Semiconductors Explained

Band Theory and the Band Gap

The electrical behavior of semiconductors is explained by band theory, which describes how the energy levels available to electrons in a solid form continuous bands rather than the discrete levels found in isolated atoms. Two bands are critical: the valence band, which is the highest energy band that is normally filled with electrons, and the conduction band, the next higher band that is normally empty. The energy difference between the top of the valence band and the bottom of the conduction band is the band gap.

In metals, the valence and conduction bands overlap, so electrons can move freely and conduct electricity with no energy input. In insulators like diamond, the band gap is so large (5.5 electron volts) that virtually no electrons can jump from the valence band to the conduction band at room temperature. Semiconductors sit in between, with band gaps typically ranging from 0.1 to 3.5 electron volts. Silicon has a band gap of 1.12 eV, germanium 0.67 eV, and gallium arsenide 1.42 eV. At room temperature, thermal energy promotes a small number of electrons across the band gap, creating modest conductivity that increases strongly with temperature, the opposite of metallic behavior.

Doping: Controlling Conductivity

The real power of semiconductors lies in the ability to control their conductivity through doping, the intentional introduction of specific impurity atoms into the crystal lattice. Silicon has four valence electrons and forms four covalent bonds with its neighbors in the diamond cubic crystal structure. When a Group V element like phosphorus or arsenic (five valence electrons) replaces a silicon atom, four electrons participate in bonding and the fifth is loosely bound, requiring only about 0.045 eV to enter the conduction band. This creates an n-type semiconductor with excess negative charge carriers (electrons). The dopant atoms are called donors because they donate electrons to the conduction band.

Conversely, when a Group III element like boron or gallium (three valence electrons) replaces a silicon atom, one bond is left incomplete, creating a hole, an absence of an electron that behaves as a positive charge carrier. This produces a p-type semiconductor. The dopant atoms are called acceptors because they accept electrons from the valence band. Doping concentrations are extremely small by most standards, typically one dopant atom per million to one per billion silicon atoms, yet they increase conductivity by factors of thousands to millions.

Modern semiconductor fabrication requires doping profiles controlled with nanometer precision. Ion implantation accelerates dopant ions to energies of 1 to 500 keV and fires them into the silicon surface, where they embed at depths determined by their energy. The implanted dose (ions per square centimeter) and energy are controlled with better than 1 percent accuracy. Subsequent thermal annealing at 900 to 1,100 degrees Celsius activates the dopants by moving them onto substitutional lattice sites and repairing the crystal damage caused by implantation.

The p-n Junction and Diodes

The most fundamental semiconductor device is the p-n junction, formed where p-type and n-type regions meet within a single crystal. At the junction, electrons from the n-side diffuse into the p-side and holes diffuse the opposite way, leaving behind ionized dopant atoms that create an electric field in a thin depletion region. This built-in electric field opposes further diffusion, establishing equilibrium. The resulting device, a diode, conducts current easily when voltage is applied in the forward direction (positive to p-side) but blocks current in reverse, enabling rectification, signal detection, and voltage regulation.

Light-emitting diodes (LEDs) are p-n junctions in direct band gap semiconductors like gallium nitride and indium gallium nitride. When forward-biased, electrons and holes recombine in the junction region, releasing energy as photons with wavelengths determined by the band gap. Gallium nitride LEDs (band gap 3.4 eV) emit blue and ultraviolet light, and when coated with a yellow phosphor, produce the white light that has largely replaced incandescent and fluorescent bulbs, reducing lighting energy consumption by over 50 percent. Solar cells operate in reverse: photons absorbed in the p-n junction generate electron-hole pairs that the built-in electric field separates, producing a photovoltage and current.

Transistors and Integrated Circuits

The transistor, invented at Bell Labs in 1947, is the building block of all digital electronics. The dominant modern transistor type, the metal-oxide-semiconductor field-effect transistor (MOSFET), uses a gate electrode separated from the silicon channel by a thin oxide insulator. Applying voltage to the gate creates an electric field that either attracts or repels charge carriers in the channel, switching it between conducting and non-conducting states. A MOSFET can switch in under a picosecond, and modern processors contain over 100 billion transistors on a single chip, each only a few nanometers in critical dimension.

The relentless miniaturization of transistors, following the trend Gordon Moore observed in 1965, has been fundamentally a materials science achievement. Each generation of scaling required new materials: silicon dioxide gate insulators were replaced by hafnium-based high-k dielectrics when oxide thickness fell below 2 nanometers and quantum tunneling leakage became unacceptable. Strained silicon channels, created by growing silicon on a silicon-germanium substrate, increased carrier mobility by 20 to 40 percent. Copper replaced aluminum interconnects because its lower resistivity reduced signal delay in the increasingly long wires connecting billions of transistors. FinFET (fin field-effect transistor) and gate-all-around architectures wrap the gate electrode around three or all four sides of the channel, maintaining electrostatic control at dimensions below 10 nanometers.

Compound and Wide Band Gap Semiconductors

Gallium arsenide (GaAs) is the most important compound semiconductor, with electron mobility roughly six times higher than silicon and a direct band gap that makes it efficient for light emission and absorption. GaAs dominates in radio frequency amplifiers for cell phones (power amplifiers in every smartphone contain GaAs transistors), fiber optic communication lasers, and high-efficiency multi-junction solar cells for space applications, where GaAs-based cells achieve efficiencies exceeding 47 percent under concentrated sunlight.

Wide band gap semiconductors including silicon carbide (SiC, band gap 3.3 eV) and gallium nitride (GaN, band gap 3.4 eV) are transforming power electronics. Their larger band gaps allow operation at higher voltages, temperatures, and switching frequencies than silicon. SiC MOSFETs are now standard in electric vehicle inverters (Tesla adopted SiC in 2018), reducing power conversion losses by 50 to 75 percent compared to silicon IGBTs. GaN high-electron-mobility transistors (HEMTs) are replacing silicon in laptop chargers, data center power supplies, and 5G base station amplifiers, enabling smaller, lighter, and more efficient power conversion.

Semiconductor Manufacturing

Semiconductor fabrication begins with growing a single crystal of ultra-pure silicon using the Czochralski method, pulling a seed crystal slowly from a crucible of molten silicon to form cylindrical ingots up to 300 millimeters in diameter and 2 meters long. The silicon purity required is extraordinary: fewer than one impurity atom per billion silicon atoms, making electronic-grade silicon one of the purest materials ever produced. The ingots are sliced into thin wafers, polished to atomic flatness, and processed through hundreds of steps involving photolithography, etching, deposition, implantation, and chemical-mechanical polishing to create the intricate patterns of a modern integrated circuit.

Photolithography defines the transistor patterns by projecting ultraviolet light through a mask onto a photosensitive resist layer. Extreme ultraviolet (EUV) lithography, using 13.5 nanometer wavelength light, is required for the smallest features in leading-edge chips. Each EUV scanner costs over 300 million dollars, weighs 180 tonnes, and represents one of the most complex machines ever built, using tin droplet targets struck by carbon dioxide laser pulses to generate the extreme ultraviolet light. The entire semiconductor manufacturing ecosystem, from crystal growth through packaging, represents one of the largest and most materials-intensive industries in the world.

The Future of Semiconductor Materials

As conventional silicon scaling approaches fundamental physical limits, new materials are being explored to continue performance improvements. Two-dimensional semiconductors like molybdenum disulfide (MoS2) and tungsten diselenide (WSe2) are atomically thin layers that can serve as transistor channels with thickness of less than one nanometer, potentially enabling continued scaling beyond the limits of bulk silicon. Carbon nanotube transistors have demonstrated switching speeds and energy efficiency superior to silicon at equivalent dimensions, though manufacturing challenges in producing aligned, sorted nanotubes at wafer scale remain significant.

Neuromorphic computing materials aim to replicate the energy efficiency of biological neural networks. Phase-change materials like germanium-antimony-tellurium (GST) alloys switch between amorphous and crystalline states to store analog weights for artificial synapses, enabling in-memory computing that avoids the energy-intensive data movement between separate processor and memory chips. Ferroelectric hafnium oxide, compatible with existing silicon fabrication, offers non-volatile memory and analog computing capabilities in a material already used in production transistors. These emerging semiconductor materials are not replacing silicon but rather expanding the palette of electronic materials available to chip designers.