Concrete joint sealant replacement field guide

Joint sealant, and the two jobs it has to do at once

Concrete joint sealant is the flexible material that fills a joint and does two jobs that pull against each other: it keeps water and dirt out of the joint, and it stretches and squeezes so the joint can move without tearing the seal. A rigid filler can do the first. Only an elastic sealant on a properly shaped joint does both, and doing both is the whole point.

The joint exists because concrete moves. It shrinks as it dries, it grows and contracts with temperature, and the slabs on either side ride that movement back and forth for the life of the structure. The sealant is the only part of that joint that has to live with the movement and stay watertight through it. When it does its job, water runs off the surface and the joint underneath stays dry. When it lets go, water gets in, and that is where the damage starts.

This guide is about the sealant in the joint, not where the joint goes. The layout, the spacing, the depth of cut, and the difference between a control, construction, and isolation joint live in the control joint layout guide, and the damage that follows a leaking joint, the spalled edges and the corroded steel, lives in the spall repair guide. Read this one when the joint is already there and the seal is what you are building or replacing.

Why does concrete joint sealant fail?

Concrete joint sealant fails one of two ways, and they look different at the joint. Adhesive failure is the sealant pulling clean off one joint wall, leaving a gap down the side you can run a blade into. Cohesive failure is the sealant splitting down its own middle while it stays stuck to both walls. Both let water in, and both trace back to the same short list of causes.

The biggest one is the sealant cannot move enough. The joint opens further than the sealant can stretch, because the sealant is bonded to the bottom of the joint as well as the two sides, so it has no room to neck down and elongate. That is three-sided adhesion, and it is the failure a backer rod exists to prevent. The next biggest is prep. Sealant put on a dusty, damp, or laitance-coated joint wall never really bonds, so it peels the first cold night the joint opens. Then there is the wrong material for the movement, a stiff sealant in a working expansion joint, and plain age, because every sealant hardens and loses its stretch over years of sun and weather.

Here is the part that connects this guide to the other two. A failed joint seal is not the end of the damage, it is the start of it. Once water gets into the joint it sits against the slab edges, it freezes and pries the edges apart, it carries chloride down to the reinforcing steel, and it undermines the subbase. The cracked, spalled edge and the corroded bar in the spall repair guide very often began as a joint seal nobody maintained. Fix the seal and you stop feeding the failure.

The joint you are sealing: control, construction, isolation, expansion

Before you pick a sealant you have to know which joint you are looking at, because the joint type tells you how much it moves, and movement drives every other decision. The control joint layout guide covers what each one is and where it goes. The short version, from the sealant's point of view, is how far the two sides travel.

A control, or contraction, joint is the sawcut weak line that steers the shrinkage crack. The two sides stay close together and move very little once the slab has done most of its shrinking, so it is the joint that often takes a stiff filler instead of a stretchy sealant on a hard floor. A construction joint is the cold joint between two pours. It moves a little more than a control joint but is still a low-movement line. An isolation joint, frequently called an expansion joint in the field, is the full-depth gap that frees the slab from a column, a wall, a footing, or an adjacent structure. It is the high-movement joint, and it is where elastic sealant earns its keep.

The naming gets loose on real jobs. People call any sealed joint an expansion joint, but a true expansion or isolation joint is the one designed to open and close the most, and it is the one that punishes a wrong sealant choice the fastest. When in doubt, look at the joint width and whether there is a compressible filler down in it. A wide joint with foam or fiberboard below the seal is built to move, and the sealant over it has to move with it.

Joint movement: why the seal has to stretch

The joint is not a static gap. It is widest on the coldest day, when the slabs have contracted and pulled apart, and narrowest on the hottest day, when they have grown and closed the gap. A sealant installed on a warm afternoon is sitting in a joint that is near its tightest, so it has to stretch every cold night just to stay attached. Install on a cold morning and the opposite is true, the joint will close on the sealant and put it in compression.

That daily and seasonal cycle, repeated for years, is the load the sealant lives under. Add the early movement from drying shrinkage on a young slab, which keeps opening the joint for weeks to months after the pour, and you see why timing matters. Seal a fresh slab too early and the joint is going to keep opening well past whatever the sealant can take, the same way a stiff joint filler splits if you put it in before the shrinkage is out. On new work, the longer you can wait before the final seal, the less total movement the sealant has to absorb over its life.

Everything downstream serves this one fact. The sealant must stretch. The material has to be elastic enough for the range the joint travels, the geometry has to let it neck down and elongate instead of locking it, and the prep has to make it stick hard enough that it stretches instead of peeling. Get the movement wrong in your head and you will pick the wrong material and build the wrong shape, and the seal will fail on schedule.

Sealant types: polyurethane, silicone, polysulfide, and hybrids

Four families cover most concrete joint work, and they trade off the same way every time: how far they stretch, how well they take traffic and abrasion, how they hold up to sun, and what they cost. Match the family to the joint, then confirm the specific product against the manufacturer's data sheet, because formulations inside each family vary widely.

Polyurethane is the workhorse for concrete joints. It bonds well to concrete, takes foot and wheel traffic, resists abrasion and tear, and handles the movement in most exterior slab and pavement joints. It is the default for horizontal joints that see traffic. Silicone stretches the most and holds up the longest against sunlight and weather, which is why it dominates on high-movement joints and on facades, but standard silicone is soft, picks up dirt, and does not take abrasion or traffic well, so it is usually the wrong call for a floor joint that gets driven on. Polysulfide is the old chemical-resistant choice, used where fuel, solvents, or chemical exposure would eat a polyurethane, common around fueling areas and some industrial and immersed joints. Hybrids, the silyl-modified polymers often sold as MS polymer or hybrid sealants, sit in the middle: good adhesion, good movement, and they can be painted, which the others resist.

The other axis cuts across all of them, and it is the one crews get wrong most: self-leveling versus non-sag. A self-leveling sealant, what ASTM C920 calls grade P for pourable, is fluid enough to flow out and level itself in a horizontal joint, so it wets the joint walls and finishes flat with little tooling. A non-sag sealant, grade NS for gunnable, is thick enough to hold its shape on a vertical or overhead joint without slumping out, but it has to be tooled in by hand. Self-leveling goes in the floor. Non-sag goes on the wall. Put non-sag in a wide floor joint and you fight it; put self-leveling on a slope and it runs to the low end and leaves the high end starved.

What is the sealant movement class under ASTM C920?

ASTM C920, the specification for elastomeric joint sealants, sorts sealants by type, grade, class, and use, and the class is the number that tells you how far the sealant can move. A class 25 sealant is tested to withstand the joint opening and closing by at least 25 percent of its width at the time of installation, so plus or minus 25 percent of movement. That is the common class for joints with moderate movement, and it is the one to recognize first on a data sheet.

The full designation reads like a code, and it pays to read it. Type S is single-component, ready to gun out of the tube or sausage. Type M is multicomponent, mixed on site, common on larger commercial jobs. Grade P is pourable, the self-leveling sealant for floors. Grade NS is non-sag, the gunnable sealant for walls. The use letters say what it is qualified for, with T for traffic and I for immersed among them. So a Type S, Grade P, Class 25, Use T sealant is a single-part, self-leveling, plus or minus 25 percent product rated to be driven on.

C920 also defines a lower class for joints that barely move and higher classes for joints that move a lot, and some high-movement silicones are rated well past 25 percent, into plus 100 and minus 50 percent territory on the products built for it. The number you need comes from the joint, not the shelf. Estimate the total movement the joint will see between its widest and narrowest, then pick a class and an installed width that keep the sealant inside its rated range. The project specification and the sealant manufacturer set the final call, and where the spec names a class, that class governs over any rule of thumb.

What is a backer rod?

A backer rod is a compressible foam rope pushed into the joint below the sealant, and it does three jobs at once. It sets the depth of the sealant so you get the right thickness instead of pouring a tube into a bottomless joint. It backs up the sealant so you can tool it and force it against the joint walls. And it breaks the bond at the bottom, so the sealant sticks to the two walls only and not to the floor of the joint. That last job is the one that keeps the seal alive.

Sealant bonded on three sides, both walls and the bottom, cannot stretch when the joint opens, because it is anchored along its whole back. It tears instead. With a backer rod, the sealant is free at the bottom, so when the joint opens the middle of the bead necks down and elongates and the seal survives the movement. The rod is doing two things people credit to the sealant: it is the bond breaker and it is the depth gauge.

Pick the rod for the joint. Closed-cell backer rod has a smooth, non-absorbent skin and is the general-purpose choice, especially where moisture is around, because it will not soak up water and feed it back to the joint. Open-cell rod is softer and breathable, used with sealants that need air to cure and in joints whose width wanders, because it compresses further without bursting. There is a hybrid with a closed skin over an open core. Match the rod diameter to the joint so it goes in under compression, roughly a quarter larger than the joint width, and do not puncture closed-cell rod, because a punctured cell can outgas a bubble up into a curing sealant. Where the joint is too shallow for a rod, a bond-breaker tape on the bottom does the same anti-bond job without setting depth.

What is the 2 to 1 rule for sealant?

The 2 to 1 rule is the joint geometry that lets the sealant move: the sealant bead should be about twice as wide as it is deep. On a joint half an inch wide, you want roughly a quarter inch of sealant depth over the backer rod. Tool that bead and it takes an hourglass shape, thick at the two bonded edges and thinner through the middle, and that shape is what makes it stretch instead of split.

The mechanics are worth carrying. When the joint opens, the thin waist in the middle of the hourglass is what elongates, and the shape shifts the peak stress away from the bonded faces and into the middle of the sealant where it belongs. A bead that is too deep, too square in section, concentrates stress right at the bond line and peels off a wall. A bead that is too thin and stretched flat across a wide joint tears straight through. The 2 to 1 width to depth, with the backer rod setting that depth, is what keeps the stress in the body of the sealant and off the bond.

There are limits on the rule. Most manufacturers set a minimum sealant depth, commonly about a quarter inch, so on a narrow joint the depth floor wins and the ratio loosens, and they set a maximum width a single bead should span before the joint needs redesign. ASTM C1193, the guide for the use of joint sealants, lays out the geometry in detail, and the sealant manufacturer's data sheet gives the depth and width limits for their specific product. Where those numbers differ from the round 2 to 1, follow the data sheet, because the product was tested at its own dimensions.

Joint prep: the make-or-break step

Sealant sticks to the joint wall or it does not, and that is decided before the sealant ever comes out of the gun. There is no bond on a dirty joint. Most resealing failures that get blamed on the sealant are prep failures wearing a sealant's name, and the inspector who knows this trade checks the prep, not the bead.

On a replacement, the first job is getting the old sealant out, all of it. Cut it loose from both walls with a utility knife or a joint-sealant blade run down each side, then pull it and scrape the walls back to clean concrete. Old sealant residue left on the wall is a bond breaker for the new material, so it has to go. For a clean, square, sound wall, the standard move is to rout or saw the joint, re-cutting it with a dry diamond blade so the new sealant bonds to fresh concrete instead of weathered, contaminated surface. On a new joint, that fresh sawcut face still carries dust and laitance that has to come off.

Then clean it and get it dry. Abrasive blast or wire-wheel the joint faces to take off laitance, curing-compound residue, and weak surface, working the top inch or so of the joint wall where the sealant bonds. Blow the joint out with oil-free compressed air right before you seal, because dust settles back in fast. The joint has to be dry, since most sealants will not bond to a damp wall, and that means waiting out the joint after a wash or a rain, not just wiping the top. Priming is where the manufacturer governs outright. Many concrete joints seal fine without a primer, but some sealants and some substrates need one to bond, and the only correct answer is the one on the data sheet. Where it calls for primer, use it, and respect the flash-off time before sealing.

Application and tooling the bead

With the joint cut, cleaned, dried, primed if the data sheet says so, and the backer rod set to depth, the sealant goes in. Gun it in one steady pass, pushing the nozzle ahead of the bead so it fills from the bottom up against the backer rod and does not trap air behind it. Overfill slightly. You want more material than the finished bead so tooling presses it into full contact with both walls.

Then tool it, and tool it wet. Tooling is not for looks. Drawing a spatula or a tool down the bead while the sealant is still wet forces it against the joint walls, wets out the bond, and shapes the slightly concave hourglass face that lets the bead stretch. A bead that is gunned and left untooled may be touching the walls without being bonded to them, and it pulls away the first time the joint opens. On self-leveling sealant in a floor joint, the material flows out and levels itself, so it needs little or no tooling, which is half the reason it is used on floors. On non-sag sealant in a wall joint, tooling is the whole skill, because the material holds whatever shape you leave it.

Mind the conditions. Seal when the joint is near the middle of its movement range if you can, not on the hottest or coldest hour, so the sealant is not installed at one extreme and asked to take the full swing from there. Watch the temperature and humidity limits on the data sheet, give multicomponent material a proper mix, and protect a fresh bead from traffic and water until it has skinned and cured. A floor sealed and driven on before it sets is a bead smeared down the aisle.

Horizontal versus vertical, and traffic joints

Orientation decides grade, and traffic decides material. A horizontal joint in a floor or a slab takes a self-leveling, grade P sealant, because the material flows out, levels, and wets the walls with little tooling, and a floor joint that gets driven on needs a traffic-rated, abrasion-tough sealant under it, which is why polyurethane runs most floor joints and standard silicone stays off them.

A vertical joint on a wall, or an overhead joint, takes a non-sag, grade NS sealant, because anything pourable runs straight out and pools at the bottom. Non-sag holds its place while you tool it. The same split shows up on a ramp or a sloped slab: a self-leveling sealant on a grade flows downhill and leaves the top of the joint starved, so on a slope you reach for a non-sag or a slope-grade product even though the joint is technically a floor.

Traffic joints carry an extra demand the others do not. A wheel rolling across a joint pounds the sealant and the joint edges, so a traffic joint wants a tougher, harder sealant or, on the hardest floors, a semi-rigid filler instead of a flexible sealant, which is the next decision. Match the material to both questions at once. Is the joint horizontal or vertical, and does it carry traffic? The wrong answer to either one shows up as a bead that slumped, tore, or got chewed off the edge.

Joint filler versus joint sealant

A joint filler and a joint sealant are different materials doing opposite jobs, and using one where the other belongs is a common and expensive mistake. A flexible sealant is soft and elastic and its job is to move with the joint while keeping water out. A semi-rigid filler is hard and its job is to support the joint edges against heavy wheels, not to move. The control joint layout guide covers when an industrial floor wants a filler at all; this is the line between the two from the sealant side.

On an interior industrial or warehouse floor, the control joints take a semi-rigid filler, a stiff epoxy or polyurea poured flush, because a hard forklift wheel rolling across an unsupported joint edge breaks the concrete arris off, and the filler backs up the edge so the wheel rolls across instead of slamming the corner. That filler cannot stretch much, so it goes in late, after most of the slab's drying shrinkage is out, commonly toward 60 to 90 days, or it splits away from the joint wall and you are back to an open, spalling joint. A flexible sealant in that same hard-traffic joint would be punched out in a week.

The flip side is just as wrong. A semi-rigid filler in a joint that actually moves, an exterior expansion joint, a joint exposed to real temperature swing, cannot take the movement and tears loose. There the elastic sealant is the only right answer. The decision is one question with two parts: does the joint mainly need to move, or mainly need to support a wheel? Moving joints get flexible sealant. Hard-traffic control joints on a stable interior floor get semi-rigid filler. A joint that does both, a moving traffic joint, is a design problem for armored edges and a specialty detail, not a single tube of anything.

Data center and clean-floor joints

On a data center or a clean industrial floor the joint seal is a contamination and durability problem as much as a water problem. The slab joints have to control dust, take constant hard-wheeled cart and lift traffic across tight flatness tolerances, and stay put, so the structural-slab control joints in these floors are usually semi-rigid filler for the edge support, not a soft flexible sealant that a wheel would track through and a vacuum would never beat.

Where flexible sealant still belongs in these buildings is the moving joints and the perimeter: isolation joints around columns, walls, and equipment pads, and any joint exposed to washdown or moisture. Those joints move, so they need elastic sealant over a backer rod, built to the same 2 to 1 hourglass as any other moving joint. The mistake is treating a whole clean floor as one material. The hard-traffic control joints want filler for the edges; the isolation joints want sealant for the movement.

The broader heavy-slab and flatness picture, why these floors run fewer, better joints and where the durability money goes, sits in the control joint layout guide. The joint-seal lesson for a data center is narrow and firm: spec the filler and the sealant per joint type, fill the traffic joints late enough that the shrinkage is out, and keep the moving joints flexible, because a tracked-out filler in a moving joint and a torn sealant in a traffic joint are both failures the operation feels at the cart and the rack.

Field example: resealing a 3/4 in exterior expansion joint

Take an exterior plaza expansion joint, 3/4 in wide, between two slabs, with the original sealant cracked down the middle and pulled off one wall, water getting in, and the slab edges starting to show damage. This is a moving joint exposed to weather and some foot traffic, so it wants a traffic-capable elastic sealant, a polyurethane at class 25 or better, grade NS tooled in or a self-leveling grade P if the joint sits dead flat.

Work the sequence. Cut and pull the old sealant, then rout the joint with a dry diamond blade so both walls are fresh, square concrete. Abrasive blast the faces, then blow the joint clean and dry with oil-free air just before sealing. Check the data sheet for a primer and apply it if it calls for one. Set a closed-cell backer rod about 1 in diameter so it seats under compression in the 3/4 in joint, deep enough that the sealant over it lands near 3/8 in, holding the 2 to 1. Gun the sealant in from the bottom up against the rod, overfilled, then tool it wet into a slightly concave hourglass against both walls.

Then write down what you built. The width, the sealant product and its C920 class, the backer rod type and the installed depth, the primer if used, and the date, because the next person resealing this joint inherits whatever you leave undocumented. Seal it near the middle of the day's temperature swing, not at an extreme, and keep traffic and water off it until it cures.

QC and inspection: depth, ratio, adhesion, cure

The inspection checks the things that decide whether the seal lasts, and most of them are set before the sealant cures, so the time to catch them is during the work, not after. The first thing a knowledgeable inspector looks at is not the surface of the bead. It is the depth and the ratio, because a bead at the wrong depth fails no matter how clean it looks.

Check the backer rod depth so the sealant lands near 2 to 1 width to depth, and confirm there is a rod or a bond breaker at all, since three-sided adhesion is the silent killer and you cannot see it once the bead is in. Check that the joint was cut clean and dry before sealing, because the prep is where the bond is won. After cure, the field test for adhesion is a hand pull-back, or a knife-and-pull check on a sample length, where you cut a slice and pull it to see whether it stretches and holds or peels off a wall. ASTM C1193 describes field adhesion checks for sealant joints. A bead that peels off the concrete failed adhesion, prep or primer; a bead that tears in its own body at low stretch failed cohesion, often wrong material or a bad mix.

Confirm the cure before the joint goes into service. A polyurethane skins in hours but takes days to cure through, and traffic or water on an uncured bead ruins it. The acceptance numbers, how much adhesion loss is allowed and how the sample is tested, come from the project specification and the sealant manufacturer, so check the bead against those, not against a number from memory.

The re-seal cycle the owner inherits

Joint sealant is a maintenance item, not a permanent part of the building. Every elastic sealant hardens over years of sun, weather, and movement, loses its stretch, and eventually cracks or pulls off a wall, and when it does the joint is open again and water is getting in. The owner who treats the seal as one-and-done is the owner who pays for spalled edges and corroded steel a few years later. The seal gets re-done on a cycle, and the cycle is part of owning the structure.

How long a seal lasts varies with the sealant, the exposure, and how much the joint moves, so put a number on it for your own joints by inspecting them, not by guessing. Walk the joints on a schedule, look for the bead pulling off a wall, splitting down the middle, going hard and chalky, or sitting below the surface where it has torn loose underneath. Catch a failing seal and re-do that run before the open joint has fed water into the slab for a winter.

The blunt version: inspect the joints, and re-seal on a cycle. The cost of cutting out a tired bead and laying a new one is small next to the cost of the joint failures it prevents, the freeze-pried edges, the chloride that reaches the steel, the undermined subbase. A sealant maintenance plan is the cheapest part of keeping a slab alive, and it is the part most often skipped until the damage shows.

What to document

Log the build of each joint, because the seal is a maintenance item that gets cut out and redone, and the only way to repeat what worked is to know what went in. When the seal fails again or the joint next to it goes, the record is what tells the next crew what was in there, what shape it was built to, and how long it lasted, so the re-seal is a repeat of what worked and not a fresh guess.

Capture, for each joint or run, the joint type and width, the sealant product and its ASTM C920 type, grade, class, and use, the backer rod type and the installed sealant depth, whether a primer was used and which one, the width to depth ratio you built, and the date sealed. If the joint takes a semi-rigid filler instead, record that and the fill date relative to the pour. The table below is the minimum a maintenance program can run on.

Common mistakes

• No backer rod or bond breaker, so the sealant bonds on three sides and tears the first time the joint opens.
• Wrong depth or width-to-depth ratio, a bead too deep and square or too thin and stretched flat.
• Sealing a dirty, dusty, or damp joint with no real prep, so the bead never bonds to the wall.
• Leaving old sealant residue on the joint wall, which acts as a bond breaker for the new material.
• Picking a sealant whose movement class is below what the joint actually travels.
• Self-leveling sealant on a slope, where it runs to the low end, or non-sag poured into a flat floor joint and fought by hand.
• Standard silicone in a traffic joint, where it has no abrasion resistance and gets chewed off the edge.
• A semi-rigid filler in a moving expansion joint, or a flexible sealant in a hard-traffic interior control joint.
• Skipping primer where the data sheet calls for it, or using it past its flash-off and trapping it under the bead.
• Treating the seal as permanent and skipping the re-seal cycle until the open joint has already damaged the slab.

Field checklist

Standards and references

The sealant manufacturer's data sheet governs the prep, the primer, the joint dimensions, the movement class, and the cure, because the warranty rides on the tested system, and the project specification governs which product and class go in the joint. Everything below is the framework those two are built on, and where the spec or the data sheet is stricter or more specific, it wins.

ASTM C920, the specification for elastomeric joint sealants, is the one to know, classifying sealants by type, grade, class, and use, and it is where the class 25 movement rating and the grade P and grade NS designations come from. ASTM C1193, the guide for the use of joint sealants, covers the joint design itself, the width to depth ratio, the hourglass geometry, the backer rod, and field adhesion checks. For pavement and slab joints, ACI floor and slab guidance and the joint-sealing practice from the concrete pavement bodies cover sawcut joint sealing and the filler-versus-sealant decision, the same ground the control joint layout guide works through. Backer rod, primer, and bond-breaker products carry their own ASTM and manufacturer specifications.

The document numbers and the specific provisions shift between editions, so confirm them against the version the project adopted before you cite a clause. Name the standard that actually controls the point, lean on ASTM C920 and C1193 and the sealant manufacturer for the sealant itself, and let the project specification override any rule of thumb when it is stricter.

Units, terms, and conversions

Joint and sealant work reads in inches and feet on US jobs and millimeters and meters elsewhere, and the same idea shows up under different names across a spec, a drawing, and a product sheet. The 2 to 1 width to depth ratio is unitless, so it travels the same in metric. Movement class is a percentage of joint width, so it travels too.

Joint width and sealant depth are inches in the US and millimeters in metric, where 1/2 in is about 13 mm and 3/4 in is about 19 mm. Movement is given as a percentage of the joint width under ASTM C920. Sealant is sold by the cartridge, the sausage, or the pail, and self-leveling versus non-sag is sometimes labeled pourable versus gunnable, the grade P and grade NS of C920. An expansion joint, an isolation joint, and a movement joint often mean the same high-movement joint in field talk.

Joint sealant

The elastic material in a joint that keeps water out while letting the joint move

Backer rod

A compressible foam rope set below the sealant that breaks the bottom bond and sets the depth

Three-sided adhesion

Sealant bonded to both walls and the bottom, which cannot stretch and tears when the joint opens

Hourglass / 2 to 1 ratio

A sealant bead about twice as wide as deep, thinned in the middle so it stretches without peeling

Movement class

The plus-or-minus percentage of joint width a sealant can take, per ASTM C920, such as class 25

Self-leveling / non-sag

Pourable grade P for horizontal joints versus gunnable grade NS for vertical and overhead joints

Adhesive / cohesive failure

Sealant peeling off a wall versus splitting through its own body

Semi-rigid filler

A hard epoxy or polyurea that supports a traffic joint edge but does not move with the joint

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• How far apart should control joints be?
• Why does spalled concrete happen?
• What is a waterstop?
• What is a building expansion joint?
• What is a concrete admixture?