Heat is a universal quantity. Since all energy turns to heat, the study of thermal phenomena is applicable over the whole range of science and engineering. For example, the performance of many modern semiconductor devices are limited by their thermal performance e.g. RF power transistors, lasers, power switches and microprocessors. At the macroscopic scale thermal effects are generally slow, because heat transport is diffusive. A characteristic of diffusive phenomena is that they speed up quadratically with changes in dimension: If a device or thermometer is made ten times smaller it will have a thermal response which is a hundred times faster. In a normal material the effects can become enormously rapid, in the GHz range for modern devices which is a rate with is comparable to their operating frequency.

At the nanometre scale, where modern devices and materials are fabricated, the transport of heat ceases to be classical and diffusive. Heat may travel "ballistically" without scattering across a small distance leading to even more rapid thermal transport in a small device and the thermal conductivity of narrow wires may be quantised by the dimensions of the thermal conductor so that classical values of thermal conductivity do not apply and prediction of the thermal performance of such systems becomes impossible: it can only be measured. Since heat is often transported by sound waves known as transverse acoustic phonons which may have a wavelength which approaches the atomic spacing in the material the influence of interfaces between materials and their roughness on the atomic scale has a huge effect on how well structured materials may conduct heat. This might be used to advantage in thermoelectric materials or thermally insulating materials, or it may be an enormous problem when trying to extract the heat of operation from a device such as a laser or microprocessor. Again, since the technological control of atomic scale interfaces is hard this makes thermal conduction across interfaces impossible to predict, so we are forced to measure it.

Scanning Thermal Microscopy is a technique using the mechanical scanning of a sensor probe over a surface to make measurements of local temperature at the nanometre scale. This is very effective but currently the sensors used are very slow in response time, because they are relatively large (about ten micrometres) and are supported by large volumes of thermally insulating material. This project is concerned with the complete re-engineering of the probe to make it much smaller and to fabricate it without the use of thick support structures. This will allow the sensor to respond rapidly to the temperature variation of the sample and permit thermal measurements with a resolution of 30nm x 30nm x 1ns in the spatial and temporal domains.

The importance of such a probe is that it will, for the first time, enable the measurement of local temperature rise in devices and materials on the scale of device operation. Crucially, this will allow us to distinguish between heat being conducted away and heat being locally "stored" by heat capacity. This information allows the path of flow of heat to be determined through a device, enabling the designer to control it, and also determines the maximum temperature achieved. Since the thermal degradation of devices is limited by "thermal activation energy" the rate of damage and hence the lifetime of a device varies exponentially with temperature: The device lifetime is determined by the peak temperature, not the average. Since the properties of a material do not scale simply with size, and since the operating frequency of devices is fixed by device size, thermal measurements of the peak temperature can only be made at the spatial and temporal scales of the actual device. The proposed tools uniquely fulfil this requirement for measurements at the very heart of modern technological achievements.