Medical ASICs, particularly for the portable and implantable device market, are largely driven by the need to reduce size, weight, and power. The general tendency is to think that more integration is better. As my colleague Andy Kelly blogged recently (Medical Device ASIC Integration: Optimize, Don’t Maximize), blindly integrating all analog, digital, and RF on a single piece of silicon (called a System-on-Chip or SOC) may increase development time and risk, reduce system flexibility, and compromise performance.
However, silicon integration may reach far beyond the traditional questions of whether to integrate digital, analog, and RF on one integrated circuit. So what's next for silicon integration? Perhaps more importantly, what types of integration will provide the most benefit for devices such as those used in portable and implantable medical?
In discussions with our customers as well as a review of their overall electronic and mechanical systems, there are some clear areas to examine with regards to future silicon integration. A generic system consists of a sensor/transducer, integrated circuit(s), power source, and passive devices including mechanical devices such as connectors. Interestingly, many of these devices are now being implemented in silicon processing technology similar to ASIC technology. This obviously raises the question, “Why not integrate all these devices onto a single piece of silicon?” Let's look at each of these devices in a bit more detail.
Sensors have already become significantly smaller with the advancement of micro-electro-mechanical devices called MEMS. MEMS are using silicon wafer processing capability to create miniature accelerometers, pressure sensors, gyroscopes, touch sensors, and other electro-mechanical devices. In the past, MEMs were offered discretely as a separate die or packaged device.
However, now we see the MEMs devices being offered as part of standard CMOS IC processing to provide a more fully integrated solution. For example, X-Fab provides its foundry customer access to relative and absolute pressure sensors on their XC10 and XH035 IC technologies at the cost of only one additional mask. This reduces chip-to-chip interconnect and overall system size.
Examples of electronic passive devices found on a printed circuit board might include capacitors, resistors, and inductors. These devices are discrete components that are often not integrated in an ASIC due to size or performance requirements. For example, microfarads of capacitance might take an unreasonable amount of die size if integrated. Similarly, a large resistor value or a resistor of acceptable tolerance (10 percent or better) might not be achievable in standard integrated circuit processing.
However, today newer IC technology is available that focuses on integration of passive devices. For example, On Semiconductor's Integrated Passive Device (IPD) technology allows for integration of copper inductors, precision resistors, and precision capacitors. Capacitor density of 100nF/mm2 with better than 10 percent tolerance is available. If you examine the PCB of an implantable medical device that requires a boost or buck converter, the inductor alone is a significant consumer of real estate. If the inductor can be integrated into an IPD chip, substantial size reduction can be achieved.
Power Source (Battery)
Implantable medical devices usually, although not always, make use of a battery that is resident with the implantable device. These batteries come in two flavors, a primary cell (used once), or rechargeable cell. Relatively new rechargeable solid-state batteries (SSB) are now readily available (Cymbet Corp.). To date the primary issue with solid-state batteries has been battery capacity. A single cell might provide 6-12μA-h of capacity. Compare this to typical coin cells with capacities in the mA-h.
But these SSBs have an important advantage — they can be stacked and connected in parallel to provide higher capacity. Solid-state batteries are fabricated on silicon wafers using the same standard semiconductor processes and equipment. The device functions much like a standard rechargeable Li-Ion battery with significant space savings. Today the batteries can be purchased in standard IC packages or in bare die form. If purchased in bare die form, the batteries can be used as part of a multi-die stack to increase integration.
However, according to Jeff Sather of Cymbet, in some medical applications x, y, and z dimensions are so constrained that even a die stack is too big. Thus full integration might be necessary. Currently for very small SSB, and for medical applications in particular, a fully integrated approach can be technically and economically feasible.
Standard implantable medical devices are relatively large, with device volumes occupying between 15cc and 50cc. In order to reduce size and produce devices that can be placed at the point of therapy with less invasive surgeries we need to reduce their volume to less than 4cc. These miniature implantable medical devices (MIMD) must leverage higher levels of integration to meet this goal.
Relatively new technologies are leveraging silicon wafer processing methods and silicon wafers substrates to make various system components such as sensors, passive devices and power sources smaller. Therefore it would seem reasonable to assume that such components could eventually all be integrated onto a single piece of silicon. However, as my colleague has pointed out, we need to understand the trade-offs in the quest for a fully integrated solution. “Smart integration” should win out over “maximum integration” every time.
- Medical Device ASIC Integration: Optimize, Don’t Maximize
- Medical Electronics: Put All the Functionality in the IC
- Medical Implantable Electronics: Promises & Challenges
- Let’s Chat: Medical Implantable Electronics Integration
- The Future of Medicine Is Here
- What Can You Do With a Programmable Current Source?