Faster by design

A European consortium has set its sights on developing a transistor that would enable the design of cheaper and faster products for applications ranging from communication to imaging.

The £11m DOTFIVE project, co-ordinated by France-based STMicroelectronics' Gilles Thomas, aims to create an integrated circuit with advanced silicon-based bipolar transistors that operate at a frequency of 500GHz, or 0.5TeraHertz. Other industry partners in the 15-strong consortium include Germany's Infineon Technologies and Belgium's IMEC. Academic partners include the University of Naples and the Johannes Kepler University of Linz, Austria.

'The objective is to develop silicon/germanium heterojunction bipolar transistors (SiGe HBTs) at 500GHz, which could allow the design of products with circuits that could reach up to 160GHz,' said Thomas. This compares with devices such as today's GSM phones, which operate at a standard frequency of 2GHz in Europe. Moreover, SiGe HBTs are widely used in all applications above 1GHz, including all GPS systems.

'Depending upon the complexity of the application and the capability of the circuit designer in terms of layout and integration, what you reach at the single device level compared with what you reach at the complete integrated circuit level is usually a factor of three to four,' said Thomas. This means that a GSM phone transmitting at 2GHz would typically use circuit technologies with a frequency of 10GHz — a factor of four.

A key part of the new transistor will be the silicon germanium compound, built on a substrate of pure silicon, which allows the band gap of the semiconductor material to be widened. This is designed to enable higher mobility of the electrons and consequently higher transistor frequencies. The researchers also plan to achieve this goal by exploring new ways of building the transistors.

'We need to define new transistor architecture, new ways of realising extremely thin transistor bases — in the 10nm range,' said Thomas. At present, the engineers are aiming for an improvement of at least a factor of three in the base thickness compared with what is currently being done.

'Also, we need to reduce extremely old parasitics that you have inside the transistor by having a structure that allows self-alignment and to avoid all parasitic resistance to make the transistor work at the highest possible frequency.'

Existing SiGe HBTs achieve a maximum operating frequency of about 300GHz at room temperature, and it is possible for the engineers to reach 500GHz, but only at extremely low temperatures — for example, that of liquid helium or very close to absolute zero.

'This is not difficult to reach, but of course you cannot have cheap electronics working at those temperatures — you cannot have cryogenic systems creating a temperature of a few degrees permanently around the chip,' said Thomas. The researchers are therefore aiming to design cost-effective circuits by creating 500GHz transistors that will work at room temperature.

According to Thomas, the transistors will allow for applications that are currently only available using very costly techniques — either electronics comprising expensive compound materials like gallium arsenide or indium phosphide, or by using discrete electronics (devices with thousands of transistors on the same chip).

'You could have applications such as proximity radars, which currently work at 77GHz and are being developed on very high-end cars, and radars with a range of distances to increase safety, especially in poor visibility,' said Thomas.

'You could also, for example, cheaply build airport security equipment to detect weapons on people without them having to remove any clothing,' he said.

The researchers are hoping to move to even higher frequencies in the future, saying that the number of possible applications increase as they progress in the development of the optimum transistor.

The 36-month project is scheduled to end in 2010, when the researchers hope to demonstrate the individual key circuit functions — such as noise amplifier or transmission capabilities — to prove that the technology can reach a certain level of frequency on simple circuit blocks.