Jadnashua
Retired Defense Industry Engineer xxx
Many people do not understand the difference in the relationship between bonding with thinset with the use of a membrane such as Kerdi or Ditra from Schluter systems, and a thinset when used over something like a cement board, or even plywood. Very different mechanical things are going on.
First, thinset does not stick to either Kerdi or Ditra. What, you say, how does it hold anything together? With Kerdi, and the fleece on the bottom of the Ditra, it bonds because, when properly installed, the fleece (which IS bonded to the membrane) is embedded by the semi-fluid thinset - IOW, the fleece is encapsulated by the mortar, and because the fleece is bonded to the sheet membrane, once the thinset cures, it's all locked together. Sort of like the difference between getting gum in your hair verses a wax job...the gum sticks, the wax encapsulates and then hardens...they both create a bond, but do it in different ways. So, the bond strength is limited by the strength of the fleece, not by the ability of the thinset to actually stick to the membrane. Knowing this, it should become clear why it is VERY important to use the right thinset and to mix it up properly...it must be viscous enough so that when you press the membrane into the stuff, the thinset actually flows around all of the fleece...this is CRITICAL to the proper bonding. Letting your thinset skin over, or get its moisture sucked out by a dry substrate before embedding the membrane is the major user error. Knowing how it works, and it's easy to check (pull a section back off and look for full saturation and coverage of the membrane!), and you can avoid that pitfall.
Kerdi has fleece on both sides, and the same action is what is bonding the membrane to both the wall, floor, or the tile attached...the thinset surrounds the fleece, but it does need to bond in the traditional way to the tile you install.
As to Ditra, the top surface does not have any fleece, and as was mentioned earlier, thinset doesn't stick to it. So, how can a tile be anchored to a surface that thinset does not stick to or bond to? The shape of the Ditra membrane has square dovetailed pockets in it...when you spread the thinset, you pack it into the pockets, and then comb out enough over the top of it to provide a bond to the tile. The thinset to tile bond is 'normal', but the bond to the membrane itself is entirely mechanical...the pillars of thinset that hold the tile up are locked into the dovetails. That the tile is actually supported by those pillars to the bottom of the pockets (and the whole membrane is bonded to the floor) and the fact that there are gaps between those pillars of thinset is the reason why there is a minimum size requirement when using Ditra - you must have enough pillars holding it up so a point load won't have a chance to tip the tile...that works out to a minimum of 2" square tile. Anything smaller and you MIGHT dislodge things with a point load (it might work fine, but Schluter doesn't like to take chances - they're conservative).
The cement in thinset gets it strength, mechanical bond strength, by the crystalline structure of the cement when it cures...this creates interlocking spikes of cement spires, or crystals. The sand and other fillers minimize the amount of cement needed and provide density and strength to hold the tile up while the thinset cures (i.e., so it doesn't sink). It is the water you use to mix the thinset with that activates the chemical process that starts the curing process, and maintaining excess moisture means that all of the cement can become fully hydrated and cure, creating the maximum number and longest crystals possible. The environment between the waterproof membrane and the potentially impervious tile provides the optimum environment for the mortar to reach its highest possible strength. Those cement crystals actually grow into the microscopic depressions and pores of the tile, and lock it in place.
So, given that the actual bond to the membrane is relying on a different mechanism than say tile over a cement board (cbu), it is a logical conclusion that the use of a dryset mortar (unmodified) to hold the tile in place is entirely adequate, and preferred. Without getting into the full chemical characteristics, a modified thinset must not only cure, but it must also DRY the modifiers for it to reach its full capability. When used over a waterproof membrane, that can take literally MONTHS to happen, where a dryset (unmodified) will cure at a predictable rate, and actually works best when it is kept wet during that process.
A very different scenario exists when trying to bond a tile to cbu or plywood, or nearly anything else, and a modified thinset can often provide a stronger bond than an unmodified. With Ditra over a plywood subfloor, because the wood moves MUCH more than the thinset or the membrane with the tile on top, they specify a modified thinset to help maintain the bond to the plywood. The wood also allows any excess moisture to be wicked away, so drying isn't a factor like it would be on top of Ditra with a dense tile on top. The differences in expansion/contraction between the thinset and plywood means the modifiers used also allow a little more flex than when an unmodified is used.
First, thinset does not stick to either Kerdi or Ditra. What, you say, how does it hold anything together? With Kerdi, and the fleece on the bottom of the Ditra, it bonds because, when properly installed, the fleece (which IS bonded to the membrane) is embedded by the semi-fluid thinset - IOW, the fleece is encapsulated by the mortar, and because the fleece is bonded to the sheet membrane, once the thinset cures, it's all locked together. Sort of like the difference between getting gum in your hair verses a wax job...the gum sticks, the wax encapsulates and then hardens...they both create a bond, but do it in different ways. So, the bond strength is limited by the strength of the fleece, not by the ability of the thinset to actually stick to the membrane. Knowing this, it should become clear why it is VERY important to use the right thinset and to mix it up properly...it must be viscous enough so that when you press the membrane into the stuff, the thinset actually flows around all of the fleece...this is CRITICAL to the proper bonding. Letting your thinset skin over, or get its moisture sucked out by a dry substrate before embedding the membrane is the major user error. Knowing how it works, and it's easy to check (pull a section back off and look for full saturation and coverage of the membrane!), and you can avoid that pitfall.
Kerdi has fleece on both sides, and the same action is what is bonding the membrane to both the wall, floor, or the tile attached...the thinset surrounds the fleece, but it does need to bond in the traditional way to the tile you install.
As to Ditra, the top surface does not have any fleece, and as was mentioned earlier, thinset doesn't stick to it. So, how can a tile be anchored to a surface that thinset does not stick to or bond to? The shape of the Ditra membrane has square dovetailed pockets in it...when you spread the thinset, you pack it into the pockets, and then comb out enough over the top of it to provide a bond to the tile. The thinset to tile bond is 'normal', but the bond to the membrane itself is entirely mechanical...the pillars of thinset that hold the tile up are locked into the dovetails. That the tile is actually supported by those pillars to the bottom of the pockets (and the whole membrane is bonded to the floor) and the fact that there are gaps between those pillars of thinset is the reason why there is a minimum size requirement when using Ditra - you must have enough pillars holding it up so a point load won't have a chance to tip the tile...that works out to a minimum of 2" square tile. Anything smaller and you MIGHT dislodge things with a point load (it might work fine, but Schluter doesn't like to take chances - they're conservative).
The cement in thinset gets it strength, mechanical bond strength, by the crystalline structure of the cement when it cures...this creates interlocking spikes of cement spires, or crystals. The sand and other fillers minimize the amount of cement needed and provide density and strength to hold the tile up while the thinset cures (i.e., so it doesn't sink). It is the water you use to mix the thinset with that activates the chemical process that starts the curing process, and maintaining excess moisture means that all of the cement can become fully hydrated and cure, creating the maximum number and longest crystals possible. The environment between the waterproof membrane and the potentially impervious tile provides the optimum environment for the mortar to reach its highest possible strength. Those cement crystals actually grow into the microscopic depressions and pores of the tile, and lock it in place.
So, given that the actual bond to the membrane is relying on a different mechanism than say tile over a cement board (cbu), it is a logical conclusion that the use of a dryset mortar (unmodified) to hold the tile in place is entirely adequate, and preferred. Without getting into the full chemical characteristics, a modified thinset must not only cure, but it must also DRY the modifiers for it to reach its full capability. When used over a waterproof membrane, that can take literally MONTHS to happen, where a dryset (unmodified) will cure at a predictable rate, and actually works best when it is kept wet during that process.
A very different scenario exists when trying to bond a tile to cbu or plywood, or nearly anything else, and a modified thinset can often provide a stronger bond than an unmodified. With Ditra over a plywood subfloor, because the wood moves MUCH more than the thinset or the membrane with the tile on top, they specify a modified thinset to help maintain the bond to the plywood. The wood also allows any excess moisture to be wicked away, so drying isn't a factor like it would be on top of Ditra with a dense tile on top. The differences in expansion/contraction between the thinset and plywood means the modifiers used also allow a little more flex than when an unmodified is used.