Ultrasound Transducers: Modalities and Operation
During the last several decades, ultrasound devices have become ubiquitous in daily life for various airborne and immersion applications such as automobile, parking sensors, and medical imaging. Traditional piezoelectric transducers were previously used mainly in ultrasound applications however, in the past two decades, micromachined ultrasound transducers (MUTs) have been developed and used in several medical imaging and consumer electronics applications such as handheld/catheter-based medical devices and fingerprint sensors. In general, MUTs operate in 2 different mechanisms, capacitive force (CMUT) or piezoelectric (PMUT) sensing-actuation. See figure 1 and ref. [1] [2].
The pulse-echo mechanism in ultrasound systems
Most of the ultrasound systems operate in pulse-echo mechanism in which, an ultrasound wave is sent to a target (transmit mode), hits the object and the reflected wave is detected by a receiver transducer (receive mode), as shown in figure 2. The received signal is usually small, thus needs an amplifier for further signal processing. In order to detect the pulse-echo signal, the received signal is needed to be above the device noise level, which is mainly the transducer’s thermal-mechanical noise, see ref. [3]. Moreover, the transmitted ultrasound wave is faced with three types of losses: medium attenuation, diffraction loss during travelling, and reflection loss from the object. The dominant loss varies and depends on the medium (immersion, airborne), object (soft tissue, hard wall), and the travelling distance. However, the medium attenuation is usually prevailing and it’s a function of frequency, see ref. [4]. Smaller (larger) transducers correspond to higher (lower) frequencies resulting in higher (lower) loss, and lower (larger) resolution length and penetration depth. For example, a transducer suitable for intracardiac echocardiography (ICE) application usually operates at around 5-15 MHz and penetration depth of about 5-10 cm and ~250 µm axial resolution, while for a transducer suitable for intravascular echocardiography (IVUS) application, the operation frequency is at around 20-40 MHz with 0.5-1 cm penetration depth and ~100 µm axial resolution.
Choosing the piezoelectric transducer
As a general ultrasound system comprises a transmitter (Tx) and a receiver (Rx), in transmit operation the goal is to maximize the output pressure, while in the receiver the aim is to optimize the sensitivity. There are several criteria to choose among different types of ultrasound transducers (bulk piezoelectric, CMUT, or PMUT): pressure, bandwidth, Tx and Rx sensitivity, coupling coefficient, manufacturing process and CMOS integration. Piezoelectric transducers are based on wave propagation (d33-mode) in the surrounding medium and therefore, requires other layers (backing and matching layers) to mitigate the acoustic impedance mismatch between the transducer and the medium. MUTs have been introduced as strong competitors to the conventional piezoelectric transducers due to their various advantages including the ease of integration with front-end electronics (CMOS integration), impedance matching, wide bandwidth, and high coupling coefficient. With respect to conventional piezoelectric transducers, MUTs operate based on the flexural vibration caused by electrostatic force (CMUT) or d31/d33-mode (PMUT) hence do not require matching layers, see ref. [1]. Moreover, MUTs are fabricated using conventional low-temperature micro-machining techniques that make the CMOS-MUT integration process much easier than bulk piezoelectric-CMOS integration.
Although PMUTs and CMUTs are both in the category of flexural transducers, they have pivotal differences. MUTs convert electrical energy to mechanical energy and vice versa. The ratio of electromechanical converted energy to the total input energy is defined as a positive quantity, the coupling coefficient we mentioned before. Piezoelectric transducers’ coupling coefficient is limited by the material properties, hence cannot exceed a certain value. CMUTs coupling coefficient can be close to 1 (or 100%), which corresponds to wide fractional bandwidth and better image resolution in the medical imaging application. Although high coupling coefficient requires high DC bias applied on the electrodes, CMUTs have been used into the handheld medical devices such as Butterfly Network iQ and Hitachi probes [5-7]. Besides, CMUT arrays have been successfully used in several ongoing researches in catheter-based devices for intracardiac and intravascular echocardiography (ICE and IVUS) applications. For PMUTs, the active piezoelectric layer (Lead zirconate titanate or PZT, aluminum nitride or AlN) is deposited with PECVD or sputtering technique thus the quality of the materials due to degradation results in an even lower coupling coefficient. Several techniques have been recently employed to increase the PMUTs coupling coefficient, see ref. [8], and transducer efficiency. Despite high efficiency, CMUTs suffer from high DC bias voltage particularly in Rx mode results in device charging, failure and non-linear drive effects.
Recent advances in PMUT design and manufacturing have made PMUTs a strong alternative for CMUTs especially in airborne applications. PMUTs are derived with only AC signals in both Tx and Rx mode hence are not faced with DC bias related issues such as device charging and reliability. Another advantage of PMUTs over conventional piezoelectric transducers and CMUTs, is the higher device capacitance results in a lower element impedance that improves the matching of the transducer to the signal cabling and cable loss in the system, see ref. [9].
Piezoelectric, PMUT and CMUT Transducer Comparison and Industry Scale
Although PZT has been widely used in piezoelectric devices due to its good piezoelectric properties, it has several disadvantages such as the lead component in the compound, ageing deformation and high annealing/deposition temperature. Therefore, the fabrication process is not CMOS compatible. AlN is recently used in many PMUT applications. AlN is lead-free and can be uniformly deposited over the entire wafer. The AlN deposition temperature is relatively low and therefore, the fabrication process is CMOS compatible.
CMUTs can be fabricated with regular low temperature micromachining fabrication process and can be easily integrated with ASICs used in CMUT-on-CMOS process. It can be designed and manufactured in small membranes suitable for high frequency applications including IVUS, photoacoustic, non-linear imaging, and underwater acoustics.
PMUTs are manufactured in several foundries including STMicroelectronics, Global foundries, and Fujifilm Dimatix. Qualcomm, TDK-InvenSense, and Chirp Microsystem have introduced products using PMUTs suitable for fingerprint and range finding applications. Recently, MEMS foundry SilTerra has unveiled a monolithic PMUT-on-CMOS platform, see ref. [10]. We have recently demonstrated OnScale’s capability for PMUT arrays simulations in partnership with SilTerra, see ref. [11].
Currently, CMUT arrays are manufactured by Phillips, Global foundries, and Vermon. CMUT arrays are currently used in medical probes by Hitachi, Verasonics, KOLO Medical, and Vermon, and handheld devices by Butterfly iQ. Butterfly Network has so far raised $350M on its CMUT-based probes making the CMUTs market dominant in the medical field. Siemens Healthineers, Phillips, and GE Healthcare are also actively working on CMUT-based medical devices, see ref. [7]. In a research done by Dausch et al. at Duke university, see ref. [9], PMUTs are also used for real-time 3-D intracardiac echocardiography (ICE) application.
In the following table, the characteristics of CMUTs, PMUTs and bulk piezoelectric transducers in both Tx and Rx mode are compared.
Simulating PMUTs and CMUTs
CMUT and PMUT arrays for both airborne and immersion applications can be easily simulated with OnScale. Structural mechanics, piezoelectric and acoustic physics are optimally coupled in the multiphysics solvers resulting an efficient time domain simulation platform. Several examples of PMUT, CMUT and bulk piezoelectric transducers can be found on our website.
References
[1]Y. Qiu, J. Gigliotti, M. Wallace, F. Griggio, C. Demore, S. Cochran and S. Trolier-McKinstry, “Piezoelectric Micromachined Ultrasound Transducer (PMUT) Arrays for Integrated Sensing, Actuation and Imaging,” Sensors 2015, 15, vol. 15, pp. 8020-8041, 2015. |
[2]K. Brenner, A. Sanli Ergun, K. Firouzi, M. Rasmussen, Q. Stedman and B. Khuri–Yakub, “Advances in Capacitive Micromachined Ultrasonic Transducers,” Micromachines , vol. 10, no. 2, 2019. |
[3]G. Gurun, M. Hochman, P. Hasler and L. Degertekin, “Thermal-Mechanical Noise Based CMUT Characterization and Sensing,” IEEE Trans Ultrason Ferroelectr Freq Crtl, vol. 59, no. 6, p. 1267–1275, Jun 2012. |
[4]B. Treeby, E. Zhang, E. Thomas and B. Cox, “Measurement of the Ultrasound Attenuation and Dispersion in Whole Human Blood and its Components from 0–70 MHZ,” Ultrasound in Med. & Biol, vol. 37, no. 2, pp. 289-300, 2011. |
[5]e. a. Otake. T., “Development of 4G CMUT (CMUT Linear SML44 probe),” Center for Technology Innovation, R&D Group, Hitachi, Ltd.. |
[6]”http://www.hitachi-medical-systems.eu/products-and-services/ultrasound/transducers/4g-cmut.html,” [Online]. |
[7]E. Strickland, “New “Ultrasound on a Chip” Tool Could Revolutionize Medical Imaging,” IEEE Spectrum, 2017. |
[8]Q. Wang, Y. Lu, S. Mishin, Y. Oshmyansky and D. Horsley, “Design, Fabrication, and Characterization of Scandium Aluminum Nitride-Based Piezoelectric Micromachined Ultrasonic Transducers,” J Microelectromech Syst, vol. 26, no. 5, pp. 1132-1139, 2017. |
[9]D. Dausch, K. Gilchrist, J. Carlson, S. Hall, J. Castellucci and O. Von Ramm, “In Vivo Real-Time 3-D Intracardiac Echo Using PMUT Arrays,” IEEE Trans on Ultrason Ferroelectric Freq Crtl, vol. 61, no. 10, pp. 1754-1764, Oct. 2014. |
[10]”Ultrasound Sensing Technologies for Medical, Industrial, and Consumer Applications,” Yole report, 2018. |
[11]”Reducing Time to Market For Fingerprint Sensors With Cloud Simulation,” https://onscale.com/videos/webinar/reducing-time-to-market-for-fingerprint-sensors-with-cloud-simulation/. |