Ion Implantation and Synthesis of Materials - Studium

236 15 Ion Implantation in CMOS Technology: Machine Challenges15.6 Conclusions and the Future of Ion Implantationin SemiconductorsWe have discussed some of the device and economic imperatives driving ionimplantation technology today. Ion implant remains one of the most productiveequipment sectors in a fab, despite arguably being one of the most complex. Theimplantation machine landscape looks very different than it did ca. 1974 when lowcurrent implanters used for threshold voltage adjustment comprised more than60% of a small ($14 M) market (Rose 1998). The number of implanter types hasincreased with the number and type of implants. Medium current implanterssupplanted low current implanters in the mid-1970s. This was followed by thewidespread manufacturing use of high current and high energy implanters in theearly 1980s and the mid 1990s, respectively. Figure 15.17 shows the number ofimplant steps for an individual process flow as a function of node and year. Whatwill the next 20 years bring in terms of implantation technology? This questionwill be answered by some combination of the laws of physics and the vagaries ofeconomics.The shrinking of devices over the past 30 years has been fairly accuratelydescribed by Moore’s Law, with no fundamental physical constraints encounteredin this evolution. However, in the next 5–10 years, new classes of materials anddevices will be required as physical limits are reached. The most obvious exampleof this limit is the transition from SiO 2 (SiO x N y ) gate dielectrics, which havereached their physical thickness limit, to new higher permittivity dielectrics. Thistransition and others require unparalleled investment to prevent Moore’s Lawfrom faltering. Similarly, continuous increases in wafer size have been driven bythe economic advantage of being able to produce more devices per wafer at arelatively constant manufacturing cost. However, the next projected wafer size of450 mm requires not only a tour-de-force in crystal growth technology butenormous investment in both wafer processing and wafer transport hardware.The biggest impact on implant technology will come from the widespreadadoption of SOI technologies. Devices built on SOI substrates are now inproduction. The use of oxide instead of p–n junctions to provide much of thecircuit isolation significantly reduces the number of implants required. This isespecially true for higher energy (well) implants, since the buried oxide layer is inthe region formerly occupied by the well structures. How widely SOI technologiesare eventually adopted will depend mainly on how much the substrate cost can bereduced. In the unlikely event that 450 mm substrates are ever used, only singlewaferendstations will be practical in terms of footprint. It will be interesting tosee how all these forces play out, but one thing is certain: ion implantation willcontinue to play a central role in Si technology. The unequalled precision andversatility of doping achievable by ion implantation ensures this.