|
How impossible data all adds up
|
12/05/2008
|
| |
Tom Shelley learns how modelling, extrapolating and advanced data analysis helps scientists understand surfaces and extract data from experiments under ‘impossible’ conditions
Scientists at the National Physical Laboratory (NPL) are using physics modelling to extract valuable information from experiments that cannot be undertaken in real-world conditions – and then applying advanced statistics to dig out meaningful data from the results.
NPL’s Joanna Lee is using a technique called Secondary Ion Mass Spectrometry (Sims), in conjunction with the modelling and statistical analysis, to analyse coatings and thin films. This could well prove important in fields as diverse as electronics and medicine.
Sims works by breaking molecules – such as those in a surface coating – into smaller fragments. The problem is then to make sense of it and work out which original molecules were on the surface before they were blasted to pieces.
NPL has developed a refined technique called G-Sims, which uses manganese or bismuth ions to cause different degrees of fragmentation. By comparing the spectra produced by the two ions, and understanding the physics, it is possible to extrapolate the results for a lower – and ‘experimentally impossible’ – plasma temperature, in order to identify the original molecular species. NPL has developed a computer programme to produce simulated spectra of every possible way that a molecule might break up.
Lee has also been conducting studies on using C60 'Buckyball' ionised molecules to blast the surface. “They are like bombs and blast loads of materials from the surface for us to analyse,” she says. “Metal ions are a bit like bullets and go in deep, causing a lot of damage, but emitting very little material."
She has also been applying ‘multivariate methods’ – in particular, principal component analysis (PCA) and multivariate curve resolution (MCR) – to the results. These are statistical methods for simultaneously analysing multiple variables and determining the contribution made by each variable to an observed result. In the case of Sims, these methods can be applied to the very large data sets. The methods have been successfully applied to resolving the surface distributions of very similar polymer blends.
Lee explains that the team was applying the methods to organic electronic films, which are often only 10-50 nm thick. “Sims is probably the only method with sufficient chemical specificity and surface sensitivity to do this.”
The great potential of organic electronics is that they can be inkjet printed and are potentially cheap enough to be disposable – but, for this to be successful, it is essential to understand what affects their wettability.
Another important area of research is looking at drug coatings on coronary implants. These need to release their drugs steadily over a period of time after surgery, so the engineering of the coatings and the precise distribution of the drug molecules are crucial. Solid-state microelectronics is another important area and the technology is also highly relevant to improving the functioning of adhesives.
But using physics modelling to extrapolate from ‘real’ experiments – and then deducing behaviour from ‘impossible’ experiments and using statistical analysis to deduce the individual causes of effects – could be applied to any engineering study involving experiments or the outputs of sensors.
Pointers
* Secondary Ion Mass Spectroscopy (Sims) can identify which molecules were present on a surface before they were broken up by the ion impacts – and make observations under conditions that are not feasible
* Multivariate statistical analysis can separate multiple causes and effects to produce images of the surface distribution of chemical species – even when they are closely related
* The methods are being applied to electronic polymer circuitry and other surfaces, but could be applied to almost any kind of real-world engineering system
Measured response
As well as high-tech collaborative research, NPL offers help on measurement and sensing problems to any UK company engaged in innovation – under the Department of Innovation Universities and Skills funded ‘Measurement for Innovators’ initiative.
"We can spend time with small companies, sorting out their measurement and sensing problems,” says Stuart Windsor, head of the NPL's knowledge transfer team. “This might involve helping them understand the functioning of their biosensors or looking at why their sensors suffer from cross interference. We can give up to five days of free support."
This is in addition to any part of the UK's National Measurement System being able to give free advice to any UK company for anything up to a few hours and setting up joint industry projects lasting six months to a year.
|
| |
Author
Tom Shelley
|
| |
| |
|
| Email this article |
| |
|
|
|
|