Epitaxy (MBE and MOCVD)
Growth of single-crystal layers in registry with a substrate, with composition and thickness controlled to the nanometer. The fabrication step where a laser's quantum wells are created and most of its performance is decided.
Epitaxy is crystal growth as deposition: atoms arriving at a heated single-crystal substrate arrange themselves to continue its lattice, layer by layer. Everything a heterostructure device is (barrier compositions, quantum-well thicknesses, doping profiles, strain) is written during epitaxy and only revealed by later processing. A laser's threshold current density and reliability are, to first order, epitaxy report cards.
Two growth technologies carry the industry:
MOCVD (metal-organic chemical vapor deposition; also MOVPE). Precursor gases (trimethylgallium, arsine, phosphine) flow over substrates at hundreds of mbar and ~600–750 °C, reacting at the surface. Multi-wafer reactors, high throughput, excellent uniformity, and comfortable handling of phosphides make MOCVD the production tool for telecom lasers, LEDs, and most III-V optoelectronics.
MBE (molecular beam epitaxy). Elemental beams evaporate from effusion cells toward the substrate in ultra-high vacuum ( mbar class); shutters gate each species with sub-monolayer timing, while RHEED oscillations count atomic layers in real time. Slower and single-wafer-scaled, but supreme in interface abruptness, low background impurity, and low-temperature control. MBE is the traditional tool for quantum cascade lasers (now also grown by MOCVD), quantum-dot structures, Sb-based materials, and research heterostructures generally.
Shared vocabulary of the craft: lattice matching (alloy compositions chosen so layer and substrate lattice constants agree, InGaAsP/InP's gift); pseudomorphic strain (thin layers deliberately mismatched below the critical thickness, storing strain that beneficially reshapes band structure; strained QWs cut laser thresholds); relaxation and dislocations (what happens past critical thickness, the defects that kill emitters); plus the growth-quality diagnostics of X-ray diffraction, photoluminescence mapping, and atomic-force microscopy that qualify a wafer before a single device is made.
Frontier note: quantum-dot active regions grown directly on silicon tolerate dislocations far better than quantum wells, keeping direct hetero-epitaxy of lasers on silicon (the endgame that would obsolete bonding) an active race.
References: L. A. Coldren, S. W. Corzine and M. L. Mašanović, Diode Lasers and Photonic Integrated Circuits, 2nd ed. (Wiley, 2012). H. C. Casey Jr. and M. B. Panish, Heterostructure Lasers, Part A: Fundamental Principles (Academic Press, 1978). S. M. Sze and K. K. Ng, Physics of Semiconductor Devices, 3rd ed. (Wiley, 2007).