thermal conductivity of solder

As the most important part of heat dissapation is the little bit of solder between the led and the star, I searched for the thermal conductivity of solder and found that using solder paste is not that good of an idea as the only ones I can find are around 60% tin 40% lead (SnPb).
If you look in the table below that has a rating of 50 W/mK and while I do not fully understand the W/mK rating I do know that higher is better, the best in the table is SnAg at 78 W/mK, I repeat I don’t really understand the W/mK, however 78 is half as much again than 50 and as I understand it means 50% more heat transfer, got to be an improvement, I will be getting some SnAg, when I can find it, to do my reflows.
The table below from here,


To save you looking them up like I had to :wink:
Sn - tin
Pb - lead
Ag - silver
Cu - copper
Au - gold
Si - silicon
Ge - germanium
In - indium
Ni - nickel
Sb - antimony
Bi - bismuth

Am I misunderstanding the table or on the right track ??

Cheers David

http://www.ccfltd.co.uk/Trade-Support/Calculating-U-Values

Solder has a lower heat conductivity than aluminium or copper, but it is still very good compared to non-metals. Using a more conductive solder for soldering leds will help with heat transfer, but the thickness of the solder layer under the led is limited, perhaps 0.1mm, or less, (while the led board is up to 2mm, I think would not want the entire board made of solder). This makes the solder layer only a minor bottleneck. Such a thin layer can have a significant impact if it has a much lower conductivity than solder, like the dielectric layer in the common aluminium boards.

Be careful with those thermal conductivity lists unless you know the specific alloy you're working with and that alloy is listed in the chart. Alloys can have vastly different properties that what you'd assume from looking at the properties of their constituent parts.

I am aware that alloys are a pain in the bum, that is why I only gave examples from the table :slight_smile:
Quote from the site linked to in my first post

Even though the solder layer is thin, it is still a bottle neck due to being the smallest area with the most heat to transfer so any gain is welcome.

Cheers David

Oh sure, it's just lots of folks find those charts (of materials in general, not the one specifically for different solder alloys) and think they directly apply, when they might not. Believing something is true when it's actually not can be more harmful than just not knowing at all.

I agree with djozz. The joint thickness, or lack thereof makes it's thermal resistance very low. I looked for typical SMD joint thicknesses, but didn't find too much. One research paper tested electrical conductivity of different solder joint thicknesses. The thickness range that they used was .003-.060". I am guessing, but .003-.006" is probably a good value to use if you are diligent in "lapping" both surfaces. In that range, the "R Value" which many are familiar with for thermal resistance are very, very small.

Solder W / M * K Thickness R Value
50/50 Solder 50 0.003 8.65E-08
Sn96/Ag4 Solder 78 0.003 5.54E-08
50/50 Solder 50 0.006 1.73E-07
Sn96/Ag4 Solder 78 0.006 1.11E-07

Being careful is always prudent. Even knowing the alloy does not give exact thermal properties. There are deviations in each batch of solder. What is advertised as SN96/Ag4 will most likely be close to SN95.5/Ag3.5. Manufacturers will use the least amount of the more expensive material possible while still staying within spec.

With that said, I don't think any of us are doing the math to find the bare minimum components to mod a light. All of this is just for qualitative purposes..."This is better for heat transfer than that is." I think these numbers are useful for qualitative comparison.

Been there, done that, the tee shirt says “warning idiot above” :bigsmile:

While I did think of how thin the layer is, for some reason it didn’t sink in enough, well it has now, while it is a gain it is very small and could be negated by maybe damaging the led due to the extra heat needed to reflow, 221c against 183c.

Thanks for the input, once again this forum shines. <—- pun intended :smiley:

Cheers David

Moreover, not all copper is created equal and can have significant differences in thermal conductivity.

If you are concerned with the solder layer thermal conductivity, use solder paste and a 0.003” thick stencil. Reflowing a joint that started with wire solder or did not use a stencil practically guarantees a thicker joint.

+1

Silver-bearing copper has much better thermal and electrical conductivity. <<1%

There's a few people here that do reflows with your solder stencils. I wonder, do they apply enough solder paste to fill the central solder pads of Sinkpads and Noctigons, since those are a bit deeper than the electrical connections?

Those Noctigons are a heck of a lot deeper than .003". In fact they are like a shallow well. I toss the solder in there and wait to hear it hit the bottom!

The thermal conductivity should really be of no real concern when the thickness of the solder is taken into account.

The hole purpose of Noctigons/Sinkpads is that the electrical and thermal pad connections should be at the same level and make contact with the LED chip regardless of the solder paste thickness. Anyway, all of the stars that I have done with stencils had good contacts everywhere… much better than those that used wire solder.

The reality is that the thermal pad in Sinkpads and Noctigons is not level with the + and - pads, I use a bit thicker layer of solder paste there to fill the hole. Hence the question. You would only see that there's not enough solder there when you remove the led again from the board.

The thermal pads on the latest 3XP boards is on the same level as the electrical pads. Will add some photos later. As I know all older boards have the thermal pads lower than the electrical pads, the most visible is on the XP boards.

EDIT, the photos.

Anyway if the photos are not clear for anyone I can tell you that is clear with the eye when you compare the 2 boards that the 3XP is I described it above.

I checked all of mine with a razor blade and they have all been level. I think that there may have been a bad batch that slipped through.

BTW, OSHSTENCILS.COM has said that they are going to a 0.005” stencil as standard. I think that is too thick for many fine-pitch chips. They will also do 0.003” stencils.