stack of CMOS and BiCMOS chips, the passivation layer forms the top layer to protect the

underlying layers from oxidation. However, the protection is only effective against am­

bient air and not liquid electrolytes. In an in vivo test, degradation rates of 50 nm/month

were determined for the standard passivation layer fabricated using IHP’s SG25V tech­

nology [35]. The corrosion rates observed may vary between individual technologies and

manufacturers, but measures must always be taken to protect the microchip against

corrosion.

For the interaction between sensor and actuator on one side and the biomilieu on

the other, a miniaturized window should be provided, while all other components are

housed in a hermetically sealed enclosure. The highest durability and diffusion resistance of

all materials has so far been shown by housings made of titanium. As miniaturization has

progressed, the enclosure issue has become increasingly difficult from a technological point

of view. This is because in all cases reliable separation between electrical connections and

active surfaces is required, and in many bioelectronic systems, the two areas are separated

by only a few dozen or a few 100 µm. And it is this remaining area of space in which safe

separation of the liquid biomilieu from the electrical leads must be ensured [19]. It is

therefore inevitable to include the modern methods of microsystem integration in implant

development projects.

21.4 Intelligence

The central control of measurements or sequence of stimuli, the recording of measure­

ment data, their intermediate storage and forwarding for data transmission as well as the

control of the energy supply is carried out in intelligent implants by a microcontroller

µC [18]. Depending on the length of the data words exchanged between the components

of the µC, 8-, 16- and 32-bit architectures are distinguished. Typically, they are supplied

with a working voltage of 3–5 V.

The power consumption of the µC can be reduced by choosing a small clock speed and

a simple architecture. This includes small memories (<32kB), few and general instruc­

tions, and few interfaces to the outside world (<10) as well as limiting the data bus width

to 8- or 16-bit. Interfaces to the outside world include serial ports, analog/digital con­

verters, and input/output pins. Other components are analog to digital converters for

digitizing the incoming, usually analog sensor signal, and a timer that sets the clock for

processes running on the µC.

Common microcontrollers have different power supply modes. These include the ac­

tive mode and the sleep mode, which is switched to when no tasks are pending. For

optimum system life, the microcontroller should be set to this mode whenever possible.

Between both modes, there are usually still various intermediate modes in which only

certain components are switched off.

The functional sequence in the µC, i.e. the sequence of measuring-reading-transmitting

and all other activities are controlled by the program code, which is stored in the main

memory built up from flash cells. Depending on the application of the implant, mea­

surement data are generated. In the case of a continuously operating glucose sensor, for

example, one measured value every 1.5 minutes makes sense, which is following the

physiological time constants, so that up to 1,440 values must be temporarily stored per

day, provided no data retrieval takes place.

Implantable Microelectronics

347