Fabrication and characterization of piezoelectric micromachined ultrasonic transducers with thick composite PZT films

Ferroelectric microelectromechanical systems (MEMS) has been a growing area of research in past decades, in which ferroelectric films are combined with silicon technology for a variety of applications, such as piezo-electric micromachined ultrasonic transducers (pMUTs), which represent a new approach to ultrasound detection and generation. For ultrasound-radiating applications, thicker PZT films are preferred because generative force and response speed of the diaphragm-type transducers increase with increasing film thickness. However, integration of 4- to 20-microm thick PZT films on silicon wafer, either the deposition or the patterning, is still a bottleneck in the micromachining process. This paper reports on a diaphragm-type pMUT. A composite coating technique based on chemical solution deposition and high-energy ball milled powder has been used to fabricate thick PZT films. Micromachining of the pMUTs using such thick films has been investigated. The fabricated pMUT with crack-free PZT films up to 7-microm thick was evaluated as an ultrasonic transmitter. The generated sound pressure level of up to 120 dB indicates that the fabricated pMUT has very good ultrasound-radiating performance and, therefore, can be used to compose pMUT arrays for generating ultrasound beam with high directivity in numerous applications. The pMUT arrays also have been demonstrated.     

Hybrid multi/single layer array transducers for increased signal-to-noise ratio

In medical ultrasound imaging, two-dimensional (2-D) array transducers are necessary to implement dynamic focusing in two dimensions, phase correction in two dimensions and high speed volumetric imaging. However, the small size of a 2-D array element results in a small clamped capacitance and a large electrical impedance, which decreases the transducer signal-to-noise ratio (SNR). We have previously shown that SNR is improved using transducers made from multi-layer PZT, due to their lower electrical impedance. In this work, we hypothesize that SNR is further increased using a hybrid array configuration: in the transmit mode, a 10 Omega electronic transmitter excites a 10 Omega multi-layer array element; in the receive mode, a single layer element drives a high impedance preamplifier located in the transducer handle. The preamplifier drives the coaxial cable connected to the ultrasound scanner. For comparison, the following control configuration was used: in the transmit mode, a 50 Omega source excites a single layer element, and in the receive mode, a single layer element drives a coaxial cable load. For a 5x102 hybrid array operating at 7.5 MHz, maximum transmit output power was obtained with 9 PZT layers according to the KLM transmission line model. In this case, the simulated pulse-echo SNR was improved by 23.7 dB for the hybrid configuration compared to the control. With such dramatic improvement in pulse-echo SNR, low voltage transmitters can be used. These can be fabricated on integrated circuits and incorporated into the transducer handle.     

Comparison of conventional and collapsed region operation of capacitive micromachined ultrasonic transducers

We report experimental results from a comparative study on collapsed region and conventional region operation of capacitive micromachined ultrasonic transducers (CMUTs) fabricated with a wafer bonding technique. Using ultrasonic pulse-echo and pitch-catch measurements, we characterized single elements of 1-D CMUT arrays operating in oil. The experimental results from this study agreed with the simulation results: a CMUT operating in the collapsed region produced a higher maximum output pressure than a CMUT operated in the conventional region at 90% of its collapse voltage (3 kPa/V vs. 16.1 kPa/V at 2.3 MHz). While the pulse-echo fractional bandwidth (126%) was higher in the collapsed region operation than in the conventional operation (117%), the pulse-echo amplitude in collapsed region operation was 11 dB higher than in conventional region operation. Furthermore, within the range of tested bias voltages, the output pressure monotonously increased with increased bias during collapsed region operation. It was also found that in the conventional mode, short AC pulses (larger than the collapse voltage) could be applied without collapsing the membranes. Finally, while no significant difference was observed in reflectivity of the CMUT face between the two regions of operation, hysteretic behavior of the devices was identified in the collapsed region operation.     

Element shape design of 2-D CMUT arrays for reducing grating lobes

Fully populated two-dimensional (2-D) arrays are needed to produce high quality ultrasonic volumetric images for real-time applications, but they present many challenges for their physical realization because of the large number of elements. In fact, lambda/2 and lambda minimum spacing between elements is required, respectively, for pyramidal and rectilinear scanning in order to avoid unwanted grating lobes (GLs). However, in past years, capacitive micromachined ultrasonic transducer (CMUT) technology has made possible the production of arrays with large flexibility in element shape and size. In this paper, this property is analyzed, and a new element shape, based on the concept of spatial interpenetration of adjacent elements, is proposed in order to design fully populated 2-D CMUT arrays with a low number of elements, whose beam characteristics are valid for volumetric imaging. Through the use of simulations, it is demonstrated that arrays with pitch larger than lambda (up to 3lambda) used for rectilinear scanning, have notably lower GLs than the equivalent standard arrays designed according to the classical squared element shape. As consequence, the proposed geometry has the advantage of reducing the number of elements (up to a factor of 9) and of enlarging the element size, implying an increase of the SNR relative to the single element. When beam steering is required, arrays can be designed with pitch equal to lambda, reducing the number of elements by a factor of 4 if the maximum steering angle is limited to +/-15 degrees .     

Multilayer piezoelectric ceramics for two-dimensional array transducers

In medical ultrasound imaging, 2-D array transducers have become essential to implement dynamic focusing and phase-correction in the elevation dimension as well as real-time volumetric scanning. Unfortunately, the small size of a 2-D array element results in a small clamped capacitance and a large electrical impedance near resonance. These elements have poor sensitivity because their impedance is much higher than the electrical impedance of the transmit and receive circuitry. Sensitivity can be improved by using an N layer structure of PZT ceramic with the layers connected acoustically in series and electrically in parallel. For the multilayer ceramic (MLC), the damped capacitance is multiplied by a factor of N(2) and the electrical impedance by 1/N(2) compared to a single layer element of the same dimensions. A 3x43 phased-array transducer has been fabricated using 3 layer PZT-5H material. Each element had a thickness of 0.66 mm and an area of 0.37x3.5 mm. The MLC was manufactured using thick film technology with plated-through vias to electrically interconnect the electrode layers. The completed transducer was compared to a single layer control array of similar dimensions. With a light epoxy backing and a lambda/4 matching layer, the MLC array elements had an impedance of 100 Omega at series resonance of 2.25 MHz, compared to 800 Omega for the control elements. The lower impedance of the MLC elements resulted in a minimum round-trip insertion loss of 24.0 dB, compared to an 34.1 dB for the control array elements. These results were consistent with KLM modeling. B-scan images were made of cysts in a tissue-mimicking phantom and of the left kidney in vivo. The images clearly showed a higher signal-to-noise ratio for the MLC array compared to the control. As a result, 2-D arrays made of multilayer ceramics can be used to form images at a higher frequency and greater range than single layer arrays.     

An integrated circuit with transmit beamforming flip-chip bonded to a 2-D CMUT array for 3-D ultrasound imaging

State-of-the-art 3-D medical ultrasound imaging requires transmitting and receiving ultrasound using a 2-D array of ultrasound transducers with hundreds or thousands of elements. A tight combination of the transducer array with integrated circuitry eliminates bulky cables connecting the elements of the transducer array to a separate system of electronics. Furthermore, preamplifiers located close to the array can lead to improved receive sensitivity. A combined IC and transducer array can lead to a portable, high-performance, and inexpensive 3-D ultrasound imaging system. This paper presents an IC flip-chip bonded to a 16 x 16-element capacitive micromachined ultrasonic transducer (CMUT) array for 3-D ultrasound imaging. The IC includes a transmit beamformer that generates 25-V unipolar pulses with programmable focusing delays to 224 of the 256 transducer elements. One-shot circuits allow adjustment of the pulse widths for different ultrasound transducer center frequencies. For receiving reflected ultrasound signals, the IC uses the 32-elements along the array diagonals. The IC provides each receiving element with a low-noise 25-MHz-bandwidth transimpedance amplifier. Using a field-programmable gate array (FPGA) clocked at 100 MHz to operate the IC, the IC generated properly timed transmit pulses with 5-ns accuracy. With the IC flip-chip bonded to a CMUT array, we show that the IC can produce steered and focused ultrasound beams. We present 2-D and 3-D images of a wire phantom and 2-D orthogonal cross-sectional images (Bscans) of a latex heart phantom.     

Modeling of piezoelectric multilayer ceramics using finite elementanalysis

For medical ultrasound imaging, 2-D array transducers have greater versatility than linear arrays. Unfortunately, the tiny array elements in a 2-D array have poor signal-to-noise ratio (SNR). We have previously shown that SNR is increased in 2-D array transducers made from piezoelectric multilayer ceramics. Conventional one-dimensional models provide accurate results when comparing multilayer ceramic performance relative to single layer transducers. However, these models are not accurate when comparing simulations directly to measurements. Because multilayer ceramics have a complex structure, a 3-D model, such as finite element analysis, is needed for accurate simulations. We modeled four arrays that were previously fabricated: a single layer and multilayer 1 MHz, 2-D array element, and a single layer and multilayer 2.25 MHz, 1.5-D array element that can focus and steer in azimuth but only steer in the elevation dimension. We compared the simulated and measured impedance plots for each transducer. The finite element analysis plots accurately predicted the impedance for each vibration mode. On the other hand, the one dimensional KLM transmission line model could simulate only the thickness mode vibrations and the results were inaccurate compared to measurements. We also simulated the transmit output pressure for the 2.25 MHz arrays and compared the results to measurements. The simulated pressure vs. time plots and their spectra were accurate when compared to measurements. Finally, we obtained a series of images that show the impulse response vibrations for the 2.25 MHz, arrays. These animations show the vibration modes in the complex multilayer ceramic structure. Measurements were not available to confirm the animations. Our results show that finite element analysis in three dimensions is a valuable tool to predict the performance of multi-layer transducers     

Finite-element analysis of capacitive micromachined ultrasonic transducers

In this paper, we present the results of finite-element analysis performed to investigate capacitive micromachined ultrasonic transducers (CMUTs). Both three-dimensional (3-D) and 2-D models were developed using a commercially available finite-element modeling (FEM) software. Depending on the dimensionality of the model, the membranes were constructed using plane or shell elements. The electrostatic gap was modeled using many parallel plate transducers. An axisymmetric model for a single membrane was built; the electrical input impedance of the device then was calculated in vacuum to investigate series and parallel resonant frequencies, where the input impedance has a minimum and a maximum, respectively. A method for decomposing the membrane capacitance into parasitic and active parts was demonstrated, and it was shown that the parallel resonant frequency shifted down with increased biased voltage. Calculations then were performed for immersion transducers. Acoustic wave propagation was simulated in the immersion medium, using appropriate elements in a 3-D model. Absorbing boundaries were implemented to avoid the reflections at the end of the medium mesh. One row of an array element, modeled with appropriate boundary conditions, was used to calculate the output pressure. The results were compared with a simpler model: a single membrane in immersion, with symmetry boundary conditions on the sidewalls that cause the calculations to reflect the properties of an infinitely large array. A 2-D model then was developed to demonstrate the effect of membrane dimensions on the output pressure and bandwidth. Our calculations revealed that the small signal transmit pressure was inversely proportional to the square root of gap height. We also compared FEM results with analytical and experimental results.     

High-power CMUTs: design and experimental verification

Capacitive micromachined ultrasonic transducers (CMUTs) have great potential to compete with piezoelectric transducers in high-power applications. As the output pressures increase, nonlinearity of CMUT must be reconsidered and optimization is required to reduce harmonic distortions. In this paper, we describe a design approach in which uncollapsed CMUT array elements are sized so as to operate at the maximum radiation impedance and have gap heights such that the generated electrostatic force can sustain a plate displacement with full swing at the given drive amplitude. The proposed design enables high output pressures and low harmonic distortions at the output. An equivalent circuit model of the array is used that accurately simulates the uncollapsed mode of operation. The model facilities the design of CMUT parameters for high-pressure output, without the intensive need for computationally involved FEM tools. The optimized design requires a relatively thick plate compared with a conventional CMUT plate. Thus, we used a silicon wafer as the CMUT plate. The fabrication process involves an anodic bonding process for bonding the silicon plate with the glass substrate. To eliminate the bias voltage, which may cause charging problems, the CMUT array is driven with large continuous wave signals at half of the resonant frequency. The fabricated arrays are tested in an oil tank by applying a 125-V peak 5-cycle burst sinusoidal signal at 1.44 MHz. The applied voltage is increased until the plate is about to touch the bottom electrode to get the maximum peak displacement. The observed pressure is about 1.8 MPa with -28 dBc second harmonic at the surface of the array.     

Forward-viewing CMUT arrays for medical imaging

This paper reports the design and testing of forward-viewing annular arrays fabricated using capacitive micromachined ultrasonic transducer (CMUT) technology. Recent research studies have shown that CMUTs have broad frequency bandwidth and high-transduction efficiency. One- and two-dimensional CMUT arrays of various sizes already have been fabricated, and their viability for medical imaging applications has been demonstrated. We fabricated 64-element, forward-viewing annular arrays using the standard CMUT fabrication process and carried out experiments to measure the operating frequency, bandwidth, and transmit/receive efficiency of the array elements. The annular array elements, designed for imaging applications in the 20 MHz range, had a resonance frequency of 13.5 MHz in air. The immersion pulse-echo data collected from a plane reflector showed that the devices operate in the 5-26 MHz range with a fractional bandwidth of 135%. The output pressure at the surface of the transducer was measured to be 24 kPa/V. These values translate into a dynamic range of 131.5 dB for 1-V excitation in 1-Hz bandwidth with a commercial low noise receiving circuitry. The designed, forward-viewing annular CMUT array is suitable for mounting on the front surface of a cylindrical catheter probe and can provide Doppler information for measurement of blood flow and guiding information for navigation through blood vessels in intravascular ultrasound imaging.     

Fabrication and characterization of transducer elements in two-dimensional arrays for medical ultrasound imaging

Some of the problems of developing a two-dimensional (2-D) transducer array for medical imaging are examined. The fabrication of a 2-D array material consisting of lead zirconate titanate (PZT) elements separated by epoxy is discussed. Ultrasound pulses and transmitted radiation patterns from individual elements in the arrays are measured. A diffraction theory for the continuous wave pressure field of a 2-D array element is generalized to include both electrical and acoustical cross-coupling between elements. This theory can be fit to model the measured radiation patterns of 2-D array elements, giving an indication of the level of cross-coupling in the array, and the velocity of the acoustic cross-coupling wave. Improvements in bandwidth and cross-coupling resulting from the inclusion of a front acoustic matching layer are demonstrated, and the effects of including a lossy backing material on the array are discussed. A broadband electrical matching network is described, and pulse-echo waveforms and insertion loss from a 2-D array element are measured.     

Calculation of ultrasonic transducer signal-to-noise ratios usingthe KLM model

This paper describes a method for adding thermal and amplifier noise to a KLM model for a transducer element. The model is used to compare the magnitudes of various noise sources in a 5 MHz array element typical of those used for linear array imaging with and without an amplifier. Fundamental signal-to-noise ratio (SNR) issues of importance to array and amplifier designers are explored, including the effect on SNR of effective dielectric constant of the piezoelectric material, individual element size, changing the number of elements, and adding an amplifier to an element before and after a cable. SNR is considered both for the case in which the acoustic output is limited by the maximum rarefactive pressure which is considered safe for a particular application (Mechanical Index limitation) and the case in which acoustic output is limited by the maximum transmit voltage than can he delivered by the imaging system or tolerated by the transducer. It is shown that the SNR performance depends on many controllable parameters and that significant improvements in SNR can be achieved through proper design. The implications for 1.5-D and 2-D array elements are discussed     

Sub-Structural Parameter Identification Including Cracks of Beam Structure Using PZT Patch

This paper presents the numerical simulation of damage identification of beam structures using novel One Dimensional (1-D) PZT patch model with sub-structuring applied to the PZT Patch boundaries. The 1-D PZT patch model is simple and requires less computational effort than 2-D or 3-D models. A hybrid element constituted of 1-D beam element and a PZT sensor is used with reduced material properties which is very convenient for beams and is a novel application. The accuracy of 1-D hybrid element with cracked structure is first verified experimentally. The stiffness, damping and crack parameters are identified by minimizing the deviation between the predicted and measured voltage from the PZT patch using an optimization algorithm. The numerical signals are polluted with 5% Gaussian noise to simulate experimental noise. Numerical studies are performed on a beam and a nine member frame structure for sub-structural parameter and multiple crack identification. The PZT patches are attached at either ends of beam member to be identified. Sub-structuring is done to isolate the beam with patch, thus reducing the size of model to be identified. The numerical results show significant improvement in identification accuracy compared to other methods.     

Micro-machined high-frequency (80 MHz) PZT thick film linear arrays

This paper presents the development of a micromachined high-frequency linear array using PZT piezoelectric thick films. The linear array has 32 elements with an element width of 24 μm and an element length of 4 mm. Array elements were fabricated by deep reactive ion etching of PZT thick films, which were prepared from spin-coating of PZT sol-gel composite. Detailed fabrication processes, especially PZT thick film etching conditions and a novel transferring-and-etching method, are presented and discussed. Array designs were evaluated by simulation. Experimental measurements show that the array had a center frequency of 80 MHz and a fractional bandwidth (-6 dB) of 60%. An insertion loss of -41 dB and adjacent element crosstalk of -21 dB were found at the center frequency.     

An approach for reducing adjacent element crosstalk in ultrasound arrays

A method is presented for active cancellation of crosstalk effects in ultrasonic arrays. The approach makes use of the programmable transmitter waveform generators that are now being used with growing prevalence in diagnostic ultrasound systems. The array's transmit mode transfer function is represented by a transfer function matrix. Elements of this matrix are determined by exciting a single, central element with a wideband waveform and determining the resulting pressure output from the central element and adjacent elements. The desired output then is defined (e.g., finite output from a single, central element) and zero output from all other elements. The transfer function matrix equation can be solved to determine the required excitation functions on both the central array element and its neighbors. These excitation functions result in reduced evidence of crosstalk on the output signals. Therefore, the single-element, angular-response function is improved. Using superposition, the approach can be extended to beamformed array excitation. A variety of theoretical and experimental results are shown. The method also can be used in the receive mode but with a less satisfactory solution. A transmitting mode experiment based on a prototype five-element transducer has provided results indicating that sidelobes in the angular response can be reduced using this technique.     

Volumetric ultrasound imaging using 2-D CMUT arrays

Recently, capacitive micromachined ultrasonic transducers (CMUTs) have emerged as a candidate to overcome the difficulties in the realization of 2-D arrays for real-time 3-D imaging. In this paper, we present the first volumetric images obtained using a 2-D CMUT array. We have fabricated a 128 x 128-element 2-D CMUT array with through-wafer via interconnects and a 420-microm element pitch. As an experimental prototype, a 32 x 64-element portion of the 128 x 128-element array was diced and flip-chip bonded onto a glass fanout chip. This chip provides individual leads from a central 16 x 16-element portion of the array to surrounding bondpads. An 8 x 16-element portion of the array was used in the experiments along with a 128-channel data acquisition system. For imaging phantoms, we used a 2.37-mm diameter steel sphere located 10 mm from the array center and two 12-mm-thick Plexiglas plates located 20 mm and 60 mm from the array. A 4 x 4 group of elements in the middle of the 8 x 16-element array was used in transmit, and the remaining elements were used to receive the echo signals. The echo signal obtained from the spherical target presented a frequency spectrum centered at 4.37 MHz with a 100% fractional bandwidth, whereas the frequency spectrum for the echo signal from the parallel plate phantom was centered at 3.44 MHz with a 91% fractional bandwidth. The images were reconstructed by using RF beamforming and synthetic phased array approaches and visualized by surface rendering and multiplanar slicing techniques. The image of the spherical target has been used to approximate the point spread function of the system and is compared with theoretical expectations. This study experimentally demonstrates that 2-D CMUT arrays can be fabricated with high yield using silicon IC-fabrication processes, individual electrical connections can be provided using through-wafer vias, and flip-chip bonding can be used to integrate these dense 2-D arrays with electronic circuits for practical 3-D imaging applications.     

Characterization of one-dimensional capacitive micromachined ultrasonic immersion transducer arrays

We report on the characterization of 1D arrays of capacitive micromachined ultrasonic transducers (cMUT). A 275×5600 μm 1D CMUT array element is experimentally characterized, and the results are found to be in agreement with theoretical predictions. As a receiver, the transducer has a 0.28-fm/√Hz displacement sensitivity, and, as a transmitter, it produces 5 kPa/V of output pressure at the transducer surface at 3 MHz with a DC bias of 35 V. The transducer has more than 100% fractional bandwidth around 3 MHz, which makes it suitable for ultrasound imaging. The radiation pattern of isolated single elements, as well as those of array elements are measured, and two major sources of acoustical crosstalk are identified. A weakly dispersive non-leaky interface wave (Stoneley wave) is observed to be propagating at the silicon substrate-fluid interface at a speed close to the speed of sound in the fluid. This wave causes internal reflections, spurious resonance, and radiation from the edges of the silicon substrate. The large lateral component of the particle velocity generated by the membranes at the edge of the cMUT array elements is found to be the source of this interface wave. Lowest order Lamb waves in the silicon substrate are also found to contribute to the crosstalk between elements. These waves are excited at the edges of individual vibrating membranes, where they are anchored to the substrate, and result in a narrowing of the beam profile of the array elements. Several methods, such as trench isolation and wafer thinning, are proposed and implemented to modify the acoustical cross coupling between array elements     

Evaluation of wafer bonded CMUTs with rectangular membranes featuring high fill factor

Increasing fill factor is one design approach used to increase average output displacement, output pressure, and sensitivity of capacitive micromachined ultrasonic transducers (CMUTs). For rectangular cells, the cell-to-cell spacing and the aspect ratio determine the fill factor. In this paper, we explore the effects of these parameters on performance, in particular the nonuniformity of collapse voltage between neighboring cells and presence of higher order modes in air or immersed operation. We used a white light interferometer to measure nonuniformity in deflection between neighboring cells. We found that reducing the cell-to-cell spacing could cause bending of the center support post, which amplifies nonuniformities in collapse voltage to 18.4% between neighboring cells. Using a 2-D finite element model (FEM), we found that for our designs, increasing the support post width to 1.67 times the membrane thickness alleviated the post bending problem. Using impedance and interferometer measurements to observe the effects of aspect ratio on higher order modes, we found that the (1,3) modal frequency approached the (1,1) modal frequency as the aspect ratio of the rectangles increased. In air operation, under continuous wave (CW) excitation at the center frequency, the rectangular cells behaved in the (1,1) mode. In immersion, because of dispersive guided modes, these cells operated in a higher order mode when excited with a CW signal at the center frequency. This contributed to a loss of output pressure; for this reason our rectangular design was unsuitable for CW operation in immersion.     

A planar unimorph-based actuator with large vertical displacement capability. I. Experiment

A piezoelectric actuator utilizing a planar arrangement of unimorph elements that produces displacements perpendicular to the plane of the actuator is described. The elements are connected in series mechanically so that the vertical displacement of each element adds to the vertical displacements of the other elements. Two prototype actuators were built and tested. One was stick-built from individual bars of PZT and aluminum, which were epoxied together and connected to each other with short aluminum connector bars. The other was fabricated monolithically from a single plate of PZT by bead-blasting the PZT plate through a steel shadowmask to mill out the desired actuator shape. Copper bars epoxied to the milled-out PZT bars then formed the unimorph elements. The unloaded output displacement of each actuator versus applied voltage was measured as well as the displacement versus applied force with no applied voltage. These measurements were in agreement with the predictions of an electromechanical model of the actuator that has been developed. The model predicted and measurements verified that stiffeners can be added to the basic geometry that will significantly increase the force output without affecting the displacement versus applied voltage characteristic.     

Beam steering with segmented annular arrays

Two-dimensional (2-D) arrays of squared matrix have maximum periodicity in their main directions; consequently, they require half wavelength (lambda/2), interelement spacing to avoid grating lobes. This condition gives rise to well-known problems derived from the huge number of array elements and from their small size. In contrast, 2-D arrays with curvilinear configuration produce lower grating lobes and, therefore, allow the element size to be increased beyond lambda/2. Using larger elements, these arrays have the advantage of reducing the number of elements and of increasing the signal-to-noise ratio (SNR). In this paper, the beamforming properties of segmented annular phased arrays are theoretically analyzed and compared with the equivalent squared matrix array. In the first part, point-like elements are considered in order to facilitate the field analysis with respect to the array structure. Afterward, the effect of the element size on the steered beam properties also is presented. In the examples, it is shown that the segmented annular array has notably lower grating lobes than the equivalent squared matrix array and that it is possible to design segmented annular arrays with interelement distance higher than lambda whose beam characteristics are perfectly valid for volumetric imaging applications.