Silicone Shrinkage: 2 to 4 Percent and the Tool That Must Allow It

Why silicone shrinks two to four percent

Every moulded polymer shrinks because it is formed hot and used cold. Silicone is unusual in that the shrinkage is large, consistent and driven by chemistry rather than by crystallisation, which makes it predictable in principle and awkward in practice.

The shrinkage has three parts. Thermal contraction as the part cools from the cure temperature to room temperature is the largest and the most predictable, at roughly 1.5 to 2 percent for a typical silicone. Cure shrinkage, from the network forming and pulling the chains together, adds a fraction of a percent. Volatile loss during and after cure adds a little more.

The sum lands between 2 and 4 percent for most compounds, and the width of that range is what makes tool design interesting. A tool cut at 3 percent for a compound that shrinks 4 percent produces a part that is 1 percent undersize, which on a 100 mm gasket is a millimetre.

The practical answer is to cut the tool at the middle of the range for the compound, run a trial, measure, and adjust. The adjustment is done by adding metal to the cavity, not by removing it, so a trial tool that is cut slightly oversize is cheaper to correct than one that is undersize.

Hardness sets where in the range you land

Softer compounds shrink more. On a measured set of silicone compounds, the linear shrinkage was 4.1 percent at 30 Shore A, 3.7 at 40, 3.3 at 50, 3.0 at 60, 2.7 at 70 and 2.4 at 80.

The mechanism is the oil. A soft compound reaches its hardness by carrying a high loading of low molecular weight silicone oil, and that oil is a diluent. It occupies volume in the cured part and it is partly lost over time, and the network around it is looser, so the part contracts more on cooling.

That means the shrinkage is tied to the same variable that causes blooming. A compound that blooms heavily is a compound that will shrink at the top of the range, because both are symptoms of a high mobile fraction.

The practical consequence is that a tool cannot be designed for a hardness, it has to be designed for a compound. Two 50 Shore A compounds from two suppliers can differ by 0.5 percent in shrinkage, which is enough to matter on a precision part.

Filler holds it back

A filler does not shrink. Silica, calcium carbonate, alumina and the mineral fillers used in a heat-resistant compound all occupy volume that stays constant through the cure and the cooling. The more filler, the less the part shrinks.

On the measured set, shrinkage was 3.6 percent with no filler, 3.3 at 10 phr, 3.0 at 20, 2.7 at 30 and 2.4 at 40. The relationship is close to linear, at about 0.03 percent per phr, which makes it predictable.

Filler also changes everything else. It raises hardness, raises tensile strength, reduces elongation and reduces tear strength. A compound that is filled to control shrinkage is a compound whose other properties have been moved to get there.

The interaction with hardness is the confusing part. Adding filler raises hardness, which by itself would reduce shrinkage; and the hardness rise is achieved without adding oil, which also reduces shrinkage. The two effects add, which is why a filled 60 Shore A compound shrinks much less than an unfilled 60 Shore A one.

Table 1 — Hardness, filler and the shrinkage that results

The tool allowance and the trial

A silicone tool is cut with a shrinkage allowance applied to every dimension: cavity, core, pitch, and any feature that has a tolerance. The allowance is a single number, and it is applied multiplicatively, not additively, so a 100 mm dimension and a 10 mm dimension both grow by the same percentage.

The trial is where the number gets corrected. A first shot is measured at 23 C after 24 hours, and the measured shrinkage is calculated from the tool dimension and the part dimension. The difference between the assumed and the measured value is the correction.

The correction should be applied to the tool by welding or by inserting, which is why the trial tool is cut slightly small in the cavity and slightly large on the core. A tool that is cut at the wrong allowance and cannot be corrected is a tool that has to be remade.

The measurement has to be taken on a stabilised part. A silicone part continues to cure and to shrink for days after demoulding, and a measurement at one hour gives a shrinkage that is 0.1 to 0.3 percent lower than the final value. Measure at 24 hours at minimum, and at a week if the part is critical.

Table 2 — The sequence that gets the allowance right

Shrinkage is not always the same in every direction

A compression moulded part shrinks roughly isotropically, because there is no flow to orient the network. An injection moulded liquid silicone part does not: the shear during filling orients the filler and the network, and the shrinkage along the flow is different from the shrinkage across it.

The difference is typically 0.2 to 0.5 percent, which is small in absolute terms and significant on a long, thin part. A 200 mm injection moulded silicone profile can be 1 mm different along its length than across its width, and a tool with a single allowance cannot fix both.

The response is to orient the part in the tool so that the critical dimension runs across the flow, or to specify the compound and the process so that the anisotropy is small. It is a design decision, not a tool correction.

A compression moulded part avoids the problem, which is one reason why precision silicone parts are often compression moulded even when injection would be faster.

Putting the number in the specification

A silicone part drawing should state the shrinkage allowance used, the compound, the cure cycle and the time at which dimensions are measured. Without those four, a dimensional argument cannot be resolved.

The allowance itself should be stated as a value with a tolerance, not as a range. A tool cut at 3.0 plus or minus 0.1 percent is a tool that will produce parts in tolerance; one cut at 2 to 4 percent is a tool that will produce some parts in tolerance.

The measurement time is the one that causes the most disputes. A part measured at demould and again at one week can differ by 0.3 percent, which on a 50 mm dimension is 0.15 mm and often the whole tolerance.

The last item is the compound change rule. A new supplier, a new batch or a new hardness is a new shrinkage, and the tool may need correcting. Writing that into the specification turns a surprise into a procedure.