Viscoelasticity is the mechanical behavior of materials whose deformation combines features of elasticity and viscosity. An ideal elastic solid responds to loading by storing recoverable mechanical energy, whereas an ideal viscous fluid flows at a rate determined by the applied stress and dissipates energy. A viscoelastic material both stores and dissipates energy, with the balance depending on loading duration, deformation rate, temperature, and material structure. Consequently, the same material can appear solid-like in a rapid experiment and liquid-like over a sufficiently long time.
Viscoelasticity is central to the behavior of polymers, rubber, adhesives, asphalt, biological tissues, food products, and many glasses and geological materials. It explains why a seal slowly changes shape under compression, why a pressure-sensitive adhesive can both spread over a surface and resist removal, and why soft tissues respond differently to rapid and sustained loading. Its study belongs principally to rheology and continuum mechanics, linking molecular motion and structural rearrangement to measurable forces, displacements, and energy losses.
Characteristic responses and material classification
The defining feature of viscoelastic behavior is that stress depends on deformation history rather than solely on the instantaneous deformation or deformation rate. Two specimens at the same current strain can therefore carry different stresses if they reached that strain through different loading histories. Conversely, the deformation produced by a specified force depends on how rapidly the force was applied, how long it has acted, and whether the specimen was previously loaded.
In a creep experiment, a constant stress is applied and the resulting strain is recorded. An elastic solid reaches its deformation immediately in the idealized limit, whereas a Newtonian fluid continues deforming at a constant rate. A viscoelastic material generally exhibits an evolving strain that may include an immediate elastic contribution, delayed recoverable deformation, and continuing flow. When the stress is removed, immediate and delayed recovery can occur. Any remaining deformation must be interpreted carefully: it may reflect viscous flow, plastic deformation, damage, or recovery processes longer than the observation period.
In a stress-relaxation experiment, a strain is imposed and then held fixed. The stress required to maintain it decreases as internal degrees of freedom rearrange. Stretched polymer chains, for example, can progressively change their conformations and redistribute the imposed deformation. Complete decay of stress is characteristic of an ideal viscoelastic liquid at sufficiently long times. A viscoelastic solid retains a nonzero equilibrium stress under a maintained small strain, provided its load-bearing network remains intact.
Cyclic loading reveals another characteristic: stress and strain are out of phase, producing a hysteresis loop on a stress–strain plot. The enclosed area is the mechanical energy dissipated per unit volume during one cycle. Hysteresis alone, however, does not establish viscoelasticity. Friction, plasticity, damage, and structural transformations can also generate loops. Identification requires examining how the response changes with time, frequency, amplitude, and loading history.
Whether a material behaves as a solid or liquid is partly a question of observation time. A useful measure is the Deborah number, commonly written , where is a characteristic material relaxation time and is the experimental timescale. When , the associated structure has time to relax; when , it is effectively constrained during the measurement. Real materials usually possess many relaxation times, so a single Deborah number describes only part of their response. The classification also differs from the distinction between linear and nonlinear behavior: a material can be strongly time-dependent while remaining linear at sufficiently small deformation.
Linear constitutive theory and dynamic moduli
Linear viscoelasticity applies when scaling the imposed deformation scales the stress proportionally, and responses to separate deformation histories can be added. For a material whose properties do not change with its absolute age, the Boltzmann superposition principle expresses the shear stress as
where is shear strain, is shear rate, and is the shear relaxation modulus. The equation states that each past strain increment contributes to the present stress, with its influence weighted by the elapsed time. The lower integration limit assumes a suitably defined history; experiments beginning from an equilibrated state can instead use a finite initial time with the corresponding initial conditions.
The complementary creep description is
where is the creep compliance. Stress and modulus have units of pascals, while compliance has units of inverse pascals. Although elastic compliance is the reciprocal of elastic modulus, is generally not equal to . Relaxation and creep functions are related through an integral transformation. With zero initial history, their Laplace transforms satisfy
This relation makes conversion possible in principle, but experimental noise and incomplete time coverage complicate numerical inversion.
For sinusoidal shear strain, , the steady linear stress response is
The storage modulus measures the component in phase with strain, while the loss modulus measures the component in phase with strain rate. With the corresponding complex-exponential convention, the complex modulus is . The phase angle satisfies . These quantities describe the response at a specified frequency rather than immutable material constants.
The energy dissipated per unit volume in one sinusoidal cycle is
For a viscoelastic liquid with finite zero-shear viscosity, the terminal low-frequency limit normally gives , while decreases more rapidly, commonly as when the relevant relaxation moments are finite. An equilibrium solid instead retains a nonzero low-frequency storage modulus.
A complete three-dimensional isotropic description requires separate shear and volumetric response functions. Tensile, shear, and bulk moduli cannot generally be interconverted using a single, time-independent Poisson ratio. Causality also constrains the frequency response: storage and loss components are mathematically related, rather than independently adjustable functions. Linear viscoelasticity is thus closely related to the theory of a linear time-invariant system, with the relaxation kernel encoding the material’s mechanical memory.
Mechanical analogues and relaxation spectra
Spring-and-dashpot models provide compact representations of viscoelastic constitutive laws. A spring obeys , while a linear dashpot obeys . Here, is a shear modulus and a viscosity. These elements are mathematical idealizations: a fitted spring does not necessarily correspond to a particular molecular bond, nor does each dashpot represent a separately identifiable physical flow process. Their value lies in reproducing specified features of measured behavior with interpretable parameters.
The Maxwell model connects a spring and dashpot in series. Both carry the same stress, and their strain rates add:
Following a step strain, the stress relaxes exponentially with relaxation time :
Its creep compliance is , combining instantaneous elastic deformation with unrestricted long-time flow. Its dynamic moduli are
The model is a viscoelastic liquid: it has no equilibrium shear modulus.
The Kelvin–Voigt model places a spring and dashpot in parallel, giving . Under a step stress, its compliance increases as
It captures delayed elasticity and complete eventual recovery after unloading. It does not permit instantaneous strain under finite stress, however, and an ideal step strain would require an impulsive stress contribution from the dashpot. This illustrates why no elementary model reproduces every experimental feature.
The standard linear solid adds a spring in parallel with a Maxwell element. It has both instantaneous and equilibrium stiffness and exhibits stress relaxation without unlimited steady creep. More broadly, a generalized Maxwell representation uses multiple parallel relaxing branches:
Here is the equilibrium modulus, and each branch contributes a modulus and relaxation time . For a liquid with , this model gives . Positive branch moduli and viscosities provide a convenient way to construct passive models that do not generate mechanical energy.
Broad material responses are often better understood through a continuous relaxation spectrum than through one or two characteristic times. A spectrum represents contributions distributed over logarithmic timescale intervals. Recovering it from measured moduli is an ill-conditioned inverse problem: different spectra may fit the same limited dataset almost equally well. Consequently, a multi-exponential fit can be an effective computational representation without proving that the material contains that number of distinct microscopic mechanisms.
Molecular and structural origins
Viscoelasticity arises when a material contains internal configurations that require finite time to rearrange. These may be polymer-chain conformations, particle arrangements, temporary molecular associations, crystal defects, or fluid distributions within a deformable network. Applied deformation pushes these structures away from equilibrium. Some mechanical work is temporarily stored in altered configurations, while subsequent rearrangements dissipate energy. The observed relaxation times depend on both the resistance to motion and the restoring forces driving recovery.
In flexible polymers, chain conformations provide an important source of elasticity. Stretching a chain reduces the number of accessible configurations, producing an entropic restoring force. Local bond rotations and segmental motion relax relatively rapidly; rearrangement of a whole chain generally takes longer. The Rouse model describes a chain as connected beads subject to friction and thermal motion, neglecting hydrodynamic interactions and entanglement constraints. Its hierarchy of modes explains why even a chemically uniform polymer can display many relaxation times rather than one.
Long chains in concentrated solutions and melts cannot readily pass through one another. These entanglements constrain motion and can produce an intermediate-frequency rubbery plateau. The reptation picture represents a chain as moving within an effective tube formed by surrounding chains, with long-time relaxation occurring as it escapes its original constraints. In the simplest scaling argument, the terminal relaxation time grows approximately as the cube of molecular mass. Many well-entangled polymer melts exhibit a zero-shear viscosity dependence closer to a power near 3.4 over substantial molecular-mass ranges; constraint release and other effects complicate the simplest prediction.
Permanent crosslinks change the long-time response fundamentally. An intact covalent network cannot lose all shear stress through whole-chain flow and therefore has a nonzero equilibrium shear modulus. Temporary associations, including hydrogen bonds, ionic clusters, and reversible coordination bonds, can behave as crosslinks at short times but disengage over longer periods. Such networks may therefore resemble elastic solids in one frequency range and liquids in another. Crosslink density, bond lifetime, solvent content, and network defects all influence the measured spectrum.
Not all time-dependent deformation originates in polymer chains. Colloidal suspensions can relax through particle diffusion and collective rearrangement, while concentrated systems may retain long-lived particle cages. Hydrogels combine a polymer network with solvent, so their response can include both intrinsic network viscoelasticity and poroelasticity, in which fluid migration redistributes stress. The latter often depends strongly on specimen dimensions because fluid must travel over a characteristic distance. In metals and crystalline solids, thermally activated defect motion can produce delayed recoverable strain, often described as anelasticity. Similar-looking relaxation curves can thus arise from physically different mechanisms.
Temperature, timescale, and physical aging
Temperature strongly influences viscoelasticity because molecular and structural rearrangements require thermal activation or cooperative motion. Raising temperature usually shortens relaxation times, making a material appear more fluid-like at a fixed experimental frequency. Lowering temperature usually suppresses rearrangement and increases apparent stiffness. These effects are especially pronounced near the glass transition, where segmental mobility changes rapidly over a relatively narrow temperature interval. A glass transition measured mechanically depends on the timescale and definition used; it is not simply a universal temperature independent of the experiment.
The principle of time–temperature superposition uses measurements at different temperatures to extend the accessible range of effective times or frequencies. If temperature changes all relevant relaxation times by approximately the same factor without substantially changing the spectral shape, the material is called thermorheologically simple over that range. Defining
a frequency-dependent modulus can be shifted horizontally using the reduced frequency . A vertical correction may also be required, for example to account for density and temperature dependence of the modulus scale. The combined dataset forms a master curve.
Near the glass transition of many amorphous polymers, shift factors are often represented by the Williams–Landel–Ferry equation:
Malcolm L. Williams, Robert F. Landel, and John D. Ferry introduced this relation in 1955. Its constants depend on the reference temperature and material. The frequently quoted values and are associated with a reference at the glass transition for many polymers, but are not universal constants. For processes dominated by an approximately constant activation energy, an Arrhenius form may be more appropriate:
Here temperatures are absolute, is an activation energy, and is the gas constant.
Superposition can fail when different mechanisms have different temperature dependences. Examples include materials with multiple phases, separate segmental and associative relaxation processes, or changing crystalline structure. Heating may also cause solvent loss, chemical reaction, oxidation, or degradation. Shifting curves across such changes can create a smooth-looking master curve that does not represent a single unchanged material. Extrapolation far beyond the measured range therefore depends on physical assumptions, not merely on the mathematical quality of the shift.
Physical aging introduces another timescale. Below the glass transition, an amorphous material may evolve gradually toward a lower-energy structure, changing its compliance and relaxation times even at constant temperature. The response then depends on the waiting time after preparation as well as the duration of loading. Ordinary time-translation-invariant linear theory is insufficient unless the aging during a test is negligible. Moisture and solvent uptake can further alter relaxation by plasticizing the material, while drying or additional crosslinking can move the spectrum in the opposite direction.
Experimental methods and interpretation
Dynamic mechanical analysis measures the force and displacement response of a specimen subjected to oscillatory deformation. Depending on geometry, the instrument may operate in tension, compression, bending, or torsion. Frequency sweeps characterize the distribution of timescales, temperature sweeps reveal transitions, and amplitude sweeps identify the range in which the moduli are independent of deformation amplitude. Reported values are meaningful only with the loading mode, specimen dimensions, temperature, frequency, and amplitude specified.
Rotational rheometry is widely used for liquids, melts, gels, and soft solids. Common geometries include parallel plates, cone-and-plate fixtures, and concentric cylinders. Instruments impose torque or angular displacement and infer stress and strain from the geometry. Small-amplitude oscillatory shear provides and , while steady rotation measures flow curves. Creep and stress-relaxation experiments provide complementary time-domain information. Finite instrument acceleration means that real loading steps are ramps, which must be considered when interpreting the earliest part of a transient.
Measurement artifacts can resemble material behavior. Wall slip causes the material to move relative to the fixture, reducing the apparent deformation within the sample. Fixture compliance can introduce apparent relaxation, and instrument inertia can dominate at high frequencies or during rapid transients. Evaporation changes concentration, while edge fracture and sample expulsion alter the sheared region. At low frequencies, drift and ongoing aging may become more important than the nominal oscillation. Temperature gradients are particularly consequential near transitions where moduli vary sharply with temperature.
The connection between oscillatory and steady-flow measurements requires care. The magnitude of the complex viscosity is
For some polymer systems, the empirical Cox–Merz rule approximately equates this quantity with steady-shear viscosity when and shear rate have the same numerical value. It is not a general constitutive identity and often fails for structured fluids, gels, suspensions, and materials whose structure changes under sustained flow. Agreement or disagreement can itself reveal differences between near-equilibrium fluctuations and flow-induced structural changes.
Other methods extend the accessible scale. Microrheology infers response from the motion of small tracer particles; passive methods require an equilibrium fluctuation–dissipation relation and assumptions about probe coupling and local homogeneity. Indentation methods are useful for small or spatially heterogeneous specimens but require contact-mechanics models. Acoustic measurements probe much shorter timescales, often with greater sensitivity to compressional than shear behavior. Across these techniques, calibration, reproducible sample preparation, and explicit treatment of measurement uncertainty are essential. A modulus obtained under one geometry or frequency range cannot automatically be substituted for another without establishing the relevant constitutive connection.
Nonlinear viscoelasticity and complex loading
Linear theory fails when deformation changes the structures responsible for relaxation. Polymer chains can orient strongly or stretch; particle networks can break and rebuild; soft solids can undergo large rotations and shape changes. Stress then depends nonlinearly on deformation history. A large deformation need not cause permanent damage, however: rubber can undergo substantial reversible strain while displaying nonlinear viscoelasticity. Conversely, some materials leave their linear regime at very small strains because weak structural connections are easily disrupted.
The Weissenberg number, often written in steady shear, compares a relaxation time with a deformation rate. At small , the relevant structure can relax nearly as quickly as it is deformed. At large , deformation accumulates before relaxation is complete, making orientation, stretching, and elastic stresses important. Unlike the Deborah number, which compares relaxation with an observation or process duration, the Weissenberg number specifically characterizes deformation accumulated over a relaxation time. Definitions may vary with flow geometry and the selected relaxation mode.
Viscoelastic liquids can develop normal stress differences even in steady shear. With flow, gradient, and vorticity directions denoted , , and , the first normal stress difference is
A positive , common in polymer solutions and melts, contributes to rod climbing and die swell. Elastic stresses can also destabilize flows at low inertial Reynolds number, particularly where streamlines are curved. Extensional flow is especially sensitive to chain stretching and can produce strong strain hardening, long-lived liquid filaments, and responses that cannot be inferred reliably from shear viscosity alone.
Nonlinear constitutive equations must distinguish deformation from rigid-body rotation. A material undergoing only rigid rotation should not acquire artificial stresses because of the observer’s coordinate system. Frame-indifferent models therefore use appropriate tensor measures and objective stress derivatives. The upper-convected Maxwell and Oldroyd-B models represent important idealized cases; the latter combines a viscoelastic polymer contribution with a Newtonian solvent. More elaborate models introduce finite chain extensibility, anisotropic drag, or entanglement dynamics. No single model reproduces all shear, extension, transient, and large-strain responses across every polymer system.
Large-amplitude oscillatory shear provides a controlled way to examine nonlinear behavior. Stress waveforms acquire higher harmonics, and their shape distinguishes intracycle stiffening, softening, thickening, and thinning. Interpretation must separate viscoelastic memory from thixotropy, which concerns time-dependent structural breakdown and rebuilding, and from plasticity or damage. A yield-stress material may combine all of these effects. Temperature rise caused by dissipation can further alter viscosity and relaxation during the experiment. Consequently, a successful nonlinear model must account not only for a single stress curve but also for recovery, repeated loading, and changes in the imposed flow history.
Engineering, processing, and biological materials
Viscoelasticity creates both useful functionality and design constraints. In vibration isolation and damping, an elastic component supports loads while a dissipative component reduces resonant motion. Constrained-layer damping places a viscoelastic layer between stiffer sheets, causing bending to generate shear deformation within the damping layer. Performance depends on frequency and temperature: a material with a high loss factor in one operating range may provide little damping in another. Energy dissipation also creates heat, which can shift the material’s properties during sustained operation.
In pneumatic tires, cyclic deformation of rubber dissipates energy and contributes to rolling resistance, while deformation across the roughness spectrum of a road contributes to grip. Fillers and polymer networks produce behavior extending beyond simple linear response. Seals and gaskets present a different problem: under a maintained compression, stress relaxation reduces the contact force available to maintain sealing. Long-term design therefore requires relaxation and recovery data, not only an initial hardness or elastic modulus. Asphalt likewise combines rate-sensitive binder behavior with aggregate structure, making traffic duration and temperature central to pavement response.
Pressure-sensitive adhesives must deform sufficiently during application to make intimate contact, yet retain enough cohesion to resist flow and separation. Their performance depends on relaxation over timescales spanning bonding, holding, and removal. During peeling, energy can be dissipated through bulk deformation, fibrillation, and local separation processes. A single low-frequency modulus cannot fully characterize this behavior. Polymer processing imposes another set of demands: extrusion and injection molding involve large shear and extensional rates, while die swell reflects elastic recovery after flow leaves a die. In 3D printing, deposited materials may need to flow readily through a nozzle but rapidly recover enough stiffness to preserve shape.
Long-term structural calculations often use the finite element method with generalized Maxwell or related constitutive laws. These simulations predict stress redistribution, creep deformation, and time-dependent contact forces. For reinforced polymers and layered structures, the matrix may relax while fibers remain comparatively stiff, changing how load is shared. Reliable calculations require a distinction between reversible viscoelastic response and irreversible processes such as cracking, fatigue, oxidation, or interface debonding. Fitting short-term data alone does not establish long-term durability.
Biological tissues exhibit viscoelasticity through interacting networks of proteins, cells, and fluid. Tendons, cartilage, skin, and the brain have different architectures and therefore different directional, rate-dependent, and relaxation responses. Collagen fibers can rotate and straighten, proteoglycan-rich matrices resist deformation, and fluid can redistribute through porous structures. Living cells add active remodeling and force generation, so their mechanical response is not always that of a passive material. Experiments must distinguish intrinsic solid-network relaxation from fluid-flow effects and changes occurring after tissue preparation. Engineered hydrogels and tissue scaffolds use these principles to control both stiffness and the timescale over which stresses relax.
Historical development and current modeling problems
The mathematical foundations emerged from nineteenth-century work on elasticity, fluid flow, and delayed deformation. Wilhelm Weber’s experiments on silk in the 1830s documented elastic aftereffects: deformation and recovery continued over time rather than occurring instantaneously. James Clerk Maxwell’s 1867 work, On the Dynamical Theory of Gases, introduced the relaxation concept associated with the series spring-and-dashpot model. William Thomson, later Lord Kelvin, and Woldemar Voigt contributed to the traditions represented by the parallel spring-and-dashpot model, while Ludwig Boltzmann formulated a hereditary approach to mechanical response in the 1870s.
Twentieth-century polymer science connected these mathematical descriptions to molecular architecture. The recognition that molecular mass, chain flexibility, crosslinking, and entanglements determine mechanical timescales made viscoelastic measurements a structural probe rather than merely a means of fitting engineering curves. John D. Ferry’s Viscoelastic Properties of Polymers, first published in 1961, became a major reference for experimental and theoretical work. Pierre-Gilles de Gennes introduced reptation in 1971, and the subsequent work of Masao Doi and Sam Edwards developed a systematic tube-based description of entangled-polymer dynamics.
An important continuing problem is model identifiability. A relaxation spectrum inferred from a finite frequency window is not unique, and parameters that fit shear data may predict extension poorly. Physical constraints help narrow the possibilities: constitutive laws should respect causality, material symmetry, and the thermodynamic requirement that passive materials do not generate net mechanical work. Independent experiments in creep, relaxation, oscillatory shear, and extension can discriminate between models that appear equivalent in a single dataset. The distinction between fitting and prediction is particularly important when extrapolating to long service times.
Fractional calculus offers compact descriptions of broad, approximately power-law memory. An idealized fractional element may obey
where is a specified fractional time derivative and has units of . It interpolates mathematically between a spring and a dashpot and can represent behavior that would otherwise require many exponential modes. Such models require explicit derivative definitions, history conditions, and physically suitable short- and long-time limits. A power law fitted over a finite window is not evidence that it persists indefinitely.
Current research also addresses materials whose relaxation is altered by the deformation itself: reversible polymer networks, densely packed suspensions, living tissues, and actively driven cellular systems. Coupled models may need to include solvent transport, bond exchange, damage, heat generation, and structural aging. Data-driven methods can accelerate parameter estimation, but a low fitting error does not ensure frame indifference, stability, or thermodynamic consistency. Establishing which internal variables are necessary—and which can be inferred from available experiments—remains a central difficulty in connecting molecular mechanisms to reliable predictions under complex loading.