Trends and Applications of Surface and Bulk Acoustic Wave Devices: A Review

The past few decades have witnessed the ultra-fast development of wireless telecommunication systems, such as mobile communication, global positioning, and data transmission systems. In these applications, radio frequency (RF) acoustic devices, such as bulk acoustic waves (BAW) and surface acoustic waves (SAW) devices, play an important role. As the integration technology of BAW and SAW devices is becoming more mature day by day, their application in the physical and biochemical sensing and actuating fields has also gradually expanded. This has led to a profusion of associated literature, and this article particularly aims to help young professionals and students obtain a comprehensive overview of such acoustic technologies. In this perspective, we report and discuss the key basic principles of SAW and BAW devices and their typical geometries and electrical characterization methodology. Regarding BAW devices, we give particular attention to film bulk acoustic resonators (FBARs), due to their advantages in terms of high frequency operation and integrability. Examples illustrating their application as RF filters, physical sensors and actuators, and biochemical sensors are presented. We then discuss recent promising studies that pave the way for the exploitation of these elastic wave devices for new applications that fit into current challenges, especially in quantum acoustics (single-electron probe/control and coherent coupling between magnons and phonons) or in other fields.


Introduction
In recent decades, the rapid development of microwave wireless communication technology has led to very rapid developments in other fields, such as mobile communication systems (CDMA, UMTS, GSM, etc.), global positioning systems (GPS, Galileo, etc.), data transmission systems (such as WLAN, Bluetooth, etc.), satellite communication, and other military communication systems [1]. The most important parts of the components that make up these systems, such as filters, duplexers, voltage-controlled oscillators, frequency meters, and tunable amplifiers, are microwave resonators [2]. With the remarkable advances in microelectronics and microfabrication technologies, researchers have paid more attention to the miniaturization and integration of devices while improving their performance. They have succeeded in integrating multiple devices on a single chip and in accommodating active and passive devices or MEMS devices in a single package [3].
Conventional microwave resonators use electromagnetic waves as energy carriers. In the RF/microwave frequency range, the required resonators have dimensions that can hardly meet the integration requirements for applications such as the current fifthgeneration (5G) telecommunication standard. With wave velocities four to five orders of magnitude slower than that of electromagnetic waves, acoustic waves allow a reduction of device dimensions in the same proportions [4]. As a consequence, acoustic wave devices can meet the current miniaturization and integration requirements [5].
This paper aims to present an overview of surface and bulk acoustic waves technologies and of major application fields in a comprehensive manner, especially for the This paper aims to present an overview of surface and bulk acoustic waves technol- 46 ogies and of major application fields in a comprehensive manner, especially for the benefit 47 of novice readers, because a profusion of literature is associated with such quickly ex- 48 panding technologies. We first present the basic principles of SAW and BAW devices, 49 typical structures, main characterizations, potential limitations, and possible optimization 50 methods. Regarding BAW devices, we give particular attention to the film bulk acoustic 51 resonator (FBAR), which holds a higher operation frequency and a better integration ca- 52 pacity compared with SAW devices. We then propose an overview of major current ap- 53 plications, which is divided into three parts: RF signal filters, physical sensors and actua- 54 tors, and chemical and biochemical sensors. Finally, we conclude this review with a brief 55 introduction to current trends associated with quantum acoustics and some applications. 56 57 Acoustic components are based on acoustic waves generated from the piezoelectric 58 effect, one of the most exciting material properties. The term derives from the Greek word 59 piezein (to press), and the effect is based on an interaction between an electrical charge and 60 mechanical stress [6]. In 1880, the Curie brothers discovered that applying pressure or 61 tension in a certain direction to a quartz crystal results in the formation of electrical 62 charges on its surface, and the density of the electrical charge is proportional to the mag- 63 nitude of the applied external force [7]. This is the positive piezoelectric effect of piezoe- 64 lectric materials. One year later, the Curie brothers proved the inverse effect through ex- 65 periments and determined the direct and inverse piezoelectric coefficients of quartz crys- 66 tal [8]. 67 Acoustic waves generated from the piezoelectric effect can be broadly classified into 68 two types: BAW [9] and SAW [10]. Essentially, BAW are generated by an alternating (AC) 69 electrical signal applied between the two sides of the piezoelectric material, and the acous- 70 tic wave propagates through the entire thickness, giving rise to a stationary wave at spe- 71 cific frequencies. Depending on the crystalline orientation, the piezoelectric material can 72 expand and contract perpendicularly to the surface as in Figure 1 (a); the wave polariza- 73 tion is longitudinal, as the mechanical displacement is parallel to the propagation direc-74 tion; such a mode is called thickness-extension mode. When the piezoelectric material de- 75 forms in a direction parallel to the surface, the mode is called thickness-shear mode, as in 76 Figure 1 (b) [11]. In the case of SAW, the vibrations occur only at the surface of the mate- 77 rial. By placing interdigitated transducers (IDTs) on the surface of the piezoelectric mate- 78 rial, as in Figure 1 (c), the propagating SAW will be generated when an AC signal is ap-79 plied; the resonance frequency, amplitude, and propagation characteristics are deter- 80 mined by the design of the IDTs, the dimensions of the electrodes, the material properties, 81 and the applied electrical signal. As for the BAW, several types of polarization and modes 82 exist; some of the most significant ones are described in the following part.

Basic Structures
In 1885, Lord Rayleigh discovered the "Waves Propagated along the Plane Surface of an Elastic Solid" and defined the first discovered SAW, which were named Rayleigh waves [12]. Rayleigh waves cause surface particles to move elliptically in planes perpendicular to the surface and parallel to the propagation direction. They are typically involved in seismic movements. In 1911, Augustus Edward Hough Love mathematically predicted the existence of so-called Love waves, which can propagate when the surface of a semiinfinitely thick elastic body is covered with a layer of a medium with lower sound velocity. The Love wave is a guided shear horizontal surface wave (SH-SAW) with a displacement of the particles parallel to the surface [13]. In 1965, Richard Manning White et al. proposed to directly deposit interdigital transducers (IDTs) onto the surface of piezoelectric materials in order to generate, transmit, and receive SAW efficiently [14]. As represented in Figure 2a, in a standard SAW resonator with a delay line, when the electrical signal arrives at the input IDT (left side) through a feedline, here with matching dipole, acoustic waves are generated by the inverse piezoelectric effect and acoustic resonance occurs, at specific frequencies of constructive waves, for which acoustic waves travel across the propagation path. Arriving at the output IDT (right side), the acoustic signal is therefore converted back into an electrical signal through the output feedline by the piezoelectric effect.
Micromachines 2022, 13, x FOR PEER REVIEW 3 of 21 Figure 1. Profiles of (a) thickness-extensional mode and (b) thickness-shear mode of a BAW device; 84 (c) SAW. 85 86 In 1885, Lord Rayleigh discovered the "Waves Propagated along the Plane Surface of 87 an Elastic Solid" and defined the first discovered SAW, which were named Rayleigh 88 waves [12]. Rayleigh waves cause surface particles to move elliptically in planes perpen-89 dicular to the surface and parallel to the propagation direction. They are typically in- 90 volved in seismic movements. In 1911, Augustus Edward Hough Love mathematically 91 predicted the existence of so-called Love waves, which can propagate when the surface of 92 a semi-infinitely thick elastic body is covered with a layer of a medium with lower sound 93 velocity. The Love wave is a guided shear horizontal surface wave (SH-SAW) with a dis-94 placement of the particles parallel to the surface [13]. In 1965, Richard Manning White et 95 al. proposed to directly deposit interdigital transducers (IDTs) onto the surface of piezoe-96 lectric materials in order to generate, transmit, and receive SAW efficiently [14]. As repre-97 sented in Figure 2 (a), in a standard SAW resonator with a delay line, when the electrical 98 signal arrives at the input IDT (left side) through a feedline, here with matching dipole, 99 acoustic waves are generated by the inverse piezoelectric effect and acoustic resonance 100 occurs, at specific frequencies of constructive waves, for which acoustic waves travel 101 across the propagation path. Arriving at the output IDT (right side), the acoustic signal is 102 therefore converted back into an electrical signal through the output feedline by the pie-103 zoelectric effect.  [4]; (b) the classic structure of IDT. 106 Indeed, the IDT is a typical electroacoustic conversion structure in SAW filters and 107 other SAW devices. Its basic design is shown in Figure 2 (b), with metal strips on the pie-108 zoelectric substrate intertwined and connected to the signal and the ground, alternately. 109 These interdigitated electrodes are structurally characterized by the finger width , inter-110 finger distance , acoustic aperture , spatial periodicity , and number of finger pairs 111 . When an IDT is connected to an alternating signal source, the material is deformed 112 locally due to the converse piezoelectric effect. At the resonance frequency, the waves 113 emitted by each pair of electrodes are constructive and the resulting SAW propagate in 114 the two opposite directions perpendicular to the fingers. The acoustic signals reaching the 115 second IDT are converted into electrical signals due to the piezoelectric effect. Although 116 the two transducers are reciprocal, for simplicity, the transducer connected to the alter-117 nating signal source will be referred to as the exciting or input IDT and the transducer 118 connected to the load will be referred to as the receiving or output IDT. The acoustic res-119 onance frequency is expressed as Indeed, the IDT is a typical electroacoustic conversion structure in SAW filters and other SAW devices. Its basic design is shown in Figure 2b, with metal strips on the piezoelectric substrate intertwined and connected to the signal and the ground, alternately. These interdigitated electrodes are structurally characterized by the finger width a, interfinger distance b, acoustic aperture W, spatial periodicity p, and number of finger pairs N. When an IDT is connected to an alternating signal source, the material is deformed locally due to the converse piezoelectric effect. At the resonance frequency, the waves emitted by each pair of electrodes are constructive and the resulting SAW propagate in the two opposite directions perpendicular to the fingers. The acoustic signals reaching the second IDT are converted into electrical signals due to the piezoelectric effect. Although the two transducers are reciprocal, for simplicity, the transducer connected to the alternating signal source will be referred to as the exciting or input IDT and the transducer connected to the load will be referred to as the receiving or output IDT. The acoustic resonance frequency is expressed as

Basic Structures
where v s is the velocity of the SAW, which depends on the piezoelectric substrate properties in the considered orientation, and the spatial periodicity p of the IDTs is equal to the wavelength λ of the propagating SAW. Unlike SAW resonators, BAW resonators use acoustic waves which propagate in the direction of the thickness of the piezoelectric material. The thickness of the piezoelectric plate typically corresponds to half a wavelength (λ/2) of the fundamental resonance frequency of the thickness-extensional mode (Figure 3), which can be expressed as follows, assuming infinitely thin electrodes: and the angular frequency where is the velocity of the SAW, which depends on the piezoelectric substrate prop-122 erties in the considered orientation, and the spatial periodicity of the IDTs is equal to 123 the wavelength λ of the propagating SAW. 124 Unlike SAW resonators, BAW resonators use acoustic waves which propagate in the 125 direction of the thickness of the piezoelectric material. The thickness of the piezoelectric 126 plate typically corresponds to half a wavelength (λ/2) of the fundamental resonance fre-127 quency of the thickness-extensional mode (Figure 3), which can be expressed as follows, 128 assuming infinitely thin electrodes: 129

= 2ℎ
(3) Here ℎ is the thickness of the piezoelectric layer, and is the velocity of longitudi-130 nal waves in the piezoelectric medium in the plate thickness direction. 131 132 Figure 3. Schematic diagram of BAW resonator. 133 Though such devices are widely used based on bulk piezoelectric materials, like clock 134 sources and quartz crystal microbalances (QCMs), equation (3) highlights the need for 135 very thin plates for high frequency operation, not compatible with bulk materials. As a 136 consequence, due to the limitations of the fabrication process, the performance of conven-137 tional BAW resonators remained lower compared with that of the SAW resonators. In 138 parallel, piezoelectric layer-based devices have been investigated. In 1965, Newell first 139 proposed a piezoelectric sandwich resonator structure using an acoustic Bragg reflector 140 with thickness layers of λ/4 and indicated that it was likely to be used in the high-fre-141 quency range [15]. In 1967, Sliker and Roberts proposed a CdS-based acoustic resonator 142 on a quartz wafer [16], and the theoretical model of the device was basically mature at this 143 time. In 1980, Grudkowski et al. proposed the concept of a BAW resonator filter and fab-144 ricated a ZnO-based BAW filter on a Si substrate with an operating frequency of 200~500 145 MHz [17]. In 1981, Lakin et al. clarified for the first time the application perspectives of 146 thin-film bulk acoustic resonators (TFBARs) [18].
With the development of microfabrication processes, the advantages of BAW reso-148 nators slowly began to appear by the end of the 20th century. In 1996, Ruby prepared a 149 BAW resonator based on a piezoelectric AlN film with a quality factor (Q value) of 2300 150 and an electromechanical coupling coefficient k 2 of 6% [19]. Q value is a dimensionless 151 ratio of the stored energy to the energy loss within a resonant element. He then investi-152 gated and prepared a 1900 MHz duplexer based on a film bulk acoustic waves resonator 153 (FBAR) [20]. At this time, FBARs gradually began to be commercialized, which prompted 154 more companies to conduct research on FBARs. In 2001, Agilent (i.e., Avago Avago) first 155 introduced the PCS (personal communications systems) duplexer with an operation fre-156 quency of 1900 MHz for the mobile phone market [20], which was already in mass 157 Here h is the thickness of the piezoelectric layer, and v L is the velocity of longitudinal waves in the piezoelectric medium in the plate thickness direction.
Though such devices are widely used based on bulk piezoelectric materials, like clock sources and quartz crystal microbalances (QCMs), Equation (3) highlights the need for very thin plates for high frequency operation, not compatible with bulk materials. As a consequence, due to the limitations of the fabrication process, the performance of conventional BAW resonators remained lower compared with that of the SAW resonators. In parallel, piezoelectric layer-based devices have been investigated. In 1965, Newell first proposed a piezoelectric sandwich resonator structure using an acoustic Bragg reflector with thickness layers of λ/4 and indicated that it was likely to be used in the high-frequency range [15]. In 1967, Sliker and Roberts proposed a CdS-based acoustic resonator on a quartz wafer [16], and the theoretical model of the device was basically mature at this time. In 1980, Grudkowski et al. proposed the concept of a BAW resonator filter and fabricated a ZnO-based BAW filter on a Si substrate with an operating frequency of 200~500 MHz [17]. In 1981, Lakin et al. clarified for the first time the application perspectives of thin-film bulk acoustic resonators (TFBARs) [18].
With the development of microfabrication processes, the advantages of BAW resonators slowly began to appear by the end of the 20th century. In 1996, Ruby prepared a BAW resonator based on a piezoelectric AlN film with a quality factor (Q value) of 2300 and an electromechanical coupling coefficient k 2 t of 6% [19]. Q value is a dimensionless ratio of the stored energy to the energy loss within a resonant element. He then investigated and prepared a 1900 MHz duplexer based on a film bulk acoustic waves resonator (FBAR) [20]. At this time, FBARs gradually began to be commercialized, which prompted more companies to conduct research on FBARs. In 2001, Agilent (i.e., Avago Avago) first introduced the PCS (personal communications systems) duplexer with an operation frequency of 1900 MHz for the mobile phone market [20], which was already in mass production at that time, officially initiating the commercial move of FBARs. The German company Infineon [21] then also launched its own BAW devices. Later, Intel [22], TriQuint [23] in the United States, Philips [24] in the Netherlands, and Samsung [25] in South Korea joined the development of BAW resonators.
Conventional QCM type BAW devices have been well-investigated, with a variety of fundamental and harmonic modes. We focus on the emerging thin-film-type FBAR, with some outstanding features compared with the classical BAW and even SAW. The current FBAR devices can be divided into two main types by their structures: the front-side etch or airbag type and the solidly mounted type (solidly mounted resonators; SMRs). Both are based on the same principles; the main difference is the method of energy confinement. In the FBAR, an air cavity with a length and width of 100-300 µm is etched under the bottom electrode in order to obtain a suspended piezoelectric membrane confining the acoustic energy, as illustrated in Figure 4a [26]. In the SMR structure, a "mirror" under the electrode reflects acoustic waves. As shown in Figure 4b, these acoustic Bragg reflectors alternate layers of different acoustic impedances, such as W and SiO 2 (impedance ratio of about 4:1), AlN and SiO 2 (impedance ratio of about 3:1). By reflecting acoustic waves back to the piezoelectric film, they play a role in limiting energy dissipation [27]. To design the acoustic Bragg reflectors, the acoustic impedance of each material layer can be calculated as follows: where ρ is the material density and v L is the velocity of the longitudinal wave in the film thickness direction if considering the SMR thickness-extensional mode.
Micromachines 2022, 13, x FOR PEER REVIEW 5 of 21 production at that time, officially initiating the commercial move of FBARs. The German 158 company Infineon [21] then also launched its own BAW devices. Later, Intel [22], TriQuint 159 [23] in the United States, Philips [24] in the Netherlands, and Samsung [25] in South Korea 160 joined the development of BAW resonators. 161 Conventional QCM type BAW devices have been well-investigated, with a variety of 162 fundamental and harmonic modes. We focus on the emerging thin-film-type FBAR, with 163 some outstanding features compared with the classical BAW and even SAW. The current 164 FBAR devices can be divided into two main types by their structures: the front-side etch 165 or airbag type and the solidly mounted type (solidly mounted resonators; SMRs). Both are 166 based on the same principles; the main difference is the method of energy confinement. 167 In the FBAR, an air cavity with a length and width of 100-300 μm is etched under the 168 bottom electrode in order to obtain a suspended piezoelectric membrane confining the 169 acoustic energy, as illustrated in Figure 4 (a) [26]. In the SMR structure, a "mirror" under 170 the electrode reflects acoustic waves. As shown in Figure 4 (b), these acoustic Bragg re-171 flectors alternate layers of different acoustic impedances, such as W and SiO 2 (impedance 172 ratio of about 4:1), AlN and SiO 2 (impedance ratio of about 3:1). By reflecting acoustic 173 waves back to the piezoelectric film, they play a role in limiting energy dissipation [27]. 174 To design the acoustic Bragg reflectors, the acoustic impedance of each material layer can 175 be calculated as follows: where is the material density and is the velocity of the longitudinal wave in the film 177 thickness direction if considering the SMR thickness-extensional mode. The differences in structure and acoustic reflection of the two resonators determine 180 the differences in their fabrication, performance, and applications. In terms of design and 181 fabrication, the longitudinal acoustic waves are confined in the FBAR membrane, the de-182 sign is flexible, and the processing is quite simple due to the smaller number of layers [28]. 183 In the design of SMR, not only does the lateral acoustic energy dissipation need to be 184 considered (the impact brought by the lateral acoustic modes is presented below), but the 185 design of the acoustic Bragg reflector is also critical as it directly impacts the energy dissi-186 pation and the performance of the resonator. To ensure the Q value, a fine control of the 187 thickness during processing is also required [26]. 188 As for performance, FBAR has a higher effective electromechanical coupling coeffi-189 cient ( 2 ) [28], with a larger difference in acoustic impedance at the electrode-air inter-190 face than at the junction with the Bragg reflector of the SMR structure. Furthermore, the 191 Bragg reflector adds some new loss mechanisms that reduce its Q value. However, the 192 SMR structure also offers some interests. Indeed, the multiple 2 layers of the Bragg 193 reflector have a negative temperature coefficient of frequency (TCF), which provides a 194 matching effect on the TCF of the whole SMR [26]. Second, the FBAR membrane is me-195 chanically supported at the edge of the cavity structure, which is risky during the micro-196 fabrication process, and the stress of the film must be carefully controlled to avoid 197 The differences in structure and acoustic reflection of the two resonators determine the differences in their fabrication, performance, and applications. In terms of design and fabrication, the longitudinal acoustic waves are confined in the FBAR membrane, the design is flexible, and the processing is quite simple due to the smaller number of layers [28]. In the design of SMR, not only does the lateral acoustic energy dissipation need to be considered (the impact brought by the lateral acoustic modes is presented below), but the design of the acoustic Bragg reflector is also critical as it directly impacts the energy dissipation and the performance of the resonator. To ensure the Q value, a fine control of the thickness during processing is also required [26].
As for performance, FBAR has a higher effective electromechanical coupling coefficient (k 2 t e f f ) [28], with a larger difference in acoustic impedance at the electrode-air interface than at the junction with the Bragg reflector of the SMR structure. Furthermore, the Bragg reflector adds some new loss mechanisms that reduce its Q value. However, the SMR structure also offers some interests. Indeed, the multiple SiO 2 layers of the Bragg reflector have a negative temperature coefficient of frequency (TCF), which provides a matching effect on the TCF of the whole SMR [26]. Second, the FBAR membrane is mechanically supported at the edge of the cavity structure, which is risky during the microfabrication process, and the stress of the film must be carefully controlled to avoid mechanical cracking. On the contrary, the SMR structure does not suffer from such drawbacks, provided there are stable layer interfaces [9].

Typical Characterizations
SAW devices are usually electrically characterized by their scattering parameters (Sparameters), S ij corresponding to the output power at port i divided by the input power at port j, or transmission gain from port j to port i. Figure 5 gives an example of the characterization results of a SAW delay line (two-port device) such as those used by Rubé, M et al. [29], measured with a vector network analyzer (VNA) in the frequency domain. It exhibits a resonance at about 118 MHz, with a sharp dip in the reflection parameter S 11 , which corresponds to a maximum of S 21 , the transmission being supported by an acoustic wave between ports.
Micromachines 2022, 13, x FOR PEER REVIEW 6 of 21 mechanical cracking. On the contrary, the SMR structure does not suffer from such draw-198 backs, provided there are stable layer interfaces [9]. 199

200
SAW devices are usually electrically characterized by their scattering parameters (S-201 parameters), corresponding to the output power at port divided by the input power 202 at port , or transmission gain from port to port . 203 Figure 5 gives an example of the characterization results of a SAW delay line (two-204 port device) such as those used by Rubé, M et al. [29], measured with a vector network 205 analyzer (VNA) in the frequency domain. It exhibits a resonance at about 118 MHz, with 206 a sharp dip in the reflection parameter 11 , which corresponds to a maximum of 21 , the 207 transmission being supported by an acoustic wave between ports. Among other applications, signal filtering is one of the most typical applications for 210 SAW devices, for which they are designed to meet frame specifications, in terms of pass-211 band and stopband, as well as maximum and minimum losses inside, respectively. 212 In the above, we have presented the classic design of SAW devices. However, reflec-213 tions on metal electrodes or other interfaces, material losses, and other spurious modes 214 can highly influence the performances. Such effects can be reduced by an appropriate de-215 sign, as described here. 216 The first important effect is the internal reflection at the metal strips of an IDT. This 217 is illustrated in Figure 6 (a), with an electrode pitch equal to half the center frequency 218 wavelength ( 0 /2) resulting in an in-phase addition of the unwanted reflections at this 219 frequency, which causes a significant impact on performances by causing ripples in the 220 amplitude and phase. With a split-finger IDT, as shown in Figure 6 (b), the electrode pitch 221 becomes 0 /4, so that the reflections between two adjacent electrodes cancel each other, 222 being 180° out of phase, and the overall reflected wave is cancelled. Among other applications, signal filtering is one of the most typical applications for SAW devices, for which they are designed to meet frame specifications, in terms of passband and stopband, as well as maximum and minimum losses inside, respectively.
In the above, we have presented the classic design of SAW devices. However, reflections on metal electrodes or other interfaces, material losses, and other spurious modes can highly influence the performances. Such effects can be reduced by an appropriate design, as described here.
The first important effect is the internal reflection at the metal strips of an IDT. This is illustrated in Figure 6a, with an electrode pitch equal to half the center frequency wavelength (λ 0 /2) resulting in an in-phase addition of the unwanted reflections at this frequency, which causes a significant impact on performances by causing ripples in the amplitude and phase. With a split-finger IDT, as shown in Figure 6b, the electrode pitch becomes λ 0 /4, so that the reflections between two adjacent electrodes cancel each other, being 180 • out of phase, and the overall reflected wave is cancelled. corresponding to the output power at port divided by the input power 202 at port , or transmission gain from port to port . 203 Figure 5 gives an example of the characterization results of a SAW delay line (two-204 port device) such as those used by Rubé, M et al. [29], measured with a vector network 205 analyzer (VNA) in the frequency domain. It exhibits a resonance at about 118 MHz, with 206 a sharp dip in the reflection parameter 11 , which corresponds to a maximum of 21 , the 207 transmission being supported by an acoustic wave between ports. Among other applications, signal filtering is one of the most typical applications for 210 SAW devices, for which they are designed to meet frame specifications, in terms of pass-211 band and stopband, as well as maximum and minimum losses inside, respectively. 212 In the above, we have presented the classic design of SAW devices. However, reflec-213 tions on metal electrodes or other interfaces, material losses, and other spurious modes 214 can highly influence the performances. Such effects can be reduced by an appropriate de-215 sign, as described here. 216 The first important effect is the internal reflection at the metal strips of an IDT. This 217 is illustrated in Figure 6 (a), with an electrode pitch equal to half the center frequency 218 wavelength ( 0 /2) resulting in an in-phase addition of the unwanted reflections at this 219 frequency, which causes a significant impact on performances by causing ripples in the 220 amplitude and phase. With a split-finger IDT, as shown in Figure 6 (b), the electrode pitch 221 becomes 0 /4, so that the reflections between two adjacent electrodes cancel each other, 222 being 180° out of phase, and the overall reflected wave is cancelled. In some designs, gratings on the surface are used to improve the confinement of acoustic energy in the transducer and consequently the Q value of the SAW device. In Figure 7, a one-port resonator has two such metallic side gratings, which act as two reflectors for the acoustic energy generated by the central IDT, therefore improving the Q value. Metallic gratings can not only provide mechanical but also electrical reflection, since an electrical field can also be formed by reflected acoustic waves due to the piezoelectric effect. The electrical characteristics of these reflections depend on the metal strips' electrical connections, either short-circuited as in Figure 7, or open-circuited, or a combination of them which is named as positive and negative reflection (PNR) grating, with enhanced reflection properties. Gratings are also widely used in two-port SAW delay-line structures [30]. Another way of limiting bidirectional losses lies on specific designs enabling control of internal reflections within the generating IDT itself, such as in a single-phase unidirectional transducer (SPUDT) [4]. In some designs, gratings on the surface are used to improve the confinement of 226 acoustic energy in the transducer and consequently the Q value of the SAW device. In 227 Figure 7, a one-port resonator has two such metallic side gratings, which act as two reflec-228 tors for the acoustic energy generated by the central IDT, therefore improving the Q value. 229 Metallic gratings can not only provide mechanical but also electrical reflection, since an 230 electrical field can also be formed by reflected acoustic waves due to the piezoelectric ef-231 fect. The electrical characteristics of these reflections depend on the metal strips' electrical 232 connections, either short-circuited as in Figure 7, or open-circuited, or a combination of 233 them which is named as positive and negative reflection (PNR) grating, with enhanced 234 reflection properties. Gratings are also widely used in two-port SAW delay-line structures 235 [30]. Another way of limiting bidirectional losses lies on specific designs enabling control 236 of internal reflections within the generating IDT itself, such as in a single-phase unidirec-237 tional transducer (SPUDT) [4]. 238 239 Figure 7. One-port SAW resonator with gratings, adapted with permission from [4]. 240 Spurious BAW modes are also a possible source of reduced performance. Among 241 them, so-called deep bulk acoustic waves (DBAW) can be generated by the input IDT, 242 then propagate in the volume of the piezoelectric layer, get reflected at the bottom face, 243 and propagate back to the receiving IDT. Surface skimming bulk waves (SSBW) and other 244 leaky waves are also examples of possibly interfering (or sometimes alternating) waves 245 [31]. 246 Waves undergoing reflections on the device sides or bottom can be limited by sand-247 blasting the bottom surface or biasing the sides in order to avoid constructive reflections 248 travelling back to the IDTs. Other interfering waves such as SSBW need, for intrinsic good 249 design, to take into account both longitudinal and shear acoustic waves. 250 Other particular IDT structures can also be implemented during the design process 251 in order to achieve a specific response pattern, such as apodization [32], weighted IDT 252 transducers [33], multistrip couplers (MSCs) [34], etc. 253 All these optimizations aim to reduce the impacts of spurious modes or secondary 254 effects, to better confine the acoustic energy and improve the Q value, and therefore to 255 improve the performances of SAW devices. 256 The BAW devices have a single port; they are usually characterized by measuring 257 their reflection coefficient 11 . As shown in Figure 8 (a), a minimum return loss is obtained 258 near the resonance loss with a high Q value. For a BAW device with good performance, 259 the Q value can reach several thousand [19,35,36]. Spurious BAW modes are also a possible source of reduced performance. Among them, so-called deep bulk acoustic waves (DBAW) can be generated by the input IDT, then propagate in the volume of the piezoelectric layer, get reflected at the bottom face, and propagate back to the receiving IDT. Surface skimming bulk waves (SSBW) and other leaky waves are also examples of possibly interfering (or sometimes alternating) waves [31].
Waves undergoing reflections on the device sides or bottom can be limited by sandblasting the bottom surface or biasing the sides in order to avoid constructive reflections travelling back to the IDTs. Other interfering waves such as SSBW need, for intrinsic good design, to take into account both longitudinal and shear acoustic waves.
Other particular IDT structures can also be implemented during the design process in order to achieve a specific response pattern, such as apodization [32], weighted IDT transducers [33], multistrip couplers (MSCs) [34], etc.
All these optimizations aim to reduce the impacts of spurious modes or secondary effects, to better confine the acoustic energy and improve the Q value, and therefore to improve the performances of SAW devices.
The BAW devices have a single port; they are usually characterized by measuring their reflection coefficient S 11 . As shown in Figure 8a, a minimum return loss is obtained near the resonance loss with a high Q value. For a BAW device with good performance, the Q value can reach several thousand [19,35,36].
In addition to S 11 , the electrical impedance resulting from the theory of transmission lines [37] is an important characterization property for BAW devices, which can be expressed as where Z is the electrical impedance of the device and Z 0 is transmission line characteristic impedance, typically 50 Ω for a vector network analyzer (VNA). Micromachines 2022, 13, x FOR PEER REVIEW 8 of 21 In addition to 11 , the electrical impedance resulting from the theory of transmission 2 lines [37] is an important characterization property for BAW devices, which can be ex-2 pressed as where is the electrical impedance of the device and 0 is transmission line characteris-2 tic impedance, typically 50 Ω for a vector network analyzer (VNA). 2 As is shown in Figure 8 (b), the electrical impedance reaches a minimum (tends to 2 zero) at the resonance frequency , which corresponds to a maximum of the mechanical 2 deformation caused by the piezoelectric effect; at the antiresonance frequency , the elec-2 trical impedance reaches a maximum (tends to infinity). 2 However, longitudinal deformations are accompanied by transverse ones, so that 2 some lateral acoustic modes also propagate in BAW devices [38]. These unwanted acous-2 tic modes, or spurious modes, are visible in the electrical response of the resonator in the 2 form of parasitic resonances, in addition to the main one. These lateral acoustic waves 2 travel between the boundaries of the active region of the piezoelectric layer, bounce off 2 the electrode edges, and form lateral standing waves. Since they have to share the total 2 energy, they are responsible for the degradation of the effective electromechanical cou-2 pling coefficient and of the Q value [26]. 2 There are mainly two kinds of methods to improve this degradation caused by un-2 desirable lateral modes. One method is called apodization. By using an asymmetrically 2 shaped top electrode [39], most of the standing lateral waves are smeared out between the 2 electrode edges and fewer parasitic resonances are observed in the electrical response. The 2 most commonly used shapes of BAW top electrodes are irregular squares, pentagons, and 2 circles. Another method is to build a frame around the edge of the top electrode [40]. By 2 carefully designing the width and thickness of the frame, the different orders of lateral 2 modes couple together and are vanished. This is an efficient method for suppressing un-2 wanted modes, confining energy, and obtaining a smooth electrical response and a better 2 Q factor [40]. However, due to the complexity of this structure and the difficulties in pro-2 cessing, the apodization method is commercially very successful.   2 SAW devices are widely used for signal filtering in the field of telecommunication. 2 To meet the current requirements, filters must have a large enough passband, which can 2 be adjusted with a suitable IDT design. In this perspective, as for unidirectional IDTs, an 2 appropriate design can generate specific filter templates [4]. Beyond that, much attention 2 has also been paid to filter configurations that smartly combine several one-port SAW 2 resonators, as impedance elements, with different topologies such as interdigitated 2 As is shown in Figure 8b, the electrical impedance Z reaches a minimum (tends to zero) at the resonance frequency f r , which corresponds to a maximum of the mechanical deformation caused by the piezoelectric effect; at the antiresonance frequency f a , the electrical impedance Z reaches a maximum (tends to infinity).

RF filters
However, longitudinal deformations are accompanied by transverse ones, so that some lateral acoustic modes also propagate in BAW devices [38]. These unwanted acoustic modes, or spurious modes, are visible in the electrical response of the resonator in the form of parasitic resonances, in addition to the main one. These lateral acoustic waves travel between the boundaries of the active region of the piezoelectric layer, bounce off the electrode edges, and form lateral standing waves. Since they have to share the total energy, they are responsible for the degradation of the effective electromechanical coupling coefficient and of the Q value [26].
There are mainly two kinds of methods to improve this degradation caused by undesirable lateral modes. One method is called apodization. By using an asymmetrically shaped top electrode [39], most of the standing lateral waves are smeared out between the electrode edges and fewer parasitic resonances are observed in the electrical response. The most commonly used shapes of BAW top electrodes are irregular squares, pentagons, and circles. Another method is to build a frame around the edge of the top electrode [40]. By carefully designing the width and thickness of the frame, the different orders of lateral modes couple together and are vanished. This is an efficient method for suppressing unwanted modes, confining energy, and obtaining a smooth electrical response and a better Q factor [40]. However, due to the complexity of this structure and the difficulties in processing, the apodization method is commercially very successful.

RF Filters
SAW devices are widely used for signal filtering in the field of telecommunication. To meet the current requirements, filters must have a large enough passband, which can be adjusted with a suitable IDT design. In this perspective, as for unidirectional IDTs, an appropriate design can generate specific filter templates [4]. Beyond that, much attention has also been paid to filter configurations that smartly combine several one-port SAW resonators, as impedance elements, with different topologies such as interdigitated interdigital transducer SAW (IIDT), double-mode SAW (DMS), or ladder-type [41]. The ladder-type filter is a very common configuration of such low-loss coupled SAW, represented in Figure 9, which consists of cascaded multiple stages, each one based on two SAW resonators connected in series and in parallel. Coupling them by matching the resonance frequency of a serial resonator with the antiresonance frequency of the parallel one results in a bandpass filter centered on this frequency. Such filters exhibit a relatively flat passband with low loss and good rejection of out-of-band noise [30]. Similarly, a design coupling identical SAW resonators in double symmetric and antisymmetric modes by inverting their electrical connections can result in a wider passband filter [30,42].
interdigital transducer SAW (IIDT), double-mode SAW (DMS), or ladder-type [41]. The 299 ladder-type filter is a very common configuration of such low-loss coupled SAW, repre-300 sented in Figure 9, which consists of cascaded multiple stages, each one based on two 301 SAW resonators connected in series and in parallel. Coupling them by matching the res-302 onance frequency of a serial resonator with the antiresonance frequency of the parallel 303 one results in a bandpass filter centered on this frequency. Such filters exhibit a relatively 304 flat passband with low loss and good rejection of out-of-band noise [30]. Similarly, a 305 design coupling identical SAW resonators in double symmetric and antisymmetric modes 306 by inverting their electrical connections can result in a wider passband filter [30,42]. 307 308 Figure 9. Ladder-type configuration filter for BAW and 1-port SAW devices. 309 There are some important properties to characterize a SAW filter: 310 (1) The minimum insertion loss ( 21 parameter), quantifying the power dissipation 311 caused by the device access, which depends on the input/output impedances of the device 312 itself and of the input and output circuits. In case of unmatching, calibration tests are used 313 to post-process the measurement results. Acoustic propagation can also participate in 314 losses. 315 (2) The center frequency, the arithmetic average of the two cut-off frequency values, 316 -3 dB or half-power of the minimum insertion loss level. 317 (3) The nominal frequency range, which is the usable bandwidth over which signal 318 transmission is observed, defined as the range between the two cut-off frequencies. 319 (4) The out-of-band rejection, the ratio of signals inside and outside the passband, 320 which is defined as the drop-off value between the edge of the passband and the maxi-321 mum value of the stopband. 322 (5) The Q value determines the maximum intrinsic bandwidth of a filter; it corre-323 sponds to the ratio of its center frequency to its 3 dB bandwidth. 324 With the rapid development of 5G technology, RF filters with a high frequency and 325 Q value become increasingly in demand in the mobile phone industry as well as with new 326 growth opportunities. This context gives RF SAW filters a significant market prospect. 327 From the report "Surface Acoustic Wave Filter Market" by Persistence Market Research, 328 the global SAW filter market registered a compound annual growth rate (CAGR) of 7.5% 329 between 2015 and 2020, and was expected to reach USD 5 billion in 2021 with a CAGR of 330 9% by 2031. Over 50 % of the market is shared by American and Japanese companies, such 331 as Qualcomm Technologies, Qorvo, Skyworks Solutions, Microchip Technologies, and 332 Murata Manufacturing [43,44]. 333 Like SAW, BAW devices, currently increasingly of FBAR type, are also fundamental 334 components for RF filters, requiring a wide bandwidth with low insertion loss and a stop-335 band enabling the suppression of unwanted signals. Similarly, as for the SAW compo-336 nents, the ladder-type filters are also commonly used due to advantages such as easy de-337 sign, steep filtering effect, cost, etc. Some communication systems mix ladder filters and 338 lattice ones, in which shunt elements are diagonally crossed, to achieve a good selectivity 339 of frequency bands with steeper filtering response [26,45]. Since the resonance frequency 340 of SAW mainly depends on the spatial periodicity of IDTs, which is limited by lithography 341 and patterning technology, it is quite difficult for SAW to operate above 2 GHz [46], so 342 There are some important properties to characterize a SAW filter: (1) The minimum insertion loss (S 21 parameter), quantifying the power dissipation caused by the device access, which depends on the input/output impedances of the device itself and of the input and output circuits. In case of unmatching, calibration tests are used to post-process the measurement results. Acoustic propagation can also participate in losses.
(2) The center frequency, the arithmetic average of the two cut-off frequency values, −3 dB or half-power of the minimum insertion loss level.
(3) The nominal frequency range, which is the usable bandwidth over which signal transmission is observed, defined as the range between the two cut-off frequencies.
(4) The out-of-band rejection, the ratio of signals inside and outside the passband, which is defined as the drop-off value between the edge of the passband and the maximum value of the stopband.
(5) The Q value determines the maximum intrinsic bandwidth of a filter; it corresponds to the ratio of its center frequency to its 3 dB bandwidth.
With the rapid development of 5G technology, RF filters with a high frequency and Q value become increasingly in demand in the mobile phone industry as well as with new growth opportunities. This context gives RF SAW filters a significant market prospect. From the report "Surface Acoustic Wave Filter Market" by Persistence Market Research, the global SAW filter market registered a compound annual growth rate (CAGR) of 7.5% between 2015 and 2020, and was expected to reach USD 5 billion in 2021 with a CAGR of 9% by 2031. Over 50% of the market is shared by American and Japanese companies, such as Qualcomm Technologies, Qorvo, Skyworks Solutions, Microchip Technologies, and Murata Manufacturing [43,44].
Like SAW, BAW devices, currently increasingly of FBAR type, are also fundamental components for RF filters, requiring a wide bandwidth with low insertion loss and a stopband enabling the suppression of unwanted signals. Similarly, as for the SAW components, the ladder-type filters are also commonly used due to advantages such as easy design, steep filtering effect, cost, etc. Some communication systems mix ladder filters and lattice ones, in which shunt elements are diagonally crossed, to achieve a good selectivity of frequency bands with steeper filtering response [26,45]. Since the resonance frequency of SAW mainly depends on the spatial periodicity of IDTs, which is limited by lithography and patterning technology, it is quite difficult for SAW to operate above 2 GHz [46], so FBARs usually dominate the market for filters above 2.5 GHz. This is further reinforced with the 5G New Radio (NR) systems, since a main feature is the use of high-frequency millimeter wave (mmWave) and sub-6 GHz bands [47]. As a consequence, the operating frequency band will continue to expand to high frequencies, and the working bandwidth will also increase, which can be supported by the newest FBAR-based BAW technologies.
As a result, the market for filters is expected to grow explosively. Among them, the growth of BAW filters is the fastest. Indeed, the demand for connected devices, such as vehicles, is leading to the new adoption of interface standards such as Wi-Fi, and BAW filters can also be used to establish a mobile connection with a network to enable a nextgeneration driving experience. The applications extend not only to the consumer electronics and automotive industries, but also to aerospace, defense, environment and industry, etc. These broad industrial applications provide prospects for significant and stable growth to the BAW filter market, which was estimated at USD 4.1 billion in 2020. By 2027, the global market for BAW filters is expected to reach USD 13 billion with a CAGR of 18% [48]. In terms of regional analysis, according to Maximize Market Research, Asia-Pacific has the largest market share and will continue to hold it in the future; Asian countries such as China, India, South Korea, and Japan are also the main consumers, and China holds the most important part due to its mature semiconductor manufacturing, telecommunications, and electronics industries [49]. The top five manufacturers in the BAW filter market are Avago Technologies (USA), Qorvo (USA), TDK (Japan), Skywork Solutions (USA), and Akoustis Technologies (USA) [49].

Physical Sensors and Actuators
Beside the filtering application, SAW and BAW devices are also widely used as sensors and different configurations have been developed for various physical monitoring. Among them, the QCM has been used as a way of real-time monitoring of thin film deposition thickness in microelectronics such as for evaporation equipment, based on mass-effect, with a resolution in the order of ng·cm −2 [50]. Many other applications have been studied, such as magnetic field [51], pressure and temperature [52], acceleration [53], tire-road friction [54], and gyroscopes [55], in many sectors including automobile, consumer, etc. [56]. In addition, SAW devices are also reported as motors and actuators [57].
A SAW resonator with a delay line is reported as a highly sensitive magnetic-field sensor by several studies [58][59][60]. For example, Meyer et al. [58] reported a thin-filmbased SAW magnetic-field sensor with a limit detection of 2.4 nT/ √ Hz at 10 Hz and 72 pT/ √ Hz at 10 kHz. This magnetic-field sensor is composed of an AlScN piezoelectric layer, AlCu IDTs, a SiO 2 smoothing (guiding) layer, and a magnetostrictive FeCoSiB film deposited on the delay-layer area. In the study of Schmalz et al. [60], a multimode Lovewave SAW magnetic-field sensor was designed and the first-and second-order Love-wave modes showed sufficient sensitivity. This sensor is composed of an ST-cut quartz as the piezoelectric layer, a SiO 2 layer, and a magnetostrictive (Fe 90 Co 10 ) 78 Si 12 B 10 film on the delay-line area. The BAW-based magnetic-field sensor is less used, but some designs and simulations of the BAW-based magnetic-field sensor are reported [61,62].
SAW devices are also good candidates for pressure and temperature sensing [63][64][65][66]. In the Rodríguez-Madrid et al. [66] study, a SAW-based pressure sensor was reported with a sensitivity of 0.33 MHz/bar, a working frequency between 10 to 14 GHz with high-order harmonic acoustic modes. It is a one-port resonator with AlN as the piezoelectric layer deposited on a free-standing nanocrystalline diamond (NCD) layer. Müller et al. [65] reported a SAW-based temperature sensor with a sensitivity higher than 300 kHz/ • C with an operating frequency around 5.4 GHz. It is a GaN-based SAW sensor; the detection of temperature is realized by tracking resonance frequency changes as a function of temperature. Many other studies have also reported SAW-and BAW-based sensors for high temperature detection, suitable for operation in harsh environments [67][68][69].
At the same time, SAW-based mechanical sensors are widely used in the automotive industry, with applications such as acceleration, tire-road friction sensors, etc. In this case, a wireless readout is often used as very useful facility [52,56]. For example, Wen et al. [53] reported a SAW-based acceleration sensor with a sensitivity of 29.7 kHz/g. It is a twoport SAW device with an operating frequency of 300 MHz and a very good temperature compensation system achieved by using a metal package base. SAW sensors are also used for tire pressure and tire-road friction in car and truck tires as illustrated in this study [54]. A continuous monitoring of tire pressure, which can be reduced to 50 mBar, is achieved to estimate the riding conditions (for example with braking maneuvers, over a curbstone, etc.). SAW devices have also been well-investigated as gyroscopes for several decades [70]. For example, a standing-wave-type SAW gyroscope was proposed by Kurosawa et al. [71]. This design was then confirmed by Varadan et al. [72] with experimental results. A twodelay-line SAW micro rate gyroscope was then proposed by Lee et al. [73]. Another progressive-waves-type SAW gyroscope was proposed by Oh et al. [74,75].
Besides such physical sensing applications, SAW devices can also serve as motors and actuators [57]. For example, Kurosawa et al. [76,77] proposed a Rayleigh-type SAW-based motor with an operating frequency near 10 MHz. It is a two-port SAW device with a delay line; a preloaded slider is placed on the wave propagation path and is driven by the frictional force. In addition, SAW devices are also widely used in microfluidic actuation and micro-object manipulation [78], based on travelling or standing waves between two sets of IDTs. Leaky-type travelling waves are usually used for actuation and manipulation, and Lamb-type standing waves are used for micro-actuators [79].

Biochemical Sensors
Apart from physical sensors, biochemical sensing has become a popular research topic, and a great deal of work has been carried out related to such applications of SAW and BAW for gas and liquid media [80]. Currently, such sensors can show a sensitivity with detection limits down to the ppb range [81], and SAW devices also receive intense attention in biosensing for their high sensitivity, high efficiency with label-free detection, real-time monitoring, and relatively low cost [82,83]. The sensitivity is a key factor characterizing the SAW sensor's performance [52], often improved with a sensitive layer deposited on the acoustic propagation path, which is supposed to immobilize or interact specifically with target molecules. As the acoustic wave propagation is perturbed, both the resonance frequency and the minimum loss (typically S 21 attenuation for a two-port SAW device such as a delay-line device) are modified due to an additional mass or viscoelastic property changes at the near surface. By tracking S 21 attenuation, resonance frequency shift, and phase, various vapors or liquids with different concentrations can be distinguished. Additional information on the real-time behavior of the adjacent medium can even be extracted from the electrical characterization out of acoustic resonance or on the reflection parameters [29].
Indeed, in 1979, Wohltjen and Dessy [84,85] first demonstrated the application and possibility of chemical/gas sensors based on SAW devices. Since then, by depositing different sensitive layers, SAW device-based gas sensors have been developed for detection of H 2 [86], H 2 S [87,88], NO 2 [89,90], CO 2 [89], CH 4 [91], SO 2 [92], NH 3 [93], O 3 [94], O 2 [95], CO [96], volatile organic compounds (VOCs) [97][98][99], explosives [100,101], etc. In 2022, Singh et al. [102] reported a SAW-based particulate matter (PM2.5) sensor which is wearable and shows a high detection sensitivity. Another important advantage of SAW devicebased sensors is that they are passive components with the potential to be interrogated wirelessly. Indeed, by using antennas, acoustic waves can be excited and received by RF electromagnetic signals. This allows SAW-based gas sensors to work in high-temperature, high-pressure, and toxic environments. Wen et al. [103] reported such a wireless SAW gas sensor with Teflon AF as a sensitive layer for CO 2 detection; they reported a sensitivity in phase shift of 1.98 • /ppm. Later, Lim et al. [89] developed a remotely controlled SAW sensor for the detection of CO 2 and NO 2 with simultaneous temperature measurement. The sensitive layers for CO 2 and NO 2 are Teflon AF and indium tin oxide, and the sensitivities are 2.12 • /ppm and 51.5 • /ppm, respectively. Xu et al. [104] developed a wireless SAW sensor for organophosphorus compound detection; the sensor exhibits good linearity and repeatability, and a sensitivity of 20.1 • /(mg/m 3 ).
Apart from applications in a gaseous environment, SAW-based sensors are also widely used in the liquid phase. In this case, they mostly involve waves horizontally polarized, as vertical components suffer from fast attenuation in the liquid phase, and SAW Love-mode is actually well-investigated for its high sensitivity of detection, especially in liquid [105]. Among important applications of such biosensors is the detection of deoxyribonucleic acid (DNA). Y. Hur et al. [106] reported a 15-meroligonucleotide DNA sensor in liquid solutions with a sensitivity reaching 155 ng/mL/Hz. Kim et al. [107] reported a DNA sensor with a low detection limit of 1 ng/mL and rapid response; this sensor has the potential to be used in wireless mode. Zhang et al. [108] developed a DNA sensor with a sensitive layer of deoxynucleoside transferase in order to increase the phase shift, thus lowering the detection limit down to 0.8 pM. Cai et al. [109] reported a DNA sensor with a high-order harmonic acoustic mode; the sensitivity can reach 6.7 × 10 −16 g/cm 2 for target DNA. SAW devices are also used for protein detection. Agostini et al. [110] developed a biosensor targeting the Streptavidin protein and the detection limit is down to sub-nanomolar, 104 × 10 −12 . M. Choi et al. [111] developed a SAW sensor for cardiac troponin I; the detection limit is down to 24.3 pg/mL. Zhang et al. reported a carcinoembryonic antigen (CEA) biosensor; the sensitivity was reinforced by injecting a gold staining solution, allowing a detection limit down to 1 ng/mL. Jandas et al. [112] also reported a CEA sensor; the delay-line area was coated with gold and immobilized self-assembled monolayers (SAMs) of anti-CEA antibodies. The detection limit is down to 0.31 ng/mL. They later improved their sensor with a nanomaterial thin-film bioreceptor and the detection limit reached down to 0.084 ng/mL [113]. Apart from the detection of proteins, Brugger et al. [114] reported the use of SAW for monitoring the formation of neural networks and some investigation for real-time monitoring of living matter, such as organoids or other biomaterials, along with the addition that innovative nanodrugs for efficiency and/or toxicity evaluation can also be imagined [115]. Since the COVID-19 pandemic began, some SAW-based sensors for COVID virus detection have also been reported [116,117].
Similarly, BAW devices are also widely used as chemical and biological sensors, firstly based on the classical QCM, and taking into account surrounding physical parameters such as temperature and pressure, as previously described. Its operating frequency can reach up to tens of MHz and thickness-shear mode (TSM) is mostly studied [86]. With similar advantages to SAW devices but a lower frequency, some commercial products based on QCM allowed use for a large number of applications involving mass change measurements at the nanoscale resolution. Moreover, the addition of dissipation monitoring, known as "QCM-D", allows an improved characterization of both mass and viscoelastic properties changes of the medium driven at the near surface [118]. With the rapid development of BAW devices, the FBAR is also becoming a popular topic due to its good sensitivity with a higher operating frequency ranging from several to tens of GHz. The detection field includes mass pressure, gas, liquid, chemical/biosensor, etc. [119] For example, Chen et al. [120] developed a ZnO-based FBAR as a gravimetric DNA biosensor with a working frequency of 1.67 GHz. As for SAW sensors, a sensitive layer is deposited onto the top electrode of the device and the absorbed target compounds interfere with the generation and propagation of the acoustic waves. Again, since the vertical deformation is strongly attenuated into an adjacent liquid medium, a thickness-shear-mode resonator (TSM) is usually preferred for biochemical sensing applications [121].
Since the FBAR is a one-port device, detection typically consists in tracking the attenuation changes as well as the frequency or phase shift of the S 11 parameter at the resonance [120]. The detection is also based on mechanical effects, among them the mass loading effect. Compared to the SAW sensor, whose operating frequency is typically in the range of one hundred MHz to GHz, the operating frequency of the FBAR sensor is usually in the range of sub-GHz to about 10 GHz due to the wave confinement in a very thin layer, which provides the FBAR sensor with a high sensitivity. However, a high Q factor should be ensured to accurately detect small frequency shifts.
In the field of gas sensing, Lin et al. [122] developed an FBAR-based sensor exhibiting a high sensitivity for trinitrotoluene (TNT) and 1,3,5-trinitro-1,3,5-triazacyclohexane (RDX). The detection of hydrogen (H 2 ), carbon monoxide (CO), and ethanol vapors was also reported by Benetti et al. [123], with detection limits of 2 ppm, 40 ppm, and 500 ppm, respectively. Coupling an FBAR-based sensor with a micro-preconcentrator, Yan et al. [124] showed a high sensitivity for dimethyl methyl phosphonate (DMMP), down to 2.64 ppm, with a fast response and a short recovery time. Zeng et al. [125] developed a temperature-compensated film bulk acoustic wave resonator (TC-FBAR) functionalized with a bilayer self-assembled poly (sodium 4-styrene-sulfonate)/poly (diallyl-dimethylammonium chloride) to detect and identify volatile organic compounds (VOCs), with an interesting approach based on temperature modulation as a multiparameter virtual sensor array. Gao et al. [126] also proposed a solution for VOC identification, based on a dual transduction using mass and resistance variation of a conductive polymer, poly (3,4-ethylenedioxy-thiophene) and poly (styrene sulfonate) (PEDOT: PSS), deposited on the top of the device. The detection of 380 ppm of methanol was reported.
An FBAR device is also a good candidate for sensing in the liquid phase. As the aging of the population and disease concerns have been major topics for many decades, not to mention the impact brought by the novel coronavirus (COVID- 19), biosensing technology in the liquid phase based on FBAR devices shows a wide perspective. The first FBAR-based biosensor was reported in 2003 [127]. Clear frequency shifts show DNA attachment and protein coupling. In 2004, Gabl et al. [128] reported a ZnO-based FBAR biosensor with a working frequency up to 2 GHz for the detection of DNA and protein [128]. In 2006, Weber et al. [129] showed experimentally that in the liquid phase, shear-mode FBARs have much better performance than longitudinal-mode FBARs due to higher quality factor (Q value) and lower noise level, as expected in the liquid phase. DNA sequences were also successfully detected by Zhang et al. [130], using a gold-top electrode FBAR and monitoring the shift of resonance frequency when DNA hybridization occurred. In 2011, Auer et al. reported DNA detection in a diluted serum (1%) with a resolution of 1 nM. Apart from the detection of DNA, FBAR devices have also shown their capacity to detect prostate-specific antigen (PSA), alpha-fetoprotein (AFP), and CEA [131]. Previously in 2011, Lin et al. [132] reported an FBAR sensor with a detection limit of PSA of 25 ng/cm 2 . Zhao et al. [133] reported a sensor of human prostate-specific antigen (hPSA) with a sensitivity of 1.5 ng/cm 2 . Chen et al. [134] reported a sensor of AFP with a detection limit of 1 ng/mL. Zhang et al. [131] developed a sensor of CEA, and the detection limit ranges from 0.2 to 1 mg/mL. As we presented above, acoustic wave device (SAW and BAW)-based sensors have wide applications as physical and biochemical sensors; the market for these sensors also holds an important share. Based on the report "Acoustic Wave Sensor Market-Forecasts from 2021 to 2026" by Research and Markets, the global acoustic wave sensor market will grow from USD 836.17 million in 2019 to USD 2400.54 million in 2026, with a CAGR of 16.3%. The major players are Hawk Measurement Systems (Melbourne, Australia), NanoTemper technologies GmbH (München, Germany), Pro-micron GmbH (Kaufbeuren, Germany), Siemens (Munich, Germany), and Transense (Oxford, UK) [135].

New Trend: Quantum Acoustics
The discussions above proposed a review of SAW and BAW devices and key applications, up to the most recent ones in terms of major trends, namely high-frequency filters and sensors. Here, we wish to give a brief introduction about a novel stream of application: quantum acoustics. Over recent decades, the research on electrons, photons, or magnetics has led a brilliant revolution in both scientific and industrial fields, and in recent times, attentions have shifted to the phonon, a quasi-particle that describes the excitation and vibration in a periodic, elastic arrangement of atoms or molecules [136]. Since most acoustic devices employ the mechanical vibrations generated from the piezoelectric effect, acoustic devices are ideal candidates to probe and control these quantum excitations in certain conditions. This is especially true for SAW devices, since their detection and sensing applications are well developed, and some examples are reported in this part.

Single-Electron Probing/Controlling
With the development of semiconductor technology, current integrated circuits are composed of a huge number of transistors. In order to lower the power consumption while improving the performance, scientists have put a large amount of effort into the field of low-dimensional electronic conductors to single-electron electronics in recent decades. Recently, SAW devices have shown good potential for single-electron probing or controlling due to their high operating frequencies and high Q value at low temperatures [137][138][139]. In order to detect the coupling of single-electron and SAW, a single-electron transistor (SET) is necessary [140]. Compared to classic transistors, with advantages in terms of size, voltage, and sensitivity, the SET is the most sensitive electrometer, allowing single-electron control with working temperatures in the mK range.
Gustafsson et al. [137] presented local probing of SAW for the detection of single electrons, reaching the single-phonon level at a frequency of 932 MHz. For the sample layout, a SET was placed on the propagation path of a SAW resonator, which was polarized by the piezoelectric charge when SAW passed underneath. The measurement setup was well designed and SAW were detected by their rectifying effect on the SET's gate modulation curve. Propagating acoustic pulses with an extremely low magnitude were detected. After calculation, for each pulse, the SAW energy passing under the SET was less than a single phonon on average, which proved the possibility of single-shot phonon detection.
Other applications in single-electron controlling are also reported. For example, Takada et al. [138] reported a SAW-driven single-electron transfer with an efficiency of 99%, which can be used to perform quantum logic operations with flying electron qubits and is a significant step to efficient quantum computers. Hsiao et al. [141] used a SAW-driven lateral n − i − j junction, which operates in the single-electron limit, to generate single photons, and this electron-to-photon quantum transfer marks the first step for long-distance qubit transfer.

Coherent Coupling between Magnons and Phonons
As we showed above, the coupling of photon-phonon is proven in certain conditions and some SAW-driven devices are developed for single-electron controlling. Therefore, magnetic and acoustic excitations, magnons and phonons, are also expected to interact with each other; in particular, magnons are shown to undergo a strong coupling with microwave photons [142][143][144]. For this reason, increasing attention is focused on coherent interactions between magnons and phonons recently [145][146][147][148].
Weiler et al. [148] demonstrated the detection of acoustic-driven ferromagnetic resonance (FMR), which shows the magnon excitation induced by SAW. Figure 10a shows the experimental setup, with a 50 nm thick polycrystalline ferromagnetic nickel film deposited on the propagating path of a SAW resonator. On Figure 10b is represented the S 21 parameter without an applied external magnetic field; several resonances can be observed clearly and the inset highlights a 5 dB magnitude difference of the ninth harmonic resonance under two different magnetic fields. Figure 10c shows the SAW delay-line transmission parameter as a function of magnetic field strength; it exhibits a valley associated with Ni FMR, therefore proving the magnetoelastic coupling.  [148]. 606 Recently, Zhao et al. [149] succeeded in visualizing acoustic FMR by micro-focused 607 Brillouin light scattering ( − ). Casals et al. [150] showed independent imaging of 608 magnons and SAW with the synchrotron X-ray source. Besides these examples of quan-609 tum acoustics applications, the strong coupling between magnons and phonons is still 610 under investigation [146,[151][152][153][154]. Furthermore, research in other fields continues, such as 611 coherent coupling between phonons [155][156][157][158], coupling between elastic waves and single 612 quantum dots [159][160][161][162], etc. These advances in quantum acoustics will likely bring a rev-613 Recently, Zhao et al. [149] succeeded in visualizing acoustic FMR by micro-focused Brillouin light scattering (µ − BLS). Casals et al. [150] showed independent imaging of magnons and SAW with the synchrotron X-ray source. Besides these examples of quantum acoustics applications, the strong coupling between magnons and phonons is still under investigation [146,[151][152][153][154]. Furthermore, research in other fields continues, such as coherent coupling between phonons [155][156][157][158], coupling between elastic waves and single quantum dots [159][160][161][162], etc. These advances in quantum acoustics will likely bring a revolution in current electronics science and engineering.

Conclusions
In this review, we provided a global view about current acoustic devices. We presented the piezoelectric effect, basic structures, the acoustic theory of SAW and BAW components, and gave more details for thin-film-type BAW resonators or FBARs, which hold interesting features in terms of frequency operation and integrability compared with conventional QCM devices. We also presented the possible spurious modes and some optimized designs to reduce them and therefore improve the response. Acoustic devices have been developed over the past decades; they have proven their wide applications in communication systems. Today, with the rapid development of the fifth generation (5G) and telecommunication standards, these acoustic devices, especially FBARs, represent a broad market as RF filters, compared with conventional electromagnetic devices, thanks to much slower propagation velocity allowing for shorter wavelength and, thus, easy miniaturization and integration into circuits. We then presented another important field of applications of SAW and BAW/FBARs, namely as sensors and actuators. A section was dedicated for their application as physical sensors. Examples of their use for magnetic field, pressure, and temperature monitoring and detection were illustrated. In addition, their application in other fields such as mechanical (in automotive) and orientation measurements were presented. Some examples of SAW-based motors and actuators were also introduced. We then focused on SAW/BAW-based biochemical sensors, which are receiving increasing attention in the research field. Indeed, because of their performances, among them a high sensitivity, a versatile feature that makes them easily functionalized for selectivity, and low cost, they are widely used for gas, liquid, bio-sensing, etc. The sensing applications are still under development, with a rising demand especially for biosensors, since health concerns are more than ever a major topic. As of now, SAW and FBAR devices show a very good capacity for sensing DNA, RNA, proteins, and a wide variety of other bio-compounds. With the COVID-19 pandemic, several biosensors based on SAW and FBAR devices are also reported for the detection of SARS-CoV-2 virus and application for living-matter monitoring is under development, which could be helpful for fast screening of therapeutic nanodrugs, for example. Lastly, we presented current trends related to quantum acoustics, which studies the behavior of phonons and their interactions, as opportunities for new schemes to control quantum information and explore atomic physics beyond photonic systems. SAW is the ideal candidate in this emerging field with interest in both fundamental and applied research.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.