An emerging demand for the precise manipulation of cells and microparticles for applications in cell biology and analytical chemistry has driven recent development of ultrasonic manipulation technology. Compared to the other major technologies used for cell and particle manipulation, such as magnetic tweezing, optical tweezing and dielectrophoresis, ultrasonic manipulation has shown excellent capabilities and flexibility in a variety of applications with its advantages of versatile, inexpensive and easy integration into microfluidic systems, maintenance of cell viability, and generation of sufficient forces to handle cells with dimensions up to tens of microns and agglomerates of a large number of cells. This thesis reviews current state-of-the-art of ultrasonic manipulation technology and reports the development of various ultrasonic manipulation devices, including simple devices integrated with high frequency (> 20 MHz) ultrasonic transducers for the investigation of biological cells and complex ultrasonic transducer array systems to explore the feasibility of electronically controlled 2-D and 3-D manipulation. Piezoelectric and passive materials, fabrication techniques, characterisation methods and possible applications are discussed. The behaviour and performance of the devices have been investigated and predicted in virtual prototyping with computer simulations, and verified experimentally. Issues associated during the development are highlighted and discussed. To assist long term practical adoption, approaches to low-cost, wafer level batch-production and commercialisation potential are also addressed.

As opposed to current Intraoperative Ultrasound (IOUS) systems and their relatively large probes and limited superficial high frequency imaging, the use of a biopsy needle with an integrated transducer that is capable of minimally invasive and high-resolution ultrasound imaging is proposed. Such a design would overcome the compromise between resolution and penetration depth which is associated with the use of a probe on the skins surface. It is proposed that during interventional procedures, a transducer array positioned at the tip of a biopsy needle could provide real-time image guidance to the clinician with regards to the needle position within the tissue, and aid in the safe navigation of needles towards a particular target such as a tumour in tissues such as the breast, brain or liver, at which point decisions surrounding diagnosis or treatment via in vivo tissue characterisation could be made. With this objective, challenges exist in the manufacturing these miniature scale devices and theirincorporation into needle packages. The reliable realisation of miniature ultrasound transducer arrays on fine-scale piezoelectric composites, and establishing interconnects to these devices which also fit into suitably sized biopsy needles are two such hurdles. In this thesis, the fabrication of miniature 15 MHz ultrasound transducers is presented. The first stage of development involved the production of single element transducers in needles ~2 mm inner diameter, using various piezoelectric materials as the active material. These devices were tested andcharacterised, and the expertise developed during their fabrication was used as the foundation upon which to design a wafer-scale fabrication process for the production of multiple 15 MHz transducer arrays. This process resulted in a 16 element 15 MHz array connected to a flexible printed circuit board and integrated into a breast biopsy needle. Characterisation tests demonstrated functionality of each of the 16 elements, both individually and combined as an array. To explore potential applications for these devices, the single element transducers were tested in fresh and Thiel embalmed cadaveric brain tissue. Plasticine targets were embedded in these brain models and the needle transducers were tested as navigational real-time imaging tools to detect these targets within the brain tissue. The results demonstrated feasibility of such devices to determine the location of the target as the needle devices were advanced or withdrawn from the tissue, showing promise for future devices enabling neurosurgical guidance of interventional tools in the brain. The application of breast imaging was also considered. Firstly, Thiel embalmed cadaveric breasts were assessed as viable breast models for ultrasound imaging. Following this, anatomical features, with diagnostic significance in relation to breast cancer i.e. axillary lymph nodes and milk ducts, were imaged using a range of ultrasound frequencies (6 – 40 MHz). This was carried out to determinepotential design parameters (i.e. operational frequency) of an interventional transducer in a biopsy needle probe which would best visualise these features and aid current breast imaging and diagnosis procedures.

Que ce soit dans le domaine médical ou en contrôle non destructif, les systèmes d’imagerie ultrasonore sont devenus de plus en plus utilisés de nos jours. Leurs applications ne cessent de s’élargir et des performances toujours plus accrues sont vivement recherchées, afin d’améliorer la qualité des diagnostics réalisés. Nous sommes donc passés de l’utilisation de systèmes à base de transducteurs ultrasonores mono-élément à des systèmes utilisant des réseaux de transducteurs à une dimension (1D) et à deux dimensions (2D) composés d’éléments de plus en plus nombreux et petits. Néanmoins, un phénomène indésirable est fortement présent dans ces réseaux de transducteurs ultrasonores : il s’agit du couplage inter-éléments tendant à limiter leurs performances acoustiques et à modifier leur diagramme de rayonnement. Tout au long de ce travail de recherche, nous avons donc cherché à comprendre ce phénomène parasite et à apporter des solutions pour le réduire voire le supprimer. En se basant sur des modélisations éléments finis 2D et 3D et grâce à la fabrication de prototypes, nous avons d’une part, mis en évidence les différents types de couplages présents dans un réseau de transducteurs (acoustique, mécanique) et d’autre part, deux méthodes de correction basées l’une comme l’autre sur l’application de tensions convenables aux différents éléments du réseau ont été testées. La première méthode utilise les déplacements normaux moyens à la surface de chaque élément du réseau pour évaluer ces tensions, tandis que la deuxième fait appel aux courants motionnels parcourant chaque élément pour les déterminer. Les résultats numériques et expérimentaux concernant les déplacements et les diagrammes de rayonnement sont en bon accord. En outre, les deux méthodes s’avèrent particulièrement efficaces pour réduire le couplage inter-éléments. / Whether in medicine or in non-destructive testing, ultrasonic imaging systems have become increasingly used nowadays. Their applications continue to expand and good performances are needed to improve the quality of the diagnosis. Moreover, significant progress has been made since these systems were originally based on single element ultrasonic transducers and are now made of mono-dimensional (1D) and bi-dimensional (2D) elements arrays ever more numerous and smaller. However, an undesirable phenomenon is strongly present in the ultrasonic transducer arrays: it is the cross-talk, which limits their acoustic performances and modifies their radiation pattern. Throughout this research, we have attempted on one hand to understand this parasitic phenomenon and on the other hand to provide solutions in order to reduce it or even remove it. To highlight the cross-talk types (acoustic and mechanical) and to test the proposed correction methods, we developed two-dimensional (2D) and three-dimensional (3D) finite element modeling and fabricated some prototypes. Both correction methods rely on the application of suitable voltages to the array elements. The first method uses the average of the normal displacements at the surface of each element to evaluate the voltages, while the second one utilizes the motional currents through each element to determine them. The numerical and experimental results concerning the displacements and the radiation patterns are in good agreement. In addition to this, both methods have been efficiently performed to reduce the cross-talk.