PSP-Based Study Of Symmetrical LDD Mosfet Modeling

Enormous ingenious and creative people have exploited an industry, which can convert sand into gold. They have create wonderful miracles in a dimension people cannot see with their bare eyes. What they have done have changed every aspects of our life, brought an upheaval in this society in past century. It is integrated circuit that underpins the modernity of this modern time.MOSFET, as the most fundamental device, plays an most unconspicuous but most significant role in this industry. The research work of modeling of it has been one of the most important and most difficult job, for the quality of a integrated circuit is strongly related to the precision of its model, a well-designed circuit may not result in a good chip, If model and manufacturing process is mismatched.The model engineers have long devoted themselves to building up a series of mathematical equation covering basic physical attributes of device, without sacrifice of too long simulation time. And it should not be difficult in application. Model, which can meet these requirement, is usually based on fundamental physics, according can depict and predict physical effects with minimal parameters. But engineers have to fit curves using empirical parameter because device physics may be too complex, or its mathematical description is too complex.Early in 1966, Pao and Sah had developed some equations to understand MOSFET, and found a basic physical parameter—surface potential. Relation between surface potential and applied voltage was also figured out. But this equation is transcendental. Due to pool computing power in those days, it costs too long time to get a relatively precise surface potential solution. After some elimination and approximation, first generation of compact model—BSIM, was developed. and gained a wide application. In this model, a conception of threshold voltage is introduced to define working regions. Different equations are used in different region, and linking function is used to link adjacent regions smoothly. So the fitting accuracy in this transitional region is poor. This is a structural flaw in BSIM, and cannot be eliminated thoroughly.Although only one equation is sufficient to describe all working regions in the invertion charge model, this kind of model still has a structural flaw: when the transistor is working in accumulative region, there is no invertion charge in the device. What's more, the invertion charge is a function of surface potential. So, the surface potential model is the unique physical model. The powerful computing capacity in nowadays makes surface potential calculation easy. An analytical solution to the original transcendental equation has been developed by some researchers. Consequentially, surface potential model gets its renaissance. A surface potential model PSP has been chosen by Compact Model Committee for industrial standard, taking place of BSIM. But, there still exists some drawbacks in PSP. This paper is dedicated to do some improvement theoretically. Symmetric LDD structure has been long adopted in foundries to mitigate the hot carrier effect. While lessen hot carriers, the LDD structure brings some retrograde in device quality, mainly due to high resistance it adds by using a light doping between source and channel. This high resistance is called series resistance, which is supposed to be constant in PSP. We start from the concept of surface potential, get out the current expression of LDD region and intrinsic channel respectively, then figure out the resistance with iteration, pointing out that this resistance is bias-dependent, and show that it is impacted by the bias through analysis. This metod is based on surface potential, compatible with surface potential model, can be uased in process monitering and new device development.Continuously shrinking channel length brings some annoying effects to device, especially DIBL, which is drain induced barrier lowing. In PSP, the DIBL coefficient is set to be constant again. In this paper, by solving two-dimensional Poisson Equation, we find out its analytical expression. What's more, it's alterable when drain voltage changes, and their relation gets more rigid as channel length shrinks. Application on a 32nm MOS reveals that the method proposed in this paper has a more accurate fitting result than PSP.