With no air movement (which should never be the case in a grow room) these run 55-60C @ the above suggested drive current which is well below the 100C maximum and most luminares that use mid-power run at around 85C. The Philips Lumileds 3030 2D we use in the razor 250 is color binned at 85C! Heat is the drivers biggest enemy. Also most Mars panels aren't even 1/2 as efficient as these. Lots of heat for the light provided. No paste is necessary as the copper layer distributes the heat evenly across the board. As can be seen in the thermal images the heat really is super even.
@robincnn has tested these with thermocouples and the camera is spot on. I think you would exceed the maximum allowable current on the diodes before you would need a thermal interface material.
2 HLG 240H 2100/1050 A version will pull 600 watts turned to full bore. I like to run these at around 550 watts to give the driver a little head room. The output is something to be seen for sure. Like I said the guy that runs Nanolux and is a grower in CO bought the razor 250s and loved them. I sent him 4 QBs on a 320 H with no heatsinks on a flat aluminum plate he wanted more. I told him I could increase the power with the addition of the heatsinks. He got 4 of those and called me and ordered 6 more. He's eventually going to switch out all 60 DEs. Woot
Hell the Tj is only 2-3C higher than the case temp!
The face of the led only runs 1-2C higher than the board. Copper moves heat 2X better than pure aluminum. Now if you consider most heatsinks are alloys that move heat even less efficiently than pure aluminum you can begin to see how these do so well.