Phased-Array Transducer for Intracardiac Echocardiography Based on 1–3 Piezocomposite

In this study, an ultrasonic phased-array transducer was proposed, which could effectively improve the imaging performance by using 1–3 piezocomposite. The piezocomposite consists of PZT and epoxy, with a pitch of 70 μm, kerf of 20 μm, and thickness of 170 μm. The phased-array transducer has 64 elements; the size of each element is 85 μm × 1.3 mm; the pitch of the transducer is 100 μm; and the kerf between the elements is only 15 μm. To minimize the transducer size, the 1–3 composite uses an encase structure, which connects the upper surface of the composite directly to the flexible circuit board bonded to the lower surface as the ground electrode. The size of the final fabricated transducer is 2 mm × 7.4 mm, and the transducer is mounted on a 9 F (3 mm diameter) catheter, which can bend in four directions and is primarily used for intracardiac echocardiography (ICE). The acoustic and electrical properties of the transducer were tested, including impedance, echo sensitivity, center frequency (9 MHz), bandwidth (BW) (55%), and consistency. Finally, the wire phantom experiments were carried out to demonstrate the spatial resolutions and imaging performance. This study shows that this transducer with compact design and construction can bring higher performance for the single-use disposable ICE catheter.


INTRODUCTION
Intracardiac echocardiography (ICE) has become a widely used imaging tool in the past decades (Hijazi et al., 2009). The visualization of the heart becomes possible due to the application of ICE and ICE has become an indispensable equipment for various percutaneous, interventional, and electrophysiological procedures (Bartel et al., 2014;Enriquez et al., 2018). ICE catheters can be divided into rotational catheters and phased-array catheters (Vitulano et al., 2015). Among them, the phased-array catheter has wider application potential because of its higher frequency range and ability of Doppler and color flow imaging. With the large-scale application of ICE, it is essential to develop a single-use disposable phased-array catheter with excellent performance and low cost (Bartel et al., 2014).
The miniaturized phased-array transducer has a complicated and challenging fabrication process because the pitch and the kerf must be small enough to avoid spurious modes.
Currently, many researchers have focused their efforts on the development of miniaturized phased-array transducer (Bezanson et al., 2014;Chiu et al., 2014;Basij et al., 2019;Cabrera-Munoz et al., 2019). The current manufacturing process of the transducer is rather cumbersome. For example, sputtering Au/Cr layer needs to be performed multiple times to connect the ground of array elements (Cannata et al., 2006), and individual element electrodes are separated by removing the gold layer by using photolithography or Cr/Au etching and carefully aligned with the electrode patterns on the flexible circuit (Chen et al., 2014;Chiu et al., 2014), using conductive epoxy to bond the bent ground electrode on the circuit to the matching layer (Cabrera-Munoz et al., 2019). All these extra processes not only increase the cost of the transducer but also cause the instability of the electrical connection.
The active material is another critical restricting factor for the ICE phased-array probe. The piezoelectric ceramic (Cannata et al., 2008;Kim H. H. et al., 2010) and single crystal (Sung Min et al., 2003;Chen et al., 2012;Wong et al., 2017) have been widely used in array transducer applications. However, these two homogeneous materials will bring lateral mode to interfere between phased-array elements (Yang et al., 2012). Besides, the piezoelectric ceramic is hard to fulfill the requirements of high performance for the phased-array probe though it has stable performance and cost effective (Kim K. B. et al., 2010). In addition, the single crystal is limited by its delicate and expensive cost for disposable catheter applications (Luo et al., 2010). At the same time, 1-3 piezoelectric composite has minimum lateral mode interference, increased electromechanical coupling efficiency, and better acoustic matching (Brown et al., 2007). It is more suitable in the application of ICE phasedarray probe.
In this study, a high-performance 1-3 composite is employed to realize an ICE 64-element phased-array transducer. The encase structure is adopted to minimize the transducer size and reduce the complexity of the fabrication process. The working frequency of the probe is chosen as 9 MHz to achieve a higher spatial resolution and ensure a penetration depth of at least 10 cm, which can cover most of the applications of ICE (Alkhouli et al., 2018). The design, fabrication, and characterization of the phased array are then presented. The wire phantom imaging experiments are carried out to demonstrate the spatial resolutions and imaging performance.

1-3 Piezocomposite Material
The 1-3 piezocomposite was processed by the "dice-and-fill" technique; a mechanical cutting saw (DAD3221, Disco Co., Tokyo, Japan) was used to cut kerfs into a piece of bulk piezoelectric ceramic; and the kerfs were backfilled with epoxy. In the 1-3 composite, the generation of spurious resonances within the designed operating bandwidth (BW) should be avoided.
To avoid the problems with lateral modes, the pillars in the composite should have a width that is lower than half of the pillar height. The generation of the first two spurious Lamb wave mode frequencies, f L1 and f L2 , can be predicted using the following equation: where υ phase is the phase velocity in the transverse direction across the piezocomposite and d is the pillar-to-pillar spacing within the composite. The phase velocity in low fractional ceramicvolume composites is predominantly determined by the shear wave velocity of the filler epoxy. To push the lateral modes outside the operating BW of the transducer, the composite in this study is designed with 50 µm height, 20 µm width (Brown et al., 2007), and 51% volume fraction. The composite was finally grounded to a thickness of 170 µm. The PZT-5H (3203HD, CTS [Tianjin] Electronics Company Ltd., Tianjin, China) was chosen as the active material of 1-3 composite because of its wide application and high price-performance ratio. A simulation model was built to analyze the piezoelectric characters of the 1-3 composite material, as shown in Figure 1, a part of the composite material structure was built, and its outer boundary was set as symmetric to simulate the whole characteristics of composite material. For the phased-array transducer with central frequency around 9 MHz, the element pitch should be less than λ (wavelength in water, 167 µm), which is 100 µm in our design, and the kerf is 15 µm. In order to make the distribution of the piezoelectric column in each array element as consistent as possible, the element pitch value needs to be designed to equal cutting kerf plus piezoelectric pillar width (Zhou et al., 2011(Zhou et al., , 2020Zhang et al., 2020), and the composite material is diced in a direction parallel to the edge of the pillars. This traditional method limits the design of element size. When the material is diced parallel to the edge of the ceramic pillar, due to the epoxy bumps at the interface between the ceramic pillar and bulk epoxy caused by the differences in the mechanical properties of ceramic and polymers (Cochran et al., 2006;Boonruang et al., 2019), the Frontiers in Materials | www.frontiersin.org bumps are easy to cause the fracture of the electrode Au/Cr layer, so a more suitable dicing method need to be designed.
In this study design, to make the distribution of the piezoelectric column in each array element as consistent as possible, it is necessary to cut the composite material along with a specific angle.
The specific angle is calculated as follows: (1) Select a 1 as the initiated centered position and draw a circle with a radius of 100 µm.
(2) Look for points that are most similar to a 1 . As shown in Figure 2A, point a 2 has the same location features as point a 1 .
(3) Draw a tangent line to the circle along point a 2 . The angle between the tangent line and the horizontal direction is the specific dicing angle.
It is straightforward that the optimal angle can be calculated as 46 • . After dicing the material and sputtering the Au/Cr layers on the surface of the 1-3 composite, the designed material was diced into a size of 1.3 × 7.4 mm, with an encase electrode layer structure, as shown in Figure 2B, which was connected along the short side from the upper surface to the lower surface.
Each element of the transducer has an area of 0.085 mm × 1.3 mm and is composed of several ceramic pillars and epoxy. The consistency of the array elements depends on the active volume of the pillars contained in each array element, and it can be calculated using AutoCAD software (Auto CAD 2012, Autodesk Inc., San Rafael, CA, United States), as shown in Figure 2C. Compared with the dicing method parallel to the edge of the pillar (0 • ) or the diagonal of the pillar (45 • ), the current method can achieve higher consistency, with a volume fraction of 50.95% ± 1.61% of piezoelectric ceramic.

Array Acoustic Stack
The phased-array transducer was fabricated using a novel transducer technology, and it had a total active azimuth aperture of 6.4 mm and an active elevation aperture of 1.3 mm.
As shown in Figure 3, first, both sides of the encase structure of composite were cleaned by applying the acetone and reagent alcohol with a cotton swab, and then the structure was bonded    (FPC), which was consisted of a polyimide base material and a layer of copper foil. After cured overnight at 50 • C, the dicing process was performed by using a 15-µm width dicing saw from the FPC side and the depth of dicing was set to 0.12 µm into the center of the composite material. Finally, a compound mixed with epoxy resin and tungsten powder was employed as the backing layer; its weight ratio was 3:7; and after removing bubbles by vacuuming, the backing material was poured on the backside of the transducer and cured at 60 • C for 12 h. It was cast on the acoustic stack. The Au/Cr layer on the upper surface of the composite as a ground electrode was connected to the FPC board bonded to the lower surface in order to lead out through the traces on the flexible circuit. The array design parameters and material properties are shown in Table 1. The acoustic impedances of the materials were measured through a pulse-echo method (Zhangjian et al., 2020).
The composite was diced in an oblique direction of 46 • , which caused multiple pillars half-cut in each element although the material is composite. However, several phased-array transducers have been designed in many studies with the half-cut technology, and the beam performance test indicated that the focusing and steering ability was satisfactory (Zhang et al., 2011;Bezanson et al., 2014Bezanson et al., , 2020. Compared with wire bonding the array elements to the thickness dimension of flexible circuit boards or other methods, the acoustic stack with encase structure and the half-cut elements reduced the interconnect size, allowing for significantly reduced cost and complexity of manufacturing.

Probe Packaging
The acoustic stack was assembled at the distal end of an ICE catheter; the pads on the FPC end were welded to the 48 American Wire Gauge (AWG) coaxial cable inserted inside the catheter, as shown in Figure 4A; and the other end of the cable was connected to the breakout board of an imaging system by a customized connector.
The outside of the transducer except for the imaging window was cast with Pebax 2533 material for medical applications. The custom-designed catheter (ICE, Xinhuajia Technology Co., Shenzhen, China) was a long, thin, flexible multilayer plastic tubes, with four pull wires spaced apart 90 • along the cross section connect the distal end to handle, and had an insertable length of 900 mm, as shown in Figure 4B. The catheter was manually manipulated by grasping the catheter handle and actuating each of the four degrees of freedom (left/right and anterior/posterior), and able to bend up to an angle of 160 • , with a minimum bending radius of 27 mm.
The fabricated ICE transducer, as shown in Figure 5A, is the structure of 1-3 composite materials under a microscope, and Figure 5B shows a customized flexible circuit board, which is mainly composed of two parts, namely the transducer assembly Frontiers in Materials | www.frontiersin.org part and the characteristic detection part, where the latter is used to test the characteristics of the transducer and will be removed in the subsequent process. Figure 5C shows the manufactured transducer, the remaining FPC board can be folded and cast with Pebax 2533 as protection, and the total structural size of the transducer is 3 mm in diameter, as shown in Figure 5D.
A bundle of 64 individual 48-AWG micro-coaxial cables with a total outer diameter of 1.9 mm is carefully soldered to the transducer as shown in Figure 6, and the other ends of the cables are divided into four groups where each group is soldered on a custom L-shaped circuit board, which can be connected to a commercial Verasonics Vantage 128 System (Vantage 128, Verasonics Inc., Kirkland, WA, United States).

Performance of the 1-3 Composite Material
The simulation result and the test result of the composite material are shown in Figure 7. The simulation impedance curves are almost the same as the measurement result. It can be noticed that the magnitude value of the simulation impedance curve is much larger than the measurement result, and the reason is due to that the area of the simulation model is much smaller than the tested area.
According to the measured electrical impedance spectrum, the center frequency of the fabricated 1-3 composite is 10.2 MHz, and its electromechanical coupling coefficient (k t ) is 0.66, which is higher than 0.55 of pure PZT-5H (CTS, 2021). Besides, according to the 1-3 composite thickness mode (Smith and Auld, 1991), the main piezoelectric properties of the 1-3 composite including longitudinal velocity, density, and acoustic impedance are listed in Table 2.

Basic Performance of Phased-Array Transducer
Each of the elements was repolarized in the polarizing fluid under an electric field of 2 kV/cm for 3 min using a high-voltage power supply before testing. Then the electrical impedance was measured using an impedance analyzer (E4991A, Keysight Technologies Inc., Santa Rosa, CA, United States) before  connecting each element to the coaxial cable assembly; both the impedance magnitude and phase angle were recorded over the range of 5-15 MHz.
The measured values of all 64 elements are shown in Figure 8, and the average value and SD of electrical impedance at 9 MHz were 288.48 ± 27.24 and 72.35 • ± 2.92 • , respectively. Since the electrical impedance was measured by connecting each pad on the transducer to the impedance analyzer, a 20-mm homemade electrical probe was used in the test and brought the smaller resonance peak located at around 6 MHz.
After completing the cable welding and fabrication of the catheter, the pulse-echo response was performed in deionized water at room temperature for the characterization. A device (DPR 500, JSR Ultrasonics Inc., Pittsford, NY, United States) with a remote pulser RP-H4 was used to excite the elements of the transducer: the voltage amplitude of the pulser was set as 330 V, the PRF was 200 Hz, and the echo was received by the device with 0 dB gain. The waveforms were recorded and processed by an oscilloscope (DPO5034, Tektronix Inc., Beaverton, OR, United States). Each element was connected in series with a 1.2 µH inductor for impedance matching. Figure 9A shows the measurement results of a representative array element, and the uniformity of the calculated center frequency (F c ), − 6 dB BW, and peak-to-peak sensitivity of the phased-array transducer are illustrated in Figure 9B, which are about 9 MHz, 55%, and 150 mV, respectively. It can be found that the acoustic performance of the phased array exhibits a good uniformity. In Figure 9A, ringing and bumpy spectrum are observed, and it might be caused by the FPC layer between the backing and composite materials. A single matching layer may also lead to insufficient BW.
For the two-way insertion loss (IL) measurement, the transducer was excited with a 5-Vpp, four-cycle sinusoidal toneburst signal at 9 MHz, and the reflected echo was received from a polished steel reflector. The echo signal (V o ) was measured by the oscilloscope with a 1 M coupling, and the driving signal (V i ) was then measured with a 50 coupling. The measured value was corrected for loss due to attenuation in water (2.2 × 10 −4 dB mm −1 × MHz 2 ) and reflection from the steel target (0.6 dB) (Cannata et al., 2006). The IL was calculated using the following equation: where f c is the center frequency and d is the distance between the transducer and steel reflector. The IL value of the array is measured to be 39.1 dB, which is somewhat higher than other  Frontiers in Materials | www.frontiersin.org reported transducers, and this might be due to the relatively low g 33 coefficient of piezoelectric composite material (Dias and Das-Gupta, 1994). Besides, it is not easy to obtain high sensitivity by monolayer matched layer. The sensitivity can be improved by increasing the piezoelectric volume ratio and adopting multilayer matching. For the cross-talk measurement, a 5-Vpp, 20-cycle sinusoidal tone-burst signal generated by a function generator (33250A, Keysight Technologies Inc.) was used to excite one representative element (element 23) in the phased array with a frequency range from 3 to 15 MHz in steps of 2 MHz. The voltages across the first, second, and third adjacent elements were measured and the cross talk was calculated as reference (Cannata et al., 2006).
The measured cross talk is shown in Figure 10. As shown in the figure, the maximum cross-talk values at the center frequency were −34.5, −36.4, and −38.9 dB for the first, second, and third adjacent elements respectively. Considering the simplicity of array design and construction, these values are considered satisfactory (Lukacs et al., 2006).
The cross talk decreased fast in the lower frequency and almost no change in the higher frequency. This phenomenon may be caused by the parasitic capacitance formed by the array elements. The parasitic capacitance is a high-pass filter for the cross talk.
The long wire electrode binding and flexible circuit affected the electrical impedance characteristics of the transducer, resulting in the lower value of peak-to-peak voltage in the measurement. The electrical impedance curve can be used to analyze the impact of wire (Jian et al., 2018) and the FPC board.
The one-way azimuthal directivity response of the phasedarray transducer was measured by exciting the one-array element (element 23) using the pulser/receiver DPR500. The element was rotated around an axis with a precise rotating device (QRP02, Thorlabs, Newton, NJ, United States) in a step of 3 • , along its center and length, and the amplitudes of the response were acquired at discrete angular positions by a hydrophone (NH0200, Precision Acoustic, Dorset, United Kingdom) with 0.2 mm diameter. For the transducer, the measurement of − 6 dB FIGURE 10 | Measured cross talk for the array. directivity was ± 30 • , as shown in Figure 11, which can serve satisfactorily in the performance of phased-array beam. Similar results were reported with directivity of approximately ±25 • for a 2-2 composite phased array (Bezanson et al., 2014;Cabrera-Munoz et al., 2019).

Imaging Performance of Phased-Array Transducer
The authors utilized Verasonics Vantage 128 System to determine the imaging capability of the phased array. The probe is driven by a single-cycle sinusoidal tone-burst signal with an amplitude of 50 V, and the DELAY-AND-SUM (DAS) beamforming method is used in the system, which is the most basic digital beamformer for medical ultrasound imaging (Friis and Feldman, 1937;Mailloux, 1982), and the phantom images are presented in a 50-dB dynamic range. The catheter can operate at frequencies of 6.5-11.5 MHz.
To evaluate the performance, the custom-made wire phantoms were used to carry out the image tests. The lateral resolution and axial resolution were detected by wire phantom (Foster et al., 2002) at the central frequency of 9 MHz of the catheter. We can calculate the theoretical spatial resolution using the following equations based on pulse-echo response: where PL is the − 6 dB spatial pulse length of the received echo and the measured value is 217 µm, and the theoretical axial spatial resolution R A is 108.5 µm.
where F# is the F number of the transducer (1.9) and λ is the sound wavelength in the medium (167 µm in water); therefore, the theoretical lateral spatial resolution R L is 317 µm.
The 50-µm diameter wires phantom was immersed in a tank and imaged with gray scale in a 50-dB dynamic range as shown in Figure 12. The spread functions of A wire image axial and lateral line were plotted and are shown in Figure 13.   The measured axial and lateral resolutions at −6 dB are 188 and 321 µm, respectively, and the lateral value is close to its theoretical value, but the axial value has a large deviation because the wire is difficult to be placed in the theoretical focus position. Figure 14 shows an ultrasonic image of a phantom with 10 evenly 10-mm-spaced tungsten wires immersed in a tank with deionized water, and it is clear to find that the wire at a penetration depth of 100 mm can be easily observed with reasonable image quality. The SNR of the wires image was analyzed according to the RF data acquired by the Vantage system, which is higher than 30 dB. The spatial resolutions and penetration performance are satisfactory and can meet the requirements of ICE imaging.
A commercial phantom (KS107BG, Institute of Acoustics, Beijing, China), as shown in Figure 15A, was also employed in the experiment. It can be observed from Figure 15B that the wires in the distal area have a superior resolution (<0.5 mm), which shows that the transducer has an excellent beam steering performance.

CONCLUSION
This study described the development of a 1-3 piezocompositebased phased-array transducer fit inside of a four-direction steerable ICE catheter. The dice-and-fill technique was used to produce the 1-3 piezocomposite material, and an encase structure and half-cut methods were developed to reduce the complexity of the fabrication process. Utilizing fabricated 1-3 piezocomposite, a transducer with a size of 2 mm × 7.4 mm with a central frequency of 9 MHz and −6 dB BW of 55% was prototyped. According to the echo and wire phantom experiments, the basic performance of the phased array was excellent. The experimental results suggest that a novel fabricated process of 1-3 composite phased array is competent for the single-use disposable ICE catheter in the future.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
ZH, ZL, XJ, and YCu conceived and designed the research work. NW, ZL, and YCh analyzed the data. ZH and XZ performed the experiments. All authors contributed to the article and approved the submitted version.