110-GHz f/sub T/ silicon bipolar transistors implemented using fluorine implantation for boron diffusion suppression

This paper investigates how fluorine implantation can be used to suppress boron diffusion in the base of a double polysilicon silicon bipolar transistor and hence deliver a record f/sub T/ of 110 GHz. Secondary ion mass spectroscopy (SIMS) and transmission electron microscopy (TEM) are used to characterize the effect of the fluorine implantation energy and dose, the anneal temperature and the germanium pre-amorphization implant on the fluorine profiles. These results show that retention of fluorine in the silicon is maximized when a high-energy fluorine implant is combined with a low thermal budget inert anneal. TEM images show that a high-energy fluorine implant into germanium pre-amorphized silicon eliminates the end of range defects from the germanium implant and produces a band of dislocation loops deeper in the silicon at the range of the fluorine implant. Boron SIMS profiles show a suppression of boron diffusion for fluorine doses at and above 5/spl times/10/sup 14/ cm/sup -2/, but no suppression at lower fluorine doses. This suppression of boron diffusion correlates with the appearance on the SIMS profiles of a fluorine peak at a depth of approximately R/sub p//2, which is attributed to fluorine trapped in vacancy-fluorine clusters. The introduction of a fluorine implant at this critical fluorine dose into a bipolar transistor process flow leads to an increase in cutoff frequency from 46 to 60 GHz. Further optimization of the base-width and the collector profile leads to a further increase in cutoff frequency to 110 GHz. Two factors are postulated to contribute to the suppression of boron diffusion by the fluorine implant. First, the elimination of the germanium end of range defects, and the associated interstitial population, by the fluorine implant, removes a source of transient enhanced diffusion. Second, any interstitials released by the dislocation loops at the range of the fluorine implant would be expected to recombine at the vacancy-fluorine clusters before reaching the boron profile.

I. Introduction

Over THE past few years there has been considerable interest in the behavior of fluorine in silicon for application in both bipolar and MOS devices. This interest was initially stimulated by the use of a implant for shallow p-n junction formation [1], but more recently by the effect of fluorine on boron diffusion [2]. Initial work gave conflicting results for the effects of fluorine on boron diffusion, with some research showing that fluorine had little or no effect on boron diffusion [3] and other research showing complete suppression of boron transient enhanced diffusion (TED) [4]–[5] [6] and also increased boron activation [2]. These contradictory results have been reconciled by careful analysis of the experimental conditions used for the fluorine and boron implants. For example, for boron implants into pre-amorphized silicon, it has been shown that fluorine dramatically increases boron diffusivity in the early stages of the anneal due to the reduction of the dangling bond concentration in the amorphous silicon by the fluorine [3]. Also the fluorine dose and energy have a strong effect on the degree that the boron diffusivity is suppressed. For example, low dose fluorine implants give reduced boron TED, whereas high dose fluorine implants also give reduced boron thermal diffusion by a factor of four [8].