Technical guide
Tile adhesive: why it stopped forgiving mistakes
Modern tile adhesive is still sand and cement, but the few per cent of additives sitting on top of that base changed what the material asks for: a thin bed, a flat substrate, controlled water and a narrow window of conditions. Where those demands came from, why the substrate rules never moved, and which habits carried over from the old material now bring tiling down.
By Mihail Cheptene · 31 August 2026
Tile adhesive today is not the material it was 25 years ago — not because polymers are new, they have been on the market since the 1960s, but because what counts as an ordinary bag changed. Classes went onto the bag in 2001, deformable C2 stopped being an upgrade and became the default, and formulations were rebuilt around large-format porcelain. The result holds where an ordinary bag of 25 years ago would not have held at all.
In return it became demanding about how it is handled — about temperature, moisture and timing. Those conditions are not recommendations. The figures printed on the bag were obtained under the standard test conditions of EN 12004-2: a fixed temperature, a fixed humidity, a set dose of water, a set open time. Miss one of them on site and you spend margin you have already paid for. Miss several and the tiling comes off.
All of this concerns cementitious adhesive — the C classes, and almost everything sold for tiling. Dispersion and reaction-resin adhesives are different materials and follow different rules.
Before the polymer, thickness did the work
Before polymer additives, a tiler covered every risk with a single move: he laid a thick bed. Weak adhesion, mortar drying out and the geometry of the room were all answered by the same decision, and it was enough to hand the job over and warrant it.
It did not work quite the way it is usually assumed. Thickness does not raise bond strength — that is measured at the interface in N/mm² and does not depend on how deep the bed is. What a thick sand and cement bed gave was something else: a volume of water large enough for the cement to hydrate before the substrate and the tile drew it away, and room to absorb geometry. Tiles then were neither calibrated nor rectified, substrates ran to wide tolerances, and the bed swallowed both. The bond itself came from elsewhere: a porous biscuit for the cement to key into, tiles soaked before fixing and a dampened substrate so that neither could rob the mortar of its water, and a cement slurry key at the face. None of that is depth.
The move had a cost. A deeper bed does not shrink by a larger percentage — that is set by the mix — but it does build up more absolute movement, and more stress at the interface with it, and it dries unevenly enough to curl. A thick bed closed the risk of drying out and opened the risk of shrinkage debonding; on a porous tile the mechanical key was simply strong enough to cover it.
The demand on the substrate never moved
The British standards covering tiling have been revised regularly, but they were revised behind new materials rather than because the logic of preparing a substrate had changed.
BS 5385-1, which covers internal wall tiling, has been revised in 1990, 1995, 2009 and 2018 since its first edition as a separate part in 1986. The current one came into effect on 30 June 2018 and brought 4 things: plywood was excluded as a background for direct tiling in favour of a proprietary backer board, separate adhesive coverage requirements were set for large format tiles, minimum grout widths were tied to tile size, and the competence of the installer was written in.
BS 5385-3, which covers floors, moved in the same steps: 1989, 2007, 2014, 2024. The 2014 edition revised the sections on heated and calcium sulfate screeds, among other things. The 2024 edition is more than a revision: it absorbed the separate standard for stone and terrazzo flooring, which was withdrawn, expanded the definitions considerably, and now calls for an intermediate layer — an uncoupling membrane, a tanking system or a backer board — wherever tiles go over plywood and other wood-based boards.
What the standard asks of a substrate — rigid, dimensionally stable, dry, flat, sound — runs through every one of those editions unchanged. What moved is which materials are still trusted to deliver it. Plywood was not expelled because the requirement changed; it was expelled because the plywood on the market stopped meeting a requirement the standard had already been making for years. The demand held still and the material fell away from it.
The base did not change, the additives did
Modern adhesive is still sand and cement. In a standard bag, cement and sand make up more than 90 per cent of the weight, exactly as they did 25 years ago, and everything that separates the new material from the old fits into the few per cent that remain. Highly deformable and lightweight products shift those proportions — more polymer, and perlite or glass microspheres in place of some of the sand — but the base is the same.
The polymer is a powder, and it is what made a bond between porcelain and a sand and cement mortar possible in a thin bed. It does not wait for drying to begin. The moment it meets the mixing water it redisperses, returning to the liquid dispersion it was spray-dried from; only as the water leaves do those dispersed particles coalesce into a film. It is the film that meets the smooth back of the tile, and the cement then binds to the polymer rather than to the tile. It is not a layered sandwich — the film runs right through the set cement and forms a single network with it — but at the boundary with a non-porous tile the load is carried by the polymer phase. The same film gives the deformability of classes S1 and S2 and the ability to bridge microcracks.
The bond line at the back of a porcelain tile. On the left, set cement arrives at a dense face with no pores to enter, so the 2 materials only touch. On the right, the polymer film covers that face and the same polymer runs on between the cement particles — one continuous phase rather than a layer of glue with cement above it. Schematic, not to scale: the film is measured in microns, the bed in millimetres.
In return the polymer makes 2 demands the old cement mortar never made. The first is water. The amount of polymer in the bag is fixed, so extra water adds none — it spreads the same quantity through a larger volume. The film comes out thinner, tears in places, and never forms a continuous layer at the tile. All that extra water then leaves, and what it leaves behind is porosity and shrinkage. Meanwhile the mix stirs more easily and spreads more pleasantly, which is the trap. The manufacturer states the prohibition directly: never add more water than specified, as it leads to shrinkage and debonding.
The second is temperature. The lower limit is 5 °C almost everywhere; the upper is usually somewhere between 25 and 30, though it varies by brand. The fine print matters more than the numbers: the range applies to the substrate as well as the air, and it applies right through curing, not only at the moment of spreading. 5 °C of air over a 2 °C screed is outside the range, whatever the thermometer on the wall says.
Cellulose ether with starch ether. The first holds the water in the mix, stopping the substrate from drawing it off before the cement has set, and it thickens the mix at the same time — the 2 jobs are not separable. The second pulls in the same direction. Between them the mix flows under the trowel and holds still under a tile that wants to slide. Where the bag carries a T, that is a tested result — 0.5 mm of slip or less — and it comes from the whole rheology package rather than from any single ingredient.
Fibre. Cellulose or mineral, working towards the same end: holding the tile where it was put until the adhesive sets.
There are several further additives in the mix, but they have no bearing on the subject of this guide.
The thin bed moved every tolerance onto the substrate
It is this adhesive that made possible what we have now. Porcelain holds where 25 years ago it would not have held at any standard of workmanship: a thin bed, an extended open time, a bond to a non-porous face.
The adhesive did not become demanding of the substrate because of the additives. The polymer is indifferent to whether what lies beneath it is flat. What became critical is the thickness the polymer allowed. A traditional bed ran to 15 or 25 mm on a wall and thicker again on a floor; a thin-bed adhesive works at 2 to 6, with isolated build-ups to around 12 where the data sheet allows it, and products formulated for thicker beds are a separate specification. A 3 mm bed cannot physically take out a 5 mm hollow. The tolerance used to be absorbed by the bed; there is nothing left to absorb it, and it has passed to the substrate entire.
The same substrate under 2 beds. A thick sand and cement bed varies in depth across the hollow and still presents a flat plane to the tile. A thin-bed adhesive holds its own thickness and follows whatever is underneath, so the hollow reappears as a void beneath a tile that has to sit flat. Schematic, not to scale: the difference between the 2 beds is nearer 7 to 1 than the drawing shows.
What follows is a list in which every item stopped being a remark about quality and became a risk of debonding.
Flatness. The background has to come in within 3 mm under a 2 m straightedge. Anything worse than that gets corrected before tiling starts, because the adhesive is no longer able to absorb it.
Moisture. There are 2 problems here and they pull in opposite directions. A substrate can be wet — residual construction moisture, shrinkage still running; on concrete walls BS 5385 asks for at least 6 weeks. Or it can be too thirsty. The cellulose ether in the adhesive exists to hold the gauging water against exactly that suction, and it does slow the loss, but an aggressively absorbent background can still win, and a 3 mm bed has no reserve to give up. That is the gap the primer closes.
Movement. This is where the old defence is most often misdescribed. A thick cement bed is rigid: it does not absorb movement by sheer mass, and a bed bonded to a moving base passes deflection into the tile much as a thin one does — it simply cracks as well. What protected traditional work was isolation. The base was formed true and smooth so that the bed above could slide over it when the two moved differently, and bonding the tiling directly to the substrate took that freedom away. What replaces it now is the deformability of an S1 or S2 adhesive and, where the movement is real, an uncoupling layer.
Primer and dust removal. The primer brings the absorbency of the substrate into a controlled state; dust removal takes away a release layer the adhesive cannot bond through at all. Both look like preparation and both sit inside the load-bearing chain.
No one of them on its own usually brings tiling down. The sum of them does.
The mistakes repeated out of habit
The most common mistakes on site are not made out of ignorance but out of the opposite — because it used to be done this way and it worked. The material changed; the move stayed.
A very thick bed. The old thick bed was a different material, not the same one laid deeper. It was mixed lean — almost dry and compacted on floors, plastic and floated on walls — with far less water than a gauged adhesive, and far less shrinkage with it. Modern adhesive is gauged somewhere in the 20s per cent by weight, a 25 kg bag taking roughly 5.5 to 7.5 litres depending on the product, and 20 mm of it shrinks on a scale a sand and cement bed of the same depth never did. That shrinkage builds internal stress, and the layer that was supposed to hold becomes the weak point itself. Hence the limits in the data sheets: a working range of 2 to 6 mm, with isolated build-ups to around 12 where the product allows it. The habit carried the thickness across but not the material.
Too much water. On the old mix, extra water was forgiven: the mortar was lean and a porous tile drew off the surplus by itself. On a polymer adhesive the feedback runs backwards. The mix stirs more easily and spreads more pleasantly, so the mistake registers as an improvement at the exact moment it is made.
Fixing to a substrate that has not dried. 2 mechanisms, and the important one is not the one usually named. A screed that has not given up its moisture is still shrinking, and it goes on shrinking under tiles that are already fixed, sending the stress straight into the bond line. The manufacturer states it as a requirement: a cementitious substrate should not continue to shrink after the tiles are installed. The second mechanism is slower and easy to overstate. Porcelain and adhesive together are close to vapour tight, so the water in the bed leaves through the grout joints and sideways rather than upward — weeks instead of days. The film the bond depends on forms only as that water departs, so the strength arrives late rather than not at all; it is denied outright only where the substrate stays wet in service. Both of these are cementitious mechanisms. Calcium sulfate screeds and gypsum backgrounds fail differently and follow their own rules.
Working without dust removal and primer. The 2 operations do different jobs and neither substitutes for the other. Dust acts as a release layer: the adhesive bonds to the dust, and the dust holds on to nothing. The strength of a joint equals the strength of its weakest plane, and here that is the dust. The primer does something else — it brings the absorbency of the substrate under control. The manufacturer puts it plainly: priming allows a cement-based adhesive to hydrate fully and preserves the open and working times. Without it, a thirsty substrate pulls the water out of a 3 mm bed before the cement has had time to react, and cement that never got its water never hardens. What is left at the contact face is a thin chalky layer that crumbles under a fingernail — in the one place the whole bond depends on. The order is fixed: dust first, primer second. Primer over dust simply glues the dust down, and the weak layer is still there.
Coming back to the work a week after priming. A primed surface is clean and slightly tacky, and over a week it collects dust and other trades. The window between priming and fixing is far tighter than is generally assumed. One major manufacturer puts it in these terms, and they are typical of the genre: if the dried primer has stood uncovered for more than 24 hours, lay a second undiluted coat; if that window is missed as well, take the primer off mechanically. Others word it differently, but every one of them sets a window.
Common questions, straight answers
Can I add extra water to make tile adhesive easier to spread?
No. The amount of polymer in the bag is fixed, so extra water adds none — it spreads the same quantity through a larger volume. The film that carries the bond comes out thinner and discontinuous, and the surplus water leaves porosity and shrinkage behind when it goes. The trap is that the mistake feels like an improvement: the mix stirs more easily and spreads more pleasantly at the exact moment it is being weakened. Gauge to the ratio printed on the bag. It differs between products, and the strength figures on that bag were measured at that ratio and no other.
How thick can tile adhesive be applied?
A standard thin-bed adhesive works at roughly 2 to 6 mm, with isolated build-ups to around 10 or 12 mm where the data sheet allows it — check yours, because the limit varies by product. Adhesive is not a levelling material. The deeper the bed, the more shrinkage movement it builds up, and the layer that was supposed to hold becomes the weak point itself. Where the substrate needs more correction than that, level it first, or use a product specifically rated for thick-bed work.
How flat does a wall or floor need to be before tiling?
Within 3 mm under a 2 m straightedge. The reason is arithmetic: a 3 mm bed of adhesive cannot physically take out a 5 mm hollow. The tolerance that a thick mortar bed once absorbed now sits entirely with the substrate, so anything worse than that gets corrected before tiling starts rather than bridged with adhesive.
How long must a new substrate dry before tiling?
For concrete walls the standard asks for a minimum of 6 weeks. For floors, a cement and sand screed dries at roughly 1 mm a day up to 40 mm thick in good conditions, and considerably longer beyond that depth or in poor drying conditions — but time is an estimate, not a test. Measure it: for tiling, a cementitious screed is looked for at 2 per cent residual moisture or below. The 75 per cent relative humidity figure often quoted comes from the flooring standard for vinyl and wood, not from the tiling standard. And the reason for waiting is not simply dampness. A substrate still giving up construction moisture is still shrinking, and it goes on shrinking under tiles that are already fixed, feeding that stress straight into the bond line.
How soon after priming should the tiling go on?
Sooner than most people assume. A primed surface is clean and slightly tacky, and over days it collects dust and the traffic of other trades. Manufacturers set a window — commonly around 24 hours uncovered — after which a further coat is required, and more drastic preparation if it has stood much longer than that. The window is on the data sheet, and it belongs to the specification rather than to the small print.
Can I tile directly onto plywood?
Not any more. For walls, BS 5385-1:2018 removed plywood as a background for direct tiling in favour of a proprietary tile backer board. For floors, BS 5385-3:2024 calls for an intermediate layer over plywood and other wood-based boards — a backer board, an uncoupling membrane or a tanking system. Plywood can still do structural work underneath. It simply no longer meets the dimensional stability the tiling standards have always demanded of the surface that gets tiled.
What do the letters and numbers on the bag mean — C2 TE S1?
They are performance classes under EN 12004, each backed by a test. C is cementitious; D is dispersion and R is reaction resin. 1 is normal bond strength, at least 0.5 N/mm²; 2 is improved, at least 1.0 N/mm². T is slip resistance — the tile moves no more than 0.5 mm on a vertical surface. E is extended open time, at least 30 minutes; F is fast setting. S1 is deformable and S2 highly deformable, measured by how far a set strip of adhesive bends before it breaks. Every one of those figures was obtained under standardised laboratory conditions, which is why the bag also specifies the conditions on site.
What temperature can I tile in?
The floor of the range is almost universally 5 °C; the ceiling varies by product, typically between 25 and 30. 2 details matter more than the numbers. The limits apply to the substrate as well as to the air — a 5 °C room over a 2 °C screed is outside specification. And they apply right through the curing period, not only at the moment of spreading. Outside the range the adhesive does not reach the strength printed on the bag. It reaches an unknown one.
Why does porcelain need a special adhesive?
Porcelain absorbs almost no water, 0.5 per cent or less, so it offers a cement mortar neither suction nor a mechanical key in a thin bed. The bond at that dense surface is carried by the polymer instead. The powder redisperses the moment it meets the mixing water, and as the water leaves the dispersed particles coalesce into a film that grips where bare cement cannot. In practice that means a C2 class adhesive as the baseline for porcelain, with S1 deformability where the floor is heated or movement is expected.
Based on BS 5385 Parts 1 and 3, EN 12004 Parts 1 and 2 for the classification of cementitious adhesives and the conditions under which they are tested, BS 8203 for the moisture figures that belong to the flooring standards, and the technical data sheets of adhesive manufacturers. Bed thicknesses, gauging water, open times and application limits are properties of the specific product: where a manufacturer's data sheet differs from anything set out here, the data sheet has the last word.