The industry standard process for the fabrication of 1-3 piezocomposites is limited to pillars with a width of greater than 100 lm. An alternative process, VP (viscous polymer) embossing, has the potential to overcome this limitation but is expensive because the mould is destroyed each time a piezocomposite it realised. Hot embossing offers the potential to cost effectively replicate piezocomposite moulds from a master so that the process can be moved out of the laboratory. In this publication, the fabrication of polymer moulds, consisting of an array of cavities, which have cavity walls of 30 lm and aspect ratios of 14 is presented. The quality of replication has been assessed, and the suitability of the process to produce moulds with smaller sized features and higher aspect ratios for high frequency piezocomposite applications is discussed. Piezocomposites for medical ultrasoundThe use of high frequency piezocomposites ([30 MHz) offers new opportunities for medical ultrasound in fields such as dermatology, ophthalmology and intravascular imaging (Lockwood et al. 1996). These opportunities are created because 1-3 piezocomposites provide an improved image quality over conventional, monolithic, transducers (Smith 1991). However, their use in high frequency medical ultrasound is limited because of their small feature size and high aspect ratio that becomes more demanding to manufacture with an increase in operating frequency.1-3 Piezocomposites consist of an array of piezoceramic pillars that are encapsulated in a polymer matrix (Tressler et al. 1999). They are typically made by a process called dice and fill (Safari et al. 1997). In this process, a pressed and sintered ceramic block has parallel slots cut into it. The block is then rotated by 90°and a further set of parallel slots are cut. This leaves an array of pillars that are upstanding from a stock, known as a bristle block. This bristle block is then backfilled with polymer, and the top and bottom faces are lapped to remove the stock and to ensure the faces are parallel.Unfortunately, dice and fill struggles to make pillars that are \100 lm wide (Safari et al. 1997) because in this size range, voids are also present in the ceramic material, which significantly reduces the survival rate of the pillars, and the yield of the devices. The spacing of the pillars, or kerf, is also limited by the width of the saw blade, and the range of designs are limited to straight sided pillars.Many researchers have developed processes that try to overcome the limitations of dice and fill, but few processes have made it out of the laboratory. One method that has shown promise is viscous polymer (VP) embossing (Abrar et al. 2004). In this process, a ceramic paste or dough, is pressed into a polymethyl methacrylate (PMMA) mould and is allowed to dry before the mould is dissolved. However, VP embossing is a lost mould technique which requires the mould to be destroyed each time the piezocomposite is realised, and is therefore expensive. One way to avoid this problem is to replicat...

1-3 Piezocomposites offer many advantages over monolithic transducers, however replicating their fine scale structure is demanding. Viscous polymer (VP) embossing is a technique that has already been demonstrated as being capable of achieving the small feature size and high aspect ratio features that are required for piezocomposites. However, the process utilises a lost mould technique and is therefore limited by the ability to make cost effective sacrificial moulds. Hot embossing has been identified as a technique that is suitable for replicating these moulds, and piezocomposites made from them are presented here. This paper dimensionally characterises the piezocomposites through each stage of the VP embossing process. The piezocomposites described here have feature sizes of 30 lm, volume fractions 0.5 and operating frequencies of 12.3 MHz. For the first time piezocomposites have been made by VP embossing into cost effective moulds.