{
  "id": "diffusion-coefficient",
  "title": "Diffusion Coefficient",
  "lang": "en",
  "source_lang": "en",
  "summary": "The diffusion coefficient quantifies the rate of spreading caused by random motion and relates diffusive flux to concentration gradients.",
  "category": "science",
  "subcategory": "physics",
  "type": "concept",
  "aliases": [
    "Diffusivity",
    "Mass diffusivity"
  ],
  "generated_at": "2026-11-04",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/diffusion-coefficient/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/diffusion-coefficient/",
      "json": "https://aiwiki.page/zh/diffusion-coefficient.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "The diffusion coefficient, usually denoted \\(D\\), is a transport property that describes how rapidly particles or substances spread by diffusion. In its simplest form, it relates the diffusive flux of a substance to its concentration gradient and determines how its mean-square displacement grows with time. Its SI unit is the square metre per second, \\(\\mathrm{m^2\\,s^{-1}}\\). A diffusion coefficient depends on the diffusing species, the surrounding medium, and conditions such as temperature, pressure, and composition; it is not generally a universal constant for a substance."
    },
    {
      "heading": "Definition through Fick’s laws",
      "text": "For an isotropic medium containing a single independently diffusing component, Fick’s first law is\n\n\\[\\mathbf{J}=-D\\nabla c,\\]\n\nwhere \\(\\mathbf{J}\\) is the diffusive flux and \\(c\\) is concentration. When \\(c\\) is an amount-of-substance concentration in \\(\\mathrm{mol\\,m^{-3}}\\), the flux has units of \\(\\mathrm{mol\\,m^{-2}\\,s^{-1}}\\). The negative sign indicates transport toward lower concentration.\n\nDiffusive flux is defined relative to a specified reference frame, such as the stationary solvent or the mixture’s mean velocity. In flowing systems, it must be distinguished from transport caused by bulk motion.\n\nCombining Fick’s first law with conservation of mass, in the absence of reactions or other sources, gives\n\n\\[\\frac{\\partial c}{\\partial t} =\\nabla\\cdot(D\\nabla c).\\]\n\nIf \\(D\\) is spatially uniform, this becomes Fick’s second law:\n\n\\[\\frac{\\partial c}{\\partial t}=D\\nabla^2c.\\]\n\nWhen \\(D\\) depends on concentration or position, it must generally remain inside the divergence. The solution also depends on the initial concentration distribution and boundary conditions.\n\nFor illustration, diffusion on an unbounded one-dimensional line from an initially localized source produces the normalized distribution\n\n\\[p(x,t)=\\frac{1}{\\sqrt{4\\pi Dt}} \\exp\\left(-\\frac{x^2}{4Dt}\\right).\\]\n\nIts width grows as the square root of time, rather than linearly with time."
    },
    {
      "heading": "Microscopic interpretation",
      "text": "The diffusion coefficient connects macroscopic transport with the random motion of individual particles. For ordinary diffusion in \\(d\\) equivalent spatial dimensions,\n\n\\[\\left\\langle |\\mathbf r(t)-\\mathbf r(0)|^2 \\right\\rangle=2dDt,\\]\n\nafter any systematic drift has been removed and on timescales where the motion is diffusive. In three dimensions, the mean-square displacement is therefore \\(6Dt\\).\n\nThis relation underlies the interpretation of Brownian motion developed by Albert Einstein in 1905 and independently by Marian Smoluchowski. It provides a way to determine diffusion coefficients by tracking particles rather than measuring concentration profiles.\n\nA characteristic diffusion time over a distance \\(L\\) is\n\n\\[t_{\\mathrm{diff}}\\sim\\frac{L^2}{D}.\\]\n\nThe numerical factor depends on geometry and the definition of the characteristic time. The quadratic dependence explains why diffusion can be effective over cellular distances yet very slow over macroscopic distances.\n\nIn statistical mechanics, the self-diffusion coefficient can also be expressed through the velocity autocorrelation function:\n\n\\[D=\\frac{1}{d}\\int_0^\\infty \\left\\langle\\mathbf v(0)\\cdot\\mathbf v(t)\\right\\rangle\\,dt.\\]\n\nThis relation assumes a stationary equilibrium system with a well-defined diffusive limit. It is widely used in molecular dynamics simulations."
    },
    {
      "heading": "Physical factors controlling diffusivity",
      "text": "Temperature and viscous resistance\n\nFor an isolated spherical particle in a continuum fluid, the Stokes–Einstein relation is\n\n\\[D=\\frac{k_{\\mathrm B}T}{6\\pi\\eta a},\\]\n\nwhere \\(k_{\\mathrm B}\\) is the Boltzmann constant, \\(T\\) is absolute temperature, \\(\\eta\\) is dynamic viscosity, and \\(a\\) is the particle’s hydrodynamic radius.\n\nThis expression assumes creeping flow, a no-slip surface, and a sufficiently dilute suspension. It predicts slower diffusion for larger particles and in more viscous fluids. Applying it to molecules comparable in size to the solvent molecules is an approximation rather than a generally exact law.\n\nIn many solids, diffusion occurs through thermally activated jumps between atomic sites. Over a suitable temperature range, the diffusion coefficient often follows an Arrhenius relation:\n\n\\[D=D_0\\exp\\left(-\\frac{E_{\\mathrm a}}{RT}\\right),\\]\n\nwhere \\(D_0\\) is a prefactor, \\(E_{\\mathrm a}\\) is a molar activation energy, and \\(R\\) is the gas constant. The apparent activation energy can include contributions from both defect formation and particle migration. A change in mechanism can produce a change in the temperature dependence.\n\nState of matter and composition\n\nDiffusion coefficients in gases are commonly of order \\(10^{-5}\\ \\mathrm{m^2\\,s^{-1}}\\) near room temperature and atmospheric pressure. Small molecules in ordinary liquids often have values around \\(10^{-9}\\ \\mathrm{m^2\\,s^{-1}}\\). Solid-state values span many orders of magnitude and depend strongly on temperature and mechanism. These are indicative scales, not universal boundaries.\n\nIn dilute gases, diffusion generally becomes faster as temperature rises and slower as pressure increases. In liquids, molecular size, solvent viscosity, and intermolecular interactions are important. In solids, vacancies, interstitial sites, grain boundaries, and other defects can provide distinct pathways.\n\nComposition also matters. A diffusion coefficient measured at infinite dilution need not describe a concentrated solution, where particle interactions and correlated motion can be substantial."
    },
    {
      "heading": "Different kinds of diffusion coefficient",
      "text": "The term covers several related quantities that should not be treated as interchangeable.\n\nSelf-diffusion coefficient. This describes the random displacement of particles in a system at equilibrium, without requiring a macroscopic composition gradient. It is a single-particle dynamical property.\n\nTracer diffusion coefficient. This is obtained by following a distinguishable minority of particles, often labeled with an isotope. If labeling does not appreciably alter particle behavior, tracer diffusion can measure self-diffusion.\n\nMutual or interdiffusion coefficient. This characterizes the relaxation of a composition gradient in a mixture. It describes collective transport and can differ from self-diffusion because it incorporates interactions between components and the thermodynamic response to a composition change.\n\nChemical diffusion coefficient. Commonly used in solids and electrochemical materials, this describes transport associated with changes in chemical composition. It combines kinetic mobility with a thermodynamic factor.\n\nEffective or apparent diffusion coefficient. This is inferred at a larger observational scale or from a particular model. It may incorporate obstacles, confinement, binding, or unresolved heterogeneity and therefore need not equal the local molecular coefficient.\n\nIn multicomponent mixtures, several coupled fluxes may respond to several composition gradients. A single scalar coefficient is then generally insufficient; transport may require a matrix of coefficients or a Maxwell–Stefan description based on friction between species."
    },
    {
      "heading": "Thermodynamic driving forces",
      "text": "A concentration gradient is a convenient driving force for simple systems, but the more general thermodynamic quantity is a gradient in chemical potential. For an ideal dilute solute under isothermal conditions, a simple mobility description gives\n\n\\[\\mathbf J=-bc\\nabla\\mu, \\qquad \\mu=\\mu^\\circ+k_{\\mathrm B}T\\ln c,\\]\n\nwhere \\(c\\) is number concentration and \\(b\\) is the mobility defined as drift velocity per applied force. These expressions give\n\n\\[D=bk_{\\mathrm B}T.\\]\n\nIn a nonideal mixture, chemical potential depends on activity, not merely concentration. The relation between mobility and a Fickian diffusion coefficient therefore includes a thermodynamic factor.\n\nConsequently, the concentration gradient of an individual component need not always point opposite to its flux. Coupled transport can produce “uphill diffusion” without violating thermodynamics: the complete system’s driving forces and fluxes determine entropy production.\n\nFor charged species, an electric field contributes to the electrochemical driving force. Transport models for ions therefore commonly combine diffusion with electrical migration rather than using Fick’s law alone."
    },
    {
      "heading": "Direction dependence and confinement",
      "text": "In an anisotropic material, diffusivity is represented by a second-rank tensor:\n\n\\[J_i=-\\sum_jD_{ij}\\frac{\\partial c}{\\partial x_j}.\\]\n\nIts principal values describe diffusion along principal directions. Layered crystals, aligned fibers, and biological structures can permit much faster motion in one direction than another.\n\nIn porous materials, an effective coefficient may depend on connected pore volume, pathway tortuosity, constrictions, and surface interactions. The precise formula depends on whether concentration and flux are defined per unit pore volume or per unit bulk volume.\n\nDiffusion must also be distinguished from permeability. For a simple homogeneous membrane, permeability depends on diffusivity together with solute partitioning and membrane thickness; it is not determined by diffusivity alone."
    },
    {
      "heading": "Measurement and applications",
      "text": "Diffusion coefficients can be determined by several complementary methods:\n\n- Concentration-profile measurements: concentration changes across a diffusion cell or interface are fitted to transport equations.\n- Tracer experiments: labeled particles or atoms are followed over time.\n- Particle tracking: trajectories yield mean-square displacements.\n- Dynamic light scattering: fluctuations in scattered light reveal diffusion-related relaxation rates under an appropriate model.\n- Pulsed-field-gradient nuclear magnetic resonance: signal attenuation measures molecular displacement over controlled observation times.\n- Fluorescence recovery measurements: recovery after local photobleaching provides an apparent diffusion coefficient when diffusion can be separated from binding and other processes.\n\nDiffusion coefficients are used to model gas mixing, separation processes, atomic transport during heat treatment, molecular motion within cells, and contaminant transport in groundwater. In flow systems, molecular diffusion must be distinguished from spreading caused by velocity variations or turbulence."
    },
    {
      "heading": "Limitations of a constant-coefficient description",
      "text": "A constant diffusion coefficient assumes an appropriate scale of observation and approximately Fickian behavior. At very short times, particle motion can be ballistic rather than diffusive. In crowded, disordered, or confining environments, the mean-square displacement may instead follow\n\n\\[\\left\\langle\\Delta r^2(t)\\right\\rangle\\propto t^\\alpha,\\]\n\nwith \\(\\alpha\\ne1\\). Such behavior is termed anomalous diffusion. A fitted generalized coefficient then has units of length squared divided by time to the power \\(\\alpha\\), and is not directly interchangeable with an ordinary diffusion coefficient.\n\nEven when Fickian fitting is useful, reported values require context: temperature, composition, measurement timescale, direction, reference frame, and the transport model can all affect the result."
    }
  ],
  "links": [
    "diffusion",
    "international-system-of-units",
    "ficks-laws-of-diffusion",
    "conservation-of-mass",
    "brownian-motion",
    "albert-einstein",
    "statistical-mechanics",
    "molecular-dynamics",
    "boltzmann-constant",
    "temperature",
    "viscosity",
    "activation-energy",
    "gas-constant",
    "pressure",
    "isotope",
    "chemical-potential",
    "thermodynamic-activity",
    "electric-field",
    "ion",
    "permeability",
    "nuclear-magnetic-resonance",
    "groundwater"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}