I’ve said this before, but if I had to pick one general feature of the current scientific literature versus that of (say) 30 years ago, I would vote for the ability to obtain data at far smaller scales (higher resolution) and the corresponding ability to more fully characterize structures and species that are far larger and more complex than the homogeneous-small-molecules-in-solution of classic spectroscopy. Our knowledge of surfaces, for example, is vastly greater thanks to the combination of scanning electron microscopy and the various single-atom-tip physical microscopy techniques (such as AFM, STM and so on). Solution-based techniques like NMR have become more capable thanks to improved pulse sequences and data handling. Super-resolution microscopy techniques have led to a revolution in imaging, revealing things that would once have been considered impossible to capture. Meanwhile, X-ray crystallography has provided a huge number of new protein structures (which can make solving further protein structures even easier), and cryo-EM has emerged to provide high-resolution structures of large and difficult proteins and protein complexes that likely could never be crystallized at all.
You can’t always get to these levels of resolution simply by making smaller versions of the instruments that you already have, because you start running into limits. That includes signal/noise at the very least, but also means difficulties in shrinking some of the physical components such as lenses. You start crossing over from the classical world into the quantum mechanical one if you keep scaling down, but that can also be an opportunity to move to modes of detection that aren’t available on the classical scale (such as the scanning tunneling microscope tip).
This new paper demonstrates one of those completely new modes in the well-traveled field of optical spectroscopy. UV/visible and infrared spectroscopy were of course the cutting edge of analytical chemistry back in the 1940s and 1950s, and modern fluorescence imaging techniques will show that these wavelengths are still tremendously important. But shrinking the benchtop cameras and microscopes is not an easy task, beyond a certain point. Another way to get such data is to use the spectral response of different detectors (semiconducting materials and the like) to reconstruct the spectrum of the light hitting them, but for that you need an array of different materials with a corresponding physical footprint large enough to produce a useful signal.
That’s realized in this work (which comes from a large multicenter team with branches in the UK, China, and Finland) by the use of a single microscale wire, grown epitaxially so that its composition varies along its length. That allows it to be sensitive to various wavelengths of light at different point along the wire itself, depending on the bandgap of the material at that point. By exposing this nanodetector to a range of preset wavelengths of light and working up an algorithm for the measured responses, the group reconstructed a broadband spectrometer out of a wire a few hundred nanometers across and a few tens of microns long. One end of it is a cadmium/sulfur mixture, and the other end is cadmium/selenium, with the composition between the two varying along the length. It sits on a silicon/silicon dioxide surface, and electron-beam lithography is used to connect a set of gold/iridium electrodes along its length, after which the whole thing is layered in aluminum oxide by atomic layer deposition.
The paper shows results from two of these devices, one composed of 30 segments of varying Cd/S/Se and the other of 38. At around 570 nm, they can distinguish light with wavelengths about 15 nm apart; at 10 nm separation that resolution is lost. This is pretty much where miniature commercial instruments are, only this one is several hundred times smaller. And that size difference allows you to do some pretty useful things – for example, the paper demonstrates spectral mapping at subcellular resolution with a layer of red onion cells. As generations of students have seen, these are mostly clear cells mixed with purple ones, and the instrument clearly pick up the spectral differences as the stage is moved across that cell barrier. The zone that’s being measured is far smaller than than cell itself – eyeballing it, I’d say it’s a grid of about fifty or sixty units/cell (see at right). Right now, the limiting step is the stepping and positioning part, not the optical measurements themselves.
That will be improved, and the authors mention a number of other areas to investigate. There are plenty of other nanowire compositions to investigate, for one thing, which could also extend the technique well into the IR and UV. There are variations on the nanowire growth and electrode connections to be explored as well. And basically anything that involves detection of optical wavelengths, from astronomy through aerial/satellite mapping through cell imaging, could take advantage of such instruments, dramatically shrinking their footprints, complexity, and power consumption. Bring ’em on.