The word "memory" in computing has, for more than fifty years, meant one of a very small number of things: a capacitor that forgets in milliseconds unless refreshed, a floating gate that forgets nothing but wears out after being written a few thousand times, or a latch built from six transistors that never forgets but never fits densely enough. Between roughly 1970 and 2010, DRAM, NAND flash, and SRAM formed a remarkably stable triangle, and the engineering discipline of memory design was, to a first approximation, the discipline of scaling these three devices to ever smaller dimensions.
That stability has ended. As charge-based devices approach the point where a handful of electrons must reliably represent a bit, the industry has turned to an entirely different class of physical mechanisms to store information: the position of oxygen vacancies in a metal-oxide filament, the relative orientation of two magnetic layers separated by a nanometer of insulator, the crystallographic phase of a chalcogenide alloy, and the displacement of ions within a ferroelectric lattice. These are the emerging memory technologies - ReRAM, MRAM, PCM, and their ferroelectric and ionic cousins - and alongside them, conventional NAND flash has itself been reinvented as a three-dimensional structure that stacks hundreds of layers vertically to keep the cost-per-bit curve bending downward.
This book was written to give students, practicing engineers, and researchers a single, coherent treatment of this landscape. It is organized to be read cover to cover as a two-semester graduate sequence, or dipped into chapter by chapter as a reference. Part I re-establishes the foundations - the memory hierarchy, the physics of conventional SRAM, DRAM, and floating-gate flash, and the specific scaling walls that motivate everything that follows. Part II is devoted to the remarkable re-engineering of NAND flash into three dimensions. Parts III and IV form the technical core of the book: resistive RAM, magnetoresistive RAM, phase-change memory, ferroelectric memory, and a chapter surveying memristors, conductive-bridge memories, and two-dimensional-material devices. Part V steps back to compare these technologies on a common footing, examine their reliability physics, and then look forward to what emerging memories make possible at the system level - storage-class memory, CXL-attached memory pools, in-memory analog computing, and neuromorphic hardware built directly from memristive synapses.
Wherever possible, device physics is presented alongside the circuit- and system-level consequences of that physics, because a memory technology's switching mechanism is only interesting insofar as it determines the speed, endurance, retention, density, and energy budget that a system designer must work within. Each chapter therefore pairs qualitative physical explanation with quantitative comparison tables, worked numerical examples, self-drawn schematic figures, and end-of-chapter review questions intended to consolidate the material.
A note on numbers: this is a fast-moving field, and any performance figure quoted for a commercial or research device is a snapshot that a subsequent process node or materials breakthrough will likely improve upon. Wherever specific figures are given, they should be read as representative, order-of-magnitude values that are broadly consistent with the published literature and public roadmaps at the time of writing, useful for building engineering intuition and for comparing technologies against one another - not as a substitute for the current datasheet of any specific commercial part, which the practicing engineer should always consult directly.
