Transversely excited acoustic film resonator with lambda-half dielectric layer

Abstract

Acoustic resonator device, comprising: a substrate (120) with a surface; a piezoelectric plate (310, 510) with a front and a back surface, wherein the back surface is attached to the surface of the substrate (120), except for a part of the piezoelectric plate (310, 510) which forms a membrane (115) spanning a cavity (140) in the substrate (120); an interdigital converter, IDT, (130) on the front surface of the piezoelectric plate (310, 510), such that nested fingers (338, 538) of the IDT (130) are arranged on the membrane (115), wherein the piezoelectric plate (310, 510) and the IDT (130) are configured such that a high-frequency signal applied to the IDT (130) excites a primary acoustic shear mode in the membrane (115); and a dielectric layer (350, 550) arranged either on the front surface or on the back surface of the piezoelectric plate (310, 510), where a thickness td of the dielectric layer (350, 550) is defined as: 0.85 λ 0,d ≤ 2 td ≤ 1.15 λ 0,d , and where λ 0,d a wavelength of the fundamental resonance of the acoustic volume shear wave in the dielectric layer (350, 550).

Description

BACKGROUND area

[0001] This disclosure relates to high-frequency filters that use acoustic wave resonators, and in particular to filters for use in communication equipment. Description of the related prior art

[0002] A high-frequency filter (HF filter) is a two-port device configured to pass some frequencies and block others, where "pass" means transmission with relatively little signal loss and "block" means blocking or significant attenuation. The range of frequencies passed by a filter is called the filter's "passband." The range of frequencies blocked by such a filter is called the filter's "stopband." A typical RF filter has at least one passband and at least one stopband. Specific requirements for a passband or stopband depend on the specific application. For example, a "passband" might be defined as a frequency range in which the insertion loss of a filter is better than a defined value such as 1 dB, 2 dB, or 3 dB.A "stopband" can be defined as a frequency range in which the suppression of a filter is greater than a defined value such as 20 dB, 30 dB, 40 dB or more, depending on the application.

[0003] RF filters are used in communication systems where information is transmitted wirelessly. Examples include the RF front ends of cellular base stations, mobile phones and computers, satellite transceivers and ground stations, IoT (Internet of Things) devices, laptops and tablets, fixed-point radio links, and other communication systems. RF filters are also used in radar and electronic warfare systems.

[0004] RF filters typically require many design compromises to achieve the best balance between performance parameters such as insertion loss, suppression, isolation, power handling, linearity, size, and cost for each specific application. Specific design and manufacturing methods and improvements can simultaneously benefit one or more of these requirements.

[0005] Improvements to the RF filters in a wireless system can have a broad impact on system performance. RF filter enhancements can be used to provide system performance improvements such as larger cells, longer battery life, higher data rates, greater network capacity, lower costs, improved security, and higher reliability. These improvements can be implemented at many levels of the wireless system, both individually and in combination, such as the RF module, RF transceiver, mobile or fixed subsystem, or network level.

[0006] The desire for greater bandwidth for communication channels will inevitably lead to the use of higher frequency bands. The current LTE™ (Long Term Evolution) specification defines frequency bands from 3.3 GHz to 5.9 GHz. Some of these bands are currently unused. Future proposals for wireless communications include millimeter-wave communication bands with frequencies up to 28 GHz.

[0007] High-performance RF filters for current communication systems typically incorporate acoustic wave resonators, including surface acoustic wave resonators (SAW resonators), bulk acoustic wave resonators (BAW resonators), film bulk acoustic wave resonators (FBAR resonators), and other types of acoustic resonators. However, these existing technologies are not well-suited for use at the higher frequencies proposed for future communication networks.

[0008] WO 2016 / 147687A1 discloses a resonator comprising a support substrate with a recess in its top surface, a thin film which may consist of a dielectric and which is arranged on the support substrate, a piezoelectric substrate which is arranged on the thin film, and an IDT electrode which is provided on the piezoelectric substrate. DESCRIPTION OF THE DRAWINGS Fig. Figure 1 contains a schematic top view and two schematic cross-sectional views of a transversely excited acoustic film volume resonator (XBAR). Fig. Figure 2 is an extended schematic cross-sectional view of a part of the XBAR. Fig. 1. Fig. Figure 3 is an extended schematic cross-sectional view of part of an improved XBAR with a "lambda-half" dielectric layer. Fig. Figure 4 is a diagram comparing the admittances of an XBAR with a lambda-half dielectric layer and a conventional XBAR. Fig. Figure 5 is a cross-sectional view of an XBAR with a half-wave dielectric layer with contours representing the voltage at the resonant frequency. Fig. Figure 6 is a diagram comparing the admittances of three XBAR with lambda-half AlN layers. Fig. Figure 7 is a diagram comparing the admittances of three XBAR with lambda-half SiO2 layers. Fig. Figure 8 is a diagram comparing the admittances of three other XBARs with lambda-half SiO2 layers. Fig. Figure 9 is a diagram of the admittance of an XBAR with an excessively thin lambda-half SiO2 layer. Fig. Figure 10 is a diagram of the admittance of an XBAR with an excessively thick lambda-half SiO2 layer. Fig. 11 is a diagram of the temperature coefficient of the frequency of an XBAR as a function of the SiO2 thickness. Fig. Figure 12 is a schematic circuit diagram and the layout of a filter using XBAR. Fig. Figure 13 is a flowchart of a process for manufacturing an XBAR with a lambda-half dielectric layer.

[0009] In this description, elements appearing in drawings are assigned three- or four-digit reference identifiers, where the two least significant digits are specific to the element and the one or two most significant digits are the drawing number in which the element is first introduced. For an element not described in connection with a drawing, it can be assumed to have the same characteristics and function as a previously described element with the same reference identifier. DETAILED DESCRIPTION Device description

[0010] Fig. Figure 1 shows a simplified schematic top view and orthogonal cross-sectional views of a transversely excited acoustic film volume resonator (XBAR) 100, as described in US 10,491,192 B1. XBAR resonators such as the Resonator 100 can be used in a variety of RF filters, including bandstop filters, bandpass filters, duplexers, and multiplexers. XBARs are particularly well-suited for use in filters for communication bands with frequencies above 3 GHz.

[0011] The XBAR 100 consists of a thin-film conductor structure formed on the surface of a piezoelectric plate 110 with parallel front and back surfaces 112 and 114, respectively. The piezoelectric plate is a thin single-crystal layer of a piezoelectric material such as lithium niobate, lithium tantalate, lanthanum gallium silicate, gallium nitride, or aluminum nitride. The piezoelectric plate is cut such that the orientation of the X, Y, and Z crystal axes with respect to the front and back surfaces is known and consistent. In the examples presented in this patent, the piezoelectric plates are Z-cut, i.e., the Z-axis is perpendicular to the front and back surfaces 112 and 114. However, XBARs can be fabricated on piezoelectric plates with other crystallographic orientations.

[0012] The rear surface 114 of the piezoelectric plate 110 is attached to a surface of the substrate 120, except for a region of the piezoelectric plate 110 that forms a membrane 115 spanning a cavity 140 formed in the substrate. The region of the piezoelectric plate spanning the cavity is referred to here as the “membrane” 115 due to its physical similarity to the diaphragm of a microphone. As in Fig. As shown in Figure 1, the membrane 115 borders the rest of the piezoelectric plate 110 around the entire circumference 145 of the cavity 140. In this context, “bordering” means “continuously connected without an intervening element”.

[0013] The substrate 120 provides mechanical support for the piezoelectric plate 110. The substrate 120 can be made of, for example, silicon, sapphire, quartz, or another material or a combination of materials. The back surface 114 of the piezoelectric plate 110 can be bonded to the substrate 120 using a wafer bonding process. Alternatively, the piezoelectric plate 110 can be grown on the substrate 120 or attached to the substrate in another way. The piezoelectric plate 110 can be attached directly to the substrate or via one or more intermediate material layers on the substrate 120.

[0014] “Cavity” has the conventional meaning of “an empty space within a solid body.” The cavity 140 can be a hole that passes completely through the substrate 120 (as shown in Section AA and Section BB), or a depression in the substrate 120. The cavity 140 can be formed, for example, by selectively etching the substrate 120 before or after attaching the piezoelectric plate 110 and the substrate 120.

[0015] The conductor structure of the XBAR 100 incorporates an interdigital converter (IDT) 130. The IDT 130 comprises a first plurality of parallel fingers, such as finger 136, extending from a first bus bar 132, and a second plurality of fingers extending from a second bus bar 134. The first and second plurality of parallel fingers are nested within each other. The nested fingers overlap over a distance AP, commonly referred to as the "aperture" of the IDT. The center-to-center distance L between the outermost fingers of the IDT 130 is the "length" of the IDT.

[0016] The first and second busbars 132 and 134 serve as terminals for the XBAR 100. A radio frequency or microwave signal applied between the two busbars 132 and 134 of the IDT 130 excites a primary acoustic mode within the piezoelectric plate 110. The primary acoustic mode is a volume shear mode in which acoustic energy propagates along a direction that is essentially orthogonal to the surface of the piezoelectric plate 110 and also perpendicular to the direction of the electric field generated by the IDT fingers. Therefore, the XBAR is considered a transversely excited film volume wave resonator.

[0017] The IDT 130 is positioned on the piezoelectric plate 110 such that at least the fingers of the IDT 130 are located on the part 115 of the piezoelectric plate that spans or is suspended above the cavity 140. As shown in Fig. As shown in Figure 1, the cavity 140 has a rectangular shape with an extent larger than the aperture AP and length L of the IDT 130. An XBAR cavity can have a different shape, such as a regular or irregular polygon. The XBAR cavity can have more or fewer than four sides, which can be straight or curved.

[0018] To simplify the presentation in Fig. Figure 1 shows that the geometric pitch and width of the IDT fingers are greatly exaggerated in relation to the length (dimension L) and aperture (dimension AP) of the XBAR. A typical XBAR has more than ten parallel fingers in the IDT 110. An XBAR can have hundreds, possibly thousands, of parallel fingers in the IDT 110. Similarly, the thicknesses of the IDT fingers and the piezoelectric plate are greatly exaggerated in the cross-sectional views.

[0019] Fig. Figure 2 shows a detailed schematic cross-sectional view of an XBAR 200, which is the XBAR 100 from Fig. 1. The piezoelectric plate 210 is a single-crystal layer of piezoelectric material with a front surface 214 and a back surface 216. The thickness ts between the front surface 214 and the back surface 216 can be, for example, 100 nm to 1500 nm. When used in filters for 5G NR (Fifth Generation New Radio) and Wi-Fi™ bands from 3.3 GHz to 6 GHz, the thickness ts can be, for example, 280 nm to 550 nm.

[0020] The IDT fingers 238 can be made of aluminum, an alloy essentially of aluminum, copper, an alloy essentially of copper, tungsten, molybdenum, beryllium, gold, or another conductive material. Thin layers (relative to the overall thickness of the conductors) of other metals, such as chromium or titanium, can be formed under and / or over the fingers to improve adhesion between the fingers and the piezoelectric plate 210 and / or to passivate or encapsulate the fingers. The busbars (132, 134 in Fig. 1) The IDT can be made of the same or different materials as the fingers.

[0021] Dimension p is the center-to-center distance or "pitch" of the IDT fingers, which can be referred to as the pitch of the IDT and / or the pitch of the XBAR. Dimension w is the width or "mark" of the IDT fingers. The IDT of an XBAR differs significantly from the IDTs used in surface acoustic wave resonators (SAW resonators). In a SAW resonator, the pitch of the IDT is half the acoustic wavelength at the resonant frequency. Furthermore, the mark-to-pitch ratio of a SAW resonator IDT is typically close to 0.5 (i.e., the mark or finger width is about one-quarter of the acoustic wavelength at the resonant frequency). In an XBAR, the pitch p of the IDT is typically 2 to 20 times the width w of the fingers. Furthermore, the pitch p of the IDT is typically 2 to 20 times the thickness ts of the piezoelectric plate 210.The width of the IDT fingers in an XBAR is not limited to a quarter of the acoustic wavelength at resonance. For example, the width of the XBAR IDT fingers can be 500 nm or more, allowing the IDT to be fabricated by optical lithography. The thickness tm of the IDT fingers can range from 100 nm to approximately equal to the width w. The thickness of the busbars (132, 134 in . Fig. 1) The IDT can be equal to or greater than the thickness tm of the IDT fingers.

[0022] A bulk acoustic wave (BAW) propagating perpendicular to the surfaces of a piezoelectric plate is reflected by the surfaces and resonates, forming a standing wave, when the thickness ts of the piezoelectric plate is an integer multiple of half the wavelength λ of the acoustic wave. The longest wavelength / lowest frequency at which such resonance occurs is the fundamental resonance of the shear BAW at a frequency f0 and a wavelength λ. 0,s , which corresponds to twice the thickness ts of the piezoelectric plate. The term “λ 0,s“ denotes the wavelength of the fundamental resonance (0th order) of the shear wave in the piezoelectric plate. The wavelength of the same acoustic wave (i.e., a shear wave propagating in the same direction with the same frequency) can differ in other materials. The frequency f0 can be determined by dividing the velocity of the shear wave in the piezoelectric plate by the wavelength λ.” 0,s The fundamental resonance of the shear-BAW of the piezoelectric plate is not the same as the resonance of the XBAR device 200, which is influenced by the IDT setup.

[0023] Fig. Figure 3 shows a detailed schematic cross-sectional view of an XBAR containing a “half-lambda” dielectric layer. Fig. Figure 3 shows in particular an XBAR 300 with a thick dielectric layer 350 on the front side (i.e. the side facing away from the substrate; the top side, as in Fig. 3 shown) of a piezoelectric plate 310. A comparable dielectric layer on the back 316 of the piezoelectric plate 310 could be used instead of the dielectric layer 350. On a larger scale, the XBAR 300 with the thick dielectric layer 350 resembles the XBAR 100 of Fig. 1. Fig. Figure 3 also shows two IDT fingers 338, as described previously. The dimension p is the center-to-center distance or “pitch” of the IDT fingers, and the dimension w is the width or “mark” of the IDT fingers.

[0024] The piezoelectric plate 310 is a thin single-crystal layer made of a piezoelectric material such as lithium niobate or lithium tantalate. The piezoelectric plate 310 is cut such that the orientation of the X, Y, and Z crystal axes with respect to the front and back surfaces 314, 316 is known and consistent. The thickness ts of the piezoelectric plate 310 can be, for example, 100 nm to 1500 nm.

[0025] The dielectric layer 350 can consist of virtually any dielectric material, such as SiO2, Si3N4, Al2O3, AlN, and other dielectric materials. As will be discussed later, particular advantages can arise when the dielectric material is or contains AlN, and when the dielectric material is SiO2.

[0026] The thickness ts of the piezoelectric plate 310 and the thickness td of the dielectric layer 350 are configured such that a shear-type oscillator propagating perpendicular to the surfaces 316 and 352 forms a standing wave with one full cycle at a predetermined frequency between the surfaces 316 and 352, which may be slightly lower than the desired resonant frequency of the XBAR device 300. In other words, the second overtone resonance of the shear-type oscillator occurs at the predetermined frequency. By definition, the thickness ts of the piezoelectric plate is half of λ. 0,s, which, as previously described, is the wavelength of the fundamental resonance of the shear-type BAW of the piezoelectric plate 310 without the dielectric layer 350. Nominally, the thickness td of the dielectric layer 350 is half of λ. 0,d , where λ 0,d The wavelength of the same volume BAW in the dielectric layer 350 is [missing information]. In this case, both the piezoelectric plate 310 and the dielectric layer 350 contain a half-cycle standing wave at frequency f0, which is now the frequency of the second overtone resonance. λ 0,d is equal to λ 0,s times the ratio of the speed of the acoustic shear wave in the dielectric layer 350 to the speed of the acoustic shear wave in the piezoelectric plate 310. For a relatively slow dielectric material such as SiO2, λ 0,d equal to or slightly greater than λ 0,sIn this case, the thickness td of the dielectric layer 350 can be equal to or slightly greater than ts. For a relatively fast dielectric material, such as Si3N4 or AlN, λ 0,d significantly larger than λ 0,s In this case, the thickness td of the dielectric layer 350 will be proportionally larger than ts.

[0027] While the dielectric layer 350 is referred to here as the "half-lambda" dielectric layer, the thickness td of the dielectric layer does not have to be exactly λ 0,d / 2. The thickness td can be determined by λ. 0,d / 2 deviate as long as the combined thicknesses of the piezoelectric plate 310 and the dielectric layer 350 are such that the second overtone resonance of the volume shear wave occurs at the predetermined frequency. Simulation results, some of which are discussed below, show that the thickness of the dielectric layer in a range defined by 0.85 λ0,d≤2 td≤1.15 λ0,d This results in an XBAR with low interference modes and uniform electromechanical coupling. Values ​​of td outside this range lead to lower electromechanical coupling and increased interference modes. Varying td within this range allows for tuning the resonant frequency of an XBAR by approximately 10%, which is sufficient for many filter applications to achieve the required frequency offset between shunt and series resonators.

[0028] In Fig. In 3, the dielectric layer 350 is deposited over and between the IDT fingers 338. In other embodiments, a half-wave dielectric layer can be formed only between the IDT fingers. The half-wave dielectric layer 350 can be a single layer or two or more layers of different dielectric materials with similar acoustic impedances.

[0029] A key advantage of incorporating the Lambda-Half Dielectric Layer 350 into the XBAR 300 is the increased diaphragm thickness. Depending on the materials used in the Lambda-Half Dielectric Layer 350, the diaphragm thickness of the XBAR 300 can be two to three times greater than the diaphragm thickness of the XBAR 100. Fig. 1. A thicker membrane is stiffer and less likely to bend or deform due to temperature fluctuations.

[0030] The thicker diaphragm of the XBAR 300 also has higher thermal conductivity, especially if the half-wave dielectric layer 350 is made of or contains a dielectric material with high thermal conductivity, such as aluminum nitride. Higher thermal conductivity results in more efficient heat dissipation from the diaphragm, which can allow the use of a smaller resonator area for a given thermal load or power dissipation.

[0031] The XBAR 300 also has a higher capacity per unit area compared to the XBAR 100. Fig. 1 (at the same IDT pitch). Resonator capacitance is a problem in circuit design. Especially in RF filters that use acoustic resonators, the input and output impedance of the filter must correspond to a defined value (usually 50 ohms). This requirement dictates minimum capacitance values ​​for some or all resonators in a filter. The higher capacitance per unit area of ​​the XBAR 300 with partial Bragg reflectors allows the use of a smaller resonator area for any required capacitance value.

[0032] An XBAR with a lambda-half dielectric layer on the back of the piezoelectric plate 310 (not shown) will have improved stiffness and thermal conductivity, but only a slightly higher capacitance per unit area.

[0033] Fig. Figure 4 is a diagram 400 comparing the admittance of an XBAR with that of a half-wave dielectric layer and a conventional XBAR. The in Fig. 4 and the following Fig. 6 and Fig. The data presented in Figure 7 are results of simulations of the XBAR devices using the finite element method. The solid line 410 is a curve of the admittance magnitude as a function of frequency for an XBAR with a half-wave dielectric layer. The piezoelectric plate is made of lithium niobate with a thickness of 400 nm. The IDT is made of aluminum with a thickness of 100 nm. The pitch and mark of the IDT fingers are 4.25 µm and 1.275 µm, respectively. The half-wave dielectric layer consists of a 350 nm thick layer of Si3N4 and a 350 nm thick layer of AlN. The resonant frequency is 4.607 GHz and the antiresonant frequency is 4.862 GHz. The difference between the antiresonance and resonance frequencies is 255 MHz, or about 5.4% of the average of the resonance and antiresonance frequencies.

[0034] The dashed line 420 is a curve of the magnitude of the admittance as a function of frequency for a conventional XBAR. The piezoelectric plate is made of lithium niobate with a thickness of 400 nm. The IDT is made of aluminum with a thickness of 100 nm. The pitch and mark of the IDT fingers are 3.7 µm and 0.47 µm, respectively. The resonant frequency is 4.71 GHz and the antiresonant frequency is 5.32 GHz. The difference between the antiresonant and resonant frequencies is 610 MHz, or about 12.2% of the average of the resonant and antiresonant frequencies. The admittance of the conventional XBAR (dashed line 420) exhibits some interfering modes between the resonant and antiresonant frequencies, which are not present in the admittance of the XBAR with the half-wave dielectric layer (solid line 410).

[0035] The incorporation of a half-wave dielectric layer into the XBAR 300 device results in a stiffer diaphragm with higher thermal conductivity and potentially lower excitation of unwanted modes compared to a conventional XBAR device. These advantages come at the cost of reduced electromechanical coupling and a correspondingly smaller difference between the resonant and antiresonant frequencies.

[0036] Fig. Figure 5 is a cross-sectional view of an XBAR 500 with a half-wave dielectric layer, the contours of which represent the voltage at the resonant frequency. The piezoelectric plate 510 is made of lithium niobate with a thickness of 400 nm. The IDT fingers 538 are made of aluminum with a thickness of 100 nm. The half-wave dielectric layer 550 consists of a 350 nm thick layer 552 of Si3N4 and a 350 nm thick layer 554 of AlN.

[0037] The voltage in the XBAR 500 at the resonant frequency illustrates a standing wave over a complete cycle between the surfaces of the device. The voltage is highest near the midpoint of the thickness of the piezoelectric plate 510 and near the midpoint of the dielectric layer 550, corresponding to the peak values ​​of the two half-cycles of the standing wave. The voltage is lowest at the surfaces of the device and near the boundary between the piezoelectric plate 510 and near the midpoint of the dielectric layer 550, corresponding to the zero crossings of the standing wave.

[0038] Fig. Figure 6 is a diagram 600 showing the use of pitch and dielectric thickness to tune the resonant and antiresonant frequencies of an XBAR with a half-wave dielectric layer. The solid line 610 is a curve of the admittance magnitude as a function of frequency for an XBAR with a pitch and mark of 4.25 µm and 1.275 µm, respectively. The piezoelectric plate is made of lithium niobate with a thickness of 400 nm. The IDT is made of aluminum with a thickness of 100 nm. The half-wave dielectric layer is a 700 nm thick layer of Si3N4. The resonant frequency is 4.513 GHz and the antiresonant frequency is 4.749 GHz. The difference between the antiresonance and resonance frequencies is 236 MHz, or about 5.1% of the average of the resonance and antiresonance frequencies.

[0039] The dotted line 620 is a curve of the admittance magnitude as a function of frequency for a similar XBAR with the same setup, except that the pitch and mark of the IDT fingers are 3.75 µm and 1.31 µm, respectively. The resonant frequency is 4.557 GHz and the antiresonant frequency is 4.795 GHz. Changing the IDT pitch from 4.25 µm to 3.75 µm increases the resonant and antiresonant frequencies by approximately 45 MHz. Varying the pitch over a range of 3 µm to 5 µm yields a tuning range of approximately 200 MHz.

[0040] The dashed line 630 is a curve of the admittance magnitude as a function of frequency for a similar XBAR. The pitch and mark of the IDT fingers are 4.25 µm and 1.275 µm, respectively, and the dielectric layer comprises a 700 nm thick Si3N4 layer and a 50 nm thick SiO2 layer. The resonant frequency is 4.400 GHz and the antiresonant frequency is 4.626 GHz. Adding the 50 nm thick "tuning layer" reduces the resonant and antiresonant frequencies by approximately 110 MHz.

[0041] Fig. Figure 7 is another diagram 700 showing the use of pitch and dielectric thickness to tune the resonant and antiresonant frequencies of an XBAR with a half-wave dielectric layer. The solid line 710 is a curve of the admittance magnitude as a function of frequency for an XBAR with a pitch and mark of 4.25 µm and 1.275 µm, respectively. The piezoelectric plate is made of lithium niobate with a thickness of 400 nm. The IDT is made of aluminum with a thickness of 100 nm. The half-wave dielectric layer consists of a 400 nm thick layer of SiO2. The resonant frequency is 4.705 GHz and the antiresonant frequency is 5.108 GHz. The difference between the antiresonance and the resonance frequency is 403 MHz, which is approximately 8.2% of the average of the resonance and antiresonance frequencies.

[0042] The dotted line 720 is a curve of the admittance magnitude as a function of frequency for a similar XBAR with the same setup, except that the pitch and mark of the IDT fingers are 3.75 µm and 1.31 µm, respectively. The resonant frequency is 4.740 GHz and the antiresonant frequency is 5.137 GHz. Changing the IDT pitch from 4.25 µm to 3.75 µm increases the resonant and antiresonant frequencies by approximately 35 MHz. Varying the pitch over a range of 3 µm to 5 µm yields a tuning range of approximately 100 MHz.

[0043] The dashed line 730 is an admittance curve as a function of frequency for a similar XBAR with the same setup, except that the pitch and mark of the IDT fingers are 4.25 µm and 1.275 µm, respectively, and the dielectric layer is 450 nm thick SiO₂. The resonant frequency is 4.512 GHz and the antiresonant frequency is 4.905 GHz. The difference between the antiresonant and resonant frequencies is 393 MHz, or about 8.3% of the average of the resonant and antiresonant frequencies. Increasing the thickness of the dielectric layer by 50 nm reduces the resonant and antiresonant frequencies by about 190 MHz without reducing the electromechanical coupling.

[0044] Fig. Figure 8 is another diagram 800 showing the use of the dielectric layer thickness to tune the resonant and antiresonant frequencies of an XBAR with a half-wave dielectric layer. The dashed line 810 is a curve of the magnitude of the admittance as a function of frequency for an XBAR with a pitch of 4.25 µm and a mark of 1.275 µm, respectively. The piezoelectric plate is made of lithium niobate with a thickness of 400 nm. The IDT is made of aluminum with a thickness of 100 nm. The half-wave dielectric layer consists of a 425 nm thick layer of SiO2. This example represents the case where the thickness td of the dielectric layer is equal to λ. 0,d / 2 is.

[0045] The solid line 820 is a curve of the magnitude of the admittance as a function of frequency for a similar XBAR with the same construction, except that the dielectric layer is made of 375 nm thick SiO2. In this case, td = 0.88 (λ 0,d / 2). The dashed line 830 is a curve of the magnitude of the admittance as a function of frequency for a similar XBAR with the same construction, except that the dielectric layer is made of 475 nm thick SiO2. In this case, td = 1.12 (λ 0,d / 2). By changing the SiO2 from 375 nm to 475 nm, the resonance and antiresonance frequencies shift by about 400 MHz, while maintaining electromechanical coupling and preventing unwanted interference modes.

[0046] Assuming a 400 nm thick piezoelectric lithium niobate plate, the range for td given in equation (1) corresponds to approximately 350 nm to 500 nm. This range can be expressed in terms of the thickness ts of the piezoelectric plate as follows: 0.875 ts≤td≤1.25 ts.

[0047] It is assumed that this range applies to any thickness of the piezoelectric lithium niobate plate.

[0048] Fig. Figure 9 is a diagram 900 illustrating the effect of an excessively thin "half-lambda" dielectric layer. The solid line 910 is a curve of the magnitude of admittance as a function of frequency for an XBAR with the same setup as the devices of Fig. 8, where the thickness of the SiO2 dielectric layer is reduced to 325 nm. In this case, td = 0.76 (λ 0,d / 2). Reducing the thickness of the dielectric layer to this extent leads to reduced electromechanical coupling and very large disturbance modes below the resonant frequency of the device.

[0049] Fig. Figure 10 is a diagram illustrating the effect of an excessively thick "half-lambda" dielectric layer. The solid line 910 is a curve of the magnitude of admittance as a function of frequency for an XBAR with the same setup as the devices of Fig. 8, where the thickness of the SiO2 dielectric layer was increased to 525 nm. In this case, td = 1.24 (λ 0,d / 2). Increasing the thickness of the dielectric layer to this extent leads to reduced electromechanical coupling and very large disturbance modes above the resonant frequency of the device.

[0050] The temperature coefficient of frequency of SiO2 and the temperature coefficient of frequency of lithium niobate have similar magnitudes but opposite signs. XBAR devices with a half-wavelength SiO2 dielectric layer exhibit significantly lower frequency fluctuations with temperature than conventional XBAR devices.

[0051] Fig. Figure 11 is a graph of the temperature coefficient of the frequency of an XBAR as a function of the SiO2 thickness. In particular, the solid line 1110 is a curve of the temperature coefficient of the antiresonance frequency for the XBAR devices whose admittance properties were previously described in Fig. 7 and Fig. Figure 8 was shown. The dashed line 1120 is a curve of the temperature coefficient of the resonant frequency of the same devices. The simulation results show that a conventional XBAR device without a dielectric layer has a temperature coefficient of frequency of approximately -113 ppm / °C. The presence of the lambda-half SiO2 dielectric layer reduces the magnitude of the temperature coefficient of frequency by a factor of approximately 3.

[0052] Fig. Figure 12 is a schematic circuit diagram of a bandpass filter 1200 with five XBAR X1-X5. The filter 1200 could, for example, be a bandpass filter for band n79 for use in a communications device. The filter 1200 has a conventional ladder filter architecture with three series resonators X1, X3, X5 and two shunt resonators X2, X4. The three series resonators X1, X3, X5 are connected in series between a first terminal and a second terminal. Fig. The first and second connections are labeled "In" and "Out," respectively. However, the 1200 filter is bidirectional, and each connection can serve as either an input or an output. The two shunt resonators, X2 and X4, are connected to ground from the nodes between the series resonators. All shunt and series resonators are XBAR.

[0053] The three series resonators X1, X3, X5 and the two shunt resonators X2, X4 of the filter 1200 can be formed on a single plate 1230 made of piezoelectric material, which is connected to a silicon substrate (not shown). Each resonator contains a corresponding IDT (not shown), with at least the fingers of the IDT positioned over a cavity in the substrate. In this and similar contexts, the term "each" means "in relation to each other," i.e., with a one-to-one correspondence. Fig. Figure 12 schematically depicts the cavities as dashed rectangles (e.g., rectangle 1235). In this example, one IDT of each resonator is arranged over a corresponding cavity. In other filters, the IDTs of two or more resonators can be arranged over a common cavity. The resonators can also be cascaded to form multiple IDTs, which can be located over several cavities.

[0054] Each of the resonators X1 to X5 has a resonant frequency and an antiresonant frequency. Put simply, each resonator is a short circuit at its resonant frequency and an open circuit at its antiresonant frequency. Each resonator X1 to X5 forms a "transfer zero" where the transfer between the filter's inputs and outputs is very low. It's important to note that the transfer at a "transfer zero" is not truly zero due to energy losses from parasitic components and other effects. The three series resonators X1, X3, and X5 create transfer zeros at their respective antiresonant frequencies (where each resonator is practically an open circuit). The two shunt resonators X2 and X4 create transfer zeros at their respective resonant frequencies (where each resonator is practically a short circuit).In a typical bandpass filter with acoustic resonators, the antiresonance frequencies of the series resonators are higher than the upper edge of the passband, so the series resonators create zero frequencies above the passband. The resonance frequencies of the shunt resonators are lower than the lower edge of the passband, so the shunt resonators create zero frequencies below the passband.

[0055] From the data of admittance as a function of frequency in Fig. 7 and Fig. Figure 8 shows that the frequency offset between the resonant and antiresonant frequencies of an XBAR with a piezoelectric plate made of 400 nm lithium niobate and a half-wave dielectric layer is approximately 400 MHz. This frequency separation is not sufficient in itself for bandpass filters for telecommunications bands such as n79 (4400 MHz to 5000 MHz) and n77 (3300 MHz to 4200 MHz). US 10,491,192 B1 describes the use of a dielectric layer applied over shunt resonators to reduce the resonant frequencies of the shunt resonators relative to the resonant frequencies of the series resonators. US 10,491,192 B1 describes filters with a very thin or no dielectric layer over series resonators and a dielectric layer thickness of about 0.25 times the thickness of the piezoelectric plate over shunt resonators.

[0056] A similar approach can be used to lower the resonant frequencies of shunt resonators relative to the resonant frequencies of series resonators when the resonators are XBARs with a half-wave dielectric layer. In this case, the dielectric layer thickness tds over the series resonators and the dielectric layer thickness tdp over the shunt resonators (parallel resonators) can be defined as follows: 0.85 λ0,d≤2 tds<2 tdp≤1.15 λ0,d

[0057] Back to Fig. 8: The solid curve 820 is the admittance of an XBAR with a 375 nm SiO2 layer over a piezoelectric 400 nm lithium niobate plate. The dashed curve 830 is the admittance of an XBAR with a 475 nm SiO2 layer over a piezoelectric 400 nm lithium niobate plate. A filter like the Filter 1200 could be fabricated with a piezoelectric 400 nm lithium niobate plate, featuring a 375 nm SiO2 layer over series resonators and a 475 nm SiO2 layer over shunt resonators. In this case, the frequency separation between the resonant frequency of the shunt resonators and the antiresonant frequency of the series resonators is approximately 800 MHz, which is sufficient for a bandpass filter for the n79 band. The frequency separation increases proportionally to the thickness of the piezoelectric plate.

[0058] The ranges for the thickness of the SiO2 layers across series and shunt resonators can be expressed as follows in relation to the thickness ts of the piezoelectric lithium niobate plate: 0.875 ts≤tds <tdp≤1,25 ts. where tds is the thickness of the SiO2 layer across series resonators and tdp is the thickness of the SiO2 layer across shunt resonators (parallel resonators). Procedure description

[0059] Fig. Figure 13 is a simplified flowchart showing a process 1300 for fabricating an XBAR with partial Bragg reflectors or a filter with such XBARs. The process 1300 begins at 1305 with a thin piezoelectric plate arranged on a sacrificial substrate 1302 and a fixture substrate 1304. The process 1300 ends at 1395 with a finished XBAR or filter. The flowchart in Fig. Section 13 contains only the most important process steps. Various conventional process steps (e.g., surface preparation, cleaning, inspection, firing, annealing, monitoring, testing, etc.) can be performed before, between, after, and during the process described in Section 13. Fig. The 13 steps shown will be carried out.

[0060] The flowchart in Fig. Figure 13 shows three variants of process 1300 for the production of an XBAR, which differ in when and how the cavities are formed in the substrate. The cavities can be formed in steps 1310A, 1310B, or 1310C. In each of the three variants of process 1300, only one of these steps is performed.

[0061] Thin plates of single-crystal piezoelectric materials bonded to a non-piezoelectric substrate are commercially available. At the time of this application, both lithium niobate and lithium tantalate plates bonded to various substrates such as silicon, quartz, and fused silica are available. Thin plates of other piezoelectric materials may be available now or in the future. The thickness of the piezoelectric plate can range from 300 nm to 1000 nm. The piezoelectric plate can be made of, for example, Z-cut, twisted Z-cut, or twisted Y-cut lithium niobate or lithium tantalate. The piezoelectric plate can also be made of a different material and / or with a different cut. The substrate can be silicon. If the substrate is silicon, a SiO₂ layer can be placed between the piezoelectric plate and the substrate.The substrate can consist of a different material that allows the formation of deep cavities through etching or other processes.

[0062] In one variant of process 1300, one or more cavities are formed in the substrate at 1310A before the piezoelectric plate is bonded to the substrate at 1330. A separate cavity can be formed for each resonator in a filter device. The one or more cavities can be formed using conventional photolithographic and etching techniques. For example, the cavities can be formed by deep reactive ion etching (DRIE). Typically, the cavities formed at 1310A do not penetrate the substrate.

[0063] At 1330, the piezoelectric plate on the sacrificial substrate 1302 and the device substrate 1304 can be bonded. The piezoelectric plate on the sacrificial substrate 1302 and the device substrate 1304 can be bonded using a wafer bonding process such as direct bonding, surface-activated or plasma-activated bonding, electrostatic bonding, or another bonding technique. Prior to the wafer bonding process, the device substrate can be coated with a bonding layer, which may consist of SiO2 or another material.

[0064] After the piezoelectric plate is bonded to the sacrificial substrate 1302 and the device substrate 1304, the sacrificial substrate and all intervening layers are removed at 1340 to expose the surface of the piezoelectric plate (the surface that previously faced the sacrificial substrate). The sacrificial substrate can be removed, for example, by material-dependent wet or dry etching or another method. The exposed surface of the piezoelectric plate can be polished at 1340 or otherwise machined to prepare the surface and control the thickness of the piezoelectric plate.

[0065] Conductor structures and dielectric layers defining one or more XBAR devices are formed at 1350. Typically, a filter device has two or more conductor layers that are deposited and structured sequentially. The conductor layers may include bonding pads, gold or solder dots, or other means of creating connections between the device and external circuitry. The conductor layers may be made of, for example, aluminum, an aluminum alloy, copper, a copper alloy, molybdenum, tungsten, beryllium, gold, or another conductive metal. Optionally, one or more layers of other materials may be placed below (i.e., between the conductor layer and the piezoelectric plate) and / or above the conductor layer.For example, a thin film of titanium, chromium, or another metal can be used to improve adhesion between the conductive layers and the piezoelectric plate. The conductive layers can include bonding pads, gold or solder points, or other means of creating a connection between the device and external circuitry.

[0066] The conductor structures can be formed at 1350 by depositing the conductor layers onto the surface of the piezoelectric plate and removing the excess metal by etching through the patterned photoresist. Alternatively, the conductor structures can be formed at 1350 using a lift-off process. Photoresist can be deposited onto the piezoelectric plate and patterned to define the conductor structure. The conductor layer can be deposited sequentially onto the surface of the piezoelectric plate. The photoresist can then be removed, thus removing the excess material and leaving the conductor structure.

[0067] At 1360, a half-wave dielectric layer can be formed on the front face of the piezoelectric layer. This half-wave dielectric layer can be deposited over the conductor structures or only between the fingers of the IDT. In some filter devices, a first dielectric layer can be deposited over / between the fingers of all IDTs, and a second dielectric can be selectively formed over a portion of the IDTs, e.g., only over the IDTs of shunt resonators. The first dielectric layer is typically thicker than the second. The first and second dielectric layers can be made of the same or different materials. Either the first or the second dielectric layer can be deposited first.

[0068] In a second variant of process 1300, one or more cavities are formed on the back side of the substrate at 1310B, after all conductor structures and dielectric layers have been formed at 1350 and 1360. A separate cavity can be formed for each resonator in a filter device. The one or more cavities can be formed by anisotropic or orientation-dependent dry or wet setting to open holes through the back side of the substrate to the piezoelectric plate.

[0069] In a third variant of process 1300, one or more cavities in the form of depressions in the substrate can be formed at 1310C by etching the substrate with an etchant introduced through openings in the piezoelectric plate and the half-wave dielectric layer. A separate cavity can be formed for each resonator in a filter device. The cavity(ies) formed at 1310C do not penetrate the substrate.

[0070] In all variations of process 1300, the filter device is completed at 1370. Actions that can be performed at 1370 include depositing an encapsulation / passivation layer, such as SiO2 or Si3O4, over all or part of the device and / or forming bonding pads, solder joints, or other means of connection between the device and external circuitry, if these steps were not performed at 1350. Other actions at 1370 may include cutting individual devices from a multi-device wafer, other packaging steps, and testing. Another action that can be performed at 1370 is tuning the resonant frequencies of the resonators within the device by adding or removing metal or dielectric material from the front face of the device. Once the filter device is complete, the process ends at 1395.

[0071] One variant of process 1300 begins with a single-crystal piezoelectric wafer (1302) instead of a thin piezoelectric plate on a sacrificial substrate made of another material. Ions are implanted to a controlled depth below the surface of the piezoelectric wafer (in Fig.(Figure 13 not shown). The portion of the wafer from the surface to the depth of ion implantation is (or becomes) the thin piezoelectric plate, and the remainder of the wafer is the sacrificial substrate. The piezoelectric wafer and the device substrate are bonded at 1330 as previously described. At 1340, the piezoelectric wafer can be split in the plane of the implanted ions (e.g., by a thermal shock), leaving a thin plate of piezoelectric material exposed and bonded to the acoustic Bragg reflector. The thickness of the thin plate of piezoelectric material is determined in part by the energy (and thus the depth) of the implanted ions. The process of ion implantation and subsequent splitting of a thin plate is commonly referred to as "ion cutting." After splitting the piezoelectric wafer, the exposed surface of the piezoelectric plate can be flattened and its thickness reduced, e.g., by...through chemical-mechanical polishing. Concluding remarks

[0072] Throughout this entire description, the embodiments and examples shown should be considered as models and not as limitations of the disclosed or claimed devices and procedures. Although many of the examples presented here involve specific combinations of process activities or system elements, it should be understood that these activities and elements can be combined in other ways to achieve the same objectives. With regard to flowcharts, additional or fewer steps can be taken, and the steps shown can be combined or further refined to achieve the procedures described herein. Activities, elements, and features discussed only in connection with one embodiment are not intended to exclude a similar role in other embodiments.

[0073] As used here, "multiple" means two or more. As used here, a "set" of elements may comprise one or more such elements. In the form used here, whether in the written description or in the claims, the terms "comprising," "including," "bearing," "having," "containing," "incorporating," and the like are to be understood as being unlimited, i.e., including but not limited to. Only the transitional phrases "consisting of" and "consisting substantially of" are closed or semi-closed transitional phrases with respect to claims. The use of ordinal expressions such as "first," "second," "third," etc.In the claims, the use of "and / or" to modify a claim element does not in itself imply a priority, precedence, or order of one claim element over another, nor does it indicate the temporal order in which the activities of a process are carried out. Rather, it is merely used as a distinguishing term to differentiate one claim element with a particular name from another element with the same name (but for the use of the ordinal expression to distinguish the claim elements). As used here, "and / or" means that the listed elements are alternatives, but the alternatives also include every combination of the listed elements.

Claims

[1] Acoustic resonator device comprising: a substrate (120) with a surface; a piezoelectric plate (310, 510) with a front and a back surface, wherein the back surface is attached to the surface of the substrate (120), except for a part of the piezoelectric plate (310, 510) which forms a membrane (115) spanning a cavity (140) in the substrate (120); an interdigital converter, IDT, (130) on the front surface of the piezoelectric plate (310, 510), such that nested fingers (338, 538) of the IDT (130) are arranged on the membrane (115), wherein the piezoelectric plate (310, 510) and the IDT (130) are configured such that a high-frequency signal applied to the IDT (130) excites a primary acoustic shear mode in the membrane (115); and a dielectric layer (350, 550) arranged either on the front surface or on the back surface of the piezoelectric plate (310, 510), where a thickness td of the dielectric layer (350, 550) is defined as: 0.85 λ 0,d ≤ 2 td ≤ 1.15 λ 0,d , and where λ 0,d a wavelength of the fundamental resonance of the acoustic volume shear wave in the dielectric layer (350, 550). [2] Acoustic resonator device according to claim 1, wherein a thickness ts of the piezoelectric plate (310, 510) is defined as follows: 2 ts=λ0,s where λ 0,s a wavelength of a fundamental resonance of the acoustic volume shear wave in the piezoelectric plate (310, 510). [3] Acoustic resonator device according to claim 1, wherein the dielectric layer (350, 550) is one or more of SiO2, Si3N4, Al2O3 and AlN. [4] Acoustic resonator device according to claim 1, wherein the piezoelectric plate (310, 510) lithium niobate is, the dielectric layer is (350, 550) SiO2, and The thickness ts of the piezoelectric plate (310, 510) and the thickness td of the dielectric layer (350, 550) are defined by the relationship: 0.875 ts ≤ td ≤ 1.25 ts [5] Acoustic resonator device according to claim 4, wherein The temperature coefficient of the frequency of the acoustic resonator device lies between -32 ppm / C° and -42 ppm / C° at a resonance frequency and between -20 ppm / C° and -36 ppm / C° at an antiresonance frequency. [6] Filter device comprising: a substrate (120); a piezoelectric plate (310, 510) with parallel front and back surfaces and a thickness ts, wherein the back surface is attached to the substrate (120); a conductor structure formed on the front surface, the conductor structure comprising a plurality of interdigital transducers, IDT, (130) of a respective plurality of resonators comprising a shunt resonator and a series resonator, wherein nested fingers of each of the plurality of IDT (130) are arranged on respective parts of the piezoelectric plate (310, 510) which are above one or more cavities formed in the substrate (120); a first dielectric layer (350, 550) with a thickness tds, deposited between the fingers of the series resonator; and a second dielectric layer (350, 550) with a thickness tdp deposited between the fingers of the shunt resonator, wherein ts, tds and tdp are connected to each other by the equations: 2 ts=λ0,s, and 0.85 λ0,d≤2 tds<2 tdp≤1.15 λ0,d, where λ 0,sa wavelength of a fundamental resonance of the acoustic volume shear wave in the piezoelectric plate (310, 510) is, and λ 0,d a wavelength of the fundamental resonance of the acoustic volume shear wave in the dielectric layer (350, 550). [7] Filter device comprising: a substrate (120); a piezoelectric plate (310, 510) made of lithium niobate with parallel front and back surfaces and a thickness ts, wherein the back surface is attached to the substrate (120); a conductor structure formed on the front surface, the conductor structure comprising a plurality of interdigital transducers, IDT, (130) of a respective plurality of resonators comprising a shunt resonator and a series resonator, wherein nested fingers of each of the plurality of IDT (130) are arranged on respective parts of the piezoelectric plate (310, 510) which are above one or more cavities formed in the substrate (120); a first SiO2 layer with a thickness tds, deposited between the fingers of the series resonator; and a second SiO2 layer with a thickness tdp, deposited between the fingers of the shunt resonator, wherein ts, tds and tdp are connected by the equation: 0.875 ts≤tds <tdp≤1,25 ts. [8] Acoustic resonator device according to claim 7, wherein a temperature coefficient of the frequency of the acoustic resonator device lies between -20 ppm / C° and -42 ppm / C° at the resonance frequencies and the antiresonance frequencies of all the multitude of resonators. [9] Method for manufacturing an acoustic resonator device on a piezoelectric plate (310, 510) with parallel front and back surfaces, wherein the back surface is attached to a substrate (120), the method comprising: Forming (1310A, 1310B, 1310C) a cavity (140) in the substrate (120) such that part of the piezoelectric plate (310, 510) forms a membrane (115) spanning the cavity (140); Forming (1350) an interdigital transducer, IDT, (130) on the front surface of the piezoelectric plate (310, 510) such that nested fingers of the IDT (130) are arranged on the membrane (115), wherein the piezoelectric plate (310, 510) and the IDT (130) are configured such that a high-frequency signal applied to the IDT (130) excites a primary acoustic shear mode within the membrane (115); and Forming (1360) a dielectric layer (350, 550) on the front or back surface of the piezoelectric plate (310, 510), where a thickness td of the dielectric layer (350, 550) is defined as: 0.85 λ 0,d ≤ 2 td ≤ 1.15 λ 0,d , and where λ 0,d a wavelength of the fundamental resonance of the acoustic volume shear wave in the dielectric layer (350, 550). [10] Method according to claim 9, wherein a thickness ts of the piezoelectric plate (310, 510) is defined as follows: 2 ts=λ0,s where λ 0,s a wavelength of a fundamental resonance of the acoustic volume shear wave in the piezoelectric plate (310, 510). [11] Method according to claim 9, wherein forming the dielectric layer (350, 550) further comprises: Deposition of one or more of SiO2, Si3N4, Al2O3 and AlN. [12] Method according to claim 9, wherein the piezoelectric plate (310, 510) is lithium niobate, and Forming the dielectric layer (350, 550) includes depositing SiO2 to a thickness td, where td is greater than or equal to 0.875 ts and less than or equal to 1.25 ts, where ts is a thickness of the piezoelectric plate (310, 510).

Images