Lateral bulk acoustic wave filter

Abstract

To provide an improved lateral bulk acoustic wave filter.SOLUTION: There is disclosed an acoustic wave filter device. The device includes a piezoelectric layer, an input electrode and an output electrode located on a top surface of the piezoelectric layer and physically separated from one another, and a counter electrode having a top surface connected to a bottom surface of the piezoelectric layer. The input and output electrodes each include a base and at least one extension extending from the base. The at least one extension of the input electrode extends along an adjacent extension of the at least one extension of the output electrode in a generally opposite direction, and is separated by a gap width from an adjacent extension of the at least one extension of the output electrode. In some embodiments, the at least one extension of the input or output electrode has a width that changes from a first end of the at least one extension to a second end.SELECTED DRAWING: Figure 8A

Description

[Technical field]

[0001] This specification relates to thin film high frequency acoustic wave filters. [Background technology]

[0002] Radio frequency ("RF") components such as resonators and filters, based on micro-acoustic and thin-film technologies, are widely used in wireless applications such as mobile phones, wireless networks, satellite positioning, etc. Their advantages over their lumped element, ceramic, and electromagnetic counterparts include small size and mass producibility. Summary of the Invention [Problem to be solved by the invention]

[0003] This application describes the technology of lateral bulk acoustic wave ("LBAW") filters. Herein, LBAW filters include input and output resonators formed from a piezoelectric layer sandwiched between two pairs of electrodes. LBAW filters are formed from a piezoelectric layer sandwiched between two pairs of electrodes. One electrode of each pair is disposed on the top surface of the piezoelectric layer and forms the input or output of the LBAW. The input and output electrodes are spaced apart by a gap. Each pair has a counter electrode disposed on the bottom surface of the piezoelectric layer. By applying an AC voltage to the piezoelectric layer of the input resonator, a mechanical resonance is created in the piezoelectric layer below the input electrode. The thickness of the piezoelectric layer and the gap between the electrodes can be designed such that this mechanical resonance is coupled across the gap to the output resonator. The frequency range over which such coupling occurs determines the achievable bandwidth (or width of the passband) of the LBAW filter. [Means for solving the problem]

[0004] In general, one innovative aspect of the subject matter described herein may be embodied in an acoustic wave filter device including a piezoelectric layer, input and output electrodes disposed on an upper surface of the piezoelectric layer and physically spaced apart from one another, and a counter electrode having a top surface connected to a bottom surface of the piezoelectric layer. Each of the input and output electrodes includes a base and at least one extension extending from the base. The at least one extension of the input electrode extends in a generally opposite direction along an adjacent extension of the at least one extension of the output electrode and is spaced apart by a gap width from the at least one adjacent extension of the output electrode. The thickness of the piezoelectric layer and the gap width between the extensions of the adjacent input and output electrodes cause application of a high frequency voltage between the input and counter electrodes to create symmetric and asymmetric acoustic thickness-extendable resonant modes in the voltage layer. At least one extension of the input or output electrode has a width that varies from a first end to a second end of the at least one extension.

[0005] Each of the above and other embodiments may optionally include one or more of the following features, alone or in combination.

[0006] The first end of the extension may be proximal to the base of the corresponding electrode, and the second end of the extension may be distal to the base of the corresponding electrode. At least one extension of one of the input or output electrodes may have a width that decreases from the first end to the second end of the at least one extension. In some examples, at least one extension of the other of the input or output electrodes has a width that increases from the first end to the second end of the at least one extension. At least one extension of the input electrode may increase in width from the first end to the second end of the extension of the input electrode, and at least one extension of the output electrode may decrease in width from the first end to the second end of the at least one extension of the output electrode. At least one extension of one of the input or output electrodes may have a width that increases from the first end to the second end of the at least one extension.

[0007] In some embodiments, the gap width between the extensions is constant. In some embodiments, the width of at least one extension of the input electrode or the output electrode varies continuously from a first end to a second end of the at least one extension.

[0008] In some embodiments, the input and output electrodes each have a comb structure with interdigitated extensions, hi some examples, the input and output electrodes have at least 10 extensions.

[0009] The acoustic wave filter may further include a Bragg reflector in contact with a bottom surface of the counter electrode.

[0010] In some embodiments, the thickness of the piezoelectric layer is at least 0.5 μm. The width of each extension may increase by 20% to 250% from the first end to the second end of at least one extension. The width of each extension may increase by 25% to 100% from the first end to the second end. In some embodiments, the width of each extension increases by 4 μm to 5 μm from the first end to the second end.

[0011] The acoustic wave filter may be a laterally coupled bulk acoustic wave filter.

[0012] One innovative aspect of the subject matter described herein may be embodied in an acoustic wave filter device including a piezoelectric layer, input and output electrodes disposed on an upper surface of the piezoelectric layer and physically spaced apart from one another, and a counter electrode having a top surface connected to a bottom surface of the piezoelectric layer. Each of the input and output electrodes includes a base and at least one extension extending from the base. The at least one extension of the input electrode extends in a generally opposite direction along an adjacent extension of the at least one extension of the output electrode and is spaced apart by a gap width from the at least one adjacent extension of the output electrode. The thickness of the piezoelectric layer and the gap width between adjacent extensions of the input and output electrodes cause application of a high frequency voltage between the input and counter electrodes to create symmetric and asymmetric acoustic thickness extendible resonant modes in the voltage layer. The gap width between adjacent extensions of the input and output electrodes varies between the base of the input electrode and the base of the output electrode.

[0013] Each of the above and other embodiments may optionally include one or more of the following features, alone or in combination.

[0014] In some embodiments, the gap width between adjacent extensions of the input and output electrodes increases between the base of the input electrode and the base of the output electrode, in some examples, the gap width between adjacent extensions of the input and output electrodes increases by between 2 μm and 3 μm.

[0015] In some embodiments, the gap width between adjacent extensions of the input and output electrodes decreases between the base of the input electrode and the base of the output electrode.

[0016] The input and output electrodes may each have a comb structure with interdigitated extensions.

[0017] The subject matter described herein may be implemented in certain embodiments that provide one or more of the following advantages: The LBAW filters described herein may be easier to manufacture since they use only a single piezoelectric layer compared to two vertically stacked bulk acoustic wave (BAW) coupled resonator filters. LBAW filters may operate at higher frequencies as surface acoustic wave (SAW) filters since operation is determined by the thickness of the piezoelectric layer rather than the dimensions of the IDT electrodes. In some embodiments, LBAW filters may also achieve wider bandwidths than BAW filters. LBAW filters may be performed as a filter with a single lithographic patterning step compared to the multiple (close to 10) lithographic steps used with BAW, and may operate without the reflectors required for SAW, and therefore at a smaller size.

[0018] In some embodiments, the LBAW filter may be designed to suppress sidebands, hi some embodiments, the LBAW filter may be designed to reduce ripple in the passband.

[0019] The details of one or more embodiments of the subject matter of this application are set forth in the accompanying drawings and the description below. Other features, aspects, and advantages of the subject matter will become apparent from the description, drawings, and claims. [Brief description of the drawings]

[0020] [Figure 1A] FIG. 1 is a schematic perspective view of a rigidly fixed LBAW filter. [Figure 1B] FIG. 1 is a schematic perspective view of a free-standing LBAW filter. [Figure 1C] FIG. 2 is a schematic plan view of an IDT electrode structure. [Figure 2A] FIG. 2 is a schematic diagram of one type of propagating plate wave mode in an LBAW piezo layer. [Figure 2B] FIG. 1 is a schematic diagram of another type of propagating plate wave mode in an LBAW piezo layer. [Diagram 3] 1 is an exemplary LBAW dispersion curve. [Figure 4A] FIG. 1 is a schematic diagram of two resonant modes in LBAW. [Figure 4B] 1 is an exemplary transmission response of LBAW as a function of frequency. [Diagram 5] 1 is an illustrative transmission curve of LBAW as a function of frequency. [Figure 6] FIG. 13 is a schematic diagram of different acoustic modes formed under multiple electrode extensions. [Figure 7] 1 is an exemplary plot of insertion loss versus frequency for an exemplary LBAW. [Figure 8A] FIG. 1 is a schematic top view of a cross section of a sector-shaped LBAW electrode with a tapered extension. [Figure 8B] FIG. 1 is a schematic top view of a sector-shaped LBAW electrode with tapered extensions. [Figure 9A] FIG. 1 is a schematic top view of a cross section of a sector-shaped LBAW electrode having a tapered gap width. [Figure 9B] FIG. 1 is a schematic top view of a sector-shaped LBAW electrode with a tapered gap width. [Figure 10A] 1 is a plot of insertion loss versus frequency for two LBAW filters having electrode extensions of constant width and gap. [Figure 10B] 10B is a plot of insertion loss versus frequency for the filter combination of FIG. 10A. [Figure 11] 1 is a plot of insertion loss versus frequency comparing the performance of LBAW with electrode extensions of constant width and gap to LBAW with sector electrodes with tapered extensions. [Figure 12] 1 is a plot of insertion loss versus frequency comparing the performance of an LBAW filter with electrode extensions of constant width and gap to an LBAW with sector-shaped electrodes. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0021] Like numbers in the different drawings indicate like elements.

[0022] 1A and 1C show an example of an LBAW filter (or resonator) 100 with an input electrode 150 and an output electrode 170 with an interdigitated shape (also called an "interdigital transducer" or "IDT" LBAW). The LBAW filter 100 includes a piezoelectric ("piezo") layer 110 having a thickness d, an IDT electrode structure 102 disposed on the top surface of the piezo layer, and a bottom counter electrode 120 disposed on the bottom surface of the piezo layer. The IDT electrode structure ("IDT") 102 includes two comb-like electrodes 150 and 170 of a conductive material, e.g., metal or polysilicon. The IDT electrodes 150 and 170 have parallel extensions 150a and 170a, respectively, that form the "teeth" or "fingers" of the "comb". The electrode 150 and the counter electrode 120 form an input resonator with the piezoelectric layer 110. The electrode 170 and the counter electrode 120 form an output resonator with the piezoelectric layer 110.

[0023] Acoustic vibrations are generated in the piezo layer 110 by applying an oscillating (or alternating) input voltage between the IDT electrode 150 and the bottom counter electrode 120 at the input port 160. The applied voltage is converted to mechanical (e.g., acoustic) vibrations via the piezoelectric effect. Under resonant conditions (e.g., in certain acoustic resonance modes, as described in more detail below), this vibration can generate a standing wave below the input electrode 150 and an evanescent wave (with an exponentially decaying amplitude) in the gap region 190. With appropriate selection of the vibration frequency and gap width G, the standing wave can be mechanically coupled across the gap 190 from the piezo region below the electrode 150 to the piezo region below the electrode 170 by the evanescent wave, generating a similar standing wave in the piezo layer 110 below the electrode 170. The standing wave below the electrode 170 results in an output signal voltage of the same frequency at the output port 180 via the inverse piezoelectric effect. The frequency range over which this coupling occurs at mechanical resonance with the strong piezoelectric coupling forms the passband (or bandwidth) of the LBAW filter 100. In some examples, the frequency range is 1.8-1.95 GHz. As described further below, the thicknesses and shapes of the various layers of the LBAW filter 100, as well as the gaps, can be adjusted to change the RF response and passband of the filter.

[0024] The reflecting structure 130 may help isolate the vibrations of the piezo layer 110 from the underlying substrate 140 and prevent acoustic leakage. The reflecting structure may be a stack of thin layers, e.g., a high acoustic impedance ("Z ac ") material layers arranged in alternating layers to form a Bragg reflector. The thicknesses of these layers can be designed so that frequencies at or near the passband of the LBAW filter are reflected back to the piezo layer 110, while all other frequencies pass through the mirror.

[0025] In some embodiments, the LBAW 100 does not directly cover the substrate 140 (as shown in FIG. 1A), but is free-standing, as shown in FIG. 1B. In such an arrangement, the substrate 140 and mirror 130 are replaced by an air gap, and the portion of the piezo that extends laterally beyond the area in which the LBAW 100 is fabricated is supported by the substrate 140.

[0026] 1C, the extensions 150a and 170a are rectangular and have a width W, a length L, and are spaced apart by a gap width G. Each electrode 150 and 170 has one or more extensions 150a and 170a, respectively. The total number of electrode extensions is designated as K.

[0027] FIG. 1C shows rectangular interdigital electrodes 150 / 170 with parallel extensions 150a / 170a of the same shape and gap G, although other electrode shapes are contemplated. Design considerations include the gap between the electrodes, the length of the electrodes, the number, if any, and the shape of the electrode extensions. The gap can be used to control the coupling between the input and output electrodes. Longer electrodes can also increase coupling. The number of extensions K can be used to control the bandwidth and / or increase coupling while conserving the area occupied by the electrodes. In some embodiments, the electrodes are composed of rectangular strips with two or more extensions (e.g., K≧2). For example, each extension can be a rectangular strip. In some embodiments, the electrodes are concentric circles or spirals with a common axis.

[0028] The piezoelectric layer 110 may be formed from a variety of piezoelectric materials. Exemplary materials include ZnO, AlN, CdS, PZT, LiNbO3, LiTaO3, quartz, KNN, BST, GaN, Sc, alloyed AlN, or the aforementioned materials doped or alloyed with additional elements. Doping may be used to improve or tune the electromechanical properties of the piezoelectric layer 110. As described in more detail below, the thickness d of the piezoelectric layer is selected such that a thickness extensional mode is generated in the piezoelectric layer near the frequency of the desired bandwidth of the LBAW filter. In some embodiments, the thickness d of the piezoelectric layer is selected to be greater than or equal to λ. Z λ is 50% to 20%, or 45% to 30%, Z is the wavelength of the piezoelectric vibration in the thickness direction. In some embodiments, d is between 1500 nm and 2500 nm, or between 1800 nm and 2200 nm.

[0029] The thin film IDT 102 may be constructed from a variety of materials. In some embodiments, the IDT electrodes 150 and 170 are metals. For example, the electrode material may include Al, Mo, Pt, Cu, Au, Ag, Ti, W, Ir, Ru, or metals doped with multiple layers of metals and / or additional materials, such as AlSi, AlSiCu, polysilicon, etc. Doping may be used to improve or tune the electrical or mechanical properties of the IDT.

[0030] FIG. 1A shows a single common counter electrode 120, but the filter 100 may include separate electrodes for the input and output resonators. Various materials are suitable for the counter electrode pole (e.g., electrode 120). For example, the electrode may include a metal such as Al, Mo, Pt, Cu, Au, AG, Ti, W, Ir, Ru, or a metal doped with multiple layers of metal and / or additional materials, e.g., AlSi, AlSiCu, etc. Doping may be used to improve or tune the electrical or mechanical properties of the IDT. For example, the electrode may be Ti+Mo, Ti+W, AlN+Mo, or Al+W. The electrode may be multi-layered. The electrode may have an extra thin seed layer deposited under the electrode.

[0031] The reflecting structure 130 may be composed of alternating layers of different materials. For example, the reflecting structure 130 may include alternating layers of two of tungsten (W), SiO2, silicon (Si), and carbon (C). For example, high acoustic impedance layers may include W, Mo, IR, Al2O3, diamond, Pt, AlN, Si3N4. Low acoustic impedance layers may include SiO2, glass, Al, Ti, C, polymers, or porous materials. The Si layer provides an intermediate acoustic impedance. Various materials such as Si or SiO2 or glass, sapphire, and quartz are suitable for the substrate 140. The material of the substrate 140 may have a high electrical resistivity. The substrate may have a thickness suitable for RF applications, such as integration into a mobile phone platform. For example, the substrate may have a thickness of less than 500 microns, or less than 200 microns. For example, Si wafers can be purchased in thicknesses of 675 and may be thinned to achieve a desired device thickness, for example for a mobile platform.

[0032] Modeling the acoustic response of the LBAW100 may provide guidance on how to adjust the design parameters of the individual elements of the structure to achieve the desired bandpass characteristics. For example, the LBAW100 may be designed to have resonant modes at specific frequencies. In general, the geometries of the various LBAW100 components may be selected to achieve different acoustic characteristics. The characteristics of the LBAW100 may depend on a combination of these geometries and may not be independent of each other.

[0033] In the piezoelectric layer 110, different bulk acoustic vibration modes can occur at different excitation frequencies f of the input voltage (e.g., at port 160). The acoustic vibrations of the piezoelectric layer 110 can propagate laterally as Lamb waves (or plate waves), with particle motion occurring in a plane that contains the wave propagation direction and the plate normal (e.g., the z-axis in FIG. 1A). Two such modes are shown in FIGS. 2A and 2B. With reference to FIG. 2A, a thickness extensional (TE or long axis) bulk mode 200 has particle displacements 210 perpendicular to the propagation direction (z-direction). With reference to FIG. 2B, a second order thickness shear (TS2) bulk mode 220 has particle displacements 230 parallel to the propagation direction (y-direction). For both modes, the lowest frequency at which resonance in the thickness direction can occur is when the thickness d of the piezoelectric layer 110 is half a wavelength λ. Z (regardless of the thickness of the electrodes 150 / 170), in other words,

number

number

[0034] The acoustic properties of the LBAW resonator 100 can be described by a dispersion curve. Referring to FIG. 3, an exemplary dispersion curve for the LBAW resonator 100 is shown with respect to the transverse wave number k of vibration. || where, as a function of frequency f,

number

[0035] In some embodiments, the LBAW 100 may be designed to produce Type 1 dispersion. For example, a material for the piezoelectric layer 110 may be selected that allows Type 1 dispersion to occur. For example, ZnO may be used. In another example, appropriate design of the acoustic Bragg reflector 130 may help achieve Type 1 dispersion. For example, using aluminum nitride ("AIN") for the piezoelectric layer 110 may initially reduce k with increasing frequency f. || decreases, then with increasing frequency f k ||This may produce a typical Type-2 dispersion, behaving non-monotonically as TE increases (similar to that illustrated by the dispersion curves in FIG. 3 with TE1 and TE2 swapped). However, in some embodiments, with proper design of the reflecting structure 130 (e.g., acoustic Bragg reflector), the LBAW 100 can use AlN in the piezo layer 100 and still achieve Type-1 dispersion. See, for example, Fattinger et al., "Optimization of acoustic dispersion for high performance thin film BAW resonators," Proc. IEEE International Ultrasonics Symposium, 2005, pp. 1175-1178.

[0036] In Figure 3, k || Positive values ​​of k indicate real wave numbers (propagating waves), || Negative values ​​of correspond to imaginary numbers (evanescent waves). For resonance to occur, the acoustic energy must be confined inside the LBAW resonator structure. In the thickness (z-axis) direction, the distance from the substrate (using the reflecting structure 130) can be used for energy trapping. In the lateral direction, evanescent waves can be generated when they are formed outside the electrode area (e.g., on the "no electrode" curve). To obtain resonant coupling between the two resonators of the LBAW (e.g., electrodes 150 / 170 and 120), a standing wave of the TE1 mode is formed in the active area of ​​the piezo layer (under the electrodes) and an evanescent wave is formed in the "no electrode" area. In other words, k ||is positive for the TE1 "electrode" curve and negative for the TE1 "no electrode" curve. According to Figure 3, this occurs in the frequency range labelled "trapping range". With type 1 dispersion, energy trapping can be achieved more easily. Without wishing to be bound by theory, we will use a monotonically increasing dispersion curve like the thick TE1 line in Figure 3. As shown in Figure 3, at the "electrode", at a single frequency within the trapping range, there is either a single virtual wave number available, or a single real wave number above the trapping range. The former means that TE1 does not propagate outside the electrode, the latter means that TE1 may couple to a propagating wave outside the electrode and "leak". Type 2 dispersion can be represented by a similar curve, with the TE1 and TS2 curves swapped. The fact that the type 2 curves are non-monotonic means that there may be several real wave numbers at a particular frequency. Having multiple wave numbers at a frequency means that propagating waves may be available outside the electrode, potentially causing "leak".

[0037] 4A-4B show the relationship between standing wave resonance modes and LBAW bandgaps. Referring to FIG. 4A, a portion of the LBAW 100 includes two adjacent electrodes 401 and 402 having a width W (e.g., corresponding to the extensions 150a and 170a of the respective electrodes 150 and 170 in FIG. 1A). The bandpass frequency response of the LBAW 100 is formed by two (or more) transverse standing wave resonance modes 410 and 420 arising within the structure. Transverse standing wave resonances can occur when plate waves are reflected from the edges of adjacent electrodes 401 and 402. In the even mode resonance 410, the piezo layers under both electrodes 150 and 170 vibrate in phase, while in the odd mode resonance 420, the phases are reversed. If the overall width of the structure is

number

number

[0038] 4B, even and odd modes 410, 420 are shown as transmission peaks as a function of input frequency f for an LBAW with Type 1 dispersion. For Type 1 dispersion, the even modes 410 are longer wavelength and lower frequency than the shorter wavelength odd modes 420. The frequency difference 430 between the modes determines the achievable bandwidth of the LBAW filter 100 and depends on the acoustic properties of the structure and the dimensions of the IDT resonators 102. The acoustic coupling strength can be defined in terms of the (resonant) frequency difference between the even (symmetric) and odd (asymmetric) resonators:

number

[0039] In some embodiments, increasing the number of extensions (e.g., 150a and 170a) of each electrode (e.g., 150 and 170) can increase the frequency difference between the even and odd modes of the LBAW, and therefore increase the bandwidth. This effect can be attributed to the fact that the transverse wavelength of the odd modes can depend on the periodicity (e.g., width W) of the electrode structure, while the even modes can depend on the total width of the structure (e.g., summing all widths W and gap G). For example, for a total number of electrode extensions K, electrode width W, and gap width G, the transverse acoustic wave length λ at the resonant frequency of the even wave mode is || teeth:

number

number

number

[0040] In some embodiments, the number K of electrode extensions is from 2 to 200, or from 10 to 60. In some embodiments, the length L of the electrode extensions can be from 50 μm to 2000 μm, or from 70 μm to 500 μm.

[0041] In some embodiments, the gap G is selected to allow coupling of the evanescent tails of standing waves formed under electrodes 150 and 170. For example, the gap G between the electrode extensions can be between 0.1 μm and 10 μm, or between 2 μm and 5 μm.

[0042] In some embodiments, the topology of electrodes 150 and 170 can be designed such that the gap width G provides sufficient coupling between the electrode extensions to create a single even mode 410 across the entire width of the structure. For example, the gap width G can be 2%-300%, or 10%-100%, of the decay length of the evanescent acoustic wave, i.e., the original amplitude A in the gap at the desired even resonant mode. o Amplitude of

number

[0043] In some embodiments, the gap width G can be defined in terms of the thickness d of the piezo layer. For example, G can be designed to be 10% to 300% of d, or 25% to 150% of d.

[0044] In some embodiments, the width of the electrode extension W can be between 0.1 μm and 30 μm, or between 2 μm and 5 μm. In some embodiments, W is equal to or smaller than the desired odd-mode resonant frequency λ odd Wavelength λ of transverse sound at || can be designed to obtain

[0045] In some embodiments, the electrode width W is designed such that no more than one half wavelength falls within the electrode width. For example, W may be less than the wavelength λ of the transverse acoustic wave in the desired odd resonant mode. || For example, λ || = λ odd It can be designed to be.

[0046] In some embodiments, the thicknesses of the various LBAW 100 components can be selected to achieve different acoustic properties and can be interdependent. For example, the thickness d (minimum and maximum) of the piezo layer 110 can be initially determined in terms of the acoustic wavelength (λ) of the piezo material at the operating frequency f. In some embodiments, the thicknesses (minimum and maximum) of the other LBAW 100 layers can be selected based on the selection of the piezo thickness d. For example, the combined thickness of the electrode (including the counter electrode 120) and the piezoelectric layer can be selected to be about half the wavelength of the mode being used, e.g., a longitudinal bulk wave in a thickness extensional mode. For the fundamental mode with N=1 (first mode, i.e., first harmonic), greater coupling is possible, but N>1 modes are also possible. For example, the thicknesses of the electrodes 150 and 170, the bottom electrode 120, and the reflecting structure 130 can be defined as a percentage of the piezo layer thickness d. In some embodiments, once all thicknesses are selected, the geometry of electrode extensions 150a and 170a, such as number K, width W, and length L, can be adjusted to match the electrical impedance of LBAW 100 to the system impedance. Without wishing to be bound by theory, impedance matching helps to avoid losses and reflections in the system.

[0047] In some embodiments, the thickness of electrodes 150 and 170 is between 1% and 30% of d, between 5% and 25% of d, or between 3% and 15% of d.

[0048] In some embodiments, the thickness of the bottom electrode 120 is between 5% and 50% of d, between 10% and 30% of d, or between 10% and 20% of d.

[0049] In some embodiments where the reflecting structure 130 is a Bragg reflector, the alternating layers of the reflector can be designed to provide the required reflectivity of the passband wavelengths. For example, the thickness of each layer can be adjusted to 1 / 2 the acoustic wavelength λ in the thickness direction to reflect the odd and even TE1 resonant modes. z In some embodiments, a single layer of the Bragg reflector can be between 15% and 80% of d, or between 20% and 70% of d.

[0050] The mass loading of the IDT 102, determined by the thickness and material of the electrodes 150 and 170, reduces the k || The frequency difference between the k = 0 frequency and the TS2 mode of the outer electrode region can be designed to be small. Without wishing to be bound by a particular theory, if the frequency difference between the TS2 mode of the outer region and the TE1 mode of the electrode region is small, the trapping range is large. More specifically, the k of the TS2 mode of the outer region can be designed to be small. || The k = 0 frequency can be 95% to 99% of the TE1 cutoff frequency in the electrode region. The k of the TS2 and TE1 modes in the outer region || =0 frequencies is designed to be, for example, 5% to 15%, for example 6.5% to 7.5%, larger than the TE1 mode cutoff frequency of the electrode region.

[0051] According to a particular embodiment of the present invention, the k of the TS2 mode in the outer region || = 0 frequency is 98% or more, or 98% to 99.5%, or 98.9% of the TE1 cutoff of the electrode area. Similarly, k of the electrode area TE1 and the outer area TS2 || =0 frequency:

number

[0052] FIG. 5 shows a curve of insertion loss IL versus frequency f for an exemplary LBAW 100. The curve shows two passbands with a peak 510 corresponding to a TE1 wave and a peak 520 corresponding to a TS2 wave. As explained above, the width of each passband is determined by the frequency difference between the even and odd modes of the respective type of wave. Here, the TS2 mode corresponds to sideband 520a (also referred to herein as the "TS2 passband"), and the TE1 mode corresponds to passband 510a (also referred to herein as the "TE1 passband"). In some embodiments, the LBAW 100 is designed to suppress the peak 520 corresponding to the TS2 mode while maintaining the characteristics of the peak 510 corresponding to the TE1 mode. Without wishing to be bound by theory, the TE1 mode operation can be selected because the piezo thin film material has a strong electromechanical coupling in the thickness direction. In other words, the TE1 longitudinal mode vibration is efficiently coupled to the electrical excitation through the thickness of the piezo layer 110.

[0053] In some embodiments, the LBAW 100 may be designed to have a passband for the TE1 mode between 0.5 and 10 GHz, or between 1 and 4 GHz. In some examples, the TE1 passband is 1.8 to 3.7 GHz. The passband limit may incorporate design considerations. For example, the dimensions of the device may be very large or small. If the dimensions are too large, it may take up too much space and be inefficient. If the dimensions are too small, it may result in poor performance due to thin and narrow electrodes, resistance and losses. In some embodiments, the LBAW 100 may be designed to have a TE1 passband width 510a of 0.5 to 15% relative to the center frequency, for example, 10%, 5%, 2%, or 1% relative to the center frequency. In some embodiments, the insertion loss in the passband is -7 dB to -0.5 dB or -5 dB to -1.5 dB.

[0054] Varying the width of the extensions or their spacing along their length can be utilized to reduce ripples in the LBAW passband, such as the ripples in the passband 510a shown in FIG. 5. These ripples are due in part to intermediate standing modes that form between the IDT electrodes. FIG. 6 is a schematic diagram of some of these modes, where the gray areas represent the electrode extensions and the curves represent the amplitudes of acoustic modes that form below the extensions in the LBAW piezo layer. The symmetric 602 and asymmetric 604 modes can form the frequency boundaries of the passband passband, while the intermediate mode 606 can contribute to ripples within the passband frequencies.

[0055] In contrast, by using extensions that have widths that vary along their length, as shown in Figure 7, the passband 710 is much flatter. Without wishing to be bound by theory, varying the extension width or the gap spacing between the extensions may cause these intermediate modes to occur at slightly different frequencies along the length of the extension, reducing the ripple effect in the passband.

[0056] The various embodiments discussed above can be implemented with tapered extensions (e.g., 150a and 170a), or tapered gap width G, or a combination of both. Referring to FIG. 8A, a section 800 of interdigitated electrodes 850 and 870 with tapered extensions (e.g., where the width changes from one end to the other) is shown. Electrodes 850 and 870 are physically separated from each other. Referring to FIG. 8B, a full set 820 of interdigitated (segmented) comb electrodes (e.g., 850 and 870) with tapered extensions is shown. Electrode 850 has a base 802 and an extension 806 extending from base 802, and electrode 870 has a base 804 and an extension 808 extending from base 804. The extensions from input electrode 850 extend in generally opposite directions along and are separated by gap width 810. The extension 806 has a smaller width 806a in a region distal to the base 802 and a larger width 806b in a region proximal to the base 802. Conversely, the extension 808 has a larger width 808a in a region distal to the base 804 and a smaller width 808b in a region proximal to the base 804.

[0057] In some embodiments, the width of the electrode extension varies monotonically, e.g., continuously, along the length of the extension between the proximal and distal regions. In some embodiments, the width varies in a stepped or non-linear manner (e.g., parabolic), or random manner (e.g., dithered, with increasing variation along the finger length). In some embodiments, the width increases or decreases along the length of the extension. In some embodiments, the width of the electrode may vary from 10 to 100% along its length. Thus, the wider end of the extension may be 10 to 250% wider than the narrower end, e.g., 10 to 25%, 25 to 50%, 50 to 75%, 75 to 100%, or 100% to 250% wider. In some embodiments, the electrode width may vary from 10 μm to 15 μm, 2 μm to 7 μm, 3 μm to 4 μm, or 3.75 μm to 4 μm, depending on the selected bandpass frequency range.

[0058] In some embodiments, the gap width remains constant between the electrodes from base 802 to base 804. In some embodiments, the gap width is between 0.6 μm and 3 μm, for example 1 μm.

[0059] Referring to FIG. 9A, a section 900 of interdigitated electrodes 950 and 970 with tapered gap widths is shown. The electrodes 950 and 970 are physically separated from one another. Referring to FIG. 9B, a full set 920 of interdigitated comb electrodes (e.g., 950 and 970) with tapered gap widths is shown. The electrode 950 has a base 902 and an extension 906 extending from the base 902, each having a width 906a. The electrode 970 has a base 904 and an extension 908 extending from the base 904, each having a width 908a. The extensions from the input electrode 950 extend in generally opposite directions along adjacent extensions of the output electrode 970 and are separated by a gap width. The gap width between adjacent extensions of the input 950 and output 970 electrodes varies between the base 902 of the input electrode and the base 904 of the output electrode. Adjacent extensions from electrodes 950 and 970 are separated by a larger gap width 910 a in a region proximal to base 902 and a smaller gap width 910 b in a region distal to base 902 .

[0060] For sector electrodes with tapered gap widths, the gap width increases or decreases along the length of the gap from one base to the other. In some embodiments, the gap varies continuously between the bases of the two electrodes (e.g., 902 and 904), for example when the electrode extensions 906, 908 have uniform widths along their lengths. In some embodiments, the gap width varies in steps. In some embodiments, the gap width can vary from 0.5 μm to 3 μm, e.g., 2 μm to 3 μm, 1.25 μm to 1.75 μm, or 0.75 μm to 2 μm, depending on the bandpass frequency range selected.

[0061] In some embodiments, the width of the extension remains constant along its length, hi some embodiments, the width of the extension is between 1 μm and 30 μm, for example 4 μm.

[0062] In embodiments where both the gap width G and the extension width W vary, the G / W ratio is between 0.25-5.

[0063] Varying the width of the extensions or the width of the gaps between extensions along the length of the gaps or extensions, respectively, can be used to select the acoustic properties of the LBAW filter (e.g., change the acoustic properties relative to a filter having both extensions and gaps of uniform width along their lengths). For example, the interdigitated electrodes can be designed to suppress the TS2 sideband, or other parasitic TS1 sidebands that do not fall into the desired LBAW passband frequencies.

[0064] Without wishing to be bound by any particular theory, the use of interdigitated electrodes with different extension widths or spacing along their length may be acoustically similar to having a series of parallel-connected LBAW filters with constant but different extension widths and spacings. This theoretical series of different LBAW filters can therefore be selected to have TS2 sidebands at different frequencies. Thus, when the LBAW filters are connected in parallel (simulating the effect of a sector-shaped LBAW electrode), the TS2 sidebands average (or "smear out") between these frequencies, and the TS2 band is generally suppressed. The effect on the TE1 passband is weaker because it is broadband and less affected by the geometry of the electrodes. A similar method can be used to suppress the parasitic TS1 band. In some embodiments, TS2 sideband suppression is preferred over TS1 sideband suppression, since the TS2 sidebands generally introduce higher insertion loss at frequencies close to the passband of the LBAW filter.

[0065] The proof of concept is shown in Figures 10A and 10B. Figure 10A is a plot of insertion loss versus frequency measured for two separate LBAW filters. Curve 1002 corresponds to an LBAW filter with regular interdigitated electrodes having a constant gap width of 1.5 μm and a constant extension width of 3 μm. Curve 1004 corresponds to an LBAW filter with regular interdigitated electrodes having a constant gap width of 1.5 μm and a constant extension width of 4 μm. The TS2 peak 1002a and the parasitic TS1 peak 1002b of curve 1002 are located at different frequencies than the TS2 peak 1004a and the parasitic TS1 peak 1004b of curve 1004.

[0066] FIG. 10B shows the frequency vs. insertion loss curves obtained from the combination of the two filters of FIG. 10A. Curve 1006 is synchronized from the measured response of the filter of FIG. 10A by cascading the filters such that the output of the first filter is connected to the input of the second filter. Curve 1008 is the measured response from a filter of a similar shape (e.g., sector-shaped) to that shown in FIG. 8B with a constant gap width of 1.5 μm and extension widths that vary from 3 μm to 4 μm along their length. The similarity of curves 1006 and 1008 indicates that the series LBAW filter performs similarly to a sector-shaped IDT LBAW filter of various extension widths. Furthermore, as shown in FIG. 10A, the TS2 peak 1010 and parasitic TS1 peak 1012 of both curves are significantly lower than when using the LBAW filters separately. In summary, the sector-shaped LBAW electrodes are effective in sideband suppression.

[0067] FIG. 11 shows additional experimental confirmation that sidebands can be suppressed using sector-shaped electrode extensions of various (tapered) widths in a plot of insertion loss versus frequency. Measured curve 1104 corresponds to a conventional LBAW filter with interdigitated electrodes having a constant gap width of 1.5 μm and a constant extension width of 4 μm. Measured curve 1002 for sector-shaped LBAW electrodes corresponds to interdigitated sector electrodes with a constant gap width of 1.5 μm and extension widths that vary from 3 μm to 4 μm. Curve 1002 has a lower insertion loss at the TS2 peak 1106 and the parasitic TS1 side peak 1108. The passband width and insertion loss can be optimized by conventional methods.

[0068] The effect of different electrode shapes can be predicted by simulation. In some embodiments, a continuously varying electrode width can be approximated by dividing the electrode into multiple sections (e.g., 3 sections) and summing the results.

[0069] Simulated insertion loss versus frequency curves are shown for three regular LBAW electrode shapes and a sector-shaped LBAW electrode shape. Curves 1202, 1204, and 1206 correspond to a regular LBAW shape with a gap width of 2 μm and 10 extensions, with constant extension widths of 4 μm, 4.5 μm, and 5 μm, respectively. Curve 1208 corresponds to an LBAW with a gap width of 2 μm between extensions, with sector-shaped IDT electrodes with extension widths that vary from 4 μm to 5 μm along the length of the extension. Curve 1208 is formed by combining the other three curves to roughly simulate a tapered LBAW structure. As expected, the lower frequency region 1208a of curve 1208 shows reduced ripple in the passband between 1920 MHz and 1940 MHz. [Explanation of symbols]

[0070] 100 LBAW resonators, LBAW filters 102 IDT electrode structure 110 Piezoelectric layer 120 Electrode, Counter Electrode 130 Acoustic Bragg Reflector 140 Substrate 150 Extension 160 input ports 170 Electrode, Output Electrode 180 output ports 190 Gap 200 Length Extension (TE) Bulk Mode 210 Particle Displacement 220 Length shear (TS2) bulk mode 230 Particle Displacement 401 Electrode 402 Electrode 410 Resonant Mode, Even Mode 420 Odd mode resonance 420 Resonant Mode, Odd Mode 430 Frequency Difference 510 Peak corresponding to TE1 wave 520 Peak corresponding to TS2 wave 602 Symmetric Mode 604 Asymmetric Mode 606 Intermediate Mode 710 Passband 800 Section 802 base 804 base 806 Extension 808 Extension 810 Gap width 820 Full Set 850 electrode 870 electrode 900 Section 902 Base 904 Base 906 Extension 908 Extension 910 Gap width 920 Full Set 950 Input electrode 970 Output electrode 1002 curve 1004 curve 1006 curve 1008 curve 1010 TS2 Peak 1012 TS1 Peak 1104 Curve 1106 TS2 Peak 1108 Parasitic TS1 side peak 1202 Curve 1204 Curve 1208 Curve

Claims

1. A piezoelectric layer; input and output electrodes disposed on a top surface of the piezoelectric layer and physically spaced apart from one another, the input and output electrodes each including a base and at least one extension extending from the base, the at least one extension of the input electrode extending in generally opposite directions along adjacent ones of the at least one extension of the output electrode and spaced apart from adjacent ones of the at least one extension of the output electrode by a gap width; and a counter electrode having a top surface connected to a bottom surface of the piezoelectric layer; An acoustic wave filter device comprising: a thickness of the piezoelectric layer and a gap width between adjacent extensions of the input and output electrodes such that application of a high frequency voltage between the input electrode and the counter electrode generates symmetric and asymmetric acoustic resonance modes in the piezoelectric layer; i) at least one extension of the input or output electrodes has a width that varies randomly from a first end to a second end of the at least one extension; or ii) at least one extension of the input electrode or the output electrode has a width that varies nonlinearly from the first end to the second end of the at least one extension, and the gap width between adjacent extensions is 25% to 150% of the thickness of the piezoelectric layer.

2. 2. The acoustic wave filter device of claim 1, wherein the first end of the extension is proximal to the base of the corresponding electrode and the second end of the extension is distal to the base of the corresponding electrode.

3. 3. The acoustic wave filter device of claim 2, wherein the at least one extension of either the input electrode or the output electrode has a width that decreases from the first end to the second end of the at least one extension.

4. 4. The acoustic wave filter device of claim 3, wherein the at least one extension of the other of the input electrode or the output electrode has a width that increases from the first end to the second end of the at least one extension.

5. 5. The acoustic wave filter device of claim 4, wherein the at least one extension of the input electrode increases in width from the first end to the second end of the extension of the input electrode, and the at least one extension of the output electrode decreases in width from the first end to the second end of the extension of the output electrode.

6. 3. The acoustic wave filter device of claim 2, wherein the at least one extension of either the input electrode or the output electrode has a width that increases from the first end to the second end of the at least one extension.

7. 2. The acoustic wave filter device of claim 1, wherein the gap width between the extensions is constant.

8. 2. The acoustic wave filter device of claim 1, wherein the width of the at least one extension of the input electrode or the output electrode varies continuously from the first end to the second end of the at least one extension, or the width varies in a dithered manner from the first end to the second end of the at least one extension, with the variation increasing.

9. 2. The acoustic wave filter device of claim 1, wherein the input electrode and the output electrode each have a comb structure with interdigitated extensions.

10. 2. The acoustic wave filter device of claim 1, further comprising a Bragg reflector in contact with a bottom surface of the counter electrode.

11. 2. An acoustic wave filter device according to claim 1, characterized in that the input electrode and the output electrode have at least ten extensions.

12. 2. An acoustic wave filter device according to claim 1, characterized in that the thickness of said piezoelectric layer is at least 0.5 μm.

13. 2. The acoustic wave filter device of claim 1, wherein the width of each extension increases from the first end to the second end by between 10% and 250%.

14. 2. The acoustic wave filter device of claim 1, wherein the width of each extension increases from the first end to the second end by between 25% and 100%.

15. 10. The acoustic wave filter device of claim 1, wherein the acoustic wave filter device is a laterally coupled bulk acoustic wave filter.

Images