High-frequency CMOS ultrasonic transducer

Abstract

In the example of the CMOS IC described, an ultrasonic transducer (1200) having terminals is formed on a substrate (1230) of the IC. CMOS circuitry (1236) having ultrasonic signal terminals is formed on the substrate. At least one metal interconnect layer (1238) covers the ultrasonic transducer and the CMOS circuitry. The at least one metal interconnect layer connects the CMOS circuitry ultrasonic signal terminals to the terminals of the ultrasonic transducer.

Description

Technical Field

[0001] The present application relates to high frequency ultrasonic transducers formed within complementary metal oxide semiconductor (CMOS) integrated circuits. Background Art

[0002] Ultrasonic transducers and sensors are acoustic devices that fall into three main categories: transmitters, receivers, and transceivers. Transmitters convert electrical signals into ultrasound waves, receivers convert ultrasound waves into electrical signals, and transceivers can both transmit and receive ultrasound waves.

[0003] Ultrasonic transducers are used in systems to assess targets by interpreting reflected signals, similar to radar. The transmitted signal consists of short bursts of ultrasonic energy. After each burst, the electronics look for a return signal within a small time window corresponding to the time it takes for the energy to pass through the medium. Only signals received during this period are used for additional signal processing.

[0004] Medical ultrasound transducers (probes) come in a variety of shapes and sizes and are used to produce cross-sectional images of various parts of the body. The transducers can be passed over surfaces and in contact with the body. Arrays of ultrasound transducers can be used to sense fingerprints to control access to devices such as laptops or mobile phones.

[0005] Ultrasonic sensors are widely used in cars as parking sensors to help drivers back into parking spaces. They are being tested for many other automotive uses, including ultrasonic person detection and assisting autonomous UAV navigation. Summary of the Invention

[0006] In the described example of a CMOS IC, an ultrasonic transducer having terminals is formed on the substrate of the IC. CMOS circuitry having ultrasonic signal terminals is also formed on the substrate. At least one metal interconnect layer covers the ultrasonic transducer and the CMOS circuitry. The at least one metal interconnect layer connects the CMOS circuitry ultrasonic signal terminals to the terminals of the ultrasonic transducer.

[0007] Figures

[0008] 1 is a cross-sectional view of an integrated circuit (IC) including a memory cell using a ferroelectric capacitor (FeCap) as a storage element.

[0009] Figure 2A 、 Figure 2B are corresponding top and side cross-sectional views of a row of FeCaps used to form an ultrasonic transducer.

[0010] Figure 3 is a cross-sectional view of a single simulated FeCap illustrating the simulated electric field around the FeCap.

[0011] Figure 4 yes Figure 3 Dispersion plot of various modes within the FeCap structure.

[0012] Figure 5 is a cross-sectional diagram illustrating simulated oscillations within a row of FeCaps.

[0013] Figure 6 is a scatter plot illustrating a plot of frequency versus FeCap pitch for guided mode operation.

[0014] Figure 7 is a cross-sectional view illustrating a simulated operation of an ultrasonic resonator formed of a row of FeCaps.

[0015] Figure 8 is a cross-sectional diagram illustrating the simulated operation of an ultrasonic transmitter formed by a row of FeCaps.

[0016] Figure 9 is a cross-sectional view of a FeCap ultrasonic transducer coupled to a water droplet.

[0017] Figure 10A 、 Figure 10B 、 Figure 10C are schematic diagrams of various example FeCap connections.

[0018] Figure 11 is a top view of an example FeCap ultrasound transducer.

[0019] Figure 12 is a cross-sectional view of an example packaged FeCap ultrasound transducer.

[0020] Figure 13 is a flow chart illustrating the operation of an ultrasound transmitter on a CMOS IC. DETAILED DESCRIPTION

[0021] In the drawings, like elements are denoted by like reference numerals for consistency.

[0022] Piezoelectric micromachined ultrasonic transducer (PMUT) array technology is currently used to perform ultrasound imaging for medical applications, capturing images of soft tissues such as muscles, internal organs, blood flow, and fetal observation. Advances in microelectromechanical systems (MEMS) have resulted in ultrasonic transducers based on plate bending modes, which offer promising improvements in bandwidth, cost, and throughput compared to conventional large-scale thickness-mode PZT sensors. However, currently available PMUT-based transducers have limited operating frequencies, typically in the range of 2-14 MHz, due to the confinement of their bending modes. Low-frequency operation results in low-resolution images. Furthermore, these arrays are typically fabricated on a substrate that is then bonded to a complementary metal-oxide semiconductor (CMOS) integrated circuit (IC), which includes the electronics for controlling the array. This leads to complex packaging and integration issues.

[0023] Ferroelectric random access memory (FRAM) is a nonvolatile memory technology that behaves similarly to DRAM (dynamic random access memory). Each individual bit can be accessed, but unlike EEPROM (electrically erasable programmable read-only memory) or flash memory, FRAM does not require a dedicated sequence to write data, nor does it require high programming voltages. Each ferroelectric memory cell contains one or more ferroelectric capacitors (FeCap). The dielectric constant of ferroelectric capacitors is typically much higher than that of linear dielectrics due to the influence of semipermanent electric dipoles that form within the ferroelectric material's crystal structure. When an external electric field is applied across a ferroelectric dielectric, the dipoles tend to align themselves with the field direction. This is caused by tiny shifts in the positions of the atoms, resulting in a shift in the charge distribution within the crystal structure. These tiny shifts in atomic positions produce the piezoelectric effect.

[0024] Ferroelectricity is a property of certain materials that they have a spontaneous electric polarization that can be reversed by applying an external electric field. Ferroelectric materials are typically in single crystal or polycrystalline form and have a reversible spontaneous polarization over a range of temperatures. One known ferroelectric material is lead zirconate titanate (PZT), which is part of a solid solution formed between ferroelectric lead titanate and antiferroelectric lead zirconate. Different compositions are used for different applications; for memory applications, PZT is preferred, being closer in composition to lead titanate, while piezoelectric applications exploit the divergent piezoelectric coefficient associated with the morphotropic phase boundary, which is found to be close to a 50 / 50 composition.

[0025] As will be described in more detail below, FeCaps can be used as a building block to generate stress and strain within the CMOS IC itself to form an ultrasonic transducer device, thereby eliminating the need for a separate array substrate. A periodic array of FeCaps is arranged to create an acoustic waveguide that can guide the generated waves along the front-end-of-line (FEOL) layers of the CMOS die. FEOL is the first part of IC manufacturing, where individual devices (transistors, capacitors, resistors, etc.) are patterned in the semiconductor. FEOL generally encompasses everything but the deposition of metal interconnect layers. Deposition of metal layers and intermediate dielectric layers and vias is performed during back-end-of-line (BEOL) processing.

[0026] By adjusting the size of the FeCaps in the array, the waveguide properties can be manipulated to create reflectors and radiators that help guide the acoustic waves generated by the piezoelectric activity of the FeCap array. The transducer can operate in transmit mode and / or receive mode. The FeCap array can produce a very high-frequency ultrasound transducer truly integrated in CMOS technology for high-resolution, short-range applications. The examples described below operate at frequencies up to 700+ MHz.

[0027] FIG1 is a cross-sectional view of a portion of a prior art integrated circuit 100 that includes a memory cell using a ferroelectric capacitor (FeCap) 110 as a storage element. Ferroelectric random access memory has been implemented in several configurations. The one-transistor-one-capacitor (1T-1C) memory cell design in a FeRAM array is similar in construction to the memory cell in the widely used DRAM, as both cell types include a capacitor and an access transistor. While a linear dielectric is used in a DRAM cell capacitor, the dielectric structure in a FeRAM cell capacitor includes a ferroelectric material, typically lead zirconate titanate (PZT).

[0028] FeCap 110 represents a large number of FeCaps included in IC 100. FeCap 110 has a bottom plate 104 and a top plate 106 patterned with conductive layers. Dielectric PZT material 108 is sandwiched between bottom plate 104 and top plate 106 to form a capacitor configuration. Vias 102 connect conductive signal lines 103 to bottom plate 104. Vias 112 connect conductive signal lines in first metal layer 114 to top plate 106. Second metal layer 116 provides an additional signal routing layer that interconnects to signal lines in first metal layer 114 using vias (not shown).

[0029] Figure 2A is a top cross-sectional view of a portion of a CMOS IC 200 including a row of FeCaps used as ultrasound transducers fabricated within the CMOS IC 200 . Figure 2Bis a side cross-sectional view of a portion of CMOS IC 200. FeCap 210 represents a linear array of FeCaps included in IC 200. FeCap 210 has a bottom plate 204 and a top plate 206 patterned with conductive layers. Ferroelectric PZT material 208 is sandwiched between bottom plate 204 and top plate 206 to form a capacitor configuration. Vias 202 connect conductive signal lines 203 to bottom plate 204. Vias 212 connect contacts 214, and therefore conductive signal lines 215 in the first metal layer, to top plate 106. Contacts 216 in the second metal layer provide an additional signal routing layer that is interconnected to signal lines in the first metal layer using vias (not shown).

[0030] A linear array of FeCaps 210 is formed along an axis 220. Each FeCap 210 has a width (w) 224 and a length (L) 222. The linear array of FeCaps is arranged with a pitch (p) 225. The array of FeCaps is arranged so that the length 222 of each FeCap 210 is perpendicular to the axis 220. As will be described in more detail below, the respective widths and pitches of the FeCaps in the linear array are selected to generate ultrasonic waves in the center portion of the linear array, prevent ultrasonic waves from being dispersed from one end of the linear array, and cause ultrasonic waves to "leak" or disperse from the other end of the linear array, thereby forming an ultrasonic emitter. In this example, the length L is 10 μm. In another example, L can be larger or smaller for corresponding signal strength.

[0031] In this example, substrate 230 is bulk silicon. An n+ well and a p+ well are formed on top of substrate 230 to provide regions for CMOS transistor implantation. During FEOL processing of CMOS IC 200, FeCap 210 is also formed on N-well layer 231 using known or later developed fabrication techniques. Contacts 214, 216 and vias 212 are formed during BEOL processing of CMOS IC 200 using known or later developed fabrication techniques. In this example, silicon dioxide layers 232, 233, 234 provide electrical insulation around and between the first and second metal layers and contacts such as contacts 214, 216. In other examples, various types of known or later developed interconnect dielectric layers may be used between the multiple metal layers.

[0032] FeCaps 210 are fabricated during the FEOL processing of CMOS IC 200 using known or later-developed manufacturing techniques. A first conductive layer, forming base plate 204, can be deposited on substrate 200. A ferroelectric layer, forming the FeCaps, is then deposited over the first conductive layer. A second conductive layer is then deposited over the ferroelectric layer. An etching process is then performed to form the individual plates of the linear array of FeCaps. In another example, each layer can be patterned and etched individually. In this example, the first and second conductive layers forming the plates of the linear array of FeCaps are metal alloys.

[0033] Figure 3 is a cross-sectional view of a single simulated FeCap 310 illustrating the simulated electric field around the FeCap. To better understand the operation of the FeCap-based ultrasonic transducer, a numerical finite element method (FEM) simulation tool was used to simulate a FeCap-based ultrasonic transducer similar to Figure 2A 、 Figure 2B A simulation was performed on the structure of the FeCap 210. Two-dimensional (2D) and three-dimensional (3D) FEM simulation tools are commercially available from several suppliers.

[0034] In this example, the simulated unit cell 300 includes a region 337 that simulates the mechanical and material properties of the selected FEOL CMOS process to simulate the structure of the FeCap 310, which includes the lower plate 304, the upper plate 306, the PZT layer 308, and the via 302. The region 337 also simulates the mechanical and material properties of the selected BEOL CMOS process to simulate the structure of the vias 312, 318 and the contacts 314, 316. The region 337 also simulates the dielectric layer surrounding the vias and contacts, such as Figure 2B Finally, a conventional IC passivation process (silicon dioxide, silicon nitride, etc.) of a CMOS process is assumed. The FeCap 310 has a width 324 and a pitch 325 to adjacent FeCaps. For this simulation, the length of the FeCap 310 is perpendicular to the view and is considered to be infinitely in and out of the plane of the view.

[0035] Regions 338 and 339 represent "perfectly matched layers" (PMLs) that are added to allow for the study of the simulated behavior of ultrasonic radiation generated by FeCap 310 in a hypothetical infinitely extending plane. PMLs 338 and 339 have corresponding simulated properties that match those of the region 337 to which they are adjacent. In this example, PML 338 matches the dielectric of region 337 above FeCap 310, while PML 339 matches the silicon substrate below FeCap 310.

[0036] Since the goal is to simulate a resonator, the wave on the left must be related to the wave on the right with a constant phase shift. Therefore, the wave vector time interval (pitch 325) must be a constant value. Given a constant phase shift, multiple eigenfrequency simulations are performed to find all possible propagation modes in the structure. Depending on the phase shift, frequency, and material properties, various traveling waves and evanescent waves are formed. An evanescent field or evanescent wave is an oscillating field that does not propagate but whose energy is spatially concentrated near the source. The shaded areas 335 and 336 illustrate the evanescent field. The simulation results can be illustrated by scatter plotting.

[0037] Figure 4 yes Figure 3 The dispersion plot of various modes within the FeCap structure of . In this plot, k x is the wave vector in the x direction, and “a” is the period of the structure. x When multiplied by "a" equals π, there is a 180-degree offset between the left and right boundaries of the unit cell. x When multiplied by "a" equals 0, there is 0 degrees of phase shift.

[0038] There are two types of waves that can propagate in solids: longitudinal waves and shear waves. Longitudinal waves have a faster speed of sound than shear waves. Each type of wave will have its own acoustic cone. Thus, one acoustic cone 401 has a steep slope for longitudinal waves in bulk silicon, one has a slightly shallower slope for shear waves in bulk silicon, and two similar cones have lower slopes for the interlayer dielectric.

[0039] In this drawing, the acoustic cone region 401 represents the acoustic field that can propagate to a substrate such as a Figure 2B Acoustic cone region 402 represents shear waves that can propagate in bulk silicon. Acoustic cone region 403 represents shear waves that can propagate in oxides such as Figure 2B The acoustic cone region 404 represents all longitudinal waves that propagate in the dielectric layers 232, 233, 234 in the oxide. The acoustic cone region 404 represents shear waves that can propagate in the dielectric layer. Region 404 has a shallower slope because the speed of sound in the oxide is lower than the speed of sound in silicon. In region 405 below region 404, plane waves cannot propagate into the bulk silicon, but they can propagate into the dielectric. The acoustic speed (c) sets the slope of the line that defines each region (acoustic cone) 401, 404. Each line is described by expression (1).

[0040] ω=ck x (1)

[0041] The plot line 406 below region 404 represents a wave that cannot propagate into the bulk silicon or into the dielectric. Therefore, this mode creates a "trapped" mode, referred to herein as a "guided mode." The FeCap creates a "slow wave" structure for very specific frequencies. This mode line cannot propagate into the bulk silicon or the bulk dielectric because it has a much slower wave velocity. In this simulation example, the specific guided mode frequency is approximately 700 MHz, as shown at 408, where k x a=π。

[0042] The result shown in the guided-mode plot line 406 is similar to Snell's law in optics, which defines the angle of refraction when a ray of light passes through a boundary separating two media with different speeds of light. In this case, because the oxide and silicon have different acoustic velocities, at the guided-mode frequency, total internal reflection occurs from both the bulk silicon on the bottom of the FeCap cell and the bulk oxide on the top of the cell.

[0043] Guidance mode plot line 407 is the second guidance mode, which begins to appear as the pitch of the FeCap becomes smaller and the resulting operating frequency becomes higher.

[0044] Reference again Figure 3 This simulation illustrates operation at a guided frequency of approximately 700 MHz. The shaded areas on the upper and lower plates 306 and 304 indicate that stress occurs only around the FeCaps and does not propagate downward into the bulk silicon or upward into the bulk oxide. The lightly shaded areas 335 and 336 indicate that the guided mode wave attempts to penetrate but is unable to do so; therefore, it cannot propagate vertically. Thus, a waveguide is created that can confine acoustic vibrations to high frequencies of approximately 700 MHz.

[0045] Figure 5 The diagram shows the guided mode frequency of about 700 MHz (see FIG. Figure 4 ). Regions 540, 543, 544, etc. represent the negative energy generated by the evanescent wave, while regions 541, 542, 545, etc. represent the positive energy generated by the evanescent wave. Thus, moving from cell to cell, there is a 180-degree phase shift, and resonant operation occurs.

[0046] Figure 6 is an example scatter plot illustrating a plot of frequency versus FeCap pitch for guided mode operation. In this example plot, the pitch of the FeCaps in the simulated linear array was varied from 1.0 micron to 2.5 microns to create a similar Figure 4 Then plot the k for each scatter plot in the series. x The operating point is equal to 1 to form Figure 6As the size of the FeCap increases, the guided modes move into lower frequencies, as indicated by the guided mode plot line 606, see Figure 4 The guidance mode plot line 607 illustrates another guidance mode that begins to appear more clearly at lower FeCap pitches, see Figure 4 Plot line 407 in . Both guided modes indicated by plot lines 606, 607 cannot propagate into the silicon and cannot propagate into the dielectric, so they are guided.

[0047] Therefore, as the pitch of the FeCaps increases, the linear array of FeCaps resonates at a lower frequency. Similarly, the acoustic cone shifts to a lower frequency. Acoustic cone 604 is a shear mode that can propagate into the dielectric. Acoustic cone 609 is a shear mode that can propagate into both the bulk silicon and the dielectric. Acoustic cone 603 is a longitudinal mode that can propagate into the dielectric. The longitudinal mode that can propagate into the bulk silicon occurs at a higher frequency and is therefore not reflected in the Figure 6 Shown in.

[0048] In this example, a horizontal line 610 is drawn on the plot at approximately 715 MHz. This line intercepts the guided mode line 606 at 1.95 μm. In a region 612 on the right, such as at 2.2 μm, the horizontal line 610 is above the guided mode line 606 and below the acoustic cone 604, so propagation is blocked. In this way, reflectors are created in the linear array of FeCaps.

[0049] The pitch of the FeCaps in the reflector section only needs to be slightly larger than the pitch of the FeCaps in the main waveguide section of the linear array of FeCaps. In this example, a FeCap pitch in the range of 2.0 μm-2.3 μm, as indicated at 612, is suitable for reflector mode operation. However, if the pitch is made too large, operation above the plot cone 604 will occur and propagation into bulk silicon and / or bulk oxide will occur.

[0050] The width and pitch of the capacitors were chosen based on numerical FEM simulations to achieve optimal energy confinement. The lithographic limitations of the process set the minimum possible capacitor size and pitch. While these minimum process dimensions may produce functional devices, they do not necessarily result in good performance. The preferred dimensions (width and spacing) vary depending on the exact layer structure of the capacitor, the surrounding dielectric, and varying layer thicknesses. Both the pitch and width of the capacitors determine the resonant frequency of the device. However, as the pitch and width of the capacitors increase, the resonant frequency decreases as the wavelength increases. This results in longer evanescent wave "tails" in the bulk and BEOL layers. Depending on the exact composition of the process layers, there is a soft upper limit on the size and pitch of the capacitors. Beyond this limit, the wavelength at the resonant frequency is so large that the waveguide structure in the FEOL appears "too thin" to achieve any practical guiding, generating significant losses in the bulk and BEOL layers due to the very large evanescent tails.

[0051] Figure 7 is a cross-sectional view illustrating the simulated operation of an ultrasonic resonator 700 formed by a linear array of FeCaps. In this example, each FeCap unit is similar to Figure 3 FeCaps 300. In the main waveguide region 701, the FeCaps have a pitch p1 that is selected to produce a guided operating mode at a selected frequency. The corresponding pitches p2 of the FeCaps in region 702 and p3 in region 703 are selected to be greater than p1, so that regions 702 and 703 act as reflectors at the selected frequency of the main waveguide region 701.

[0052] In this example, the pitch p1 of the FeCaps in the main waveguide region 701 is chosen to produce guided mode operation at approximately 715 MHz. Figure 6 , the corresponding pitch of the FeCaps in region 701 is selected to be approximately 1.95 μm. The corresponding pitch of the FeCaps in regions 702, 703 is selected to reflect the selected 715 MHz signal. Figure 6 , the corresponding pitch of the FeCaps in the reflector regions 702, 703 is selected to be greater than about 2.0 μm and less than about 2.3 μm. Therefore, a 715 MHz signal cannot propagate left or right in the reflector regions 702, 703, and cannot propagate upward or downward in the main waveguide region 701. Therefore, the example linear array of FeCaps 700 operates as a resonator and generates a resonant signal at a frequency of 715 MHz.

[0053] In theory, there is no maximum or minimum limit to the length of an array of FeCaps. The theory assumes that there are "many" periods such that the structure maintains translational symmetry (~periodicity) at every point in space. In practice, the main waveguide section 701 and reflector section 702 require a minimum of four or five periods for the device to function properly. There is no upper limit to the number of periods, but as the device approaches the theoretical case of translational symmetry at every point, the number of periods is improved. In this example, the length of the main waveguide section 701 is approximately 20 μm.

[0054] As the width of the reflector FeCap in the reflector section 702 increases, a greater impedance mismatch and greater reflection (and scattering) will occur. Therefore, a smaller number of reflector segments will be required. For the example design, the corresponding pitch / width of the reflector FeCap can be in the range of 10% to 50% wider than the main waveguide FeCap. Smaller reflectors are also possible, but a larger number is required. Larger reflectors are also possible, however, at some point the cutoff of the other mode begins to be low enough and the reflector will fail. The reflector can have a gradual pitch / width change, or an abrupt pitch / width change.

[0055] Figure 8 is a cross-sectional view illustrating the simulated operation of an ultrasound transmitter 800 formed by a linear array of FeCaps. In this example, the FeCap units in the main waveguide region 801 and the reflector region are similar to Figure 3 The FeCap 300 has a pitch selected to operate in the waveguide mode at approximately 715 MHz.

[0056] In order for the transmitter 800 to radiate the ultrasonic signal generated by the main waveguide region 801, in this example simulation, the signal must couple into a radiating medium adjacent to the transmitter device represented by region 804. In this example simulation, the radiating medium 804 is given an oxide layer (such as Figure 2B The oxide layers 232, 233, 234 and the PML region 338) have approximately the same ultrasonic properties.

[0057] If the pitch of the FeCaps in transition region 803 is chosen to be the same as the pitch within main waveguide region 801, there will be no "radiator structure." Waves from main waveguide region 801 will continue to travel in this segment. However, due to the abrupt interruption of the capacitor period at boundary 814 between transition region 803 and radiating medium 804, it will cause significant reflection and scattering. In other words, there will be a mismatch with the radiating medium, and therefore very poor radiation efficiency. This configuration may operate as a low-Q resonator, but it is not a good radiator.

[0058] The gradual transition of the FeCap size in the transition region 803 helps match the ultrasonic impedance between the transition region 803 and the radiating medium 804. In this example, the radiating medium 804 is modeled as an oxide, which may be formed in the region 337 with various oxide layers (such as 232, 233, 234 ( Figure 2B )) of the same oxide. Figure 6 , for a selected frequency of 715 MHz indicated by horizontal line 610, reducing the FeCap pitch shifts the mode of operation leftward to region 605 where plane waves cannot propagate into bulk silicon, but they can propagate into the oxide.

[0059] In this example, the pitch and width of the FeCaps in the transition region 803 are gradually reduced. For example, starting from FeCap 810, the pitch and width of the FeCaps in the transition region 803 are gradually reduced to be as small as feasible for the CMOS manufacturing process being simulated. In addition to reducing the pitch and width, the second layer vias 318 and contacts 316 ( Figure 3 ). In addition, in the portion of the FeCap indicated by 812, the first layer vias 312 and contacts 314 are also removed ( Figure 3 A few of the top contacts and bottom vias of the FeCaps within section 812 are removed to gradually reduce the physical size of the linear array of FeCaps in transition section 803. The goal is to gradually reduce the physical size of the FeCap structure, thereby gradually transitioning the ultrasonic impedance of transition section 803 to match the ultrasonic impedance of radiating medium 804.

[0060] The gradual reduction in size and pitch of the FeCaps in the transition region 803 causes the ultrasonic signal generated in the main waveguide region 801 to propagate through the transition region 803 and then smoothly launch into the radiating medium 804. Within the transition portion 803, the ultrasonic signal begins to propagate upward into the BEOL oxide layer, thereby forming a radiating structure from the FeCaps in the transition portion 803. Signal waves 816, 817, and 818 represent the ultrasonic signal launched into the radiating medium 804 while remaining above the bulk silicon substrate region 339.

[0061] Figure 9 is a cross-sectional view of a CMOS IC 930 illustrating simulated operation of an example FeCap ultrasound transducer 900 coupled to a water droplet 920. The transducer 900 includes a linear array of FeCaps 910 similar to a water droplet 920 and a radiating medium 904. Figure 2A 、 Figure 2B The linear array of FeCaps shown in FIG, and the radiating medium 904 is similar to Figure 8 The radiation medium 804 in Figure 8 As depicted, in the main waveguide portion 901 of the linear array of FeCaps, each FeCap has a corresponding pitch and width selected to produce guided mode operation for a selected frequency. In this example, a corresponding pitch of approximately 1.95 μm is selected and the width is selected to produce guided mode operation at approximately 715 MHz. The corresponding pitch of the reflector portion 902 of the FeCaps is selected to be larger than the corresponding pitch in portion 901 to provide a reflector mode of operation at the selected frequency of 715 MHz. In this example, the corresponding pitch of the FeCaps in portion 902 is selected to be in the range of 2.0-2.3 μm.

[0062] The portion 903 of the FeCap in the linear array 910 is formed similar to Figure 8 The corresponding pitch and width of the FeCap in portion 903 are gradually reduced to promote the matching of ultrasonic impedance between the transition region 903 and the radiating medium 904.

[0063] Radiating medium 904 forms a traveling wave radiator. This structure needs to be long enough to allow for multiple wavelengths along the surface and function as a true traveling wave radiator. Simulations indicate that in the upper BEOL material, the length of radiating medium 904 should be at least five to ten wavelengths to produce a good radiation pattern. The longer radiating medium 904 is, the more beams will be emitted, and the more efficiently they will be radiated. In this example, four beams 911, 912, 913, and 914 are emitted from radiating medium 904 into water droplet 920. In this example, the length of radiating medium 904 is approximately 150 μm.

[0064] However, in another application, multiple beams may not be desired.Thus, the length of the radiator medium may be selected depending on the desired radiation pattern of the target application.

[0065] In this example, the radiative medium 904 is formed from a dielectric layer formed during the BEOL processing of the CMOS device 930. Portion 905 of the CMOS IC 930 includes an oxide formed during the BEOL processing of the CMOS device 930. In one example, portion 905 also includes CMOS circuitry coupled to the transducer 900 to provide control and processing of the ultrasonic signals of the transducer 900. The substrate 906 is bulk silicon. In this example, at least two interconnect layers are fabricated over the transducer 900 and the CMOS circuitry in region 905 to provide interconnections. Contacts within the transducer 900, such as contacts 214, 216 ( Figure 2B )) can be connected to the interconnect layer.

[0066] The CMOS circuitry within region 905 includes circuitry to generate an ultrasound signal, which is coupled to transducer 900 via interconnect layers.

[0067] In application, the transducer 900 can be interfaced to various targets in order to assess the targets using ultrasound imaging techniques. For medical imaging, the ultrasound transducer 900 can be interfaced to the human or animal body in order to "see" the body. For simulation purposes, water resembles the composition of the human and animal bodies. As illustrated by waves 911, 912, 913, 914, etc., the ultrasound waves generated by the main waveguide portion 901 and the reflector portion 902 travel down the transition portion 903 and then travel to the radiation region 904. The 715 MHz ultrasound waves propagate through the dielectric that constitutes the radiation region 904 and are then emitted into the surrounding medium. The simulated water droplet 920 illustrates how the emitted ultrasound waves propagate through the water droplet 920 as waves 911, 912, 913, 914, etc.

[0068] The ultrasonic wavelength of a 715 MHz signal in water is approximately 2 μm, which is ideal for high-resolution short-range ultrasound imaging applications. For example, compare this to existing technology using an ultrasound signal with a frequency of 10 MHz; the wavelength in water is much larger, and the resulting image resolution is therefore lower than that derived from a 715 MHz ultrasound signal.

[0069] Figure 10A 、 Figure 10B 、 Figure 10C is used for Figure 9 Schematic diagram of various example FeCap connection schemes for the transducer 900. Figure 10A In the example, all FeCaps in the radiator structure in section 903, in the main waveguide in section 901, and in the reflector in section 902 have their bottom plates connected together and to the ground reference. The top plates of the FeCaps in sections 902 and 903 are left floating. The ultrasonic excitation signal is applied to alternating top plates in section 901. In this example, the excitation signal is provided by CMOS IC 930 ( Figure 9 A positive excitation signal, Vdrive+, is applied to every other FeCap in portion 901 (such as FeCaps 1011 and 1013), while a negative excitation signal, Vdrive-, is applied to the middle FeCaps (such as FeCaps 1010 and 1012).

[0070] exist Figure 10BIn FIG. 9 , all FeCaps in the radiator structure in section 903, in the main waveguide in section 901, and in the reflector in section 902 have their bottom plates connected together and to the ground reference. The top plates of the FeCaps in sections 902 and 903 are tied to the ground reference. The ultrasonic excitation signal is applied to the top plate in section 901. In this example, the excitation signal is generated by CMOS IC 930 ( Figure 9 A positive excitation signal, Vdrive+, is applied to every other FeCap in portion 901 (such as FeCaps 1011 and 1013), while a negative excitation signal, Vdrive-, is applied to the middle FeCaps (such as FeCaps 1010 and 1012).

[0071] exist Figure 10C In the example, all FeCaps in the radiator structure in section 903, in the main waveguide in section 901, and in the reflector in section 902 have their bottom plates connected together and to the ground reference. The top plates of the FeCaps in sections 902 and 903 are tied to the bias voltage reference. The ultrasonic excitation signal is applied to the top plate in section 901. In this example, the excitation signal is generated by CMOS IC 930 ( Figure 9 A positive excitation signal, Vdrive+, is applied to every other FeCap in portion 901 (such as FeCaps 1011 and 1013), while a negative excitation signal, Vdrive-, is applied to the middle FeCaps (such as FeCaps 1010 and 1012).

[0072] With the top plate of the FeCap in the reflector portion 902 kept floating Figure 10A Compared to the connection Figure 10B and Figure 10C The top plate of the FeCap in the reflector section 902 slightly increases the attenuation.

[0073] Figure 11 is used for Figure 9 FIG. 1 is a top view of a linear array of example FeCaps 910 of an ultrasonic transducer 900 in FIG. 1 . FeCaps 1110 represent all of the FeCaps in the linear array 910. FeCaps 1110 are similar to those described with respect to FIG. Figure 2A 、 Figure 2B FeCap 210 is shown in greater detail. In this example, the upper plate of each FeCap is visible, such as upper plate 1106. In this example, a continuous polysilicon layer 1108 spans all of the FeCaps in linear array 910 to form the backplane routing interconnect.

[0074] In this example, the FeCaps are interconnected, as shown in Figure 10B Schematically illustrated, Vdrive+ is coupled to the top plate of every other FeCap in the main waveguide portion 901 via signal line 1115, which is connected to a corresponding first-level contact, such as first-level contact 1114. Similarly, Vdrive- is coupled to the top plate of the intermediate FeCaps in portion 901 via signal line 1116, which is connected to a corresponding first-level contact.

[0075] Figure 12 FIG is a cross-sectional view of an example packaged IC 1240 including a FeCap ultrasonic transducer 1200. The FeCap ultrasonic transducer 1200 is similar to the FeCap transducer 900 ( Figure 9 ) and includes a linear array of FeCaps 1210 arranged into a main waveguide section, a reflector section, and a transition section. Ultrasonic transducer 1200 also includes a radiative medium 1204. Ultrasonic transducer 1200 is fabricated on a CMOS die 1230 using known CMOS manufacturing techniques. Substrate 1231 is bulk silicon. During FEOL processing, various circuitry 1236, including the linear array of FeCaps 1210, is formed on substrate 1231. Metal interconnect layers 1238 and oxide insulating layers are formed during BEOL processing.

[0076] Die bond pads, such as pad 1232 , are also formed on substrate 1231 and interconnected to circuitry 1236 through signal lines formed in one or more metal interconnect layers 1238 of CMOS die 1230 .

[0077] CMOS die 1230 is mounted on die attach pads 1244, which are part of a metal lead frame that includes lead frame bond pads (such as pads 1245) to provide connections to external circuitry. Bond wires (such as bond wire 1233) connect the die bond pads on CMOS die 1230 to corresponding lead frame bond pads.

[0078] CMOS die 1230 is encapsulated by mold compound 1242 using known or later developed packaging techniques. In this example, opening 1246 is provided in the finished package to allow the radiating medium 1204 to interface with an externally provided fluid 1220. For example, packaged IC 1240 can be immersed in a tank or other container containing fluid 1220. In another example, packaged IC 1240 can be part of a medical ultrasound imaging device. In that case, fluid 1220 can be, for example, a liquid or gel that provides an interface between the radiating medium 1204 and the human or animal body.

[0079] In this example, the packaged IC is a quad flat no-lead package. Flat no-lead packages, such as quad flat no-lead (QFN) and dual flat no-lead (DFN), physically and electrically connect the integrated circuit to a printed circuit board. Flat no-lead (also known as micro lead frame (MLF) and SON (small outline no-lead)) is a surface mount technology and is one of several packaging technologies that can connect an IC to the surface of a printed circuit board (PCB) without the need for through-holes. Flat no-lead is a plastic encapsulated package that is close to a chip-scale package made with a planar copper lead frame substrate. Peripheral lands on the bottom of the package provide electrical connections to the PCB. Other examples can be packaged using other known or later developed packaging technologies, such as quad flat packages, ball grid arrays, etc.

[0080] Figure 13 is a flow chart illustrating the operation of an ultrasound transmitter on a CMOS IC. Figure 8 、 Figure 9 As described in greater detail elsewhere, a linear array of FeCaps fabricated on a CMOS IC can be operated as an ultrasonic transmitter. The waveguide portions of the array of FeCaps located in the middle of a row each have a respective width and pitch selected to produce a guided operating mode for a selected ultrasonic frequency. The reflector portion of the array of FeCaps is located at one end of the row. Each FeCap in the reflector portion has a respective width that is greater than the respective width of the FeCaps in the waveguide portion. The radiator portion of the array of FeCaps is located at the other end of the row. Each FeCap in the radiator portion has a respective width that is less than the respective width of the FeCaps in the waveguide portion. The radiator is dielectrically coupled to the radiator portion of the array of FeCaps.

[0081] At 1301, an ultrasonic signal is applied to a waveguide portion of a FeCap. The ultrasonic signal may be generated by circuitry within the IC in which the FeCap resides.

[0082] At 1302, ultrasonic waves are generated by the waveguide portion of the FeCap. The frequency of the ultrasonic signal is selected to match the guided operating mode generated within the waveguide portion of the FeCap.

[0083] At 1303, a portion of the ultrasonic wave is reflected by the reflector portion. The reflected wave enhances the ultrasonic signal formed within the waveguide portion.

[0084] At 1304, a portion of the ultrasound signal is radiated from a radiating structure formed by the radiator portion of the FeCap. To match the acoustic impedance of the radiating structure with that of the radiating medium, the corresponding width of the FeCap in the radiator portion and the size of the structure are gradually reduced.

[0085] At 1305, the radiated portion propagates through a radiating medium, which can be used to guide the radiated wave in a particular direction. For example, the radiating medium can guide the wave to an opening in the package of the IC, such as Figure 12 As shown in the picture.

[0086] As described above, one or more ultrasound transducers and acoustic processing circuitry are integrated within a standard CMOS IC, forming a completely solid structure that contains no air gaps or membranes.

[0087] High frequency operation in the 700 MHz range can be used for very high resolution over short distances to provide high resolution ultrasound imaging.

[0088] Applications include: surface inspection, medical imaging, surgical guidance, internal imaging of veins / arteries, microfluidic flow measurement and metering, insulin / drug microflow metering, blood flow measurement, on-chip acoustic communication, isolation between voltage domains, pain detection, etc.

[0089] Other embodiments

[0090] In the described example, a single ultrasonic transducer is illustrated. However, in another example, two or more ultrasonic transducers may be fabricated on a single CMOS IC. Figure 9 As shown in FIG, the length of the ultrasonic transducer 900 is less than 250 μm, including the radiating medium 904. The plates on the linear array of FeCaps 910 are only 10 μm long. Since the overall size of the ultrasonic transducer 900 is only about 250 μm by 10 μm, many ultrasonic transducers can be manufactured on a single CMOS IC.

[0091] In the example described, the ultrasonic transducer operates at a frequency of 715 MHz. In other examples, the width of the FeCaps in the waveguide, reflector, and radiating portions of the linear array of FeCaps can be selected to produce guided ultrasonic operation across a frequency range of 600-900 MHz or higher. The lowest frequency depends on the available material thickness, while the highest frequency depends on the lithographic capabilities of the technology. Using existing technology, the range can be pushed down to 350 MHz and up to over 1 GHz.

[0092] In the described examples, a single layer of FeCap is illustrated. In another embodiment, multiple layers of FeCap and / or multiple layers of thick tungsten can be implemented using three-dimensional integration techniques. In one example, multiple layers can be fabricated directly on top of each other. In another example, the delay structure can be fabricated in a higher metal layer in the BEOL stack. Radio frequency metal-insulator-metal (RF-MIM) capacitors can be fabricated in the upper BEOL layers to move away from parasitics and losses in the substrate. Using multiple layers of FeCap, RF-MIM capacitors and / or multiple layers of thick tungsten, the frequency range can be extended down to the tens of MHz range. For example, a system can operate at frequencies of 20 MHz or lower using these techniques.

[0093] In the example described, an ultrasonic signal is transmitted from an ultrasonic transmitter. In another example, the same structure can be used to receive ultrasonic signals. By controlling the transmission and reception operations, the same structure can function as a transducer.

[0094] In some examples, only a single transmitting device or a single receiving device may be implemented. In another example, only a single transducer device may be implemented.

[0095] In the example described, contacts in two metal layers are gradually removed to form a transition structure. In another example, there may be three or more layers of interconnect above the linear array of FeCaps. In this case, contacts and / or vias from all layers can be removed in a gradual manner to help gradually reduce the physical size of the FeCap unit to create the transition structure.

[0096] In the example described above, where the metal layer is copper, copper has an acoustic impedance similar to that of oxide. Therefore, the copper contacts and vias have only a minor impact on the acoustic impedance. However, in another example, the metal layer can be aluminum, which has a significantly different acoustic impedance than oxide. In this case, gradually removing the aluminum contacts / vias to form a tapered transition structure has a greater effect. Similarly, gradually varying the size of other types of conductive structures used for interconnects can have different effects on the acoustic impedance.

[0097] In the example described, the respective widths of the FeCaps in the main waveguide section are approximately equal. However, due to normal manufacturing tolerances, the respective widths may vary. In the example, there may be a variation of up to 5%.

[0098] In this specification, the term "couple" and its derivatives refer to an indirect, direct, optical, and / or wireless electrical connection. Thus, if a first device is coupled to a second device, that connection may be through a direct electrical connection, through an indirect electrical connection via other devices and connections, through an optical electrical connection, and / or through a wireless electrical connection.

[0099] Modifications to the described embodiments are possible within the scope of the claims, and other embodiments are possible.

Claims

1. An ultrasonic transducer comprising: substrate; a linear array of ferroelectric capacitors, or FeCaps, arranged in a single row on the substrate, each FeCap having a respective length and a respective width such that the length of each of the array of FeCaps is perpendicular to an axis of the row; a waveguide portion of the array of FeCaps located in the middle of the row, wherein each FeCap in the waveguide portion has a respective first width; a reflector portion of the array of FeCaps located at a first end of the row, wherein each FeCap in the reflector portion has a respective width greater than the first width; and A radiator section of the array of FeCaps is located at a second end of the row opposite the first end, wherein each FeCap in the radiator section has a respective width that is less than the first width. 2 . The ultrasonic transducer according to claim 1 , wherein the corresponding widths of the FeCaps in the radiator portion gradually decrease from the first width to a second width.

3. The ultrasonic transducer of claim 1 , wherein each of the array of FeCaps has a respective first conductive plate and a respective second conductive plate with a ferroelectric material therebetween.

4. The ultrasonic transducer according to claim 3, wherein the first conductive plate of at least one of the FeCaps in the radiator section is connected to at least two contact pads, the first conductive plate of at least one of the FeCaps in the radiator section is connected to only one contact pad, and the first conductive plate of at least one of the FeCaps in the radiator section is not connected to a contact pad. 5 . The ultrasonic transducer according to claim 4 , wherein at least one of the FeCaps in the radiator portion does not have a first conductive plate. 6 . The ultrasonic transducer of claim 1 , wherein the corresponding width of the FeCap in the reflector portion is in a range of 10% to 50% greater than the first width. 7 . The ultrasonic transducer according to claim 6 , wherein the FeCaps in the reflector portion have corresponding widths that gradually increase.

8. A method of operating an ultrasound transmitter, the ultrasound transmitter being implemented as a linear array of FeCaps on a semiconductor substrate, the method comprising: applying an ultrasonic signal to a waveguide portion of a ferroelectric capacitor, namely, FeCap; the waveguide portion of the FeCap is located in the middle of a linear array of the FeCaps on the semiconductor substrate, generating ultrasonic waves in the waveguide portion of the FeCap through piezoelectric operation of the waveguide portion of the FeCap; reflecting a portion of the ultrasonic wave by a reflector portion of a FeCap, the reflector portion of the FeCap being located at a first end of the linear array of FeCaps; and Portions of the ultrasonic waves are radiated from a radiating structure formed by portions of the FeCaps, the radiating structure being located at a second end of the linear array of FeCaps opposite to the first end of the linear array of FeCaps.

9. The method of claim 8, further comprising propagating the radiated portion of the ultrasonic wave through a radiating medium coupled to the radiating structure.

10. The method of claim 8, further comprising guiding the ultrasonic wave along the waveguide portion of the FeCap via a guided operation mode generated by the waveguide portion of the FeCap. The method of claim 8 , wherein the ultrasound signal has a frequency in the range of 20 MHz to 900 MHz.

Images