How Science Works | Why Some Materials Expand Sideways When You Pull Them

For a small uniaxial deformation, define axial strain as the fractional length change along the loading direction and transverse strain as the fractional width change perpendicular to it. Poisson’s ratio is ν = −ε_transverse/ε_axial, with the minus sign included in the definition.

An ordinary material stretched by 1 percent might contract sideways by 0.3 percent. The ratio is −(−0.003)/0.01 = +0.3. An auxetic specimen stretched by 1 percent and widening by 0.4 percent instead has ν = −0.004/0.01 = −0.4.

The ratio is dimensionless. It does not measure the force needed to cause the deformation, and a negative value does not mean negative stiffness. It describes the relationship between two strains in a stated loading arrangement.

Direction matters. Pulling along x and measuring width along y defines one response; rotating an anisotropic specimen or measuring thickness along z can give another. A material may be auxetic for one direction pair and conventional for another.

At larger deformations, specify whether the reported number is a ratio of total engineering strains from a fixed reference or an incremental ratio at the current state. Those conventions can diverge in nonlinear materials. A single number without strain range and measurement definition can hide the mechanism rather than explain it.

Contents · Next chapter

Picture a re-entrant honeycomb: some cell walls point inward, making a bow-tie-like shape rather than an ordinary convex hexagon. When the structure is pulled, those inward-pointing ribs can rotate and open. The cell grows in width while also lengthening.

Alternatively, imagine rigid squares connected at corners by hinges. Neighbouring squares rotate in alternating directions. Pulling one way opens spaces between them and increases the spacing of the whole array in the perpendicular direction too.

These are structural mechanisms. The polymer, metal or paper forming each rib or square need not itself have a negative local Poisson’s ratio. The larger structure has an effective response arising from how its parts move together.

Auxeticity is not restricted to visibly perforated sheets. Some crystals and other microstructures can show directional negative Poisson responses through their internal arrangements. But it would be wrong to assume that every example works by the same rotating-square mechanism.

The useful scientific sequence is structure → permitted motion → predicted strains → measurement. A photograph of an unusual pattern is only the first step. If the hinges are blocked, ribs stretch instead of rotate, or the sample buckles out of plane, the response may differ sharply from the ideal sketch.

Contents · Previous chapter · Next chapter

Consider a periodic two-dimensional array of identical rigid squares, each with edge length a. Ideal hinges connect their corners, and neighbouring squares rotate oppositely. Let θ describe the square orientation relative to the initial grid direction in a convenient branch with 0 < θ < 45 degrees.

The square’s horizontal projection is a cos θ + a sin θ. In this ideal arrangement, a repeat spacing proportional to that projection is L_x = a(cos θ + sin θ). The vertical spacing has the same form: L_y = a(cos θ + sin θ). A larger repeating cell may multiply both spacings by the same constant without changing the strain calculation.

Draw a square tilted by θ, its edge projections labelled a cos θ and a sin θ, and a neighbouring square tilted oppositely. Mark the horizontal and vertical repeat distances outside the units. The empty spaces are part of the repeating geometry, not extra solid material.

As θ increases towards 45 degrees, both spacings increase together. Their infinitesimal changes satisfy dL_x/L_x = dL_y/L_y, giving an incremental in-plane Poisson ratio of −1 along these axes for the ideal mechanism.

This is a kinematic prediction: it follows from allowed motion. Ideal freely rotating hinges do not supply a realistic nonzero force–extension law. To predict force, stored energy or failure, the model must add hinge stiffness, rib compliance, contact and other physical details.

Contents · Previous chapter · Next chapter

Use the original numerical example a = 10 millimetres, with θ changing from 10 to 20 degrees. At 10 degrees, either repeat spacing is approximately 11.5846 millimetres. At 20 degrees it is approximately 12.8171 millimetres.

The engineering strain in each direction is (12.8171 − 11.5846)/11.5846 ≈ 0.10640, or 10.64 percent. Thus the finite engineering-strain ratio for this particular equal-expansion mechanism is also −1. That agreement with the incremental ratio comes from the identical x and y scaling, not from a general rule about finite strains.

The area of the repeating footprint scales as L_xL_y. Its fractional increase is (12.8171/11.5846)² − 1 ≈ 0.22412, or 22.41 percent. Simply adding the two 10.64 percent strains would miss the product term at this deformation size.

No solid square has gained area in the model. The squares are rigid, so their own areas remain fixed. The growing footprint comes from the changing spaces and arrangement between them. The fraction of the footprint occupied by solid therefore changes.

At θ = 45 degrees, the spacing reaches a√2, approximately 14.1421 millimetres. The derivative dL/dθ = a(cos θ − sin θ), with θ measured in radians for differentiation, becomes zero there. The chosen opening branch has reached its geometric maximum; it cannot continue to expand indefinitely by the same motion.

Contents · Previous chapter · Next chapter

The ideal two-dimensional mechanism’s ratio of −1 should not be inserted without thought into every formula for a three-dimensional solid. Standard bounds and modulus relations depend on dimensionality, symmetry and stability assumptions.

For a stable, homogeneous, isotropic three-dimensional linear elastic material with positive finite bulk and shear moduli, the familiar range is −1 < ν < 0.5. The endpoints correspond to limiting behaviour rather than ordinary finite positive moduli of all types.

Within that model, Young’s modulus E, shear modulus G and bulk modulus K obey G = E/[2(1 + ν)] and K = E/[3(1 − 2ν)]. Negative ν inside the allowed range is compatible with all three moduli being positive.

For a separate hypothetical material with E = 12 megapascals and ν = −0.5, the relations give G = 12 megapascals and K = 2 megapascals. The material resists shear and uniform compression with positive stiffness. Its negative Poisson ratio does not imply that pulling it produces energy for free.

Anisotropic materials need direction-dependent elastic constants, and the simple isotropic bounds do not constrain every directional Poisson ratio in the same way. Likewise, a perforated sheet with ideal hinge mechanisms is not automatically a homogeneous three-dimensional isotropic solid. The model must match the question before its formulas are used.

Contents · Previous chapter · Next chapter

Suppose a fictional specimen has an initial gauge length of 100 millimetres and gauge width of 40 millimetres. Under a small tensile load, those dimensions become 102 and 40.32 millimetres. The axial engineering strain is 0.02 and the transverse strain 0.008, so the measured strain ratio is −0.4.

These dimensions refer to a central gauge region, not necessarily the distance between machine grips or the widest point near a fixture. Grip constraints can suppress sideways motion locally, producing a nonuniform field that is not representative of the interior mechanism.

A camera-based measurement can track a grid of marked points and estimate local displacement and strain. The analysis should distinguish actual width change from rigid-body rotation, perspective changes, out-of-plane motion and camera distortion. A changing silhouette alone is insufficient.

Resolution matters most when transverse changes are small. If the width-change uncertainty were about ±0.40 millimetres, larger than the observed +0.32 millimetres, the sign would not be securely resolved by that measurement alone. If it were ±0.02 millimetres, the same effect would be much more clearly distinguished from zero, subject to the rest of the uncertainty budget.

Suppose a later reading is 104 millimetres long and 40.40 millimetres wide. Relative to the original specimen, the total strains are 4 percent and 1 percent, giving a secant ratio of −0.25. That differs from the first interval’s −0.4. The changing ratio is useful evidence about the deformation path, not a reason to average the two numbers without explaining their reference states.

A full report specifies loading direction, gauge region, specimen geometry, temperature, rate, strain definition and uncertainty. It shows transverse strain against axial strain over the tested range instead of presenting a single negative number detached from the conditions that produced it.

Contents · Previous chapter · Next chapter

Real hinges are not perfect pins. A thin connecting ligament bends and may stretch. Making it thicker can improve load capacity while restricting rotation; making it thinner can promote rotation while concentrating stress. Geometry and material behaviour interact rather than contributing independently.

As an array opens, units can reach geometric limits, touch neighbours or transfer more load into stretching the solid. The effective Poisson response can change with strain and may even change sign. A small-strain auxetic measurement does not establish auxeticity throughout a large deformation.

Loading and unloading can follow different paths. Friction, polymer viscoelasticity, plastic deformation or damage can create hysteresis and residual shape change. A reversible-looking photograph at one load does not establish durable elastic recovery over many cycles.

Finite specimens also have edges. An array only a few cells wide may behave differently from a large periodic model, especially near grips or boundaries that prevent rotation. Testing different sample sizes and reporting central versus edge deformation helps reveal such effects.

Compression may introduce buckling and out-of-plane modes absent in tension. A design can be auxetic in a useful tensile regime yet unstable under compression, or develop auxetic behaviour only after an instability. Each claim should identify the loading path and operating range rather than treating the material name as a universal guarantee.

Contents · Previous chapter · Next chapter

Lakes’s 1987 original foam work, described with links by the author’s laboratory, demonstrated structures that become laterally wider under tension. The cellular architecture was crucial to the unexpected macroscopic response. This established a tangible counterexample to the everyday expectation that stretching always makes a material narrower.

Grima and Evans’s 2000 original rotating-square paper analysed rigid units joined at vertices and the negative Poisson response of their cooperative rotation. Our numerical angles and dimensions are a new teaching example of that idealised mechanism, not reported measurements from their paper.

These studies play different roles. A fabricated foam demonstrates a physical material response. A geometric model explains a mechanism and makes clean predictions under its assumptions. Neither alone supplies the performance specification for a new product made from a different material or cell design.

To test a proposed rotating-unit explanation, scientists should observe the rotations as well as the boundary strain. If a sample widens but the units do not rotate as predicted, another mechanism may dominate. If rotations occur only near a grip while the bulk contracts, the global interpretation needs revision.

Comparisons are strongest when alternative explanations are measured rather than merely dismissed. Images of cell motion, local strain maps, force–displacement curves and repeated loading together reveal more than one attractive before-and-after photograph. Agreement should persist across the intended operating conditions, with uncertainty and failures reported.

Contents · Previous chapter · Next chapter

A June 2026 original study in Nature Communications investigated hybrid auxetic platforms for stretchable displays. Its central challenge was that a framework’s rotating units can widen the overall structure while locally distorting a continuous elastomeric layer attached to it. The layer can also restrict the rotations responsible for auxeticity.

The researchers designed selective connections between framework and substrate, together with boundary features, to improve expansion uniformity. Their experiments and modelling assessed local-to-global deformation and display behaviour under repeated stretching. These are results for the tested architectures, not a claim that any negative-Poisson sheet automatically preserves every image.

A simple original comparison makes the distinction clear. Imagine a panel whose total width and height each grow by 10 percent, so its outer aspect ratio is preserved. Inside it, one square image region might stretch 20 percent horizontally but only 2 percent vertically. That local region’s aspect ratio changes by 1.20/1.02 ≈ 1.1765, or about 17.65 percent.

Global equal expansion therefore does not guarantee local shape preservation. Rigid islands, rotating cells and compliant connectors can distribute deformation unevenly even when the outside rectangle looks perfect. The useful performance measurement depends on where the image, sensor or other functional component actually sits.

This example also shows why auxeticity is a means rather than a complete design objective. A display needs suitable optical, electrical and fatigue performance; a cushion needs an appropriate force and energy-absorption response. Negative Poisson ratio alone does not settle any of those additional requirements.

Contents · Previous chapter · Next chapter

Sketch four alternating rotating squares and label their corner hinges. Mark the x and y directions, choose an angle between zero and 45 degrees, and calculate the two repeat spacings. A drawing or digital sketch is enough; no cutting, heating or high-force fabrication is needed for this worksheet.

Reproduce the 10-to-20-degree example and verify the 10.64 percent length and width increases. Then calculate the area increase directly from the product of the two scale factors. Explain where the additional footprint comes from without claiming that the rigid squares gained material.

Next compare the ideal mechanism with the fictional measured specimen whose strain ratio is −0.4. A departure from −1 need not refute auxetic behaviour. It may indicate hinge stiffness, solid deformation, boundary constraints or a different architecture. The explanation should predict which additional observation would distinguish those possibilities.

Keep four questions separate in the conclusion. Does the sample widen under the specified pull? How much force does that require? How much load can it survive? Does the behaviour serve the intended application over its required lifetime? The first concerns auxetic response; the others require further evidence.

For terminology and unit preservation, the existing elastic-modulus and Poisson-ratio translation guide provides a complementary reference. It does not replace the geometry and mechanism analysis developed here.

Return to Materials Science and the How Science Works roadmap. The surprising sideways expansion becomes understandable once the internal motion, loading direction and measurement scale are made explicit.

Contents · Previous chapter · Continue to Materials Science