Hardness Drift in Silicone: Cure State, Post-Cure and Ageing

Hardness is a cure state, not a fixed property

A silicone compound is specified by its hardness, and the hardness is treated as a material property like density. It is not. Hardness is a measure of crosslink density, and crosslink density keeps changing after the part leaves the tool.

The moulding cycle cures the part enough to be handleable and dimensionally stable. It does not finish the reaction. Residual Si-H groups, residual catalyst activity and unreacted vinyl all remain, and they continue to react slowly at room temperature for weeks.

Each new crosslink raises the modulus, and a higher modulus reads as a higher hardness on a durometer. That is why a part measured at 50 Shore A on the day of moulding reads 52 a month later and 56 a year later, and why the customer measures it and reports a non-conformance on a batch that was correct when it shipped.

The practical consequence is that hardness has to be specified with a time reference. A specification that says 50 plus or minus 5 without saying when is a specification that will be argued about.

Most of the drift happens in month one

The post-cure reaction is fast at first and then slows, because it is limited by diffusion of the remaining reactive groups through an increasingly tight network. On a measured 50 Shore A compound with no post-cure, the hardness change was 0 as moulded, 1.6 at one month, 2.8 at two, 3.6 at three, 4.9 at six, 5.6 at nine and 6.0 at twelve months.

Two thirds of the total drift happened in the first three months. That is useful, because it means a part that has been in storage for a quarter is close to its final value, and a part measured at dispatch is not.

The same compound with a two hour post-cure drifted 2.9 points over the year; with six hours it drifted 0.8. The post-cure does not prevent drift, it moves the reaction into the oven where it can be controlled.

That is the whole argument for a post-cure on a hardness-critical part. It costs oven time and it converts an uncontrolled room-temperature reaction into a controlled high-temperature one that finishes before the part is packed.

Post-cure temperature and the knee at 200 C

A post-cure is a second bake, usually in air, at a temperature above the service temperature and below the degradation temperature. For a peroxide-cured silicone the usual range is 180 to 220 C, and the time is two to eight hours depending on section.

On the measured compound, a post-cure at 150 C left a drift of 3.4 points after twelve months, 180 C left 1.6, 200 C left 0.8 and 220 C left 0.9. The curve has a knee at 200 C: below it the reaction is incomplete, above it the extra temperature buys nothing and starts to cost properties.

The cost above 200 C is degradation. A long bake at 220 C raises the yellowness index, reduces tear strength and can reversion-harden the surface. The 0.9 at 220 C against 0.8 at 200 is inside the measurement noise, and the damage is not.

The time matters as much as the temperature, and it scales with section. A 2 mm part reaches 200 C in twenty minutes; a 10 mm block takes over an hour. A six hour bake that is right for a gasket is not necessarily right for a thick block.

Table 1 — What each post-cure does to drift and to the as-cured hardness

What else moves with the hardness

Hardness is the visible symptom, but the same crosslinking reaction changes everything else. Tensile strength rises, elongation at break falls, compression set improves and tear strength usually falls. A part that has drifted 6 points is a different material, not the same material with a different number.

Compression set is the one that matters most in sealing. A part that was cured to 20 percent set as moulded can reach 12 percent after a year, which is better, and a part that was under-cured can go the other way if the network is still relaxing. Either way the seal performance at one year is not what was tested at dispatch.

Dimensions move too. Additional crosslinking shrinks the part slightly, typically 0.1 to 0.3 percent, which is inside most tolerances and outside a few. A precision gasket that was at the top of its tolerance band can drift out of it.

The practical answer is to qualify the part after the drift, not before. A 12 month room-temperature hold is too slow for a development programme, so the standard substitute is a 200 C bake for four hours on a sample, which brings the part close to its final state in an afternoon.

Table 2 — What a year of cure does to the rest of the datasheet

Measuring it so the number means something

A Shore A reading is only valid on a specimen thick enough to prevent the indenter from sensing the support. The standard requires at least 6 mm of material, and a stack of thin parts is not a substitute unless the layers are flat and in contact.

The reading also depends on when it is taken. A durometer read immediately gives a lower number than one read after the specified delay, because the material creeps under the indenter. The standard delay is one second for Shore A, and a reading taken at five seconds can be two points lower.

Temperature matters. A part at 10 C reads harder than the same part at 23 C, by up to 3 points on a soft compound. A warehouse inspection in winter and a lab inspection in summer will disagree, and both will be right.

The last variable is the operator. A durometer held at an angle, pressed too hard or read on a curved surface gives a number that is repeatable for that operator and wrong. A fixture that holds the gauge perpendicular is worth more than a better gauge.

Specifying a hardness that is still true later

A hardness specification should carry the value, the tolerance, the standard, the specimen thickness, the reading delay and the time after cure at which it is measured. Six items, and most drawings carry two.

The time reference is the one that changes behaviour. A specification that says 50 plus or minus 3 measured 24 hours after post-cure is checkable. One that says 50 plus or minus 3 measured at any time is not.

The second item worth writing is the post-cure itself. If the drawing specifies a 200 C four hour post-cure, the drift is bounded and the tolerance can be tight. If it specifies nothing, the moulder will do whatever the cycle allows and the drift is whatever it is.

The third is the acceptance test after ageing. A sample baked at 200 C for four hours simulates the drift, and a hardness check on that sample is a check on the whole cure, not just on the day.