Refractory material

Diamond is metastable and can turn into graphite at too high temperatures.
To do consistent design for external limiting factors other diamondoid materials like the carbides of the titanium vanadium and chromium group (interstitial carbides) can be used for high temperature applications since they are refractory. (complete sets of DMEs are needed). Stability of free or mutual or environmentally contacting passivated surfaces (that are possibly strained) will reduce the allowed temperatures well below the bulk material melting points though. Interstitial diffusion may too be a limiting factor.

4th period:

• TiC (3,160 °C; 5,720 °F; 3,430 K; abundant elements)
• VC (2810 °C; 9-9.5 Mohs)
• Cr₃C₂; Cr₇C₃; Cr₂₃C₆ (1,895 °C; 3,443 °F; 2,168 K; extremely hard; very corrosion resistant)

5th period:

• ZrC (3532 °C; extremely hard; highly corrosion resistant; very metallic)
• Nb₂C (3490 °C; extremely hard; highly corrosion resistant)
• Mo₂C (2692 °C) [1]; MoC; Mo₃C₂ [2]

6th period:

• HfC (3900 °C; very refractory; low oxidation resistance)
• TaC_X (3880 °C (TaC) 3327 °C (TaC_0.5); extremely hard; metallic conductivity)
• WC (2,870 °C; 5,200 °F; 3,140 K; ~9 on Mohs scale)

mixed:

• Ta₄HfC₅ (record holder: 4,215 °C; 7,619 °F; 4,488 K)

Note: Many elements here are neither abundant nor prime targets for mechanosynthesis.

[Todo:]

• add notes on SiC
• add notes on recycling and disassembly
• add notes on self repair

Nanscale limitations

That a material does not melt does not mean that it shows no surface diffusion. For really high temperature applications minimal sized DMEs will thus likely not work. Bigger scale (interlocking) refractory tiles will still be usable though. But they'll need regular replacement before they fuse together. Disposal of them might proove diffecult.