Nichia 1/2 watt High CRI LED bonded directly to sapphire window:

Driving the LED at 80 mA, I measured the temperature of the LED and window with a FLIR. Although in the image above, you can see a captured picture on the FLIR showing the window at 51.6 C, after leaving the LED powered up for a longer period until presumably steady state was reached, I then measured 54.5 C on the face of the window and 56.4 C on the back of the LED package. With the window removed, the temperature of the LED rose to 65 C. When installed in the head of the titanium AAA host, below, the head of the light does get warm to the touch indicating that further heat is spread to the light itself. Once the window and LED were installed in the head, I added further Norland 61 UV curing adhesive behind the LED essentially potting it in the head and adding thermal mass and transfer capability.

Some comparisons with the Sapphire hosting the Nichia 310DS LED:

First up, a beam comparison:

The comparative images below showing the color checker were taken with the camera set on daylight white balance:

The tint seen on the "white wall" behind the ColorChecker is in keeping with what the naked eye sees but a bit exaggerated

I introduced both of the light samples to my integrating sphere and the comparative spectrum with some data overlay below is the result. I can not vouch for the accuracy but suspect the data is within bounds.

The sample 036 LED used here was from a bin of 4500 CCT which is in agreement with the measurements taken. The minimum claimed CRIa on these is 85 so that is also reasonable in comparison with the resultant measurements. The 3 mm 310 DS is being over driven in the Sapphire whereas the 036 LED is driven below spec. The current drive level could explain some of the deviation from expected results.

Below is a clipping from the Nichia spec sheet on the 036:

Since the 036 is a single die LED, I decided to see how it would fare in the same titanium host and driver but with a reflector in front of it.