Is Moore’s Law Still Ticking?
May 23, 2022
Blog
Moore’s law, the prediction that the number of transistors packed into an integrated circuit (IC) would double every two years while computing costs are halved, has had a remarkably long run. When Gordon Moore, co-founder and chairman emeritus of Intel, made this bold observation in 1965, few could have imagined that state-of-the-art microchips would one day contain a billion times more circuits. Thanks to ever-shrinking circuits, computing devices have steadily grown smaller, faster and cheaper, year after year, for more than five decades.
But Moore’s law has run up against the laws of physics. Leading foundries and chipmakers are reaching the physical limits of process technology, at least as we know it today. Migration from the 7-nanometer (nm) node to 5 nm and smaller has become exponentially difficult and costly. Gains in IC performance, density and cost reduction are now incremental. As Moore’s law continues to evolve, meaningful leaps in compute power and efficiency will be achieved through advances in multicore architectures, software, AI and machine learning, interconnects, packaging and materials science.
Meanwhile, Moore’s law still has an afterlife for many types of semiconductor devices manufactured on the trailing edge of process technology including those containing analog circuitry or no transistors at all. MEMS timing devices are a prime example. Let’s see how.
Timing is the heartbeat of electronic systems, delivering an accurate, stable signal – like a human heartbeat – that provides a reference for all digital components in the system. Timing devices include passive resonators, active oscillators, and integrated clock generators and buffers, each of which provides different functionality for systems. There are two basic components inside a timing device: a resonator that vibrates at a resonant frequency and an analog IC that converts these vibrations into electrical signals and distributes them. These components are combined in a system-in-package (SIP) device to form an integrated timing solution.
Most resonators are based on crystalline quartz, which requires a precise cut during manufacturing to achieve the required resonant frequency. While quartz is a mature, widely used technology that has served the electronics industry for more than 70 years, it has limitations such as size, fragility, susceptibility to mechanical stressors, and aging effects over time and temperature. Moreover, Moore’s law doesn’t apply to quartz technologies due to these inherent limitations of manufacturing and packaging constraints.
While Moore’s law doesn’t apply directly to the design and manufacturing of silicon MEMS timing devices, the benefits of migrating from quartz to MEMS are comparable: exponentially smaller size, lower cost, and higher performance. Let’s see how.
Moore’s law is enabled by advancements in semiconductor manufacturing that increase transistor density with each new generation of process technology, particularly photolithography. In contrast, advances in MEMS aren’t derived from process manufacturing, but rather innovations in MEMS technology and IC design. While Moore’s law doesn’t apply directly to the design and manufacturing of silicon MEMS timing devices, the benefits of migrating from quartz to MEMS are by analogy comparable: the ability to scale production and achieve exponentially smaller size, lower cost and higher performance. Let’s see how.
In recent years, silicon-based microelectromechanical system (MEMS) technology has emerged as a superior alternative to quartz resonators (see Figure 1). Compared to their quartz counterparts, silicon MEMS-based resonators are undergoing exponential improvements in technology and device performance, resulting in higher performance, lower power, smaller size, and superior programmability. They are also typically more robust and reliable in harsh environments, including changing temperatures, shock, and vibration.
Figure 1. Similar to how Moore’s law has doubled transistor density while halving power, SiTime’s MEMS-based timing devices continue to improve with each generation for critical timing metrics.
For these reasons, SiTime’s MEMS-based timing solutions are rapidly displacing quartz across a wide range of markets including consumer electronics, IoT, computing, 5G infrastructure, industrial automation, automotive, and aerospace-defense. How quickly will today’s $8 billion timing industry transition from quartz to silicon MEMS technology? Time will tell.
