Image profiler

US researchers have developed a technique for measuring the surface roughness of metal used to shape car bumpers and other body parts that they claim is more accurate than current procedures.

Instead of the traditional method of measuring with a linear profilometer — which can only view sections of an image — the team, from the

(NIST) uses data from a scanning laser confocal microscope (SLCM). This has the advantage of being able to look at an entire image.

Surface roughness is a key issue that goes beyond cosmetics. Faint striations and other marks that appear when metal is shaped can indicate residual stresses that could cause the part to fail. They also lead to extra wear and early retirement for expensive stamping dies used to form sheet metal into body parts.

Measuring this roughness can help predict friction and the metal's 'springback' — the amount it will unbend after being stamped. Springback has to be known and controlled to build accurate dies for complex metal shapes.

A profilometer has a probe that is tracked in a line across a surface to record the peaks and valleys. The process is repeated several times at intervals across the surface, and the results are averaged into a 'roughness' figure.

But NIST researchers found that these measurements may be misleading. Uncertainties and statistical errors are compounded when 2D lines are used to infer the roughness of an entire surface.

So their method uses data from an SLCM, which builds a point-by-point image of a surface in 3D. The data from a single SLCM image — representing an area of about 1,000 x 800 x 20 micrometres deep — is analysed using mathematical techniques that treat every point in the image simultaneously to produce a roughness measure that considers the entire 3D surface rather than a collection of 2D stripes.

'This technology has been around for a while,' said the NIST programme leader Mark Stoudt. 'We just found a way to look at an entire image instead of individual sections of an image.'

The confocal microscope is one of a few available that operate in a completely reflective mode. It stores the topographic data as a raw depth map in tagged image file format (TIFF). The team then wrote a computer code that converts the TIFF information into a simple matrix of surface heights.

'Once we have it in matrix form we are able to deal with the information in 3D,' said Stoudt.

One early finding is that the generally accepted linear relationship between surface roughness and material deformation is wrong, at least for the aluminium alloy the group studied. The data from the 3D analysis shows that a more complicated relationship was masked by the large uncertainties of the profilometers.

Stoudt said this information could be a big help to car manufacturers who have increasingly been using more aluminium in their models.

'With all the emphasis on increased fuel economy, they've been trying to go to lighter weight alloys, and five to 10 years ago they decided aluminium was one of the ones they wanted,' he said. 'This has a different crystal structure to other metals previously used in cars and it deforms differently for the same amount of strain. You get a totally different looking surface which affects the friction inside the die.'

The researchers hope the technique will lead to more accurate models of the effects of strain on new alloys and, eventually, lower development and tooling costs for metalworking industries.

Stoudt said the NIST technique could become an industry standard and the team's next big challenge will be to help the car industry to integrate the new method.