown in Fig. 2b, when L_{s} and L_{p} are given a different value while C_{p} is determined by Eq. (4) L_{s} and L_{p} have different effects on the filter bandwidth and outofband rejection. In general, larger L_{s} and L_{p} lead to a greater fractional bandwidth (FBW), while L_{p} has a greater impact on outofband rejection than L_{s}, and a smaller L_{p} corresponds to greater outofband rejection. Q_{s} of the resonator and the lumped elements together determine the IL of the filter. By leveraging L_{s} and L_{p}, it is possible to achieve wide bandwidth and excellent outofband rejection simultaneously.
As an example of the tunability, Fig. 2a demonstrates three filters (named Filters A, B, and C) with different matching networks (for quantification purposes, we assume f_{s} = 229 MHz and f_{p } = 231 MHz based on a measured MEMS resonator; k_{t}^{2} is then computed to be 1.7%). The values of the threelumped elements L_{s}, L_{p}, and C_{p} are listed in Fig. 2c. Among the three filters, Filter A has the largest L_{s} and largest L_{p}. C_{p} is therefore the smallest. As explained above, a larger L_{s} and L_{p} produce a wider BW, and a larger L_{p} leads to smaller outofband rejection. Therefore, Filter A is assumed to have the largest BW and the smallest outofband rejection, which is verified by the simulation in Fig. 2a. Filter C has the smallest L_{s} and L_{p}, so it has the smallest BW and the largest outofband rejection. The comparison of the performance of the three filters is concluded in Fig. 2c. Figure 2b illustrates the simulated filtering FBW performance versus different L_{s} and L_{p} values; the dashed lines show the outofband rejection values for guidance.
The proposed firstorder filters stand out for their wide FBWs. To justify their bandwidth widening capabilities, the three filters (A, B, and C) are compared with existing widely used ladder and lattice filtering topologies. The bandwidth widening factor (BWF), defined as the quotient of the FBW and the resonator’s electromechanical coupling, is adopted for evaluation. Conventional ladder and lattice topologies^{27}, which are shown in Fig. 1e, f, typically achieve FBWs of 1/3–1/2 of the resonator’s coupling^{27,28}. Taking the assumed MEMS resonator that has an electromechanical coupling of 1.7%, for example, the FBWs of the ladder and lattice topologies are 0.78% and 0.95%, respectively. Figure 2d compares the BWFs of the three firstorder filters and the ladder and lattice topologies. Apparently, the proposed firstorder filters have the largest BWF of 4.2, which is 9.1 and 7.5 times higher than the ladder and lattice topologies, respectively.
Though showing a large FBW, the rolloff and outofband rejection of the firstorder wideband filter still needs enhancement. For further improvement, a secondorder wideband filter is then proposed.
Secondorder wideband filter
As shown in Fig. 3a, the secondorder wideband filter is composed of two series of firstorder wideband filters. The two firstorder wideband filters are designed to be image symmetrical for simultaneous matching of the two ports. In order to have more freedom of matching impedance, one side of the matching network is changed to be L_{s2}, L_{p2}, and C_{p2}. The roles of L_{0}, L_{s1}, L_{p1}, C_{p1}, L_{s2}, L_{p2}, and C_{p2} are the same as those described in the firstorder wideband filter, and they are also regulated by Eqs. (2) and (4). The secondorder wideband filter features a much better performance of outofband rejection and rolloff. In addition, the FBW has also been improved. The simulated response of the secondorder wideband filter in Fig. 3b demonstrates a large FBW of 6.7% and an extremely high outofband rejection of over 60 dB. Moreover, the rolloff is also significantly enhanced, and the 30dB shape factor is only 1.05.
The secondorder wideband filter enjoys favorable performance flexibility since it has four tuning elements (L_{s1}, L_{p1}, L_{s2}, and L_{p2}), while the firstorder filter has only two tuning elements (L_{s} and L_{p}). Similarly, to demonstrate the tuning capability, another three secondorder wideband filters (Filters D, E, and F) are constructed using different matching networks. The values of the lumped elements of each filter are listed in Fig. 3c. The functions of L_{s1}, L_{s2}, L_{p1}, and L_{p2} are similar to those in the firstorder case. Filter D has the largest L_{s1}, L_{s2}, L_{p1}, and L_{p2}, so it has the widest FBW but the smallest outofband rejection (Fig. 3b). Filter F has the smallest L_{s1}, L_{s2}, L_{p1}, and L_{p2}, which corresponds to the smallest FBW but the highest outofband rejection. Filter E has intermediate values of L_{s1}, L_{s2}, L_{p1}, and L_{p2}, and its FBW and outofband rejection are between Filter Ds and Filter Fs. Overall, the secondorder filters exhibit much better outofband rejection and rolloff than the firstorder filters.
The bandwidth widening effect of the secondorder wideband filters is also researched and compared in Fig. 3d. Filter D has the highest BWF of 4.6 because it uses the largest L_{s1}, L_{s2}, L_{p1}, and L_{p2}. This BWF could be further increased if higher L_{s1}, L_{s2}, L_{p1}, and L_{p2} were applied. Of course, the compromise will be the degradation of outofband rejection. Filter F has the smallest BWF of 2.2, but its outofband rejection is as high as 76 dB. Compared with the firstorder wideband filter C, which has the same BWF of 2.2, the secondorder wideband filters achieve much better outofband rejection performance.
The MEMS resonators in the proposed firstorder and secondorder filters are constructed by the MBVD model. There are no specific requirements of the types of resonators. The MEMS resonators could be various MEMS resonators, such as AlN Lamb wave resonators, SAW resonators, FBARs, or LiNbO_{3} A1 resonators. To validate the feasibility and achieve multiplefrequency filters, the following section will replace the general MEMS resonator with a concrete AlN Lamb wave resonator.
Material and device design
AlN Lamb wave resonator
AlN has emerged as the most suitable material for the transduction of acoustic waves because of its excellent performance and manufacturability. Engineered AlN has shown desirable properties of large phase velocity, high thermal conductivity, low acoustic loss, and relatively small temperature coefficients of frequency (TCFs)^{19}, which make it preferable for resonator applications. AlN S0 Lamb wave resonators have attained great success in terms of high resonant frequency, great powerhandling capability, high Q, and lowfrequency drift^{30,31,32}. Most importantly, Lamb wave resonators make it possible to integrate multiplefrequency resonators^{26} or filters^{33} on piezoelectric films of the same thickness. However, the relatively low electromechanical coupling (< 2%) of the AlN S0 Lamb wave resonators limits their use in RF wideband filters. As analyzed, the proposed wideband filtering topologies can significantly extend the FBWs, so in this section, AlN MEMS wideband filters will be designed and implemented.
As illustrated in Fig. 4a, the designed AlN S0 Lamb wave resonator is comprised of top interdigitated transducers (IDTs), a suspended AlN thin film, and a bottom electrode (BE). The top IDTs are alternatingly connected to the ground and RF signal, and the BE is electrically floating. The resonator is designed with four tethers at the four corners to increase the structure robustness and powerhandling capability. The resonant frequency of the intended S0 mode is designed to be approximately 450 MHz. The BE and top IDTs are chosen to be 100 nm Pt and 150 nm Al, respectively. The middle AlN thin film is approximately 1 μm thick. The pitch width of the IDTs is 10 μm, which forms the designed resonant frequency of approximately 450 MHz. The parameters of the AlN S0 Lamb wave resonator are described in Fig. 4c.
Figure 4b shows the fabrication process of the AlN S0 resonator. It starts with a highresistivity Si wafer. Pt (100 nm) is deposited by evaporation and then lifted off with negative photoresist AZ5214E. Then, 1 μm thick AlN is reactively sputtered, followed by IDT lithography with photoresist SPR220. The 150 nm Al is sputtered and lifted off as the top IDTs. A hard mask of SiO_{2} is chosen to define the AlN film by inductively coupled plasmareactive ion etching (ICPRIE). After the ICPRIE, the remaining SiO_{2} is removed in HF.
Finally, the resonator is exposed to XeF_{2} for release. Figure 5a–d shows the scanning electron microscope (SEM) images of the fabricated device and its zoomedin structures.
The fabricated resonator was measured in dry air with an Agilent PNAL 5230A at room temperature. The measured admittance result is shown in Fig. 5e. The resonator has a measured resonant frequency at 446 MHz, an electromechanical coupling of 0.94% and a Q of 1500. The resonator’s measured response is then fitted by the MBVD model. The fitted MBVD L_{m}, C_{m}, R_{m}C_{0} parameters are listed in the inset of Fig. 5e and will be used in the AlN S0 first and secondorder hybrid wideband filters.
AlN S0 firstorder wideband filter
The principle of the firstorder wideband filter has been introduced in the Method and Modeling section. In this section, we will replace the general MEMS resonator in Fig. 1d with the fabricated AlN S0 Lamb wave resonator, as indicated in Fig. 6a. The AlN S0 Lamb wave resonator is modeled by using the MBVD circuit lumped elements listed in the inset of Fig. 5e. According to Eq. (2), the parallel L_{0} is calculated to be 188 nH. For the other threelumped elements, L_{s} and L_{p} are chosen to be 160 nH and 145 nH, respectively, and C_{p} is computed to be 1.7 pF (shown in the inset of Fig. 6b). Generally, inductors and capacitors are lossy due to their finite Qs. According to the datasheets of current lumpedelement vendors, the Qs of L_{0}, L_{s}, L_{p}, and C_{p} are assumed to be 87, 87, 92, and 400, respectively.
The simulation result of the AlN S0 firstorder wideband filter is illustrated in Fig. 6b. The filter is centered at 446 MHz with a low IL of 1.8 dB, a high FBW of 5.6%, and an outofband rejection of 24 dB. As noted, the electromechanical coupling of the AlN S0 resonator is only 0.94%. However, it reveals a very high BWF of 6 after implementing the firstorder filter topology.
The outofband rejection of the AlN S0 firstorder wideband filter is moderate. For improvement, an AlN S0 secondorder wideband filter is then evaluated below.
AlN S0 secondorder wideband filter
The topology of the AlN S0 secondorder filter is similar to that of the secondorder filter in Fig. 3a. Shown in Fig. 6c, the AlN S0 secondorder wideband filter consists of two series of AlN S0 firstorder filters. L_{0} is the same as that in the AlN S0 firstorder filter. The lumped elements (L_{s1}, L_{p1}, C_{p1}, L_{s2}, L_{p2}, and C_{p2}) are adjusted accordingly for better impedance matching. Figure 6d shows the simulated performance of the AlN S0 secondorder filter. The values of the lumped elements used in the filter are provided in the inset of Fig. 6d. The secondorder filter is also centered at 446 MHz. It has an IL of 2.6 dB, a wide FBW of 5.1%, and very high outofband rejection (e.g., 43 dB at 300 MHz). The 2.6 dB IL is higher than the 1.8 dB in the AlN S0 firstorder filter because of the additional cascaded loss from the lumped elements. This 2.6 dB IL can be reduced if highQ lumped elements are available or at higher frequencies where highQ lumped elements are attainable. In short, the secondorder filters have much higher outofband rejection than the firstorder filters, and the payoff is the complexity of the circuit as well as the cost of extra lumped components.
Results and discussion
Implementation of the AlN MEMS filter
Due to the resource limitation and the similarity among the proposed filters, we choose only the AlN S0 firstorder wideband filter to implement. The dimensions of the fabricated AlN S0 firstorder wideband filter are listed in Fig. 4c. The filter is implemented by first fabricating the AlN S0 Lamb wave resonator and then integrating it with the matching networks on a printed circuit board (PCB). The fabrication of the AlN S0 resonator utilizes a process described in detail in Fig. 4b. The AlN S0 resonators are fabricated on a resonator chip. It is then cut into smaller pieces for later PCB mounting.
As shown in Fig. 7a, the filter assembly process starts with PCB preparation, which includes size planning and surface cleaning. The chosen PCB is based on an FR4 with a thickness of 1.524 mm, a dielectric constant (ε_{r}) of 4.8, a dielectric loss tangent tan (δ_{D}) of 0.008, and a copper cladding thickness of 18 μm. Integration generally requires wire bonding (step 7) at 120 °C. However, the copper cladding would be oxidized very quickly and make the bonding fail. To avoid this, 50 nm platinum (Pt) and 250 nm gold (Au) are continuously evaporated on the PCB. The Pt works as an adhesion layer, while the Au layer has the twofold purpose of antioxidation and easy bonding (a gold substrate usually provides the best bonding ease). Subsequently, the copper is milled to form the signal traces and bonding pads. For easy bonding, a bonding box with a depth of 600 nm is opened in the center of the PCB. The bounding box helps the top surfaces of the resonator chip and the PCB stay on the same plane. It also ensures that the bonding wires do not touch the corner of the chip. The lumped elements of the surface mount inductors and capacitors are then soldered on the PCB with a hot air gun and soldering paste. The resonator chip is attached to the bonding box and then wire bonded to the signal traces on the PCB. Finally, two SMA connectors are installed for measurement.
The PCB assembly has considerable influence on the filter. There are several aspects we need to monitor so that minimal PCB influence is introduced. First, the parasitic effect from the PCB should be reduced. In our circuit model in Fig. 6a, the filter contains the lumped elements and AlN MEMS resonators only. There is no parasitic content in the filter circuit. However, in the real case, when we assemble the filter to the PCB, the parasitic effect always exists and influences the filter performance. For lowfrequency filters, the parasitic effect is not that significant and can be compensated by tuning the threelumped elements (tuning C_{p} would be a good option). For highfrequency filters, the parasitic effect becomes dominant, which requires an electromagnetic environment to be applied to the filter circuit. Second, the grounding pads on the PCB are key. Grounding is critical because it generates different levels of parasitic capacitance. To achieve a good PCB assembly, the grounding pads on the PCB must be carefully designed so that minimal parasitic capacitance is generated. Third, the bonding wires introduce parasitics. Bonding wires can be thought of as a kind of inductor, which can also exert an influence on the filter. One has to plan the assembly beforehand so that the least length of bonding wire is needed.
Figure 7 demonstrates the assembly process and images of the fabricated filter with some integration details. The fabricated filter (Fig. 7b) has dimensions of 1.85 cm × 0.85 cm. This dimension is smaller than that of most microstrip filters but still larger than those of pure acoustic filters (SAW or BAW filters). The main causes of the relatively large size are dielevel bonding and manual mounting, which require additional operation space. By using a more compact flipchip approach that involves chipscale lumped elements, the form factor of the filter can be substantially improved. Figure 7c shows the AlN resonator chip featuring arrays of resonators. The target AlN S0 resonator is at the top center as circled. Figure 7d is the magnified view of the bonded AlN S0 resonator. The two bonding wires connect the two ports (input and output ports) of the resonator. The two wires start from the resonator and end on the bonding pads on the PCB, as shown in Fig. 7e.
Characterization results
The fabricated AlN S0 firstorder wideband filter is measured by an Agilent PNAL 5230 A analyzer in dry air and at room temperature. The measurement results are shown in Fig. 8a. The filter demonstrates a low IL of 1.84 dB, a wide FBW of 5.6%, and outofband rejection of 24 dB. Overall, the measurement agrees well with the simulation results shown in Fig. 6b. As noted, there are two mismatches: the depth of the two TZs and the spurious response. The measurement has shallower TZs than the simulation. This can be attributed to the fact that the actual Q of L_{0} is lower than the simulated Q provided by the vendors. Generally, a higher Q_{L0} gives a deeper TZ depth. The spurious response in the measurement comes from the spurious modes of the resonator. The simulation adopts the MBVD model, which does not incorporate spurious modes. This explains why we see the spurious response in the measurement only. Techniques have been reported to suppress the spurious modes in AlN Lamb wave resonators^{34,35,36,37}. By applying these spurious mode suppression techniques, the spurious response of the filter could be eradicated.
Figure 8b plots the measured response of the filter over a wider frequency range between 200 and 1200 MHz. As seen, the outofband rejection on the left has no fly off. This advantage overcomes the longexisting fly off issues in microstrip filters or purely acoustic filters. In addition, the outofband rejection on the right continuously increases up to 75 dB. Additionally, there are no spurious harmonic responses. This merit overtakes its counterparts of microstrip filters and purely acoustic filters.
Finally, it is worthwhile to notice the bandwidth widening capability of the achieved filter. The coupling of the AlN S0 Lamb wave resonator is 0.94%, and the achieved FBW of the filter is 5.6%. Therefore, the BWF is calculated to be 6.0, which is approximately 12 times higher than that of the conventional ladder or lattice topologies. As predicted from Eq. (3), a higher FBW can be achieved by implementing this bandwidth widening technique on resonators with a higher piezoelectric coupling factor.
Conclusion
Wideband MEMS filters have been analyzed, designed, fabricated and characterized in this work. Two filtering topologies (first and secondorder) are first proposed and researched. Then, AlN S0 Lamb wave resonators are applied to the filtering topologies and show excellent performance of low IL, wide bandwidth, and high outofband rejection. To validate the simulated filters, the AlN S0 firstorder filter is chosen for implementation, which progresses from the resonator fabrication to the final filter assembly. The demonstrated AlN S0 firstorder filter has a high FBW of 5.6%, a low IL of 1.84 dB, and an outofband rejection larger than 24 dB. Considering the low coupling (0.94%) of the AlN S0 resonator, the BWF is as high as 6, which is approximately 12 times higher than that of the ladder or lattice topologies. Furthermore, the demonstrated AlN S0 firstorder filter features farfrequency suppression without spurious responses. The presented hybrid filters hold great potential for various filtering applications, especially for the current 5G NR bands of n77, n78, and n79.
Moreover, very recently, highfrequency onchip rolledup inductors with miniaturized sizes have been reported^{38,39}. The selfrolledup inductor can achieve a maximum Q factor of over 12 at 3.5 GHz and 10 nH inductance with a footprint area of only 15 × 19 μm^{2}, which is 0.1% of that of planar spiral inductors^{38}. Their fabrication process could be incorporated with that of the Lamb wave resonators. In addition, it has been shown that piezoelectric resonators can be integrated with capacitors on the same membrane^{40}, especially on AlN^{41}, since the AlN thin film sputtering and resonator device fabrication process is compatible with CMOS ICs. Therefore, it is possible to achieve the inductances L_{0}, L_{s}, and L_{p} with these rolledup inductors and the capacitance C_{p} with the AlN Lamb wave resonators and finally attain multiplefrequency onchip hybrid filters. This will be our future development.
References
 1.
Plessky, V. et al. 5 GHz laterallyexcited bulkwave resonators (XBARs) based on thin platelets of lithium niobate. Electron. Lett.55, 98–100 (2019).
 2.
Gao, A., Zou, J. & Gong, S. A 3.5 GHz AlN S1 Lamb mode resonator. In 2017 IEEE International Ultrasonics Symposium 1–4 (2017).
 3.
Cassella, C., Hui, Y., Qian, Z., Hummel, G. & Rinaldi, M. Aluminum nitride crosssectional Lamé mode resonators. IEEE J. Microelectromech. Syst.25, 275–285 (2016).
 4.
Cassella, C. et al. Super high frequency aluminum nitride twodimensionalmode resonators with kt2 exceeding 4.9%. IEEE Microw. Wirel. Compon. Lett.27, 105–107 (2017).
 5.
Colombo, L. et al. Investigation of 20% scandiumdoped aluminum nitride films for MEMS laterally vibrating resonators. In 2017 IEEE International Ultrasonics Symposium 1–4 (2017).
 6.
Zhu, Y. et al. ScAlNbased LCAT mode resonators above 2 GHz with high FOM and reduced fabrication complexity. IEEE Electron Device Lett.38, 1481–1483 (2017).
 7.
Wang, N. et al. Over 10% of keff2 demonstrated by 2GHz spurious modefree Sc_{0.12}Al_{0.88}N laterally coupled alternating thickness mode resonators. IEEE Electron Device Lett.40, 957–960 (2019).
 8.
Psychogiou, D., Gómezgarcía, R., Sánchez, R. & Peroulis, D. Hybrid acousticwavelumpedelement resonators (AWLRs) for highbandpass filters with quasielliptic frequency response. IEEE Trans. Microw. Theory Tech.63, 2233–2244 (2015).
 9.
Psychogiou, D., Gómezgarcía, R., Sánchez, R. & Peroulis, D. Couplingmatrixbased design of highbandpass filters using acousticwave lumpedelement resonator (AWLR) modules. IEEE Trans. Microw. Theory Tech.63, 4319–4328 (2015).
 10.
Lu, X., Mouthaan, K. & Soon, Y. Wideband bandpass filters with SAWfilterlike selectivity using chip SAW resonators. IEEE Trans. Microw. Theory Techn.62, 28–36 (2014).
 11.
Kadota, M., Ogami, T. & Kimura, T. Wide band resonators using SH Mode of plate wave on LiNbO_{3}. In 2012 IEEE International Ultrasonics Symposium. 563–567 (2012).
 12.
Yang, Y., Gao, A., Lu, R. & Gong, S. 5 GHz lithium niobate MEMS resonators with high FoM of 153. In 2017 IEEE International UltrasonicsSymposium. 942–945 (2017).
 13.
Song, Y. & Gong, S. Spurious mode suppression in SH0 lithium niobate laterally vibrating MEMS resonators. In 2015 IEEE International Electron Devices Meeting. 18–5 (2017).
 14.
Song, Y. & Gong, S. Wideband spuriousfree lithium niobate RFMEMS filters. J. Microelectromech. Syst.26, 820–828 (2017).
 15.
Kadota, M. & Tanaka, S. Solidly mounted ladder filter using shear horizontal wave in LiNbO_{3}. In 2016 IEEE International Ultrasonics Symposium. 1–4 (2016).
 16.
Mateen, F., Boales, J., Erramilli, S. & Mohanty, P. Micromechanical resonator with dielectric nonlinearity. Microsyst. Nanoeng.4, 1–7 (2018).
 17.
Mateen, F., Maedler, C., Erramilli, S. & Mohanty, P. Wireless actuation of micromechanical resonators. Microsyst. Nanoeng.2, 1–6 (2016).
 18.
Boales, J. A., Mateen, F. & Mohanty, P. Optical wireless information transfer with nonlinear micromechanical resonators. Microsyst. Nanoeng.3, 1–6 (2017).
 19.
Qian, Z. et al. Graphene–aluminum nitride NEMS resonant infrared detector. Microsyst. Nanoeng.2, 1–7 (2016).
 20.
Fischer, A. C. et al. Integrating MEMS and ICs. Microsyst. Nanoeng.1, 1–16 (2015).
 21.
Sano, K., Karasawa, R. & Yanagitani, T. High electromechanical coefficient kt2=19% thick ScAlN piezoelectric films for ultrasonic transducer in low frequency of 80 MHz. In 2017 IEEE InternationalUltrasonics Symposium. 1–4 (2017).
 22.
Carlotti, G., Sadhu, J. & Dumont, F. Dependence of the different elastic constants of ScAlN films on Sc content: a Brillouin scattering study with polarization analysis. In 2017 IEEE International Ultrasonics Symposium. 1–5 (2017).
 23.
Matloub, R. et al. Piezoelectric Al1−xSc_{x}N thin films: a semiconductor compatible solution for mechanical energy harvesting and sensors Appl. Phys. Lett.102, 152903 (2013).
 24.
Piazza, G., Stephanou, P. & Pisano, A. Piezoelectric aluminum nitride vibrating contourmode MEMS resonators. J. Microelectromech. Syst.15, 1406–1418 (2006).
 25.
Lin, C. M., Yantchev, V., Zou, J., Chen, Y. & Pisano, A. Micromachined oneport aluminum nitride Lamb wave resonators utilizing the lowest order symmetric mode. J. Microelectromech. Syst.23, 78–91 (2014).
 26.
Zou, J., Lin, C.M., Gao, A. & Pisano, A. P. The multimode resonance in AlN Lamb wave resonators. IEEE J. Microelectromech. Syst.27, 973–984 (2018).
 27.
Yang, Q., Pang, W., Zhang, D. & Zhang, H. A modified lattice configuration design for compact wideband bulk acoustic wave filter applications. Micromachines7, 133 (2016).
 28.
Gong, S. & Piazza, G. Design and analysis of lithium–niobatebased high electromechanical coupling RFMEMS resonators for wideband filtering. IEEE Trans. Microw. Theory Tech.61, 403–414 (2013).
 29.
Zheng, X. Q., Lee, J. & Feng, P. X. L. Hexagonal boron nitride nanomechanical resonators with spatially visualized motion. Microsyst. nanoengineering3, 17038 (2017).
 30.
Gao, A. & Gong, S. Harnessing mode conversion in AlN laterally vibrating resonators for spurious mode suppression. IEEE J. Microelectromechanical Syst.25, 450–458 (2016).
 31.
Yantchev, V. & Katardjiev, I. Design and fabrication of thin film Lamb wave resonators utilizing longitudinal wave and interdigital transducers. In 2005 IEEE International Ultrasonics Symposium. 1580–1583 (2005).
 32.
Gao, A., Zou, J. & Wu, T. Narrowband impedance transformer with extremely high transformation ratio of 200. IEEE Electron Device Lett.40, 1820–1823 (2019).
 33.
Piazza, G., Stephanou, P. J. & Pisano, A. P. Singlechip multiplefrequency ALN MEMS filters based on contourmode piezoelectric resonators. IEEE J. Microelectromech. Syst.16, 319–328 (2007).
 34.
Li, W.C., Lin, Y., Kim, B., Ren, Z. & Nguyen, C.T. Quality factor enhancement in micromechanical resonators at cryogenic temperatures. In 2009 IEEE International SolidState Sensors, Actuators and Microsystems Conference. 1445–1448 (2009).
 35.
Olsson, R. H., Nguyen, J., Pluym, T. & Hietala, V. M. A method for attenuating the spurious responses of aluminum nitride micromechanical filters. J. Microelectromech. Syst.23, 1198–1207 (2014).
 36.
Giovannini, M., Yazici, S., Kuo, N. K. & Piazza, G. Apodization technique for spurious mode suppression in AlN contourmode resonators. Sens. Actuators A: Phys.206, 42–50 (2014).
 37.
Zhang, H. et al. Transverse mode spurious resonance suppression in Lamb wave MEMS resonators: theory, modeling, and experiment. IEEE Trans. Electron Devices62, 3034–3041 (2015).
 38.
Huang, W. et al. Onchip inductors with selfrolledup SiN_{x} nanomembrane tubes: a novel design platform for extreme miniaturization. Nano Lett.12, 6283–6288 (2012).
 39.
Yu, X. et al. Ultrasmall, highfrequency, and substrateimmune microtube inductors transformed from 2D to 3D. Sci. Rep.5, 9661–9667 (2015).
 40.
He, Y., Bahr, B., Si, M., Ye, P. & Weinstein, D. A tunable ferroelectric based unreleased RF resonator. Microsyst. Nanoeng.6, 1–7 (2020).
 41.
Liang, J., Zhang, H., Zhang, D., Zhang, H. & Pang, W. Lamb wave AlN micromechanical filters integrated with onchip capacitors for RF frontend architectures. IEEE J. Electron Devices Soc.3, 361–364 (2015).
Acknowledgements
This work was supported by the Science and Technology Commission of Shanghai Municipality (Shanghai Pujiang Program 18PJ1408300), Natural Science Foundation of Shanghai (19ZR1477000) and National Science Foundation of China (61874073).
Author information
Affiliations
Contributions
A.G. performed experimental design, simulation and analysis; K.L. performed simulation, analysis, and writing; J.L. and T.W. discussed the manuscript and analysis.
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Gao, A., Liu, K., Liang, J. et al. AlN MEMS filters with extremely high bandwidth widening capability. Microsyst Nanoeng 6, 74 (2020). https://doi.org/10.1038/s41378020001835
Received:
Revised:
Accepted:
Published:
Further reading

A flexible high qualityfactor bulk acoustic resonator enabled with transferred singlecrystal piezoelectric thin film for sensing applications
Sensors and Actuators A: Physical (2021)

Characterization of AlN and AlScN film ICP etching for micro/nano fabrication
Microelectronic Engineering (2021)