The manufacturing method used to make silicon chips, which is measured by the size of the transistor's elements. The driving force behind the design of integrated circuits is miniaturization, and process technology boils down to the never-ending goal of "make it smaller." As transistors get smaller, they switch faster and use less energy. Smaller also means more computing power per square inch that can be placed into ever tighter quarters. See
EUV machine and
digital perfection.
Feature Size Measured in Nanometers
The size of the elements that make up a transistor are measured in nanometers. For example, a 32 nm process technology refers to features 32 nm in size. Also called a "technology node" and "process node," early chips were measured in micrometers (see table below).
The feature size may refer to the length of the silicon channel between source and drain in the transistor. However, it may also be measured as the space between wires on a wiring layer. For the different types of transistors, see
FET.
All Chips Today Are Not Smaller
The smallest feature sizes are found on the latest, high-end CPU, SoC and AI chips that retail from several hundred to several thousand dollars apiece. However, 8-bit and 16-bit microcontrollers (MCUs) are used by the billions every year and sell for as little as a dollar and less in quantity. They use far fewer transistors and have feature sizes like the high-end chips a decade or two earlier. See
microcontroller,
CPU,
SoC and
AI chip.
Unbelievable How Small They Are!
Almost impossible to fathom, using state-of-the-art process technology, one square millimeter holds more than 100 million transistors. That is an area roughly equivalent to the head of a pin! See
transistor density.
The following table of feature sizes does not mean every chip manufacturer improved its chips in the same years. Nevertheless, it shows the progression over the decades.
Note that high-voltage chips cannot be as miniaturized as other chips, and microcontrollers (MCUs) are always a few generations behind the most advanced designs. See
microcontroller,
active area,
half-node and
feature size.
CHIP FEATURE SIZES
Nanometers Micrometers
Year (nm) (µm)
1957 120,000 120.0
1963 30,000 30.0
1971 10,000 10.0
1974 6,000 6.0
1976 3,000 3.0
1982 1,500 1.5
1985 1,300 1.3
1989 1,000 1.0
1993 600
1996 350
1998 250
---------------------------------------------
1999 180 Still used for high-voltage
2001 130 applications and MCUs.
2003 90 See microcontroller.
----------------------------------------------
2005 65
2008 45
2010 32
----------------------------------------------
2012 22 Chips always improve due to
2014 14 new designs, but not smaller
2017 10 features. Since the 2010s,
2018 7 chips are still branded
2020 5 using ever-decreasing
2022 3 nanometers. For example, in
2024 2 2025, chips have features
2025 1.8 around 21 nm, not 1 or 2 nm.
Half a Micrometer in Five Years
In the 1990s, feature sizes of these AMD CPUs were reduced from 800 nanometers (0.8 µm) on the left to 350 nanometers (0.35 µm) on the right. That may not seem like much, but half a micrometer back then was huge. See
transistor density.
(Images courtesy of Advanced Micro Devices, Inc.)