۱- Introduction

۲- Experimental

۳- Results and discussion

۴- Conclusions

References

Abstract

Thin film bulk acoustic resonator (FBAR) plays a very important role in radio frequency (RF) filters used in cell phone and other wireless systems. Although FBAR is commercialized, the design/process interactions on the frequency response variation in FBAR device are still lacking. Design and fabrication are two crucial aspects affecting FBAR device performance. In this report, various solidly mounted resonators (SMR) were designed, fabricated and analyzed to study wafer-level site-to-site RF variation on design and fabrication process. As a key process step for SMR FBAR, the optimization process of Mg-doped ZnO piezoelectric thin film deposition was studied by varying thin film sputtering conditions using various sputtering targets and by post annealing treatment after the deposition. The quality of this crucial layer was verified by XRD on its (0 0 2) crystallization and wafer-level FBAR RF characterization. FBAR devices with high quality were fabricated with an excellent resonant behavior near 2 GHz and a maximum return loss of −۱۵ to 25 dB. Quality factor Q ranges from 400 to 800, with a coupling coefficient keff2 of 1.5–۳%. Wafer-level and wafer-to-wafer variation of central frequency are within 1.8–۲٫۱ GHz. Computer simulation verified that this frequency variation correlates to the piezoelectric film variation of 1.6–۱٫۹ μm. Process control on this piezoelectric thin film is essential to maintaining the resonator frequency-controlled value when building duplexer RF circuits. The dependency of RF performance on FBAR size, density and orientation is not obvious, compared to that of the wafer-level FBAR device variation on fabrication process. Regarding to the Mg-doping effect in MgxZn1-xO piezoelectric film, the amount of Mg in MgxZn1-xO film during the sputtering process must be properly controlled within 30% to keep the piezoelectric quality. The average acoustic speed of the Mg-doped ZnO film is 6870 m/s with the estimated range of 5760–۷۹۸۰ m/s, which is better than that of pure ZnO film (6330 m/s).

Introduction

Film bulk acoustic resonator (FBAR) is a promising RF device at the frequency ranging from 0.5 to 6.0 GHz due to its small size, high device power and strong potential application in both the cell phone market and sensor application [1–۶]. Compared to surface acoustic devices (SAW) [7], FBAR is more powerful and advantageous especially for higher frequencies on 4G and 5G applications. Among the two types of FBAR devices, solid mounted resonator (SMR) is much more robust and therefore chosen in this work. FBAR built on silicon with Bragg reflector is also closer to CMOS integration [8] than SAW although it may also suffer from temperature sensitivity issues [9]. Just like a CMOS circuit that requires the MOSFET to have a controllable threshold voltage variation, FBAR device variation control is also essential to building an RF circuit such as duplexer in 4- and 5G mobile telecommunication systems. A duplexer demands that each FBAR in the circuit should have wafer-level homogeneity on the central frequency in RF performance as well as Q factor, and so on [10]. Although the FBAR product is commercially available in market, academic studies on the wafer-lever FBAR variation and process/design optimization are still relatively weak. In this paper, process variation effects on FBAR device performance were studied via various thin film depositions of Mg-doped MgxZn1−xO piezoelectric thin film by optimizing the sputtering power, pressure and ambient gas flow ratios as well as by altering sputtering targets (uniform MgxZn1−xO sputtering target with different Mg ingredient and dual sputtering targets of Mg and ZnO with independent control of sputtering power) and by post annealing treatment after the deposition. Design/process interactions and impacts on RF performance variation were studied via multi-FBAR device layouts in different sizes, locations and orientations on the same mask/wafer.