Work by a team at the University of Oxford and the University of Exeter may well become recognized as the first steps on the road to a new and bright optoelectronic future for phase-change memory materials.
The modest description of their work as an “optoelectronic framework” by a team at the University of Oxford and the University of Exeter1 hides what may well become recognized as the first steps on the road to a new and bright optoelectronic future for phase-change memory (PCM) materials. This is a future, not just for displays, but also for optoelectronic memory devices, optical switches, and modulators for intra- and inter-chip communications.
Almost everybody is familiar with the colored rainbow of interference patterns that are observed in sunlight when a very thin film of oil is floating on water. What if you could place another thin film on top of the oil, which could by electrical means be used to modulate the colors, enhancing some, reducing others, and then make it all in solid state. That simplistic description is what the Oxford University team has done. And then they went the proverbial “extra mile.”
As well as the large change in resistance that accompanies the transition between the amorphous and crystalline states for chalcogenide glasses, there is also a change in the optical refractive index. It is this refractive index change that has been exploited by the Oxford team.
The embryo structure for the reflection version (there is also a transmission option) of the proposed chalcogenide-based optical modulator or display, in this case employing a Ge2Sb2Te5 (GST) composition, is illustrated in Figure 1. From the bottom up, the active part of the device consists of a suitable substrate and a dielectric film overlaid with platinum film to serve as a reflector, a transparent film of conducting indium tin oxide (ITO), a film of GST, and a top film of transparent conducting ITO.
The two most important films in the stack are the bottom ITO electrode, with its thickness (t) a key variable, and the active memory material GST. In a typical stack, the thickness of the films would be as follows (10nmITO/7nmGST/t variable of ITO/100nm Pt) all deposited on a suitable substrate. By varying the thickness of the bottom ITO electrode, a particular color can be reinforced. In addition, others can be reduced when the surface of the stack is illuminated with white light and the GST is in its amorphous state (see Figure 1).
Switching the GST to its crystalline state changes its refractive index and, by interference, changes the color that is reinforced in the reflected beam. The films involved are all very thin compared with the wavelength range of visible light. Figure 2 shows the impressive range of color changes that can be achieved over large areas with ITO thicknesses of 50 nm, 70 nm, 120 nm, and 150 nm. The wavelengths of visible incident light ranges from 350 to 750 nm.
To PCM memory watchers, the next step will be familiar. Make an orthogonal X-Y display formed of discrete memory devices. The initial form of a possible display is illustrated in Figure 3. Here the two colors (enhanced blue and red) would be obtained by partially switching the memory material so that the GST becomes, in effect, two films of different refractive index. Alternatively, the array could be fabricated with each row having a different GST thickness.
ITO is not usually used as an electrode for PCM memory devices. It is normally used for data storage memory applications where the electrodes are a refractory metal and crystallized GST.
The Oxford/Exeter team initially answered the question as to the suitability and compatibility of ITO electrodes for phase-change switching by creating a large sheet of the sandwich structure. Then, they used the conductive tip of an atomic force microscope (CAFM) to step across the surface, to selectively switch regions under the point to create the pixels for a series of pictures in various colors, as shown in Figure 4.
If displays are the target application, then a transmission version will also be required in addition to a reflective one. In addition to displays, some of the other application options suggested by the Oxford team included contact-lens-type displays, smart eyeglasses, windshield displays, and possibly even synthetic retina devices.
Achieving an optical transmission structure requires constructing a PCM device without the platinum mirror and fabricating the device on a transparent substrate. Here again the CAFM technique was used to create large-area images on transparent quartz and (because the films are very thin) also on flexible plastic substrates.
Two of the many images demonstrated using the reflective technique are shown in Figure 4.
Clearly, for an optical matrix operating in transmission mode, a gap-type PCM device would appear to be an option. To this end, an ITO/GST/ITO gap structure was evaluated that also served as a second vehicle for the evaluation and suitability of ITO electrodes. The demonstrated device had an active area of 300 x 300 nm, a threshold voltage of 2 volts, a maximum current of 20 uA, an on-to-off resistance ratio of 350, and a write/erase lifetime of 140 set/reset cycles. It was recognized that it will be necessary to switch all of the material in the gap and not just a filamentary region.
This new work from the team at Oxford/Exeter was reported as a starting point or “framework” and acknowledged that, to date, outside of the CAFM work, no discrete planar device with ITO electrodes has been constructed or tested.
Therefore, to move this optoelectronic concept for PCM forward it will now be necessary to provide the support for the full evaluation and development of PCM memory devices with ITO electrodes. This will require a full characterization, with special attention to write/erase lifetime, switching times, power dissipation, and data retention, added to which must be the optical characterization for each step in the development -- all of which is a non trivial undertaking.
The history and literature of the development of PCM devices is replete with examples of claims of levels of performance and characteristics that, while achieved in the laboratory, are never reproduced in production. Care must be taken not to rely too heavily on that data except to learn from past mistakes.
For data memory, PCM past experience suggests to avoid initial threshold switching problems and drift from films deposited in the amorphous state. Experience shows that it is better to start with the active material in its crystallized state.
That same experience base also suggests that having one electrode of crystallized material results in a more reliable operation. For example, in an ITO/GST/ITO structure, that would require leaving a small layer of crystallized material on the ITO surface as a nucleating site. Such an arrangement should still allow the memory cell to operate as an optical modulator. All that will be necessary is to build the additional thickness of the remnant crystallized active material into the calculation as an additional phase shift.
Pixels do not have to be as small as PCM memory devices for data storage. With a need to switch all the material between the electrodes, the problem is how big can you make a pixel before a filament will form and cause only part of the device to be crystallized.
As the ITO/GST/ITO discrete planar device characterization work described above provides the detailed data with which to design a matrix, considerations will have to be given to providing a non-linear element to avoid any sneak path problems. In a reflective version of the PCM opto display, one possibility would be to build the optoelectronic stack on silicon. Then use the topside of the platinum as a mirror and its underside as a contact to create a Shottky diode to X-or-Y line connections of the matrix in the silicon. This would require dividing the platinum into a series of individual mirror/contacts as well as maintaining a flat surface.
Worthy of support
With the potential to spawn whole new families of optoelectronic phase-change-based switching displays and modulators, this optoelectronic venture certainly looks like a project that deserves further support and investigation. With PCM active film thickness on the order of 7 nm, new work in this area may well contribute to advancing the scaling efforts of PCM memory development where this work has its origins.
Politicians worldwide, including in the UK where this work originated, are always voicing their views on the need to support science and build from it an associated manufacturing industry. Further support for this work might provide them with an opportunity to turn their words into deeds. It is certainly deserving of that support.
1. Letter – An optoelectronic framework enabled by low- dimensional phase-change films, by Peiman Hosseini, C. David Wright & Harish Bhaskaran; Nature, Vol 511, 10 July 2014.