5.6 Practical Considerations

The doping profile optimizations in this work are performed without any restrictions by manufacturability issues. It is the authors opinion that, before one can look for an optimized realistic doping profile, there should be a good understanding of the theoretical limits. It has often been proven in the past that goals which looked quite unreachable in the first place could be achieved after a fundamental research of the involved theoretical limits. Once it is known that something can work in theory, the motivation is much higher to actually search for methods to realize it. However, the involved costs to change a production line or to introduce new process steps might, in most cases, make people look for ways to accomplish goals using already existing tools and methods.

The PCD device features a clearly asymmetric device structure which might lead to rejection of process and circuit designers in the first place. However, the advantages have been demonstrated in this work and by many experimental devices which can be found in the literature.

Some of them are called ``Graded Channel MOS'' (GCMOS) [35] others ``Laterally-Doped Channel'' (LDC) [38], ``Lateral Asymmetric Channel'' (LAC) [12], or ``Asymmetrically Doped Buried-layer'' (ADB) structures [40]. Even complete circuits, for example, an ultra low-power micro-controller in GCMOS technology [72], have already been implemented using asymmetric device architectures with very satisfying results.

Asymmetric devices also have superior analog performance though this is not the scope of this work. Their investigation started in the early 90's when the DMOS, usually used for high-voltage applications at that time, was shrinked to less than 1 $\mu$m gate length and adopted for high performance analog circuits [26,60].

A 0.13 $\mu$m CMOS technology for ultra low-voltage applications has been reported lately featuring an asymmetric channel doping profile [45].

The PCD device structure has been experimentally verified using a Focused Ion Beam (FIB) implantation method (Fig. 5.14) and studied with emphasis on the superior analog characteristics of this structure [50]. For mass production this method is not convenient due to the enormous amount of time being necessary to ``scan'' the whole chip area. Therefore, possible modifications to the PCD structure are investigated in the following which might simplify its production.

Figure 5.14: The FIB implantation method to generate the PCD structure.