3.4 GHz strip-type thickness shear mode solidly-mounted bulk acoustic wave resonator using X-cut LiTaO₃

Surface and bulk acoustic wave (SAW and BAW) filters are widely used in smart and mobile phones. ^1–7) Recently, SAW and BAW filters covering a higher-frequency range up to 6 GHz are demande. ⁸⁾ It is commonly understood that the BAW devices are suitable for high-frequency applications compared with the SAW devices. The piezoelectric material of the BAW device is polycrystalline AlN or ScAlN. ^5,6,9–12) On the other hand, monocrystalline LiNbO₃ (LN) and LiTaO₃ (LT) thin plates are attracting attention for high-frequency BAW devices because their thinning technique has become available.

LN has been considered for BAW resonators (BAWRs) with high electromechanical coupling. Before LN thinning technique was established, epitaxially-grown c-axis LN was used for a 3 GHz BAWRs. ^13,14) Later, polished 45°Y and 36°Y LN thin plates were used for thickness extension (TE) mode BAWRs. ^15–17) For thickness shear (TS) mode BAWRs, X and 163°Y LN thin plates were used. ^18,19) LT has moderate electromechanical coupling factors and better temperature characteristics compared with LN. TE mode 2.49 GHz and TS mode 1.64 GHz BAWRs using 42°YX LT and X LT have been reported, respectively. ²⁰⁾ All of them were composed of a self-suspended structure with a cavity.

For GHz range BAWRs, the thickness of LN and LT is submicron. Such thin self-suspended LN and LT are very fragile, which makes their practical use challenging. To address this problem, a solidly-mounted (SM) type of BAWR (SM-BAWR) using LN and LT has been studied. To date, 0.5 μm thick 41°–43° Y LN and 1 μm thick 20°Y LN were used for 3.18 GHz and 3.31 GHz TE mode resonators, respectively. ^21–23) LT showed a better temperature coefficient of frequency (TCF) and a high impedance (Z) ratio. A 1.19 GHz TS mode SM-BAWR with a BW of 7.9% and Z ratio of 61 dB was demonstrated using 1.5 μm thick X LT. ²⁴⁾ The SM-BAWR has a wider bandwidth (BW) and higher Z ratio compared with the above-mentioned self-suspended TS mode 1.64 GHz BAWR using X LT, ²⁰⁾ suggesting the high potential of SM-BAWRs. Where the BW is defined as [anti-resonance frequency (f_a) − resonance frequency (f_r)]/f_r).

Fujiwara and Wakatsuki reported strip-type TS mode BAWR using a 0.32 mm thick X LT, which had a high mechanical Q of 5000 and suitable coupling factor k² of 18.5% at a low frequency of 6 MHz. ^25,26) To compare the fabricated SM-BAWRs, we fabricated a similar strip-type TS mode BAWR using a 0.35 mm thick X LT plate as shown in Fig. 1. Figure 2 shows the frequency characteristic. Although the center frequency is as low as 5.6 MHz, the Z ratio reaches 104 dB, which is a product of a moderate BW of 10% and high Q factors of 4000 at resonance and 9100 at anti-resonance. Note that ripples between the resonance and anti-resonance are transversal modes, which are enhanced by low energy leakage (i.e. high Q) in the lateral direction.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Based on the above works, the combination of SM and strip-type structures was studied. A strip-type SM-BAWR shown in Fig. 3 was fabricated using a SiO₂/Ta Bragg reflector and 2 μm thick X LT bonded with Si by adhesive. ²⁷⁾ In this process, the thickness variation of the adhesive used to bond the Si before polishing the LT was large, so the thickness variation of the LT after polishing was also large.

Zoom In Zoom Out Reset image size

Figure 4 shows the top view and cross-section of the fabricated SM-BAWR and the measured frequency characteristic without de-embedding. ²⁷⁾ The SM-BAWR has a center frequency of 1 GHz, a BW of 6.7%, and a Z ratio of 62 dB. As far as we know, this was the first report on the SM-BAWR using single crystal LT working in the GHz range with a Z ratio higher than 60 dB. The device direction ψ of 53° is away from the best angle of 37°, which was confirmed by the finite element method (FEM). ²⁷⁾

Zoom In Zoom Out Reset image size

In 2021, we prototyped and reported a higher frequency strip-type TS mode SM-BAWRs using a thinner 0.56 μm X LT plate, which had the center frequency of about 3.4 GHz, using a fully metallic Bragg reflector composed of Al and Ta films instead of SiO₂/Ta, because a metal multilayer is easier to fabricate. ²⁸⁾

In this study, in addition to them, we study the temperature coefficient of frequency (TCF) and the effects of the combination of acoustic film of Bragg reflector on SM-BAWR frequency characteristics.

When the used LT thickness is thin, the LT with a small variation of thickness is important. Table I shows the fabrication parameters of the present and previous works. Figures 5(a) and 5(b) show the present and previous fabrication process. ^27,28)

Zoom In Zoom Out Reset image size

First, a 250 μm thick 4 inch X LT substrate and a 300 μm thick 4 inch Si substrate are directly bonded. ^28,29) To reduce the thickness variation of LT after polishing, the Si substrate with the total thickness variation (TTV) less than 0.2 μm was used. Then, the LT substrate is polished and thinned to a target thickness of 0.6 μm. As the Bragg reflector, five layers of 140 nm thick Al and Ta are alternately deposited on the polished plane of LT instead of previously used SiO₂/Ta, because a metal multilayer is easier to fabricate. Another Si substrate (B) for support is bonded to the Bragg reflector with polymer adhesive, and then the first Si substrate (A) is removed by dry etching. Two 100 nm thick Al electrodes are formed on the LT surface by photolithography. The Al low acoustic impedance first layer acts as the lower electrode. Finally, grooves are formed on both edges of each resonator using a saw dicer. ²⁸⁾

The thickness of LT after polishing is within 0.55–0.63 μm in 70% area of the substrate. In this study, an area of 0.56 μm thick LT was used. The fabricated device is a pair of resonators connected in series. This structure can be fabricated just by patterning double upper electrodes more easily than a single resonator, ^13,14,24,27) because the etching of LT for bottom electrode contact is not necessary. As shown in Table I, the last process of LT surface, LT thickness, ψ, and low acoustic film are different.

In order to simplify the process and avoid damage from the dry etching of the LT surface, it is desirable to directly bond the Bragg reflector deposited on the LT substrate and the support substrate without using adhesives. It is difficult to bond directly the uneven surface of the deposited SiO₂ film and the support substrate. On the other hand, since the surface of the deposited Al film is not uneven, bonding with the support substrate is considered to be advantageous.

Figure 6 shows a schematic figure, cross-section, and top view of fabricated strip-type SM-BAWR. ²⁸⁾ The surfaces of the fourth to tenth layer's acoustic films are not flat, but rather irregular. The irregular acoustic films are thought to affect the frequency characteristic. Two types of SM-BAWRs were fabricated using (90°, 90°, 37°) (X37°Y) and (90°, 90°, 127°) (X127°Y) LT. ²⁸⁾ The dimensions of the fabricated resonators range 60–500 μm in width (W), 80–470 μm in length (L), and 150–340 μm in the gap (G). The best performance was obtained when G was 150 μm, which was the minimum in this study, for both types. A shorter gap should be investigated in the future.

Zoom In Zoom Out Reset image size

Figure 7 shows the measured frequency characteristics. The X37°Y LT BAWR (W = 150 μm, L = 90 μm, and G = 150 μm) exhibited a resonance frequency (f_r) of 3.250 GHz, an antiresonance frequency (f_a) of 3.463 GHz, a bandwidth (BW) of 6.6%, and the Z ratio of 48 dB. The X127°Y LT BAWR (W = 150 μm, L = 140 μm, and G = 150 μm) has a similar characteristic with f_r of 3.153 GHz, f_a of 3.367 GHz, a BW of 6.8%, and a Z ratio of 46 dB. FEM simulation suggested that the X37°Y LT BAWR with one upper electrode had about 17 dB higher Z ratio than the X127°Y LT BAWR, ²⁸⁾ but the measurement results of the SM-BAWRs with double upper electrodes showed an advantage of only 2 dB.

Zoom In Zoom Out Reset image size

The both BWs obtained in this study are almost the same as the previously reported BW shown in Fig. 4. ²⁷⁾ However, it is narrower than the measured BW of the 5.6 MHz self-suspended type BAWR shown in Fig. 2 and the simulated BW in Fig. 8 shown later. On the other hand, the Z ratio obtained in this study is lower than the measured Z ratios in Figs. 2 and 4, and the simulated Z ratio shown in Fig. 8. The cause will be discussed later.

Zoom In Zoom Out Reset image size

The f_r and f_a of the strip-type X-37°Y LT SM-BAWR were measured at 25 °C, 45 °C, 65 °C, and 85 °C. Figure 8 shows the frequency shifts as a function of temperature. The TCF was defined as [f(85 °C) − f(25 °C)/(60·f(25 °C)]. The f is f_r or f_a at 25 °C or 85 °C. The measured TCFs at f_r and f_a were −21 and −37 ppm °C⁻¹, respectively. The frequency shifts are slightly large compared with the previously reported strip-type SM-BAWR composed of SiO₂ and Ta acoustic films, ²⁷⁾ because the Al acoustic film with negative TCF was used instead of SiO₂ film with positive TCF.

The strip-type TS mode 6 MHz BAWR using 0.32 mm thick X LT with a self-suspended structure reported in Ref. 25 shows a mechanical Q of 5000. As the loss of LT, material Q_m of 4000 was assumed as a mechanical Q of 5000 was obtained for the 6 MHz BAWR. The material Q_m of acoustic films of Al, Ta, SiO₂, and W films was used to be infinite. A resistance of the Al electrode was used as zero.

We simulated the frequency characteristics SM-BAWRs of the structure with a single upper Al electrode shown in Fig. 8 using 6 layers of four different combinations of Bragg reflector films of SiO₂/Ta, SiO₂/W, Al/Ta, and Al/W. ²⁸⁾ Their film thickness is 0.2 wavelength each. Figure 8 shows their characteristics. All of their BWs are almost the same, but their Z ratio largely different. Figure 9 shows the BW and the Z ratio as a function of the acoustic impedance ratio of high and low acoustic impedance films. The BW doesn't dependent on the acoustic impedance ratio, but the Z ratio largely depends. The SM-BAWR using SiO₂/W show the highest Z ratio, while Al/Ta the lowest Z ratio. If Al film is used as a low acoustic impedance, it is better to use W film as a high acoustic impedance.

Zoom In Zoom Out Reset image size

Both of the measured (6.6%–6.8% in Figs. 4 and 7) and simulated BWs (8.0%–8.4% in Fig. 9) of SM-BAWRs are narrower than that (10%) of self-suspended type 5.6 MHz BAWR in Fig. 2. It is reported that an effective coupling factor, which is proportional to the BW, of the SM-BAWR using AlN film is 9%–16% smaller than that of the self-suspended type one with cavity. ³⁰⁾ The SM-BAWRs using LT show the same result as that using AlN film. However, the measured (6.6%–6.8%) and simulated BWs (8.0%–8.1%) of SM-BAWRs are different. It might be due to that the Bragg reflector parameters were not optimum, the material Q_m of the acoustic films was simulated as infinite, and the quality of the deposited acoustic film is insufficient.

On the other hand, the measured Z ratios of the SM-BAWRs (62 dB in Fig. 4 and 48 dB in Fig. 7) are largely lower than the measured one of self-suspended 5.6 MHz BAWR (100 dB in Fig. 2) and the simulated ones (102 dB using SiO₂/Ta and 96 dB using Al/Ta in Fig. 9). The difference between the SM-BARW and self-suspended BAWR is due to the same cause as that using the AlN film in Ref. 30, which reports that the mechanical Q of SM-BAWR is 67% lower at Q_s than that of the self-suspended BAWR. The difference between measured 1 GHz and 3.4 GHz and simulated SM-BAWR are 40 and 48 dB, respectively. It might be due to the acoustic films with infinite material Q_m and Al electrode with resistance of zero used in simulation, and the insufficient quality of the deposited acoustic films as same as above-mentioned.

The Z ratio of the 3.4 GHz SM-BAWR using Al/Ta films obtained in this study is 14 dB lower than that of the previously reported 1 GHz strip-type SM-BAWR using SiO₂/Ta films in Fig. 4, although the frequencies are different. The difference might be due to a difference of used acoustic films and SiO₂/Ta and Al/Ta (6 dB difference in Fig. 9), the frequency dependence of the material Q_m of the LT and deposited acoustic films, and irregular surface of fourth to tenth acoustic films.

Zoom In Zoom Out Reset image size

The resistivity of the double upper Al electrodes deposited in our laboratory is as high as 44 nΩ m, which is 1.7 times higher than the theoretical value of 26.5 nΩ m, which causes major degradation in Z ratio, especially in SAW resonators. ³¹⁾ Another concern is the plasma damage of LT by the dry etching removal of the first Si substrate, which will be investigated in the future, in addition to acoustic impedance combination.

As mentioned above, the measured BW and Z ratio are highly dependent on the material Q_m of the LT and the deposited thin film.

We fabricated two kinds of strip-type TS mode SM-BAWRs working about 3 GHz using 0.56 μm thick X37°Y and X127°Y LT composed of Bragg reflector of Al and Ta films. The X37°Y LT SM structure BAWRs exhibited a resonance frequency (f_r) of 3.250 GHz, an antiresonance frequency (f_a) of 3.463 GHz, a bandwidth (BW) of 6.6%, and a Z ratio of 48 dB. The X127°Y LT BAWR has a similar characteristic with f_r of 3.153 GHz, f_a of 3.367 GHz, a BW of 6.8%, and a Z ratio of 46 dB. The X37°Y LT BAWR showed an advantage of 2 dB compared with the X127°Y LT one. But the advantage is not so prominent as predicted by FEM simulation. TCFs at f_r and f_a were −21 and −37 ppm °C⁻¹, respectively. Frequency characteristics of the strip-type SM-BAWRs using four combinations of acoustic films of SiO₂/Ta, SiO₂/W, Al/Ta, and Al/W were simulated. The BW is independent on the acoustic film combinations, but the Z ratio largely depends on them. The Z ratio using a combination of Al and Ta films shows the lowest Z ratio and the combination of SiO₂ and W films the highest Z ratio. The difference was 12 dB. The optimization of the combination of acoustic layers, an electrode design, and the fabrication process is necessary for better performance in addition to using a high-quality Al electrode and acoustic films.

This work was partly supported by Ministry of Internal Affairs and Communications, SCOPE #JP195002001. We would like to thank Dr. Hideki Takagi at National Institute of Advanced Industrial Science and Technology (AIST), and Mr. Shigeharu Matsumoto and Mr. Yuya Nonaka at SHINCRON Co., Ltd. for their technical support.