We are looking for a material that will give us good thermal insulation and thus low temperatures at the location from which we're reading, but also has low thermal lag so we can quickly read the steady state temperature.
Hmm, I'm not sure if that'll work. To give more detail, in our design we have the insulation sandwiched between two sheet metal steel plates. The top plate gets up to 750F and at steady state we're measuring 400F at the bottom plate with good insulation, but we want to be able to measure those temps more quickly as the top surface gets up to temp. The low specific heat insulators will still have low thermal conductivity which will result in lots of thermal lag, right? Also, the low specific heat insulators will get hotter easier and increase the temp of the bottom at steady state (which we don't want.) I understand that this could be an impossible problem, but I'm just wondering if there is something I'm missing.
Modern aircraft engines are thermally separated from the airframes using aluminium thermal jackets.
There are two types in common use.
1 Bubble types that use a welded construction similar to plastic bubble wrap.
2 Expanded metal, where molten metal is forced to bubble and quickly frozen. similar to the chocolate bars known as aero in the UK.
Both types are expensive but readily available in aircraft markets.
Both would easily meet your specifications as above.
As an afterthought - if you want to keep costs down and your construction is simple, you could separate the steel sheets with the 'honeycomb matrix' used in carbon fibre composite construction. This uses thin walled aluminium formed into hexagonal cells and is very rigid. the heat transfer would be reduced massively as the surface area of the aluminium in contact with the steel is tiny.
Essentially it's a heater. The top plate is the heating element and the insulation is there between the bottom plate to create a temperature difference, so in the case of some sort of software failure if the heater gets stuck on a thermostat on the bottom plate (in line with the current) will cut off power. (We need a temperature difference since disc bi-metal thermal switches only go up to 550F)
The problem is that at steady state (heater temp 750F, bottom plate 450F) we do not want the thermostat to cut off power, but when the heater runs off towards its max temp of 1500F it takes too long for the thermostat/bottom plate to get to 475-500F and cut it off.
Our dilemma is that if we decrease the insulation thickness, the steady state temp is too high, but our response is better. If we increase insulation thickness to keep our bottom plate
With respect, I would suggest that you may easily change your measurement / protection regime to advantage.
Use an IR diode mounted in a tube and pointed at the hot plate but remote from it to monitor the heat rise. Use a comparator (which could easily be a simple transistor but an integrated op-amp type is preferred) to switch off the power in event of over-heating. Your electronics technicians will find this a simple challenge.
This will resolve the insulation difficulty and make the system much more responsive.
I regularly 'knock up' similar devices at virtually zero cost from my stash of electronic surplus.
Silica sand has similar qualities to those you ask in your edit, but may not fall in your temperature range - you could end up with glass at 1500F.
(mind, the emissivity might change over time: dust and progressive oxidation of the plate?)
Zohair: thermocouples exist that can tolerate 800deg C: even the humble K type.
Why not just mount a TC on the heater?
(there may be a host of reasons: I admit)
I cannot think of any naterials that conduct heat less well at high temperatures exist - I think that DeBye might have something to say about that - but one could engineer a composite, but it would be messy and single-shot.
I confess that I've not used radiometric thermography (except for those cheap-as-chips point and shoot thermo-guessing devices), and so feel a bit more comfortable with wired TCs.