Sticky tape holds your weight in shear and peels off with a fingernail. The adhesive is not what changed

A strip of packing tape stuck across a box will hold a surprising load if you pull along the plane of the tape. Lift one corner with a fingernail and the same strip comes away in one continuous motion, almost effortlessly.

Nothing about the adhesive changed between those two experiments. What changed is the direction of the load, and that difference — which most people discover by feel and never examine — is one of the two things that actually govern whether a glued joint holds. The other is that a joint that fails has usually not had its glue break at all.

Adhesion is the science of what happens in a layer perhaps a few molecules thick, and it is one of the clearest cases where the interface, not the material, is the engineering.

Read the broken joint before blaming the glue

When a bonded joint comes apart, the fracture surface tells you what went wrong, and it takes a glance.

If both faces are covered in adhesive — the glue split down its own middle — the failure was cohesive. The bond to both surfaces held, and the adhesive itself was the weakest part. That is a case for a stronger or tougher adhesive, or a thinner bond line.

If one face is bare and clean, and all the adhesive is sitting on the other, the failure was adhesive: the glue simply let go of that surface. No stronger glue will fix this, because the glue was never the limiting element. The answer lies in surface preparation — cleaning, treatment, priming — and buying a more expensive adhesive is money spent on the wrong side of the problem.

This single diagnostic resolves most arguments about glue, and in practice adhesive failure is by far the more common of the two.

Why the surface matters more than the adhesive

For anything to stick, its molecules have to get close enough to the other surface for intermolecular forces to act — a fraction of a nanometre. That means a liquid adhesive must wet the surface: spread across it rather than beading up on it.

Whether it does is governed by surface energy. High-energy surfaces — clean metal, glass, ceramics — pull liquids out flat and are easy to bond. Low-energy surfaces repel them, and the notorious cases are exactly the plastics everyone tries to glue: polyethylene, polypropylene, PTFE and silicones. They are not chemically hostile so much as energetically uninviting; an adhesive drop sits on polypropylene as a bead, touching very little of it, and a bond can only form where contact was made.

This is why those plastics require surface treatment rather than better glue. Flame treatment, corona discharge and plasma treatment all work by oxidising the top few nanometres to raise its surface energy, and they work well — but the effect decays over hours to days as the surface reorganises, so treated parts have a shelf life that is easy to forget about.

Contamination does the same damage in reverse. A film of oil one molecule thick is enough to prevent bonding, because the adhesive ends up stuck to the oil rather than to the substrate. Fingerprints qualify. So does mould release agent on a moulded plastic part, which is specifically designed to stop things sticking and is present on the surface precisely because the moulder wanted it there. Degreasing is not a preliminary; on many jobs it is the step that decides the result.

Roughening is the other standard preparation, and it is more conditional than its reputation suggests. It increases the real contact area and offers mechanical interlocking, which helps — provided the adhesive can flow into the texture. If it is too viscous, or wets poorly, roughening traps air in the valleys and produces a joint with less contact than a smooth surface would have given.

A bonded joint is only as good as the last monolayer on each surface. Almost every surprising adhesive failure is an interface problem wearing a material problem's clothes.

The mechanisms, including the one with no chemistry at all

Adhesion is not one phenomenon. Four contribute, in varying mixtures.

Mechanical interlocking — adhesive flows into pores and texture and hardens there. Important on wood, on abraded metals and on porous substrates.

Adsorption and van der Waals forces — the universal contribution, present whenever two materials are in genuine molecular contact. Individually feeble, collectively substantial when the contacting area is large.

Chemical bonding — actual covalent links across the interface, which is what coupling agents and primers are for. Silane primers, for example, bond to glass or metal oxide at one end and to the polymer at the other, and they are why a properly primed glass joint outperforms an unprimed one by a wide margin.

Diffusion — relevant when both sides are compatible polymers. Solvent cement for PVC is not really glue: it dissolves the surface of both parts so their chains intermingle, then evaporates, leaving one continuous piece of PVC. The joint has no interface left to fail at, which is why it is so strong.

The most interesting case relies on the second mechanism alone.

What the gecko does, and why it matters to materials science

A gecko climbs glass using no adhesive whatsoever. Its toes are covered in millions of hairs called setae, and each seta branches into hundreds of flattened tips — spatulae — a couple of hundred nanometres across. The attachment force is ordinary van der Waals attraction, the same interaction that is too weak to be interesting between two flat surfaces.

It becomes interesting through contact splitting. Dividing one large contact into a very large number of small ones increases total adhesion substantially, for two reasons. Compliant fine structures conform to a rough surface and make real contact over far more of it, where a rigid flat pad would touch only at high points. And adhesive failure spreads as a crack — but a crack that would run across a continuous pad has to be initiated separately at every individual spatula, so subdivision removes the easy failure path.

That first reason should look familiar. The friction article explained that two apparently flat surfaces touch across roughly one per cent of their area. The gecko's answer to that problem is to give up on being flat and to become millions of independently compliant tips instead — the same real-contact-area insight, applied in the normal direction rather than the tangential one.

Synthetic gecko-inspired adhesives genuinely work, and are genuinely not on sale as a general product. The reason is mundane: dust. A gecko sheds contaminated setae and its structures self-clean as it walks; a manufactured microfibre array fouls, and fouled tips do not touch anything. Add durability under repeated use and the cost of patterning hundreds of millions of features per square centimetre, and the picture is the one this site keeps describing — the physics is sound, the laboratory demonstrations are real, and the manufacturing is the wall.

Peel is what actually breaks joints

Back to the tape. The reason it resists shear and surrenders to peel is about where the stress goes.

Pull a bonded lap joint along its plane and the load is distributed across the whole bonded area — every square millimetre carries a share, so total strength scales with area. Peel it instead, and none of that area participates. The stress concentrates almost entirely at the line where the bond is currently separating, so the joint is being destroyed a narrow strip at a time, and the area behind that line contributes nothing until its own turn comes.

This is why adhesive joints are designed to be loaded in shear or compression and never in peel or cleavage, why a wide thin bond outperforms a narrow thick one, and why designers add fillets at the edges of a bond line — the fillet spreads the stress concentration at the point where peel would otherwise start.

It also explains a counterintuitive rule: a thinner bond line is usually stronger. More adhesive is not more strength. A thick layer offers more material to deform, contains more voids and defects, shrinks more on curing and puts the stress further from both substrates. The typical optimum is a film only a fraction of a millimetre thick, which is why clamping matters and why a joint flooded with glue is usually weaker than a joint properly wetted with very little.

Pressure-sensitive adhesives — tape, labels, sticky notes — solve a contradiction by being viscoelastic. To bond on contact they must flow like a liquid and wet the surface in the moment you press them, which is what "pressure sensitive" actually means. To hold afterwards they must resist like a solid. The requirement has a number attached: the Dahlquist criterion says a PSA needs an elastic modulus below roughly 0.1 MPa at the timescale of bonding, because above that it cannot wet a real surface in the time available. Tapes that lose tack in the cold have simply become too stiff to satisfy it.

Where nanomaterials genuinely enter

Two roles are real and one is mostly marketing.

Toughening fillers. Nanoparticles and exfoliated nanoclays dispersed into an adhesive can improve fracture toughness and creep resistance by deflecting cracks and blunting them, and at low loadings without destroying the viscosity the adhesive needs in order to wet surfaces in the first place. This is an established industrial use, and the usual caveat applies: the benefit depends on dispersion, and a badly dispersed filler is a population of defects.

Conductive adhesives. Filling an adhesive with silver flake, carbon nanotubes or graphene produces a material that bonds and conducts — used for die attach, for grounding, for repairs where soldering is impossible, and for the conductive gaskets that keep EMI shields electrically continuous across a seam. Here the filler is the function rather than an additive.

Retail "nano glue" is usually a claim without a mechanism. Adhesion is limited by interfacial contact and by joint geometry, and no filler changes the fact that a contaminated low-energy surface will not bond. If a product promises to solve that, the question worth asking is which of the two — surface energy or contact — it claims to be changing.

Why it matters for students and researchers

Adhesion is interfacial science made visible, and it teaches a habit that transfers everywhere: when a system fails, look at the join. This site has now arrived at the same conclusion from four directions — masks leak at the fit rather than the filter, shields leak at the seam rather than the metal, heat stalls at the interface rather than in the copper, and bonded joints fail at the surface rather than in the glue. That is not a coincidence about four subjects; it is a general fact about engineered systems.

It is also a field where measurement is unusually treacherous. Adhesion has no single number. A lap shear result, a peel strength and a fracture energy measure different things on different geometries, and they cannot be converted into one another or compared across test methods. A supplier's headline figure is meaningful only alongside the substrate, the surface preparation, the test geometry, the rate and the temperature — and a great deal of published adhesion data omits at least one of those.

The open problems are practical. Bonding low-surface-energy plastics without surface treatment remains genuinely difficult and commercially valuable. Treatments whose effect does not decay within days would remove a real production headache. Debondable adhesives — strong in service, releasable on command by heat or light — are increasingly important for repairability and recycling, since a permanently bonded assembly is an unrecyclable one. And dry adhesives that resist fouling would finally move gecko-inspired attachment out of the demonstration stage.

Frequently asked questions

Why won't glue stick to some plastics?

Because polyethylene, polypropylene, PTFE and silicones have low surface energy, so adhesives bead up instead of wetting them and can only bond where actual contact occurs. The fix is to raise the surface energy — flame, corona or plasma treatment, or a specific primer sold for polyolefins — rather than to buy a stronger adhesive. Note also that treated surfaces revert over hours to days, so bond soon after treating.

Does using more glue make a stronger joint?

No, and it usually makes a weaker one. Thin bond lines are stronger: a thick layer contains more voids, shrinks more as it cures, deforms more under load and places the stress further from the substrates. The goal is complete wetting of both surfaces with as little adhesive as achieves it, which is what clamping is for.

Why does superglue stick my fingers instantly but not always the parts?

Cyanoacrylate cures by reacting with traces of moisture, and skin provides ideal moisture and perfect contact. Engineering parts are often drier, sometimes contaminated, and frequently do not fit closely — cyanoacrylate is poor at filling gaps and needs surfaces nearly in contact. A poorly fitting joint with a visible glue line is the wrong application for it, regardless of how well it bonded your fingertips.

What is the strongest glue?

The question has no answer without the substrates, the direction of loading, the temperature and the service environment, because those decide the outcome far more than the adhesive's headline strength does. A structural epoxy on degreased, abraded metal loaded in shear is enormously strong; the same epoxy on untreated polypropylene loaded in peel will fail almost at once. Choose for the joint, not for the number on the tube.

Why do stickers leave residue?

Because the adhesive failed cohesively — it split internally, leaving some on the surface and some on the label. Pressure-sensitive adhesives are deliberately soft in order to wet surfaces on contact, and over time and warmth they flow further, bond more firmly and become harder to remove cleanly. Gentle heat softens the residue and lets it release in one piece, which is generally more effective than scraping.