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.
ThriftBooks sells millions of used books at the lowest everyday prices. We personally assess every book's quality and offer rare, out-of-print treasures. We deliver the joy of reading in recyclable packaging with free standard shipping on US orders over $15. ThriftBooks.com. Read more. Spend less.