Printing prowess - printed silicon technology
In electronics, fabricating smaller components is a major objective. Professor David Britton, from the NanoSciences Innovation Centre at the University of Cape Town, South Africa, outlines the progress made in printed silicon technology.
Traditional electronics manufacturing processes and materials can be broadly divided into four categories – discrete electronics, microelectronics, thin-film electronics and thick-film electronics.
Microelectronics involves combinations of electron beam or photolithography, physical and chemical vapour deposition techniques, to produce integrated circuits on crystalline semiconductor substrates. In thin-film electronics the same processes are used, but the semiconductor wafer is replaced by a cheaper alternative such as glass, and, more recently, polymer film such as PET and polyimide. Thin-film electronics is mainly used in large area applications, such as solar cell modules, transistor arrays for driving displays, and increasingly, touch sensitive displays.
In a typical consumer or industrial electronics product, a mobile phone, for example, active elements produced using both technologies are combined with discrete passive components, such as resistors and capacitors, and thick-film conductors and insulators that are produced by printing. The most widely used material for the active semiconductor is silicon.
Printed electronics is a new approach with the potential to replace thin-film electronics, and, ultimately, microelectronics. Rather than applying a complex combination of deposition, annealing and etching processes, it is possible to deposit materials layer by layer in a device or circuit under ambient conditions.
This is a paradigm shift away from traditional concepts of electronic devices as a circuit board in a box. It enables the use of lightweight, rollable, or stretchable materials, to the ubiquitous integration of smart functions in everyday objects such as print media, packaging, clothing and office furniture.
To achieve this through conventional printing, novel semiconductor inks that cure without high temperature processing need to be developed. Until recently, the focus of most R&D has been on polymers such as the polythiophenes and polyanilenes, which are generally p-type semiconductors. There have already been several notable successes, including organic transistor back-planes developed by Plastic Logic, California, USA, and organic solar modules produced by Konarka, Massachusetts, USA. However, neither of these devices is fully printed, and their prices are not yet competitive with conventional silicon alternatives.
Besides the high material costs, organic semiconductors have other disadvantages compared to silicon. In general, polymer materials are less stable and more sensitive to environmental factors than covalently bonded inorganic materials. Recently, there has been renewed interest in developing printable, or, more generally, solution processable, inorganic semiconductor inks. These, unlike early chalcogenide thick-film inks, do not contain harmful heavy elements.
Pen to paper
There are two main approaches to producing a silicon ink. Our method at the University of Cape Town, is to use silicon nanoparticles to replace the pigment in a graphics ink. The second, more elaborate method, pioneered by Seiko-Epson and US start-up Kovio, is to use a silicon-containing precursor, typically cyclopentasilane, and convert it to silicon after printing. However, conversion involves pyrolysis and high temperature annealing, restricting its use to high temperature metal and ceramic substrates. At Cape Town, we have produced working devices without high temperature post-processing to sinter the silicon nanoparticle network or to pyrolise the binder phase in the dry layer. The key to this success lies in both the surface characteristics of the individual silicon nanoparticles and the printed layer’s microstructure, which in turn is governed by the ink composition and printing processes.
Characterisation has illustrated the suitability of silicon nanopowders for use in electronic inks. Two of these techniques are simple investigations of the surface chemical characteristics and the near absence of any luminescent activity. Silicon is a direct band-gap semiconductor that only emits light from transitions involving strongly localised electronic states. These states compete directly with the free conduction of charge.
Other techniques used are high resolution electron microscopy and direct determination of the oxygen concentration in the powder using X-ray emission spectroscopy, and at the surface using photoelectron spectroscopy. It is essential that the particle’s surface is free of any oxide, or other capping layer, and is stable against further oxidation. See micrographs for examples of (a) a poor quality particle, with a thin oxide layer which is electrically insulating, and (b) a good qualityparticle with a clean surface.
The particles shown (see images, above) were both produced by milling bulk silicon. The oxidised particle was produced by conventional low energy ball milling, used in ceramics, over an extended period, whereas the second particle was produced by a novel high energy milling process, developed at the University of Cape Town. For both powders, the particles are similar, being polycrystalline with an average size of around 100nm and a grain size up to 20nm. Both powders had been stored under normal laboratory conditions for approximately a year before the micrographs were taken.
Similar results have been obtained for nanoparticles produced by other methods. If the particles are clean and free of oxide when they are produced, they remain in that state almost indefinitely under normal conditions.
Making a connection
To maintain these properties during and after printing, the particle surface cannot be modified to enable dispersal in the ink. Indeed, for the ink to form a conducting network of particles after drying, it is actually better if the particles form a high density of clusters, and that the ink contains as few dissolved solids as possible. This means that the formulation cannot contain additives, such as surfactants, and needs a high particle loading.
In addition, the ink’s rheology must match the chosen printing process. For screenprinting, for example, formulations have been achieved that contain up to 90% silicon by dry weight using oil and aqueous acrylic emulsion bases.
In the micrograph above, the microstructure of a silicon layer is shown, screen-printed from acrylic ink onto paper. Most of the features seen are not individual particles, but clusters of particles that form larger clusters in a hierarchical or fractal structure. This packing is characteristic of both acrylic and oil-based inks, and is key to understanding the material’s electronic properties, which are subtly different to both crystalline and thin-film silicon, and are the subject of ongoing research. The particle clusters form a percolation network for charge conduction, which in turn is limited by the transfer of charge carriers between and within the particles.
The images above show the design of a field effect transistor optimised for low resolution printing. This is a simple bottom-gate structure consisting of four printed layers – a metallic gate electrode, an insulating ink to form the gate dielectric, followed by the silicon semiconductor layer, and finally the metallic source and drain electrodes. Top-gate transistors are also produced by reversing the print sequence.
Device characteristics are shown in the charts, left. Typical field effect mobilities of the order of 1cm2/Vs have been determined. The results have been validated independently at the Motorola Central research labs in Schaumburg, Illinois, USA.
Yet, printed silicon electronics should not simply attempt to reproduce the earlier successes of microelectronics and thin-film electronics. There are new applications which cannot be met by conventional semiconductor fabrication. These include optical, thermal and capacitative sensors, which have already been demonstrated, and are currently undergoing validation by an independent company.
David Britton and Margit Härting, NanoSciences Innovation Centre, Department of Physics, University of Cape Town, Rondebosch 7701, South Africa.
Tel: +27-21 650 3327