Silicon germanium chips

Anyone who has used GPS, Wi-Fi or a smartphone recently — which is to say, nearly everyone — owes a debt of gratitude to Bernard Meyerson’s intellect, instincts, tenacity and ... a momentary bout of clumsiness.

In 1979, Meyerson was a doctoral student at the City College of New York. While in the lab pursuing semiconductor research, he inadvertently dropped a piece of silicon, which he had just cleaned in hydrofluoric acid, on the floor. When rinsing the silicon under a faucet, he noticed that it repelled water. It was an odd sight, considering that silicon typically formed a water-absorbing layer when exposed to air. Meyerson was curious, but also on deadline, so he filed the event away in his brain for future exploration.

The subsequent year, he joined IBM Research and became fixated on an idea to mix silicon with germanium, thinking a SiGe (pronounced “sig-ee”) alloy could serve as a base for more efficient and powerful chips than silicon alone. It could also be produced on IBM’s existing fabrication lines at a fraction of the cost of gallium arsenide, which was commonly used in chips in high-frequency applications like consumer electronics. It was a prescient notion, one that would eventually expose an enormous opportunity and reveal avenues for growth.

But getting there would require the insight and persistence to tackle a nagging chemistry problem.

Creating a basis for silicon and germanium layers required first removing a contaminating dioxide, which Meyerson did by baking samples to 1,000 degrees Celsius. But the SiGe alloy couldn’t stand up to the heat. Recalling the incident from three years prior, he cleansed a sample in hydrofluoric acid and watched as a protective sheath of hydrogen formed on the silicon. As it turned out, that sheath, which is what made the chip repel water, dissipated at a much lower temperature, just south of 600 degrees — creating a clean slate on which the oxide could form while the chip remained intact.

“That was the epiphany,” Meyerson later explained. “Simply growing materials below 600 degrees avoided the whole problem. It was the most bizarre finding I’ve ever had, but it also gave us a 10-year head start on the rest of the world because nobody understood the effect, so away we went.”

Meyerson soon began producing flawless, uncontaminated silicon germanium transistors at 550 degrees that were orders of magnitude faster than silicon chips. He assumed a team of 100 technologists focused on SiGe research and development for mainframe computers. But within months, IBM pivoted to complementary metal-oxide semiconductor (CMOS) transistors and raided Meyerson’s SiGe team. While Meyerson insisted the future of SiGe chips was in communications devices, not mainframes, IBM lacked any meaningful presence in the space and so opted not to further fund his research.

In 1992, Meyerson and the sole remaining member of his team, an electrical engineer named David Harame, decided to seek funding outside of IBM for their cause. Meyerson became a one-man sales force, forging alliances and financial arrangements with several communications firms that paid IBM to first develop and then manufacture SiGe chips. IBM raced through development into production and, virtually overnight, brought new fields of wireless technology to life.

In 1996, IBM made its first internal investment in silicon germanium, establishing the SiGe Product Development and Manufacturing group and two new telecommunications design centers. The next year, the company published a briefing titled “SiGe RFIC Home Run Strategy,” calling for significant investment into radio frequency integrated circuits, including SiGe, with an eye on potentially huge returns.”

“Our Home Run strategy focuses on accelerating market credibility, RF product capability and share prior to expected competitor entry,” the report read. “The Home Run will require significant investments above our base plan. In return, the revenue opportunity is significant — over USD 1 billion for IBM within 5 years — if we are successful in realizing the full potential. Further, SiGe leadership can enhance several other core IBM businesses, including high-speed network devices, portable wireless devices, un-tethered computing and high-speed processor peripherals.”

IBM quickly established dominance in SiGe manufacturing, setting nearly every device performance record in silicon technology. SiGe’s reliability, speed and low cost spurred rapid growth in various wired and wireless networks, shrinking the size and power needs of Wi-Fi, smartphones, GPS systems and many other products. Within six years of the Home Run Strategy, the company had shipped more than 100 million SiGe chips. Shortly thereafter, both Meyerson and Harame were awarded the company’s highest technical distinction, IBM Fellow.

IBM continues to explore the utility of SiGe across new and emerging semiconductor applications. In 2016, the company published research demonstrating that it had developed the world’s first 7 nm node test chips by applying extreme ultraviolet (EUV) lithography — a technique of using light to etch patterns into other materials — and using SiGe as the channel material in the transistor. The test chips improved performance over a 10-nanometer chip by 40 percent and sent signals into the computing community that Moore’s law may not be done yet. SiGe may also prove useful to the company’s quantum computing ambitions. In 2020, IBM researchers published a paper, Investigating Microwave Loss of SiGe using Superconducting Transmon Qubits, exploring the fabrication of transmon quantum bits (qubits) on SiGe layers.

While these areas of research are still in their formative stages, suffice to say whatever ambivalence Meyerson may have faced at the company about SiGe’s utility, it no longer exists.