Cubism at the nanoscale

Some of the most famous images in nanoscience and technology have been produced by scanning probe microscopes but, as Chris Toumey explains, there is much more to these images than meets the eye.

Many discoveries and developments in science have been accepted by other scientists and the world at large because it was possible to see what had been discovered. Examples include the discovery that other planets have moons and developments in osteology (as the study of bones is known) in the age of X-rays. The science was plausible because seeing was believing¹. The same is true of nanoscience and nanotechnology, which have earned much of their credibility and profile from the pictures they produce. Because a scientist can present us with detailed before-and-after pictures of a surface that has been modified, or of atoms that have been picked up and placed elsewhere, we can believe that he or she has truly done those operations.

But there is a crucial difference because the pictures of the nanoscale created by a scanning probe microscope are not optical views of matter in the way that images produced by telescopes, conventional microscopes or X-ray plates are. On the contrary, pictures of the nanoscale taken with scanning probe microscopes are the end result of a multistage process that begins with touching the nanoscale, not seeing it, and converting tactile sensations into data, which are later converted into visual sensations². Artificial colour, artificial shading, artificial three-dimensionality and other devices are often added.

These issues have been explored by a number of authors over the past few years^1,3,4, and also at conferences in Columbia, South Carolina, in March 2004, and again at Bielefeld in Germany in May 2005. These conferences raised two kinds of concerns. Epistemologically, how can we know how faithful an image is to its nanoscale object? And in terms of credibility, will people feel deceived if they later learn that pictures of nanoscale phenomena do not have the kind of optical veracity we expect of microscopes, telescopes, photographs and X-rays?

This is not the first time that people have asked about relations of object, process, and image. One episode is especially instructive: the early development of cubist theory. Almost exactly one hundred years ago, in late 1907, Pablo Picasso finished the first cubist picture: Les Demoiselles D'Avignon (Fig. 1). Although it was not publicly exhibited until 1916, the cubist painters and collectors who saw Les Demoiselles when it was created understood that it demanded we develop a new way of seeing objects. Without going into all the subtleties of early cubist thought, we can concentrate on the most important of the cubist principles (which is known as simultaneité), and then borrow three techniques based on this principle. We do so for the benefit of enhancing the epistemology and credibility of images produced by scanning probe microscopy.

Figure 1: Picasso's Les Demoiselles D'Avignon, the first cubist painting, represented a new way of seeing objects by showing the same subject from a number of different angles. (Courtesy of the Bridgeman Art Library.)

© Succession Picasso/DACS 2007

Why should our visual knowledge of an object be limited to one moment or one perspective?

Picasso, Georges Braque and the other early cubists felt strongly that all painting from the Renaissance through to the Impressionists was artificially limited to the three Euclidian dimensions of height, width and depth. A successful picture was a two-dimensional 'illusion' of those three dimensions: a snapshot of one moment viewed from one perspective. But why should our visual knowledge of an object, a person or a landscape be limited to one moment or one perspective? Consider your favourite head-and-shoulders portrait from before cubism, whether by Rembrandt, Van Gogh or anyone else. A skilled artist will show what the face looks like, and perhaps the top of the head and one side of the head. But the back of the head was just as much a part of the person. Moreover, the person whose portrait was being painted might move their head or readjust their position, and for portraits that take months to complete, they might look and dress very differently from the moment captured in the picture.

The cubists challenged themselves to create new techniques for increasing visual knowledge of an object depicted in a two-dimensional picture: that is, they wanted to show more of the reality of the object. They called this 'the fourth dimension' and they invented several ways of adding another dimension to the Euclidian three. Their most notable technique — the device that amazed everyone a hundred years ago, thrilling some people and deeply upsetting others when it was revealed in Les Demoiselles D'Avignon — was to show the same object from multiple perspectives at the same time⁵.

In the same spirit, I commend three ways to add more perspectives to images of the nanoscale for the benefit of their epistemology and credibility.

First, add a temporal perspective. Some well-known images are frozen in time, literally. At the nanoscale, mild thermal vibrations are equivalent to massive earthquakes, and an atom is not able to stand still to have its picture taken. In the cases of the 35 xenon atoms spelling “IBM” in 1990 and the 48 iron atoms defining the quantum corral in 1993, it was necessary for Donald Eigler and his co-workers at the IBM Almaden Research Center in California to reduce the temperature within their scanning tunnelling microscope (STM) chamber to just 4 Kelvin (refs 6,7). In both cases, Eigler and co-workers included a sequence of images to show the process of creating the final picture. A panel of six images was included in the Nature paper⁶ that included the “IBM” picture, and the group's website contains a panel of four images⁸ showing the construction of the quantum corral, which was reported in Science in 1993 (Fig. 2 and ref. 7). Sequences like these help the viewer see the temporal relation between the object and the image. They also help to protect the image from a suspicion that the creator has concealed the artificial conditions for creating the image.

Figure 2: Duchamp's Nude Descending a Staircase showed that it was possible to include a temporal dimension in paintings. (Courtesy of the Bridgeman Art Library.)

© Succession Marcel Duchamp/ ADAGP, Paris and DACS, London 2007 AAAS IBM

To make this false-colour STM image of the quantum corral (top right) more convincing, Don Eigler and co-workers also published greyscale STM images of the corral as it was being constructed.

Incidentally, the early cubists urgently wanted to add a temporal dimension to their paintings, but found it especially challenging. One of the more notable successes was Marcel Duchamp's Nude Descending a Staircase (Fig. 2), and one could argue that the four- and six-panel sequences published by the IBM group are the nano equivalents of Duchamp's masterpiece.

Most nanoscale phenomena literally have no colour of their own.

The second method is to add a colour perspective. Most nanoscale phenomena are smaller than the wavelength of visible light, so they literally have no colour of their own. The human eye, however, needs to see some colour, or at least shades of grey, in order to make sense of an image. Some nanoimages have many colours, and some have only a few; but they are all artificial in the sense that they are added and selected arbitrarily.

At the workshop on imaging the nanoscale in Bielefeld, the 'nano flower bouquet' was discussed at length. This is a structure of silicon carbide, created by Ghim Wei Ho, Mark Welland and colleagues at Cambridge University in 2004, that has petals and other flower-like parts⁹. The picture, made with a scanning electron microscope (SEM), truly looks like a bouquet. SEM images are monotones, usually in grey, but they can be rendered in any colour with values from light to dark. The nano flower bouquet is usually seen as a blue object. At the Bielefeld workshop, one person commented that the bouquet had the shape but not the colour of a bouquet of flowers because blue is cold and lifeless. Another said no, its colour was very reminiscent of certain blue flowers, such as the hydrangea. So what is the right colour of the nano flower bouquet? It could be blue, violet, yellow, green, or any other colour (Fig. 3). By observing the same object in multiple colour perspectives simultaneously, the viewer can see that all these colours are equally artificial. The scientific value of the nano flower bouquet is in its silicon carbide structure, not its colour.

Figure 3: This scanning electron micrograph of a silicon carbide nano flower was originally a greyscale image.

M. Welland and G. W. Ho, Nanoscience Centre, University of Cambridge

By looking at the different false-colour versions simultaneously, the viewer can see that all these colours are equally artificial.

The third way to add more perspectives to images for the benefit of their epistemology and credibility is to add a tactile perspective. We ordinarily think that atomic surfaces and the objects that lie on them are far too small for us to touch. Instead, we have ultra-sensitive instruments like the STM and the atomic force microscope (AFM) to feel the surface for us. However, there is a device that enables one to feel nanoscale objects with the help of haptic technology. The NanoManipulator (Fig. 4), made by 3rdTech of Durham, North Carolina, begins with an AFM, which can indeed touch and scan an atomic surface and objects on that surface. Data from the AFM scan can be manipulated so the viewer is not limited to the usual top-down two-dimensional view; the NanoManipulator converts the data to a continuous three- dimensional image of the sample's surface.

Figure 4

The NanoManipulator adds a three-dimensional perspective and a tactile dimension to an AFM scan of an atomic surface.

After seeing the atomic surface in three dimensions (3D), the operator can switch from the scanning mode to a haptic mode (derived from the Greek word for touch). The last scan, shown in 3D, is held static, and the operator can use a haptic device — a feedback system involving the data from the scan and a hand-held tool like a stylus — to create the sensation of moving the tip of the AFM back and forth on the surface. This gives the operator a virtual-reality feel of the shapes of molecules, viruses and other objects captured on the AFM scan.

The device also has a third mode that enables the operator to push the AFM tip against the surface and move the objects on the surface. After doing this, the operator can return to scanning mode to see how their actions have changed the surface of the sample. So this is a process of first seeing the nanoscale in 3D; then feeling it in 3D; then changing it with a precise 3D force; and, finally, seeing the results immediately in 3D (ref. 10).

We know that the STM and the AFM can make images of nanoscale objects. We also know that they can manipulate atoms and other objects. Ordinarily, however, they do so in a hands-off mode after being programmed to perform a certain manipulation. But with the NanoManipulator, one can feel and change an atomic surface almost in real time.

Cubist painting deeply bothered some people a hundred years ago. According to the generation of Picasso and Braque, cubism should not be judged according to the degree to which it bothers, but rather according to its success or failure at enlarging the viewer's knowledge of the reality of the object in a picture. This is also a good way to judge the works of people like Eigler and others: not whether their amazing pictures throw us off balance, but whether pictures enriched with time, colour, touch and other perspectives, add to our empirical knowledge of the objects that exist at the nanoscale.

In Thesis next month:

Richard Jones on nanotechnology and poverty