Interfacial oxide in a transistor

Active solid-state devices (e.g. – transistors – solid-state diode – Bipolar transistor structure

Reexamination Certificate

Rate now

  [ 0.00 ] – not rated yet Voters 0   Comments 0

Details

C257S183000, C257S197000, C257S588000, C257S593000

Reexamination Certificate

active

06787879

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to the field of fabrication of semiconductor devices. More specifically, the invention relates to the fabrication of transistors and, in particular, fabrication of silicon-germanium transistors.
2. Related Art
In a heterojunction bipolar transistor, or HBT, a thin silicon-germanium layer is grown as the base of a bipolar transistor on a silicon wafer. The silicon-germanium HBT has significant advantages in speed, frequency response, and gain when compared to a conventional silicon bipolar transistor. Speed and frequency response can be compared by the cutoff frequency which, simply stated, is the frequency where the gain of a transistor is drastically reduced. More technically, the current gain approaches a value of one as the frequency of operation of the transistor approaches the cutoff frequency. Cutoff frequencies in excess of 100 GHz have been achieved for the HBT, which are comparable to the more expensive GaAs. Previously, silicon-only devices have not been competitive for use where very high speed and frequency response are required.
The higher gain, speeds, and frequency response of the HBT have been achieved as a result of certain advantages of silicon-germanium not available with pure silicon, for example, narrower band gap, and reduced resistivity. Silicon-germanium may be epitaxially grown on silicon wafers using conventional silicon processing and tools, and allows one to engineer device properties such as the band gap, energy band structure, and mobilities. For example, it is known in the art that grading the concentration of germanium in the silicon-germanium base builds into the HBT device an electric field, which accelerates the carriers across the base, thereby increasing the speed of the HBT device compared to a silicon-only device. One method for fabricating silicon and silicon-germanium devices is by chemical vapor deposition (“CVD”). A reduced pressure chemical vapor deposition technique, or RPCVD, used to fabricate the HBT device allows for a controlled grading of germanium concentration across the base layer. As already noted, speeds in the range of approximately 100 GHz have been demonstrated for silicon-germanium devices, such as the HBT.
A polycrystalline silicon emitter can be formed above the epitaxially grown single crystal silicon-germanium base. There are several possible methods of forming a polycrystalline silicon emitter. For example, one approach is to form a layer of some material which can be selectively etched relative to the single crystal silicon-germanium base and open a “window” in that material in which to deposit the polycrystalline silicon for the emitter. After the polycrystalline silicon is deposited for the emitter, the excess material is etched away selectively to the silicon-germanium base, forming the polycrystalline silicon emitter above the single crystal silicon-germanium base.
Prior to formation of the emitter, there is typically a thin layer of silicon oxide that is grown on the surface of the silicon-germanium base. This thin layer of silicon oxide is generally desirable to remain on the surface of the base. For example, if this thin layer of silicon oxide is completely removed from the surface of the base and the silicon for the emitter is deposited directly on top of the single crystal base, the silicon aligns with the underlying crystal structure and a single crystal epitaxial emitter is formed rather than the desired polycrystalline emitter. Furthermore, the resulting bipolar transistor has unacceptably low gain. Gain, simply stated, is the ratio of collector current, I
c
, divided by base current, I
b
, i.e. gain equals I
c
/I
b
.
As stated above, by forming a thin silicon oxide layer on top of the single crystal base before depositing silicon for the emitter, the desired polycrystalline emitter is formed. Because the thin silicon oxide layer is formed at the interface between the single crystal base and the polycrystalline emitter, it is also referred to as “interfacial oxide.” The interfacial oxide has the effect of increasing the gain of the bipolar transistor. The interfacial oxide across the emitter opposes the flow of minority carriers so that the base current in one direction is reduced, while the collector current is largely unaffected. Therefore, I
c
/I
b
, which is the gain of the bipolar transistor, is increased.
In general, making the interfacial oxide layer thicker increases the gain of the bipolar transistor, and conversely, making the interfacial oxide layer thinner decreases the gain. As remarked above, if no interfacial oxide is formed, the gain is unacceptably low. An interfacial oxide layer that is too thick increases the gain but adversely affects the cutoff frequency, causing the cutoff frequency to be too low. For the silicon-germanium HBT, an optimum gain is approximately 100.0. Therefore, it is desirable to fabricate a silicon-germanium HBT with the thickness of the interfacial oxide in an optimum range, neither too low nor too high, such that the gain of the HBT is approximately 100.0.
Other attributes, besides thickness, of the interfacial dielectric that affect gain of the bipolar transistor are the composition of the interfacial dielectric, and its density. For example, in the silicon-germanium HBT, the composition is silicon oxide. Other dielectric materials could be used, for example, silicon oxynitride. Density can be measured by secondary ion mass spectrometry (“SIMS”). For example, with interfacial oxide, a sample of the interfacial oxide is bombarded with ions to remove ions from the sample. The ions are counted to determine the atomic density of the interfacial oxide. The density is measured as “area density” in units of atoms per square centimeter. In general, there is a trade-off between thickness and area density of the interfacial oxide with respect to the gain of the bipolar transistor. If a technique for forming the interfacial oxide produces a lower density oxide, the oxide must be made thicker to achieve the same gain as the gain achieved by an interfacial oxide that is thinner and higher density. Moreover, lower density oxide is more prone to cause epitaxial re-alignment of the polycrystalline emitter. Because of this, it is not desirable to form very low-density oxides. Thus, by altering the composition, thickness, and density of the interfacial dielectric, the gain of the bipolar transistor can be controlled.
Conversely, variation in composition, thickness, and density of the interfacial dielectric can result in unacceptable variation in values for the gain of the bipolar transistor. Variation in values for the gain can cause unpredictable performance, which makes circuit design difficult. Thus it is desirable to form the interfacial dielectric in a uniform manner so that composition, thickness, and density are consistent for devices on the same wafer, and are also consistent and repeatable from wafer to wafer.
For example, one method for forming interfacial silicon oxide uses a wet bench process which comprises flowing ozone in water across the wafer surface; the ozone reacts with the wafer surface to form oxide on the wafer surface. Variation in gain from approximately 50.0 to 300.0 for devices on the same wafer using the wet bench process was found to be unacceptable. Efforts to improve the ozone distribution in the wet bench process to form a more uniform interfacial oxide have not provided the desired uniformity across the wafer or from wafer to wafer.
Another method for forming interfacial oxide is by controlling oxidation during the “drive-in” or “push-in” oxidation step of the doping process. Oxygen or water vapor is provided in the chamber used to heat the wafer for the “push-in” diffusion of dopants, and reaction of the oxygen with the wafer surface oxidizes the surface. However, heating of the wafer involved during “push-in” oxidation can contribute to unwanted device changes. For example, boron used to dope the silicon-germanium base of the HBT diffuses rapidly; excess diff

LandOfFree

Say what you really think

Search LandOfFree.com for the USA inventors and patents. Rate them and share your experience with other people.

Rating

Interfacial oxide in a transistor does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Interfacial oxide in a transistor, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Interfacial oxide in a transistor will most certainly appreciate the feedback.

Rate now

     

Profile ID: LFUS-PAI-O-3270588

  Search
All data on this website is collected from public sources. Our data reflects the most accurate information available at the time of publication.