Making Phones Better

Beneath a smartphone’s slick packaging is some interesting, highly sophisticated technology that makes the user experience what it is today. Much of that experience relies on satisfying our ever growing desire for more data capacity for video, social media and the like. Providing that capacity relies on robust filtering to receive just your data stream amongst many nearby other streams. But that filtering is about to get harder for the next generation.

For instance, the iPhone 6 had 22 filters comprised of about 100 RF-MEMS resonators known as FBARs—or more generally, BAW resonators [1]. But future phones will have even more if future designs can meet the coming tighter requirements.

What motivates the tighter requirements? The LTE-Advanced protocol will use carrier aggregation (CA) to get more bandwidth out of the current LTE standard. While LTE uses a single band to transmit and receive data, LTE-Advanced carrier aggregation will use multiple bands at once to achieve higher data rates. This will put even more demand on the numerous transmit and receive filters not to interfere with each other.

At the International Microwave Symposium (http://ims2016.org) in May, the major industrial players presented recent progress in device and circuit design in a session titled “Acoustic Multiplexers for Carrier Aggregation.” Chaired by Qualcomm/TDK, and with talks from Qorvo, Murata, Taiyo Yuden, Avago, and EPSCOS, pretty much all the players in this field were represented. Each gave their strategy to address the needs of carrier aggregation with FBAR, BAW, and even SAW resonators.

Why FBAR/BAW? The BAW style resonators have traditionally met the demand with high quality-factor resonators (>4000) that give low insertion loss and a steep skirt for a minimum guard band. BAWs have enabled the use of multiplexers for multiple transmit and receive channels off the same antenna simultaneously. The BAW resonator is a thin film resonator either over a cavity as commercialized by Broadcom (an FBAR) or surface-mounted as commercialized by Qorvo. They are considered MEMS because they typically use fabrication processes similar to MEMS. And because acoustic waves in solids have very short wavelengths compared to electromagnetic waves, these resonators are very small in size at carrier frequencies. See [2] for a nice introduction.

But can FBAR/BAW continue to meet the need? Maybe, but supporting many bands with BAW will be expensive because multiple frequencies cannot be easily supported on one chip. The reason: BAW resonators vibrate in the thickness direction of a thin film of piezoelectric material. This means that the thickness defines the frequency of resonance. Therefore, supporting multiple frequency resonators on one chip is not possible without complex post-processing of the deposited piezoelectric layer.

An alternative has been to excite the acoustic wave in-plane rather than in the thickness. Then the frequency of each resonator can be defined by the dimensions within the lithographic mask rather than the thickness of the process. But unfortunately, the best material for high-volume BAW is aluminum nitride, and the purely in-plane modes of vibration cannot meet the typical filter bandwidth requirements. Such resonators have typically been considered only for high-frequency oscillator applications [3] as a result.

But at the recent Solid-State Sensors, Actuators and Microsystems Workshop in Hilton Head, South Carolina., a group at Northeastern University [4] cleverly demonstrated an aluminum nitride resonator whose mode of vibration was simultaneously both in-plane and in the thickness as illustrated in the figure below. By partly using the thickness direction, the resonator met the bandwidth requirements and by partly using the in-plane direction, the frequency could be shifted about 40% lithographically, thus giving a new avenue for single aluminum nitride multiband chips.

Source: C. Cassella, Northeastern University

Tighter filter requirements have also seen stronger concern in controlling intermodulation distortion both for BAWs and the in-plane-mode resonators. Accurate modeling to predict intermodulation has been challenging. At the Solid-State Sensors, Actuators and Microsystems Workshop, Coventor presented a novel method of accurately modeling thermal nonlinearity of acoustic resonators via the finite-element method [5]. Thermal nonlinearity as illustrated in the figure below for rising power levels, has been identified as one of the key contributors to intermodulation distortion for in-plane-mode resonators and is important for thickness BAW as well.

Source: Coventor, Inc.

Nonlinearity can also be your friend as evidenced in the work from University of Illinois [6]. For high-frequency oscillators, aluminum nitride MEMS resonators still had not shown the phase noise performance of their bulky quartz or SAW counterparts due to temporal fluctuation in their resonance frequency. However the group at U of Illinois used parametric resonance to accomplish “noise squeezing” to reduce the phase noise of a piezoelectric oscillator. Parametric resonance uses nonlinear coupling between modes to amplify the resonance frequency by ‘pumping’ at twice the resonance as illustrated in the figure below. The phase noise was reduced by 25 dB at 1 kHz offset as compared to conventional oscillators, thus opening a new path for AlN resonators as frequency references.

Source: R. Lu, University of Illinois

All these recent developments indicate RF-MEMS acoustic resonators have a long runway for greater participation in the hardware of future mobile devices. And even better, they will greatly improve the user experience across this sector.

References

[1] R. Colin Johnson, “MEMS Market: Ups and Upstarts,” EE Times, 11/24/2015, http://www.eetimes.com/document.asp?doc_id=1328333

[2] Robert Aigner, “SAW, BAW and the future of wireless,” EDN Network, 5/6/2013 http://www.edn.com/design/wireless-networking/4413442/SAW–BAW-and-the-future-of-wireless

[3] G. Piazza, V. Felmetsger, P. Muralt, R. H. Olsson III, and R. Ruby, “Piezoelectric aluminum nitride films for microelectromechanical systems,” MRS Bull., vol. 37, pp. 1051–1061, Nov. 2012.

[4] C. Cassella, G. Chen, Z. Qian, G. Hummel, and M. Rinaldi, “UNPRECEDENTED FIGURE OF MERIT IN EXCESS OF 108 IN 920 MHZ ALUMINUM NITRIDE CROSS-SECTIONAL LAMÉ MODE RESONATORS SHOWING KT2 IN EXCESS OF 6.2%,” presented at the Solid-State Sensors, Actuators, and Microsystems Workshop, Hilton Head, SC, 2016.

[5] R. Jhaveri, R. Lu, S. Gong, and M. Kamon, “DISTRIBUTED AND THERMO-ACOUSTICALLY COUPLED MODELING FOR ACCURATE PREDICTION OF THERMAL NONLINEARITY IN PIEZOELECTRIC MEMS RESONATORS,” presented at the Solid-State Sensors, Actuators, and Microsystems Workshop, Hilton Head, SC, 2016.

[6] R. Lu, A. Gao, and S. Gong, “ALN PIEZOELECTRIC PARAMETRIC OSCILLATORS WITH LOW PHASE NOISE,” presented at the Solid-State Sensors, Actuators, and Microsystems Workshop, Hilton Head, SC, 2016.