thermal conductivity is a measure of moving heat away, so aluminum isn’t faster. the perception you mentioned seems based on copper’s higher density and ability to store more heat…but despite that fact, the LED temperature will remain lower.
here’s my experience with copper vs aluminum
large brass pill (torchlite 2x26650), both copper and aluminum sinkpads (i built 3 of these lights)
lapped to the pill and then mounted to the heatsink with screws
copper sinkpad is very difficult to solder leads to…i have to use a high power iron and the whole pill gets HOT
aluminum, not nearly as much - in fact, pretty easy to solder leads.
no one is always right. experiment, find out what doesn’t work, try something else until it does. skip some experiments if others have done them, or measurements (in this case cost, thermal conductivity) indicate they’re not worthwhile.
Touch a heated aluminum (of substantial mass) surface and initially the temperature will feel the same as or slightly cooler than an equally heated chunk of copper. The copper will continue to feel hotter because the surface heat is quickly being replenished by the core’s heat. The aluminum will feel cooler sooner because the heat within the core doesn’t move to the surface as fast. In other words, it will have a good temperature gradient at it’s surface only but copper maintains a good gradient throughout it’s mass. This is probably why people think aluminum gives up it’s heat more readily.
The problem is the aluminum can’t be compared fairly since we’re talking about volume instead of weight. An aluminum heat sink should be 3 times larger or copper thinner but it’s not fair to make comparisons based on equal volume. By weight, aluminum could be good but by volume, copper easily wins.
Give it a try, and report back on how it works out. Arguing about what should or shouldn’t work without actual testing won’t prove anything other than that you know how to use Google.
I am receiving a few 20mm alu and copper Sinkpads from rollinstone, going to compare them with a xml2. I am sure there will hardly be a performance difference, but we'll see.... :-)