Heterogeneous microtechnology – an area for smart designers
Heterogeneous microsystems are more energy-efficient, and cheaper to mass-produce than their macroscopic counterparts. And because of their microsize and tight integration, they often offer extra possibilities; think of "leave-behind" sensors or body implants. Fabricating these systems may prove not to be the hardest task. After all, the semiconductor industry does have 40 years of experience in IC and MEMS processing. But designing systems so that they profit maximally from the integration may be the real challenge, second only to making them so reliable that you'd have no afterthoughts if they would control the plane you're in, or your blood sugar level.
In heterogeneous microtechnology, various types of technology are combined and integrated in a microsized electronic solution, a single microchip for example. Such combinations are not new, of course—take a sensor that measures humidity, or the presence of gasses. This sensor can be combined with a controller IC that analyzes the measurements, and with a wireless radio that transmits the results. These technologies exist, and it's fairly straightforward to combine them on a printed circuit board (PCB), connected to the electricity grid, or to a battery. Such a package may weigh a pound or two, and require a battery charge every other day.
The systems described above, however, are a far cry from the vision of smart, ambient technology: invisible, microsized systems that measure and control, and that require no maintenance. To make such a system, you'd have to combine all technologies on a microchip, or in a micro-package that contains a number of strongly integrated microchips – a system-in-a-package (SiP).
SiPs, if you want to use them as leave-behind sensors, as implantable body sensors, or embedded in buildings, have to be extremely energy-efficient. Ideally, they should even be autonomous, harvesting the little energy they need from the environment. They must also be reliable in circumstances that are totally different than those of ICs in a run-of-the-mill laptop and they have to withstand, for example, extremely humid and corrosive environments, or high temperatures and pressures.
Monitoring cars, the environment, us
There is already a large body of expertise on hand in the industry and in research centers, with respect to electronic systems that go a long way toward implementing such heterogeneous microsystems. These systems are, for example, used in abundance in modern cars; e.g., they have up to a hundred microsystems that control the power train, the car's traction, and energy management. They are small and reliable, and they do their job in the background. They combine sensors, actuators, logic ICs and memory.
Going a step further, work is underway on applications that are close to the living world, that make a connection with the environment, with our skins, or even with our blood, brain and muscles. A recent issue of The Economist  issued a special edition titled "Medicine Goes Digital." According to the magazine, the convergence of biology and engineering is turning healthcare into an information industry, a change that will be disruptive, but also hugely beneficial for patients.
This convergence will be driven by heterogeneous microsystems, in the form of small, comfortable, cheap networks of body sensors, i.e., sensors that continuously measure body parameters and transmit the results wirelessly and that allow monitoring aging people from their homes. These sensors will be comfortable, without installation hassles, or having to change or charge batteries. Examples are wireless monitors for blood pressure, brain waves, or heart rhythm.
Highly miniaturized systems can also be used in implantables, such as hearing aids (cochlear implants), vision aids, or neuroprobes for stimulation and measurement.
To demonstrate what is possible in that area, our research lab recently built a prototype microprobe for use in deep brain stimulation therapy (used to treat patients with Parkinson's disease or severe depression). The technology uses electrodes measuring only 10µm and compared to current systems that have electrodes measuring 4mm, such a microprobe would stimulate much more precisely, with less side effects.
The prototype was designed through finite-element modeling of the electrical field distribution around the brain probe using multi-physics simulation software (COMSOL). This tool also enabled investigating the mechanical properties of the probe during surgical insertion and the effects of temperature. The results indicate that adapting the penetration depth and field asymmetry allow steering the electrical field around the probe, which results in high-precision stimulation. Also key to the design approach is developing a mixed-signal compensation scheme enabling multi-electrode probes capable of stimulation as well as recording; such a scheme is needed to realize closed-loop systems (Fig. 1).
|Figure 1. Prototype probe for stimulation of brain cells.|
The challenge: making smart, holistic designs
To design the newest generation of logic ICs, you'd need an army of well-disciplined engineers, and because they have highly automated design methods and start out from a restricted set of base elements, they are able to design the IC, laying out billions of transistors in complex levels circuits and connections. That is not the case when we design heterogeneous systems. The technologies and parameters to choose from make for a near infinite number of combinations, and there are no design methods or tools to support this process from start to end. That is why designing and fabricating such a heterogeneous system is a highly creative process. A process, moreover, that calls for a broad and deep knowledge of engineering to get the most out of the tight integration.
Taking the traditional view, a heterogeneous application would be designed block per block, technology per technology, optimizing and testing each block separately. In our body networks, for example, we use a sensor, a signal amplifier, signal processing, and a wireless radio to transmit the results. We started out by optimizing these systems separately. We succeeded, for example, to bring down the energy use of our processing IC from 1mW down to 50µW. And we have designed a wireless solution that consumes only a few tens of microWatts. Optimizing each technology separately results in an order of magnitude gain in efficiency, but that is still not enough to yield the optimal solution.
To arrive at an optimal solution, the technology blocks should be optimized together, tuning all blocks to each other. You can minimize the power you need for a wireless radio by first analyzing and processing the data, and then transmit only those data relevant for the application, and shut down the radio when there is nothing to transmit. To that end, an ultra-low power wake-up receiver with a fast response time can be placed in parallel with a conventional radio to switch it on when data needs to received or transmitted.
Using such holistic design approach, it is possible to gain another order of magnitude in efficiency, arriving at systems with a total power budget ~100µW. Then you have a system that is so efficient, you could start thinking of powering it with an energy harvester, e.g., a solar cell.
Smaller, cheaper, smarter
Heterogeneous microsystems will spawn a new era in microelectronics, allowing many new and exciting applications. Applications in the medical domain, for example, that will replace and extend the much bulkier commercial systems of today. Think for example of a wearable, lightweight, wireless system for measuring brain waves. Such system, for which prototypes are built and tested today, will implement the functionality of a full-blown EEG-scanner (electroencephalogram). In contrast to hospital scanners, patients can wear these new wireless systems in the comfort of their living environment, allowing them to pursue their daily activities.
|Figure 2. Flexible, wireless, light weight, miniaturized sensor node.|
By using new materials, we can arrive at new applications, such as pliable, stretchable electronics (Fig. 2) that can be integrated invisibly in clothing. Or the use of biocompatible materials that allow the making of body implants with integrated electronics.
Heterogeneous microsystems – as they are envisioned – are smarter than their macroscopic counterparts. They are more powerful and energy-efficient. With the techniques, processes, and infrastructure of micro-electronics, they can be produced cheaper. The resulting systems are more reliable, easier to model, and easier to test.
1. V. Vaitheeswaran, "Medicine Goes Digital," The Economist, special report on health care and technology, April, 2009 (http://www.economist.com/specialreports/displayStory.cfm?story_id=13437990)
Jan Provoost received master's degrees in languages and information science from the U. of Leuven and is a science writer at imec, Kapeldreef 75, B-3001 Heverlee, Belgium; ph.: +32 16 28 14 34; email firstname.lastname@example.org
Chris Van Hoof received his PhD in electrical engineering from the U. of Leuven and is the director of heterogeneous integration at imec.