Fabrication of SiGe HBT BiCMOS Technology

John D. Cressler

• Publisher: CRC Press
• Year: 2018
• Print ISBN: 9781420066876
• eBook ISBN: 9781351834780
• Topic: Technology & Engineering
• Subtopic: Electrical Engineering & Telecommunications
• Index: Technology & Engineering

SiGe HBT BiCMOS technology is the obvious groundbreaker of the Si heterostructures application space. To date virtually every major player in the communications electronics market either has SiGe up and running in-house or is using someone else's SiGe fab as foundry for their designers. Key to this success lies in successful integration of the SiGe HBT and Si CMOS, with no loss of performance from either device. Filled with contributions from leading experts, Fabrication of SiGe HBT BiCMOS Technologies brings together a complete discussion of these topics into a single resource.

Drawn from the comprehensive and well-reviewed Silicon Heterostructure Handbook, this volume examines the design, fabrication, and application of silicon heterostructure transistors. A novel aspect of this book the inclusion of numerous snapshot views of the industrial state-of-the-art for SiGe HBT BiCMOS technology. It has been carefully designed to provide a useful basis of comparison for the current status and future course of the global industry. In addition to the copious technical material and the numerous references contained in each chapter, the book includes easy-to-reference appendices on the properties of Si and Ge, the generalized Moll-Ross relations, integral charge-control relations, and sample SiGe HBT compact model parameters.

1

The Big Picture

John D. Cressler

Georgia Institute of Technology

1.1 The Communications Revolution

1.2 Bandgap Engineering in the Silicon Material System

1.3 Terminology and Definitions

1.4 The Application Space

1.5 Performance Limits and Future Directions

1.1 The Communications Revolution

We are at a unique juncture in the history of humankind, a juncture that amazingly we engineers and scientists have dreamed up and essentially created on our own. This pivotal event can be aptly termed the “Communications Revolution,” and the twenty-first century, our century, will be the era of human history in which this revolution plays itself out.

This communications revolution can be functionally defined and characterized by the pervasive acquisition, manipulation, storage, transformation, and transmission of “information” on a global scale. This information, or more generally, knowledge, in its infinitely varied forms and levels of complexity, is gathered from our analog sensory world, transformed in very clever ways into logical “1”s and “0”s for ease of manipulation, storage, and transmission, and subsequently regenerated into analog sensory output for our use and appreciation. In 2005, this planetary communication of information is occurring at a truly mind-numbing rate, estimates of which are on the order of 80 Tera-bits/sec (10¹²) of data transfer across the globe in 2005 solely in wired and wireless voice and data transmission, 24 hours a day, 7 days a week, and growing exponentially. The world is quite literally abuzz with information flow—communication.^* It is for the birth of the Communications Revolution that we humans likely will be remembered for 1000 years hence. Given that this revolution is happening during the working careers of most of us, I find it a wonderful time to be alive, a fact of which I remind my students often.

Here is my point. No matter how one slices it, at the most fundamental level, it is semiconductor devices that are powering this communications revolution. Skeptical? Imagine for a moment that one could flip a switch and instantly remove all of the integrated circuits (ICs) from planet Earth. A moment’s reflection will convince you that there is not a single field of human endeavor that would not come to a grinding halt, be it commerce, or agriculture, or education, or medicine, or entertainment. Life as we in the first world know it in 2005 would simply cease to exist. And yet, remarkably, the same result would not have been true 50 years ago; even 20 years ago. Given the fact that we humans have been on planet Earth in our present form for at least 1 million years, and within communities having entrenched cultural traditions for at least 15,000 years, this is truly a remarkable fact of history. A unique juncture indeed.

Okay, hold on tight. It is an easy case to make that the semiconductor silicon (Si) has single-handedly enabled this communications revolution.^* I have previously extolled at length the remarkable virtues of this rather unglamorous looking silver-grey element [1], and I will not repeat that discussion here, but suffice it to say that Si represents an extremely unique material system that has, almost on its own, enabled the conception and evolving execution of this communications revolution. The most compelling attribute, by far, of Si lies in the economy-of-scale it facilitates, culminating in the modern IC fabrication facility, effectively enabling the production of gazillions of low-cost, very highly integrated, remarkably powerful ICs, each containing millions of transistors; ICs that can then be affordably placed into widgets of remarkably varied form and function.†

So what does this have to do with the book you hold in your hands? To feed the emerging infrastructure required to support this communications revolution, IC designers must work tirelessly to support increasingly higher data rates, at increasingly higher carrier frequencies, all in the design space of decreasing form factor, exponentially increasing functionality, and at ever-decreasing cost. And by the way, the world is going portable and wireless, using the same old wimpy batteries. Clearly, satisfying the near-insatiable appetite of the requisite communications infrastructure is no small task. Think of it as job security!

For long-term success, this quest for more powerful ICs must be conducted within the confines of conventional Si IC fabrication, so that the massive economy-of-scale of the global Si IC industry can be brought to bear. Therein lies the fundamental motivation for the field of Si heterostructures, and thus this book. Can one use clever nanoscale engineering techniques to custom-tailor the energy bandgap of fairly conventional Si-based transistors to: (a) improve their performance dramatically and thereby ease the circuit and system design constraints facing IC designers, while (b) performing this feat without throwing away all the compelling economy-of-scale virtues of Si manufacturing? The answer to this important question is a resounding “YES!” That said, getting there took time, vision, as well as dedication and hard work of literally thousands of scientists and engineers across the globe.

In the electronics domain, the fruit of that global effort is silicon–germanium heterojunction bipolar transistor (SiGe HBT) bipolar complementary metal oxide semiconductor (BiCMOS) technology, and is in commercial manufacturing worldwide and is rapidly finding a number of important circuit and system applications. In 2004, the SiGe ICs, by themselves, are expected to generate US$1 billion in revenue globally, with perhaps US$30 billion in downstream products. This US$1 billion figure is projected to rise to US$2.09 billion by 2006 [2], representing a growth rate of roughly 42% per year, a remarkable figure by any economic standard. The biggest single market driver remains the cellular industry, but applications in optical networking, hard disk drives for storage, and automotive collision-avoidance radar systems are expected to represent future high growth areas for SiGe. And yet, in the beginning of 1987, only 18 years ago, there was no such thing as a SiGe HBT. It had not been demonstrated as a viable concept. An amazing fact.

In parallel with the highly successful development of SiGe HBT technology, a wide class of “transport enhanced” field effect transistor topologies (e.g., strained Si CMOS) have been developed as a means to boost the performance of the CMOS side of Si IC coin, and such technologies have also recently begun to enter the marketplace as enhancements to conventional core CMOS technologies. The commercial success enjoyed in the electronics arena has very naturally also spawned successful forays into the optoelectronics and even nanoelectronics fields, with potential for a host of important downstream applications.

The Si heterostructure field is both exciting and dynamic in its scope. The implications of the Si heterostructure success story contained in this book are far-ranging and will be both lasting and influential in determining the future course of the electronics and optoelectronics infrastructure, fueling the miraculous communications explosion of our twenty-first century. The many nuances of the Si heterostructure field make for some fascinating subject matter, but this is no mere academic pursuit. As I have argued, in the grand scheme of things, the Si heterostructure industry is already reshaping the global communications infrastructure, which is in turn dramatically reshaping the way life of planet Earth will transpire in the twenty-first century and beyond. The world would do well to pay close attention.

1.2 Bandgap Engineering in the Silicon Material System

As wonderful as Si is from a fabrication viewpoint, from a device or circuit designer’s perspective, it is hardly the ideal semiconductor. The carrier mobility for both electrons and holes in Si is comparatively small compared to their III–V cousins, and the maximum velocity that these carriers can attain under high electric fields is limited to about 1 × 10⁷ cm/sec under normal conditions, relatively “slow.” Since the speed of a transistor ultimately depends on how fast the carriers can be transported through the device under sustainable operating voltages, Si can thus be regarded as a somewhat “meager” semiconductor. In addition, because Si is an indirect gap semiconductor, light emission is fairly inefficient, making active optical devices such as diode lasers impractical (at least for the present). Many of the III–V compound semiconductors (e.g., GaAs or InP), on the other hand, enjoy far higher mobilities and saturation velocities, and because of their direct gap nature, generally make efficient optical generation and detection devices. In addition, III–V devices, by virtue of the way they are grown, can be compositionally altered for a specific need or application (e.g., to tune the light output of a diode laser to a specific wavelength). This atomic-level custom tailoring of a semiconductor is called bandgap engineering, and yields a large performance advantage for III–V technologies over Si [3]. Unfortunately, these benefits commonly associated with III–V semiconductors pale in comparison to the practical deficiencies associated with making highly integrated, low-cost ICs from these materials. There is no robust thermally grown oxide for GaAs or InP, for instance, and wafers are smaller with much higher defect densities, are more prone to breakage, and are poorer heat conductors (the list could go on). These deficiencies translate into generally lower levels of integration, more difficult fabrication, lower yield, and ultimately higher cost. In truth, of course, III–V materials such as GaAs and InP fill important niche markets today (e.g., GaAs metal semiconductor field effect transistor (MESFETs) and HBTs for cell phone power amplifiers, AlGaAs- or InP-based lasers, efficient long wavelength photodetectors, etc.), and will for the foreseeable future, but III–V semiconductor technologies will never become mainstream in the infrastructure of the communications revolution if Si-based technologies can do the job.

While Si ICs are well suited to high-transistor-count, high-volume microprocessors and memory applications, RF, microwave, and even millimeter-wave (mm-wave) electronic circuit applications, which by definition operate at significantly higher frequencies, generally place much more restrictive performance demands on the transistor building blocks. In this regime, the poorer intrinsic speed of Si devices becomes problematic. That is, even if Si ICs are cheap, they must deliver the required device and circuit performance to produce a competitive system at a given frequency. If not, the higher-priced but faster III–V technologies will dominate (as they indeed have until very recently in the RF and microwave markets).

The fundamental question then becomes simple and eminently practical: is it possible to improve the performance of Si transistors enough to be competitive with III–V devices for high-performance applications, while preserving the enormous yield, cost, and manufacturing advantages associated with conventional Si fabrication? The answer is clearly “yes,” an...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Dedication
• Foreword
• Preface
• Editor
• Table of Contents
• 1 The Big Picture
• 2 A Brief History of the Field
• 3 Overview: Fabrication of SiGe HBT BiCMOS Technology
• 4 Device Structures and BiCMOS Integration
• 5 SiGe HBTs on CMOS-Compatible SOI
• 6 Passive Components
• 7 Industry Examples at the State-of-the-Art: IBM
• 8 Industry Examples at the State-of-the-Art: Jazz
• 9 Industry Examples at the State-of-the-Art: Hitachi
• 10 Industry Examples at the State-of-the-Art: Infineon
• 11 Industry Examples at the State-of-the-Art: IHP
• 12 Industry Examples at the State-of-the-Art: ST
• 13 Industry Examples at the State-of-the-Art: Texas Instruments
• 14 Industry Examples at the State-of-the-Art: Philips
• A.1 Properties of Silicon and Germanium
• A.2 The Generalized Moll–Ross Relations
• A.3 Integral Charge-Control Relations
• A.4 Sample SiGe HBT Compact Model Parameters
• Index