between surface and tip atoms, and thus interact with the electron cloud of the atoms or atoms on the tip. Measurements of the tip’s electron clouds through voltage thus presents a way to understand the surface atoms through the behavior of the behavior of the atoms. Use of electron tunneling as a measurement technique in the STM is part of a broader trend in creating images from non-optical data, and has implications for what is able to be visualized with the instrument.
Electron tunneling is a relatively new idea; the incorporation of electron tunneling into the STM shows how the dynamic fits into the larger story of the development of non-lens-based visualization technologies. In 1960, Ivar Giaever first published the results of demonstrated electron tunneling (Giaver 147–48). For his research, he received the Nobel Prize in 1973. However, scientists did not apply electron tunneling to instrument development until the early 1970s, when Russell Young, John Ward, and Fredric Scire created a machine called the “topografiner” that, like the STM, used electron tunneling and three-dimensional scanning to measure “the microtopography of metallic surfaces,” but used a field emitter instead of a tip to create tunneling conditions (Young, Ward, and Scire 999). The topografiner was not very successful in achieving measurements due to interference from outside vibrations, caused by people walking in the building, for example. In the early 1980s, STM inventors Gerd Binnig and Heinrich Rohrer, along with Christoph Gerber and Edmund Weibel, reduced outside vibrations enough to measure the tunneling and develop the STM (Binnig et al. 178–180).29 The story of how electron tunneling became a measurement technique illustrates the complex mediation required of some visualization technologies, mediation anchored in the practices of a larger community.
The use of electron tunneling to visualize atoms affects what can be measured as well as the relations between the different atoms interacting at the interface of the vacuum. What is measured is the atomic movement that enables electron tunneling, not an object such as an atom. Recording the interactions of electrons with other electrons in a vacuum also transforms the distance between tip and surface electrons into a dynamic interface. To create data points, then, the tip passes across the area to be imaged, sampling the changes in voltage produced by the different densities of electron interactions at different spots. Therefore, the STM maps encounters, local events of tunneling.
One effect of the use of electron tunneling to create measurements is that both tip and sample can affect the interaction (and thus the measurement). For example, the tip’s characteristics can affect the sample and the image produced from the sample, establishing a multi-directional affordance that both creates and structures the dynamic between tip, sample, and resulting image. As STM textbook author Chunli Bai explains: “The size, nature and chemical identity of the tip influence not only the resolution and shape of a STM scan but also the electronic structure to be measured” (9). For example, the conical shapes that form the rim of the Quantum Corral (Figure 4) image are not “what atoms look like;” instead, the shapes are effects of the tip’s V-shape, and reflect the tip’s traverse from one level to another while, at the same time, moving across the sample surface. In short, the shape is more like a graph of the tip’s movements over time (Russell). One result of the mutual influence of tip, sample, and image is that the shape of the tip becomes important: Ideally, one atom at the tip end should protrude slightly (even just an Ångstrom) from the others so that the applied current can flow through that one small point (J. Foster 17).30 The sample, too, can affect the tip if the atoms of the sample strongly attract and “pull off” some of the tip’s atoms. In the space of the interaction, both tip and sample meet in the vacuum, and are equally able to affect the interaction.
The use of electron tunneling also creates the possibility of continued interaction, like MRI or PET scanning, as neither sample nor tip is damaged in collecting measurements (unlike, for example, electron microscope samples that are destroyed in the imaging process). The fact that the sample is not destroyed allows researchers to collect data repeatedly over the same space, and thus track dynamics over time; the user can experience atoms as a series of movements. A Journal of Physical Chemistry B article provides an example: S. A. Kandel and P. S. Weiss state that “by comparing sequentially recorded images, [they] can see that the size and shape of the clusters [on the sample] change over time” (8103). Kandel and Weiss explain the effect of the tip on the measurement: “the mobility shown in the rearrangement of these clusters is likely (at least in part) induced by the STM tip” (8103). The dynamic that occurs between the tip and the sample as the STM collects data is similar to the HCI interaction type of “flow,” where a user experiences a merging with the computer. In this case, of course, the interaction occurs between atoms. Although the tip-sample interaction does not directly involve the user, electron tunneling affects the user’s experience of atomic phenomena through a multi-directional affordance and the related possibilities for repeated interaction. The structured dynamic of the tip-sample interaction that also produces nanoscale measurements affects what can be shown in an image and affects experiment design, just as the use of electron tunneling allows researchers to manipulate the surface using the microscope’s probe affordances. Thus, the expectation of a certain kind of multidirectional interaction inheres at the level of collecting data.
Figure 4. “Quantum Corral” image. From Science 262, 5131 (8 Oct., 1993). Cover illustration. Image originally created by IBM Corporation. Reprinted with permission from AAAS.
Raster Scanning and Z-direction Moves
The STM tip moves in three dimensions (x, y, and z, in the Cartesian system), forms other constitutive components, and works with electron tunneling to shape the multi-directional affordance of the STM. The microscope’s design, like that of the scanning electron microscope, relies on a mobile tip. The fact that the tip moves—and must move—to survey the sample affects how the STM measures each point. The tip moves in three different directions: in the x and y directions in a raster pattern to collect data about points on the surface, and in the z (up and down) direction to create the tunneling conditions in which tip and surface electrons interact. Movements in x, y, and z directions coordinate with tip-surface tunneling interactions to collect data about the nanoscale; the x, y, and z movements also are part of broader traditions of dynamics used by other visualization technologies. In the STM, the x, y, and z directional dynamics structure interactions that encourage manipulation by the user.
The process of rastering, or scanning the surface with a back-and-forth pattern, from which the STM builds an image from data, forms a key dynamic in other visualization technologies. Derived from radar’s creation of an image from a signal, rastering is perhaps best known as the method that allowed cathode ray tubes, such as those used in non-digital televisions, to build an image from a linear signal. In microscopy, using rastering to image a sample had been considered since the early twentieth century, but use of rastering was first demonstrated in 1972 (Wickramasinghe 78). The STM-like topografiner also scanned microscopic surfaces and collected measurements to create an image of the surface, although the specific pattern the scanner makes is not described (Ward, Young, and Scire 999). The incorporation of rastering into scientific visualization instruments such as the topografiner in the early 1970s also fits into the trend of developing visualization technologies that Galison discusses. The STM, then, is partly formed by the “tradition” of dynamics of rastering.
In the STM, rastering only works in conjunction with measurement of tip-surface interactions in electron tunneling, because the dynamics involved in raster scanning are structured around the challenge of movement in relation to time. To create anything other than a blur, a sample would need to remain relatively still, and the tip would need to move relatively quickly. As Lev Manovich observes, one implication for images produced using rastering is that “It is only because the scanning is fast enough and because, sometimes, the referent remains static, that we see what looks like a static image” (100). However, atoms do not slow down enough to become referents; so, following Manovich’s explanation, rastering would not work in the STM unless rastering is combined with the tip-surface interactions that provide the “stability” of a measurement of movement. Even so, STM researchers often need to correct for
