Electronics Guide

Why Physics Underpins Every Circuit

The behavior of electronic devices is governed by physical law. Understanding those laws provides insight into how components work, why they have the characteristics they do, and where their limits lie. From the quantum mechanical behavior of electrons in a semiconductor crystal to the electromagnetic fields that carry a radio signal across a continent, physics forms the foundation on which all electronics rests.

Practical work rarely requires solving the underlying equations. Engineers design amplifiers with small-signal models, lay out boards with impedance targets, and select parts from data sheets. Those abstractions succeed because they are careful summaries of the physics, and they fail in predictable ways when their assumptions break down. A logic gate that works perfectly on a breadboard misbehaves on a fast board because the interconnect stopped being a wire and became a transmission line. A power stage that simulates cleanly overheats because the thermal path was never modeled. In each case the recovery begins with the physics.

This category gathers the physical principles that repay study most directly: the solid-state physics that explains semiconductors, the quantum mechanics that explains the solid-state physics, the measurement theory that governs how quantum states are read out, the acoustics that couples electronics to the mechanical world, and the plasma physics that both manufactures modern chips and drives the pursuit of fusion energy.

Articles in This Category

From Physical Law to Circuit Model

Electronics is built on a stack of models, each one a deliberate simplification of the layer beneath it. Quantum mechanics describes electrons in the periodic potential of a crystal lattice and predicts the allowed and forbidden energy bands that distinguish a conductor from a semiconductor from an insulator. Band theory combined with Fermi-Dirac statistics yields carrier concentrations and mobilities. Those quantities feed drift-diffusion transport, which describes current flow in a doped material. Device physics applies transport to specific structures, the p-n junction and the metal-oxide-semiconductor capacitor above all, and produces the compact models that circuit simulators evaluate. Only at the top of the stack do Kirchhoff's laws appear, treating each device as a two-terminal or three-terminal black box with a defined current-voltage relationship.

Every layer carries a validity boundary, and the useful skill is recognizing when one has been crossed. Kirchhoff's laws are themselves an approximation to Maxwell's equations, valid only while the circuit is electrically small, meaning its physical dimensions are a small fraction of the shortest wavelength present. Small-signal transistor models hold only for excursions small enough that the discarded higher-order terms remain negligible. The ideal diode equation assumes an ideality factor near unity and ignores series resistance and high-level injection. None of these approximations is wrong; each simply has a domain, and the physics in this category is what defines those domains.

Moving down the stack costs effort and buys accuracy. A designer who can drop from lumped circuit theory to field theory when the edge rates demand it, or from a compact transistor model to band diagrams when a device behaves strangely at temperature, holds a durable advantage over one who can only work at a single level.

Charge Carriers and the Solid State

Silicon dominates electronics because of a fortunate combination of properties rather than any single superlative. Its diamond-cubic lattice packs roughly 5 × 10²² atoms per cubic centimeter, and its indirect bandgap of about 1.12 eV at 300 K is wide enough to keep thermally generated carriers negligible at ordinary operating temperatures, yet narrow enough that junction voltages well under a volt remain practical. Intrinsic silicon contains on the order of 10¹⁰ carriers per cubic centimeter, which makes it a poor conductor; deliberate doping in the range of 10¹⁴ to 10²⁰ impurity atoms per cubic centimeter raises conductivity by many orders of magnitude. At the light end that amounts to fewer than one dopant atom per hundred million lattice atoms, which is why semiconductor manufacturing demands purity and contamination control unmatched in any other industry.

Two carrier types share the work. Electrons occupy the conduction band and holes represent vacancies in the valence band, and the two respond differently to an applied field. In lightly doped silicon at room temperature, electron mobility is roughly 1400 cm²/(V·s) while hole mobility is closer to 450 cm²/(V·s). That roughly threefold asymmetry has direct layout consequences: an n-channel MOSFET drives more current than a p-channel device of the same geometry, so CMOS designers routinely make the p-channel transistor wider to balance rise and fall times. Mobility also falls with heavy doping, because ionized impurities scatter carriers, and with rising temperature, because lattice vibrations do.

Other materials win where silicon's properties fall short. Gallium arsenide has a direct bandgap near 1.42 eV, so an electron can recombine with a hole and emit a photon without requiring a lattice vibration to conserve momentum, which is what makes direct-gap compounds the basis of laser diodes and efficient light-emitting diodes. Wide-bandgap materials serve power electronics: 4H silicon carbide has a bandgap near 3.26 eV and gallium nitride near 3.4 eV, and both sustain critical electric fields roughly ten times higher than silicon. A higher critical field permits a thinner, more heavily doped drift region for a given blocking voltage, which is the physical reason a silicon carbide or gallium nitride switch achieves far lower on-resistance and faster switching than a silicon device of the same rating.

Thermal Energy, Statistics, and Noise

Temperature enters electronics through a single quantity that appears everywhere: the thermal voltage kT/q, about 25.9 mV at 300 K. It sets the scale on which carrier populations respond to potential. Because carrier energies follow Boltzmann statistics, a forward-biased diode with an ideality factor of one changes current by a factor of ten for roughly every 60 mV of applied voltage. That steep exponential is why a silicon diode's forward drop barely moves across decades of current, and therefore why treating it as a fixed 0.6 to 0.7 V works as well as it does.

The same statistics impose a hard limit on digital logic. The subthreshold swing of a conventional MOSFET, meaning the gate voltage needed to change drain current by one decade, cannot fall below ln(10) × kT/q, which is about 60 mV per decade at 300 K. That floor is thermodynamic rather than technological. It means a transistor cannot be turned off arbitrarily sharply, which in turn means the supply voltage cannot be reduced arbitrarily while preserving an acceptable on-to-off current ratio, which in turn keeps switching energy above a floor. This chain of reasoning explains a great deal about modern computing, including why clock frequencies stopped climbing in the mid-2000s while transistor counts kept rising, and why steep-slope devices such as tunnel field-effect transistors and negative-capacitance devices attract sustained research attention.

Thermal agitation also sets the noise floor. Johnson-Nyquist noise gives any resistor an open-circuit noise voltage of √(4kTRB); a 1 kΩ resistor at room temperature has a noise voltage density of about 4 nV/√Hz, independent of the material or construction of the resistor. Shot noise, √(2qI), arises because current consists of discrete charges arriving at random times. Neither is a defect to be engineered away. Together they define the ultimate sensitivity of amplifiers, radio receivers, and instrumentation, and they are the reason low-noise design is a matter of choosing impedances, bandwidths, and sometimes operating temperatures rather than simply buying better parts.

Heat removal is thermal physics applied under commercial constraint. Power dissipated in a die must reach the ambient environment through a chain of thermal resistances measured in kelvins per watt, from junction to case to heat sink to air. Silicon conducts heat reasonably well, near 150 W/(m·K), but package interfaces and air films dominate the chain. Junction temperature limits of roughly 125 to 175 °C for silicon devices exist because failure mechanisms accelerate with temperature in Arrhenius fashion; electromigration, dielectric breakdown, and solder fatigue all speed up sharply as the junction runs hotter. At the far end of the scale sits the Landauer limit, the minimum energy of kT ln 2, about 2.9 × 10^-21 J at 300 K, that thermodynamics requires for erasing a single bit. Real logic dissipates many orders of magnitude more, but the limit is genuine and clarifies which inefficiencies are engineering problems and which are not.

Fields, Waves, and Interconnects

Maxwell's equations govern every circuit, but their effects stay hidden while the circuit remains electrically small. Once physical dimensions approach a meaningful fraction of a wavelength, or once signal transition times approach the propagation delay along a conductor, the lumped abstraction collapses and interconnect must be treated as a transmission line with characteristic impedance, reflections, and delay.

The numbers involved are easy to carry in the head. A signal travels about 30 cm in a nanosecond in vacuum, and considerably slower in a dielectric. On a typical FR-4 board, whose relative permittivity falls between about 4.0 and 4.5, a microstrip trace delays a signal by roughly 150 ps per inch and a stripline trace by roughly 180 ps per inch, the buried stripline being slower because all of its field lives in the dielectric rather than partly in air. A common rule of thumb treats an interconnect as a transmission line once its one-way delay exceeds about a sixth of the signal's rise time. With edge rates measured in tens of picoseconds on modern digital interfaces, that threshold falls to a few millimeters, which is why controlled impedance, length matching, and return-path continuity became mandatory practice rather than specialist concerns.

Frequency also redistributes current within a conductor. The skin effect confines high-frequency current to a surface layer of depth √(ρ/πfμ), which in copper is roughly 66 µm at 1 MHz and about 2 µm at 1 GHz. Because the effective cross-section shrinks as frequency rises, AC resistance grows approximately with the square root of frequency; this is why litz wire subdivides conductors in switching magnetics, why surface finish and plating metallurgy affect gigahertz loss, and why a thick conductor buys nothing above a certain frequency.

Fields that are not confined to their intended path become interference. Capacitive coupling between adjacent traces, inductive coupling between current loops, and shared return-path impedance produce crosstalk, ground bounce, and radiated emission. Electromagnetic compatibility work is applied field theory: reducing loop areas, providing continuous return paths, controlling common-mode currents on cables, and shielding where the field cannot otherwise be contained.

Quantum Effects at the Device Scale

For most of the transistor's history, quantum mechanics explained why devices worked. In advanced processes it now constrains how small they can be made. Gate dielectric scaling is the clearest illustration: silicon dioxide was thinned to roughly 1.2 nm, about five atomic layers, at the 65 nm generation, at which point electrons tunneled directly through the insulator in numbers that wasted significant power. The industry's answer was to change materials rather than dimensions, adopting hafnium-based high-k dielectrics with metal gates, introduced by Intel at its 45 nm node in 2007, which restore the required capacitance using a physically thicker film that tunneling penetrates far less readily.

Electrostatic control drove a parallel change in geometry. Planar transistors gave way to FinFETs, in which the gate wraps three sides of a raised channel, and FinFETs are now giving way to gate-all-around nanosheet structures in which stacked horizontal channels are surrounded on all sides. TSMC's N2 process, the company's first to use gate-all-around nanosheets, entered volume production in the fourth quarter of 2025. Each step exists because short-channel effects and leakage, both fundamentally quantum and electrostatic in origin, could no longer be suppressed by the previous geometry.

Tunneling is not only a nuisance. Flash memory programs and erases its cells by driving electrons through a thin oxide barrier, storing charge that would be classically trapped forever. The tunnel diode and the resonant-tunneling diode exploit tunneling to produce negative differential resistance. The scanning tunneling microscope converts the exponential dependence of tunneling current on distance into atomic-resolution imaging. Josephson junctions, in which Cooper pairs tunnel through a thin barrier between superconductors, underpin both the primary voltage standard and the superconducting qubits used in quantum computers. Quantum confinement is likewise put to work in the quantum-well active regions of laser diodes, the two-dimensional electron gas of a high-electron-mobility transistor, and the quantum dots that narrow the emission spectra of display backlights.

Where quantum states carry information rather than merely explain conduction, measurement becomes a design problem in its own right. Reading a qubit disturbs it, readout fidelity and speed trade against one another, and dispersive and weak measurement techniques exist precisely to manage that trade. The same theory governs quantum sensing, where devices such as superconducting quantum interference devices and nitrogen-vacancy centers in diamond reach sensitivities that no classical instrument approaches.

Plasma, Light, and the Fabrication Environment

Plasma physics reaches electronics from an unexpected direction: it is the working medium of semiconductor manufacturing. Reactive-ion etching accelerates ions toward the wafer surface to produce the near-vertical sidewalls that purely chemical wet etching cannot achieve, and it is the reason deep, narrow features can be patterned at all. Plasma-enhanced chemical vapor deposition supplies the energy for film growth electrically rather than thermally, allowing dielectrics to be deposited at temperatures that already-formed structures can survive. Sputtering uses ion bombardment of a target to deposit metals. Ion implantation places dopants at controlled depth and dose, replacing the diffusion furnaces of earlier decades.

Lithography depends on plasma as well. Extreme ultraviolet exposure at 13.5 nm has no conventional light source; instead, droplets of molten tin are struck by a high-power carbon dioxide laser, and the resulting hot, dense plasma radiates in the required band, with the emission collected by multilayer mirrors because no material transmits usefully at that wavelength. The most advanced patterning tool in the industry is, at its heart, a plasma light source.

The same discipline supports the pursuit of fusion energy, where hydrogen isotopes must be heated to conditions at which nuclei overcome mutual repulsion. Magnetic confinement approaches such as the tokamak and the stellarator hold the plasma with strong shaped fields; inertial confinement compresses a fuel capsule with laser or particle-beam drive. Electronics appears on both sides of this effort, as the fast control systems, magnet power supplies, and diagnostic instrumentation that make experiments possible, and as a prospective beneficiary of the resulting energy supply.

Acoustic and Mechanical Coupling

Electronics rarely stays purely electrical. The piezoelectric effect, discovered by Jacques and Pierre Curie in 1880, converts mechanical strain into charge and an applied field into strain, and it provides the bridge between electrical circuits and the mechanical world. Quartz resonators built on that effect give oscillators their stability: the 32.768 kHz tuning-fork crystal in watches and real-time clocks is chosen because it is a power of two, so fifteen stages of binary division yield exactly one pulse per second, while megahertz crystals clock microcontrollers to within a few parts per million.

Acoustic devices dominate radio-frequency filtering for a reason rooted in wave physics. Acoustic waves travel through a solid at a few thousand meters per second, roughly five orders of magnitude slower than electromagnetic waves, so an acoustic resonator at a given frequency is about a hundred thousand times smaller than an electromagnetic one. Surface acoustic wave filters and bulk acoustic wave or thin-film bulk acoustic resonator filters exploit that compression to deliver steep, low-loss channel selectivity in packages of a few square millimeters. A modern smartphone supporting many cellular bands contains dozens of them, and without acoustic filtering the radio front end would not fit in a handset.

Ultrasound extends the same physics to sensing. Medical imaging typically operates between roughly 1 and 15 MHz, trading penetration depth against resolution, since higher frequencies resolve finer structures but attenuate faster in tissue. Industrial applications include non-destructive testing for cracks and thickness, flow measurement by transit-time difference, level sensing, and the short-range ranging used in parking sensors. Microelectromechanical systems carry the coupling further still, etching accelerometers, gyroscopes, microphones, pressure sensors, and timing resonators directly into silicon, so that mechanical physics and integrated electronics share a single die.

Physics as a Design Constraint

Each subcategory here addresses a boundary that engineering practice eventually reaches. Semiconductor physics explains what a device can do and why its parameters drift with temperature and doping. Quantum mechanics explains the band structure that semiconductor physics assumes, and increasingly sets the leakage and scaling limits of leading-edge processes. Quantum measurement theory governs how information is extracted from states that cannot be observed without disturbance, which matters wherever quantum sensing or quantum computing is involved. Acoustics and ultrasonics supply the electromechanical coupling behind timing references, radio filters, and imaging. Plasma physics operates the fabrication environment in which chips are made and frames the long effort toward fusion power.

The recurring lesson is that the abstractions of circuit design are honest summaries with stated limits, and that progress in electronics has repeatedly consisted of noticing which limit was about to bind and changing the physics rather than the circuit. Thinner oxide gave way to a different dielectric material. Planar channels gave way to wrapped gates. Silicon power switches gave way to wide-bandgap devices where the electric field, not the current, was the constraint. Practitioners who understand the underlying principles read such transitions in advance instead of encountering them as surprises.

Related Topics

The principles collected here are applied throughout the rest of this guide: