Abstract
Traditionally bipolar transistors have monocrystalline emitters that are contacted by metal, usually aluminum. However, the current gain of conventional BJTs does not reach the highest values predicted by theory. This is due to the high doping effects which limit the emitter injection efficiency and/or high minority carrier recombination in the emitter Silicon bipolar technology has reached a state of advancement that the device characteristics and circuit performance are not only determined by the doping profiles but also by the emitter contact technology. In the last few years polycrystalline silicon has been used increasingly as the emitter contacting material. Polysilicon contacted devices have made it possible to achieve much greater emitter injection efficiencies, and possess the ability to greatly increase the current gain a t a given base impurity doping concentration. The performance of bipolar transistors has been considerably enhanced by the use of polysilicon as both a diffusion source and a contact for shallow emitter; devices. Improvements in packing density and switching speed have resulted from the self-aligned structure [2], which has reduced device parasitics, and the lower base current as compared to metal contacted shallow emitter devices. With a lower base current, the base doping level can be increased to reduce the intrinsic base resistance without sacrificing the current gain of the original device [3]. Several researchers have investigated enhanced gains in polysilicon emitter devices, suggested various models to explain their operations, fabricated devices, and obtained good results. However, none of them reported reproducible devices or data from the devices they made in terms of beta variability. The objective of this thesis lies not only in demonstrating that polysilicon emitter transistors have higher current gains than the conventional shallow emitter aluminum contacted devices but also in showing that the polysilicon emitter devices can be manufactured in a consistently reproducible manner. In fabricating n+pn transistors, either arsenic or phosphorus can be used as the dopant for the emitter region in monocrystalline silicon and for the polysilicon contact. Arsenic was chosen for our process due to the superior shallow doping profile that could be obtained. The shallow emitter was formed in the monocrystalline substrate before the polysilicon was deposited on that region to make a polysilicon contact, which is also doped with arsenic. The emitter is then composed of both a monocrystalline and polycrystalline region. The base currents of these shallow emitter devices are controlled by the material, which is polysilicon contacting the emitter, and the interface between the contacting material and the emitter region under the contact. There are three major different theories proposed to explain the improvement in emitter injection efficiency and hence beta of polysilicon contacted transistors. These theories and a model of the conduction mechanisms in polysilicon are discussed in chapter II. Polysilicon emitter contacted bipolar transistors were fabricated by the introduction of two extra masking steps into an existing four mask conventional shallow emitter bipolar process excluding isolation. The basic process and process development are discussed in chapter III. Before devices could be fabricated it was necessary to predict the device performance from the proposed fabrication sequence. The process simulators SUPREM II and SUPREM III have been useful in the design and optimization of integrated circuit technologies. SUPREM II, however, does not model structures that utilize polysilicon. SUPREM III, on the other hand, is an improved process simulator that can model up to five material layers, including polysilicon, and was available in the Engineering Computer Network at Purdue University. Using SUPREM III, the proposed bipolar junction transistor (BJT) structure was modeled and optimized with the existing implants, oxidations, and design rules. The program has predicted that an acceptable profile can be obtained by varying those parameters. This is also included in chapter III. Other processes that were performed for the purpose of developing the polysilicon emitter contacted devices are described. Their characteristics are explained and compared with the test results. Basic electrical measurements were made on both conventional devices and polysilicon emitter contacted devices that were fabricated in the same wafer and conditions except for the polysilicon contact part. Mainly enhanced current gain in the polysilicon emitter contacted devices, the deviation in the current gain values, and resistance values for the contacts over numerous devices are used as the evaluating criteria. The measurement method and results of measurements are discussed in chapter IV. Conclusions and recommendations are made in chapter V.
Date of this Version
3-1-1988