CMOS, short for complementary metal-oxide-semiconductor, is the method used to build nearly all of today’s chips, from laptop processors to the sensor in a phone camera. The word describes a design idea rather than a single part: tiny switches are always used in opposite-acting pairs. That means a CMOS chip spends power mainly when it changes state, not while it sits still, which is what made billions of switches on one battery-powered chip practical.1GeeksforGeeks. CMOS Inverter
How a CMOS Switch Pair Works
A transistor is a microscopic switch controlled by voltage rather than by a moving part. CMOS uses two kinds. The first, called NMOS, switches on when its input voltage is high. The second, PMOS, switches on when its input is low.2GeeksforGeeks. CMOS Inverter Wire one of each to the same input and exactly one is on at any moment.3GeeksforGeeks. CMOS Inverter
The simplest circuit built this way is the inverter. When the input is low, the PMOS turns on and connects the output to the supply voltage, so the output reads high, or 1.4Allele Electronic. What Is CMOS? Semiconductor Technology Explained When the input is high, the NMOS turns on and connects the output to ground, so it reads low, or 0.5Allele Electronic. What Is CMOS? Semiconductor Technology Explained The output is always the opposite of the input, which is why the circuit is called an inverter.6GeeksforGeeks. CMOS Inverter Every more complex gate in a chip is built from the same complementary pairing.
The name fits this picture too. “Complementary” is the opposite-acting pair. “Metal-oxide-semiconductor” describes the layers of a switch: a control electrode, a thin insulating layer, and silicon. Modern chips no longer use plain metal or oxide exactly, but the name stuck.
The two halves are deliberately made different sizes. The positive charges that carry current in PMOS move roughly three times more slowly than the electrons in NMOS.7Snubber. What Is CMOS? Semiconductor Technology Explained So designers make the PMOS about 2 to 3 times wider so it can push the output up as fast as the NMOS pulls it down.8Medium. Why is PMOS Wider than NMOS in a CMOS Inverter? Matched speed in both directions keeps signals clean and avoids delays.9Snubber. What Is CMOS? Semiconductor Technology Explained
The flip itself is not instant. For a brief moment while the input is changing, both switches are partly on, so a small burst of current can pass from supply to ground. Afterward, the newly connected switch has to fill or drain the tiny electrical storage at the output, which takes a short but real amount of time. Repeated through long chains of gates, that charging time sets how fast a chip can run.
Why CMOS Took Over the Chip Industry
Low idle power was the decisive advantage. Because one switch in each pair is always off, no steady current flows from supply to ground while a CMOS gate holds a value.10GeeksforGeeks. CMOS Inverter The older TTL logic family, built from a different kind of transistor, draws significant power even when idle.11Grasp. TTL vs. CMOS: Electrical Characteristics
The reason a single-type design wastes power is its load. In an NMOS-only gate, the output is pulled high through a resistor-like load that stays connected to the supply all the time. Whenever such a gate holds a low output, current flows steadily from supply to ground through that load. In CMOS the load is replaced by the PMOS switch, which turns off exactly when the NMOS turns on, so the path to ground never stays open. A little wasted current per switch is harmless with a thousand switches, but multiplied by billions it would melt a chip or drain a battery in minutes.
CMOS is also robust. Its outputs swing all the way to the supply voltage and all the way to ground, leaving wide margins between a 0 and a 1.12Grasp. TTL vs. CMOS: Electrical Characteristics Stray electrical noise is therefore less likely to flip a bit by mistake.13Grasp. TTL vs. CMOS: Electrical Characteristics Early TTL was faster, though modern CMOS families now match or beat it.14Grasp. TTL vs. CMOS: Electrical Characteristics
Where CMOS Chips Spend Their Power
Each time a gate flips, it charges or drains a tiny amount of stored charge, and that takes energy.15IntechOpen. Power Consumption in CMOS Circuits Every wire and switch input on a chip must be filled with charge to represent a 1 and emptied to represent a 0. Filling it draws charge from the supply, and emptying it sends that charge to ground, so each full cycle costs a small, fixed amount of energy. The energy per flip is minuscule, but a modern processor performs billions of flips per second across many gates, so the total becomes watts of heat.
Switching power therefore grows in step with clock speed and with the amount of circuitry being switched.16IntechOpen. Power Consumption in CMOS Circuits It grows with the square of the supply voltage, so halving the voltage cuts this power to about a quarter.17IntechOpen. Power Consumption in CMOS Circuits Lowering voltage is the most powerful lever designers have, which is why chip voltages fell from around 5 volts toward about 1 volt.
Leakage became the hidden cost. It is current that trickles through a switch that is supposed to be off.18Federal University of Rio Grande do Sul. Subthreshold Leakage Modeling and Estimation of General CMOS Complex Gates Below about 100 nanometers, thinner insulating layers and lower switch-on voltages made leakage rise sharply.19Federal University of Rio Grande do Sul. Subthreshold Leakage Modeling and Estimation of General CMOS Complex Gates
How CMOS Chips Are Made
Chipmaking starts with a pure silicon crystal sliced into thin wafers.20Semiconductor Physics Course. CMOS Fabrication A thin oxide layer is grown in a furnace and coated with a light-sensitive material.21Semiconductor Physics Course. CMOS Fabrication Light shone through a patterned mask then marks where material will be etched away or where atoms will be implanted to change the silicon’s behaviour.22Semiconductor Physics Course. CMOS Fabrication After exposure, a chemical rinse removes either the lit or the unlit coating, leaving protected and bare areas on the wafer. The next step, etching or implanting atoms, affects only the bare areas, so the mask pattern is transferred into the silicon. This cycle repeats about 12 to 20 times to build up a CMOS chip.23Semiconductor Physics Course. CMOS Fabrication A simple teaching process needs just seven patterned layers to create both switch types and wire them into gates.24Montana State University. EELE 414 – Introduction to VLSI Design Module #4 CMOS Fabrication
Both switch types need their own regions, because NMOS and PMOS need silicon of opposite types. A “well” of different doping is created so both can sit side by side. This is part of why CMOS takes more steps than older single-type processes.
Materials changed as switches shrank. The silicon dioxide insulator under each switch’s control electrode became so thin that current leaked straight through it.25Materials Science and Engineering R: Reports. High-K materials and Metal Gates for CMOS applications Chipmakers replaced it with a thicker layer of a hafnium-based material and swapped silicon control electrodes for metal.26Materials Science and Engineering R: Reports. High-K materials and Metal Gates for CMOS applications At Intel’s 45nm generation this cut leakage by more than 25 times in NMOS and over 1,000 times in PMOS while speeding circuits up.27Intel. High-k and Metal Gate Transistor Research
The Limits of Shrinking CMOS
For decades, shrinking switches also lowered their power, keeping heat per area constant. Around 2005 to 2007 that broke down because leakage and switch-on voltage stopped shrinking with size.28Wikipedia. Dennard scaling Chips started producing more heat per area as they shrank, a barrier called the power wall.29Wikipedia. Dennard scaling In 2004 Intel cancelled planned processors that would have run too hot.30Wikipedia. Dennard scaling Heat also raises leakage, which raises heat further, a risk of runaway heating.31Wikipedia. Dennard scaling
Chips also wear out. At nanometer sizes, transistor ageing has become a major reliability concern.32Proceedings of the great lakes symposium on VLSI. Input and transistor reordering for NBTI and HCI reduction in complex CMOS gates Over years of use, effects such as heat and high voltage stress slowly change how much current a switch passes and the voltage that turns it on.33Facta Universitatis, Series: Electronics and Energetics. EVALUATING NBTI AND HCI EFFECTS ON DEVICE RELIABILITY FOR HIGH-PERFORMANCE APPLICATIONS IN ADVANCED CMOS TECHNOLOGIES
Even the label on a chip is less literal than it sounds. A process node such as “3nm” is a generation name, not a measurement of one feature. Smaller numbers mean denser, more efficient chips, and each step is harder and costlier than the last.
New Transistor Shapes Keeping CMOS Alive
Flat switches lost control of their current as they shrank, leaking and switching poorly.34PubMed Central. Comprehensive Review of FinFET Technology: History, Structure, Challenges, Innovations, and Emerging Sensing Applications The industry moved to FinFETs, where the channel stands up as a fin with the control electrode on three sides.35PubMed Central. Comprehensive Review of FinFET Technology: History, Structure, Challenges, Innovations, and Emerging Sensing Applications These switches turn off more completely, run faster and behave more consistently.36PubMed Central. Comprehensive Review of FinFET Technology: History, Structure, Challenges, Innovations, and Emerging Sensing Applications Below about 3nm, fins reach their limits, and gate-all-around designs surround the channel on all sides with stacked sheets or wires.37Digital Applications in Networking and Informatics. The Development of Gate-All-Around FET (GAA FET) Technology as a Successor to FinFET: Architecture, Fabrication Challenges, and Future Opportunities
The next step stacks the complementary pair itself. In a design called CFET, the NMOS and PMOS of each pair sit on top of each other instead of side by side.38imec. CFET (complementary FET) It can fit roughly 1.5 to 2 times more circuitry into the same space.39IEEE Xplore. Complementary Field-Effect Transistor (CFET) Demonstration at 48nm Gate Pitch for Future Logic Technology Scaling The catch is that it makes manufacturing more complex and expensive.40IEEE Xplore. Complementary Field-Effect Transistor (CFET) Demonstration at 48nm Gate Pitch for Future Logic Technology Scaling
CMOS in Cameras and Everyday Devices
A “CMOS sensor” in a camera spec means the light-sensing chip is built with the same process as other chips. Each pixel has its own small amplifier, which makes the sensor fast and easy to scale up.41Wikipedia. Active-pixel sensor Compared with older CCD sensors, CMOS sensors generally use less power, show less image lag and can be made in ordinary chip factories.42Wikipedia. Active-pixel sensor The sensor and image-processing circuits can therefore sit on a single chip.43Wikipedia. Active-pixel sensor CMOS sensors were once noisier, and engineers had to work to match CCD image quality.44SPIE Proceedings. CMOS active pixel image sensor with CCD performance
Shrinking CMOS to around 1 volt has a cost for the analog circuits that often share these chips, like those handling radio and audio: they become harder to design because of more noise and mismatch between parts.452007 IEEE Asian Solid-State Circuits Conference. Analog design challenges in nanometer CMOS technologies
On a PC, the “CMOS” battery keeps settings and the clock alive in a low-power memory chip. “Clear CMOS” erases the saved firmware settings stored in that chip, which people do to fix boot problems. On many older PCs, removing the coin battery for a while does the same, though some newer boards keep settings in memory that doesn’t need the battery.
Where CMOS Is Heading Next
Quantum computers need millions of control signals, too many wires to run from room temperature.46Proceedings of the 54th Annual Design Automation Conference 2017. Cryo-CMOS Electronic Control for Scalable Quantum Computing Researchers are building CMOS control chips that work inside the freezer next to the quantum bits.47arXiv. Cryogenic Control and Readout Integrated Circuits for Solid-State Quantum Computing One such chip ran at 3 degrees above absolute zero and controlled silicon quantum bits as accurately as room-temperature lab instruments.48PubMed. CMOS-based cryogenic control of silicon quantum circuits Designers still lack reliable models of how transistors behave at such cold temperatures, though.49arXiv. Cryogenic Control and Readout Integrated Circuits for Solid-State Quantum Computing
New materials are opening another path. Carbon nanotube films can form both halves of a CMOS pair.50ACS Nano. Monolithic Three-Dimensional Integration of Carbon Nanotube Circuits and Sensors for Smart Sensing Chips Because they are made at low temperature, logic and sensors can be stacked in layers on one chip.51ACS Nano. Monolithic Three-Dimensional Integration of Carbon Nanotube Circuits and Sensors for Smart Sensing Chips Ultra-thin 2D materials are being explored for the same kind of 3D stacking.52Nano Research. A review on monolithic 3D integration: From bulk semiconductors to low-dimensional materials

