Thermal experiments on aerogel and micro-aerogel

[Zarquan]

As many people already know, normal aerogel (solid tiles with 0.1 g mass), as described here, is no longer perfectly insulating as of the March 2026 bug-fix update.  It still highly insulating, but it will now transfer heat and melt.

This made me sad as I watched my abyssalite flaker's walls vaporize and flood my base with high pressure 3500 C hydrogen.  To avenge my fallen build, I went on a quest to find the holy grail of modern aerogel: The 1 microgram (or 0.000001 g) aerogel.  I surpassed my goal and created 0.116 g mcg aerogel, as well as a process for making it in survival in a repeatable fashion (i.e. a machine that will output appropriate liquid packets and send them out of a pipe).  I call this aerogel "micro-aerogel".

I intend to do a how-to guide for making your own micro-aerogel and other micro-packets in the future, but for now, I want to verify a claim I asserted:  The 0.116 micro-aerogel is actually better than the 0.1 g (or 100000 mcg) aerogel.

TL;DR:  Micro-aerogel is equivalent with or better at everything when compared to normal 0.1 g aerogel.  If you put a cool radiant pipe behind it, it can be used to perfectly insulate even the hottest things (9999 K) with the thermally worst aerogel (refined carbon) and up to 5000 K with simple glass aerogel (for one adjacent tile).

First discovery, insulated insulite is NOT perfectly insulating with aerogel or micro-aerogel!  20 C insulated insulite tiles can and will decrease the temperature of the aerogel and, thus, can make aerogel less thermally conductive (two surrounding insulated insulite tiles reduced the thermal transfer from normal aerogel from 11.0904 to 3.531 C, which is a significant decrease) and can make micro-aerogel not melt when touching things that would otherwise make it melt.  This is why I am not using it except for in tests that call for it.  Neutronium is actually perfectly insulating and does not interact with aerogel.

All temperature numbers were determined using the Sample tool from Debug mode.

First test:  Basic conduction between 0.1 g aerogel and 0.116 mcg aerogel:

Spoiler

I created these setups in identical starting conditions, with 300 K thermium on one side and 2500 K thermium on the other, all 10 kg.  For thoroughness, I am checking in all 4 directions.  These are the results (with the ones marked in green being micro-aerogel and the ones in red being normal aerogel), with the cold thermium being under the blue diamond:

Spoiler

After one cycle, I measured the cold 10 kg thermium tiles (blue diamond) on each trial.  Orientation did not matter. 

1. The thermium on the micro-aerogel trial was unchanged at 300 K according to the sampling tool.  Similarly, the hot tile was still 2500 K, so no heat moved
2. The thermium on the normal aerogel trial was 311.0904 K and the hot tile 2465.177 K, meaning a small amount heat was lost (220.188 DTU).  And it was not absorbed by the glass, as the glass would have to have been heated by over 2600 C, which did not happen.  But this is a small amount of heat lost.

Conclusion:  At these temperatures, we don't see thermal transfer through micro-aerogel, but we most certainly do with normal aerogel.

Second test, figure out how mass affects heat transfer at low masses:

Spoiler

To get some data points, I recreated the experiment, but only with heat going left to right and with aerogel masses from 0.1 g to 0.02 g, at increments of 0.02 g.  For more transfer, I increased the initial thermium temperature to 2500 K and reran it for one cycle.  I graphed my results, and got a nice line!  As I predicted, at this mass, the thermal transfer rate is linear.

Spoiler

I believe the mechanism for why the thermal transfer lowers with mass at this point is saturation.  Imagine that heat transfer happens in two steps:  First heat is transferred between the two tiles on the left and right.  Since the aerogel of any variety has such a low mass, it's thermal conductivity allows it to be heated right up to the point of that tile (in the worst case, not actually true in ONI, but I don't know the exact mechanism).  Then it would do the same with the other exchange.  What this does is cap the heat transfer through the aerogel to it's thermal mass.

 Determine relationship between normal aerogel's actual contact thermal conductivity and specific heat capacity.

Spoiler

I would have done this with micro-aerogel, but I've only seen it conduct heat in one case which shows up later.  I created a setup like the previous one.  From top to bottom, I used glass, tungsten, thermium, and refined carbon.  The left thermium started as 10 kg 2500 K thermium and the right started as 300 K 10 kg thermium. 

I also got another pretty graph that indicates my conclusion about specific heat capacity being the actual measure of thermal conductivity at these masses to be correct.

Now, I will experiment on resistance to melting based on my experience with my abyssalite flaker.

Spoiler

Below are four images, one when the 3500 K 100 kg hydrogen, with the green-surrounded sections containing the hydrogen with micro-aerogel and the red-surrounded one having 0.1 g aerogel.  One set has insulite tiles surrounding them and the other does not.  The first two images are taken three ticks apart, the third a few seconds later, and the fourth was taken to confirm a hunch.  I will walk through what I saw tick by tick.

First tick after start:  All the aerogel heats up.  The non-surrounded ones and the surrounded normal aerogel one heat up to exactly 1100 K, whereas the surrounded one heats up to 1094.862 K.  I believe that this smaller number is only related to the larger one in that it is smaller, as (293.15+293.15+293.15+3500) / 4 = 1094.8625, and the insulite tiles are all the default 20 C, or 293.15 K.

Second tick after start:  The non-surrounded aerogel heats up to 1700 K, whereas the surrounded normal aerogel is at 1699.996 K, probably due to conduction with the insulated insulite.  The surrounded micro-aerogel reads 1094.863 K now, but I bet that's just because it's bouncing around some floating point near 1094.8625 K and randomly rounding up or down.

Third tick after start (second picture):  The glass around the non-surrounded aerogel and the surrounded normal aerogel melts.  On the surrounded aerogel, the hydrogen is now in the top tile, probably due to the 0.1 g molten glass falling (swapping with the hydrogen) before being deleted.  In the non-surrounded trials, 100 kg is on the left and 800 kg is above and 100 kg above and to the right of the aerogel.  (I'm ignoring the lower one, as it interfered.)  I suspect this is because things are being calculated left to right and the hydrogen moves, then gets swapped, then spreads in the other direction.  I see the same behavior when four beads of water are in place of the aerogel.

A few seconds later, the hydrogen fills the room and the surrounded micro-aerogel is still intact. 

I then remove one of the insulated insulation tiles around the micro-aerogel and see a predicted increase in temperature to 1362.1 K, which is precisely (293.15+293.15+3500)/3.

Spoiler

 

Experiment with debris and buildings (the one with the breakthrough in the TL;DR):

Spoiler

Careful with bridges, temp shift plates, and hot debris, as they will both heat up the micro-aerogel and melt it.  In the pictures below, I put hot tungsten and an automation wire bridge in contact with the high pressure hydrogen.  Both the tungsten and hydrogen are at 2500 K.  Note that while the bridge successfully melted the micro-aerogel, the debris heated the micro-aerogel to 2500 K, well above the melting point of glass.  I personally thought the devs fixed this, but it will melt when you load the game, as seen in the third picture.

I bet what's happening is that melting is determined between the tile thermal conduction and debris thermal conduction.  The tiles reduce the temp, melting gives it a "no", and then the debris heats it up to the display temperature.  I tested it and, without the neighboring tiles, the tile melts.

Spoiler

  

However, there is an upshot.  While hot debris and buildings cause a disproportionate heating effect, cool buildings, debris, and pipes have a disproportionately strong cooling effect in comparison to solid tiles.  I found that radiant liquid pipes and radiant gas pipes have a stronger effect than normal pipes.

In the picture below, I have four glass aerogel tiles successfully containing 100 kg of 4000 K hydrogen.  The left-most one is normal aerogel and the others are microaerogel.  They all have an aluminum radiant liquid pipe behind them. 

After one cycle, the left-most aerogel, which was 0.1 g, and its associated pipe was heated up to 318.7069 K.  The micro-aerogel was still at exactly 300 K according to the sample tool. 

Spoiler

For reasons I don't quite understand, this fails somewhere between 5000 K and 6000 K, but I don't think that matters too much for normal gameplay, as those temps are barely achievable in survival and have no practical uses. 

EDIT:  Actually, I believe that this is because at those temperatures, it hits the melting point of glass mid-tick and melts it before the pipe temperature change gets to register.  This means that a more robust aerogel material, like steel or tungsten, can contain such things.

Conduction through temp shift plates:

Spoiler

Since aerogel interacts differently with buildings, I decided a repeat of my basic conduction experiment, except with temp shift plates as middle-men.

Spoiler

In the top trial, there is a tungsten radiant liquid pipe behind the aerogel preheated to the temperature it eventually reaches after a long time, 1500 K.  On the bottom, nothing.  The hot thermium was set to 10000 kg at 1500 K and the cool thermium was 10 kg at 300 K.  The temp shift plates are diamond and are pre-heated to the temperature of the adjacent thermium. The

After 1 cycle, the micro-aerogel trials is still locked in at 300 K.  The normal aeorgel trial's thermium is at 303.15 K

Pushing it to the limit:

Spoiler

It really doesn't make sense for this to ACTUALLY be perfectly insulating.  So, I pushed it to the limit:  I build the image below with 1 gram of thermium, a micro-aerogel made of refined carbon with no pipe, and a 100 kg block of 4500 K of refined carbon.  And I saw temperature change in the thermium!  It is a whopping 300.5493 K!  I did a similar experiment with tungsten, and it was still locked at 300 K, whereas my specific heat capacity experiment predicts a change of about 0.0418 K, but I bet the increments are lost in rounding.

I replaced the thermium with lead (for its low specific heat capacity) on the bottom one, and I detected heat transfer.  But the delta T is so high and the amount of heat transferred is so low (barely perceptible), that I can't imagine this mattering in the slightest.

I will do an over-night trial to see if 9999 K hydrogen is capable of heating a refined carbon aerogel (worst case) with a tungsten pipe (tungsten having the lowest specific heat without potentially melting) in it at all.  I will do the same with steel (at 9000 K, as 9999 K causes the walls to fail).

EDIT: Experiment: The absolute worst case, holding back the worst the universe has to offer:

Spoiler

For this experiment, I created the setups in the images below and ran it overnight on full speed (at least 166 cycles).  The micro-aerogels containing setups are as follows (top to bottom):  Refined carbon (9999 K hydrogen), tungsten (9999 K hydrogen), and steel (9000 K hydrogen).  Each aerogel contains a tungsten radiant pipe for its low thermal capacity and high melting point.  I attached a 1 g lead tile to it, as it is pretty sensitive to these changes.  

I saw no discernable energy transfer in that time.  All the micro-aerogel and all the lead tiles are exactly the same temperature as they were at the start (293.15 K) Therefore, I officially declare micro-aerogel + radiant liquid pipe perfectly insulating.  I would declare it safe, although weird things happen, so I would still contain your containment system with insulated tiles to protect your base from weirdness, just in case something weird happens with a load or something and it explodes.

There are three other tests I would like to do:  micro-aerogel surrounded by hot hydrogen, and with a temp-shift plate next to it, and how small you actually need to get your micro-aerogel to get this effect (keeping in mind that any micro-aerogel larger than 0.116 mcg would be multiple times more conductive than the optimal, as 0.166 mcg is the quantum of the meter valve used in my system).

Initial tests show that surrounding the aerogel decreases the maximum protected temperature from instantly melting significantly, but does not seem to effect the perfectly insulating properties (but I can't test to as high of a temp as I did before).  Glass aerogel with a tungsten radiant pipe in it failed between 3000 K and 3500 K when surrounded by two of these tiles.  That means you would want to avoid inside corners in your aerogel-contained builds that get that extremely hot, or perhaps make them out of a material with a higher melting point like steel, which failed in the exact same conditions between 5000 K and 5500 K. 

I have a theory and will probably do some math later and create a table of materials and how hot they can get before melting for a certain number of tiles surrounding them.  That's a little complicated, though, as if there is an atmosphere on the other side of the wall, it should impact the results significantly.

Edited Monday at 07:30 PM by Zarquan

[Gurgel]

And the legacy of IEEE754 abuses continues 

[Zarquan]

New observation:  The temperature increase on micro-aerogel from a hot surface is constant for a constant hot surface.  Steel and glass both increased from 293.15 K to precisely 969.8625 K in one tick when in contact with one 3000 K hydrogen tile.  That makes my next experiment easier, as I don't have to measure all the different materials in that way.  Increasing to 4000 K increased the first-tick temperature to 1219.863 K on both tiles. 

I suspect that when this first-tick temp goes above melting, that is when the micro-aerogel with a radiant pipe in it melts.  Since that is material independent, I think I can create a calculator that can take the temperature of the pipe and the temperature of the gas and calculate the first-tick temperature.  But that will have to wait...

Edited Monday at 07:11 PM by Zarquan

[Zarquan]

On 4/13/2026 at 4:43 AM, Gurgel said:

And the legacy of IEEE754 abuses continues 

Yup!  I was thinking that they might give and take so little heat energy that it would be lost in the rounding.

So, I figured out the equation for the first-tick temp of a micro-aerogel tile.  I believe what it does it is calculates the average of all the tiles around it and sets that as its temperature to that value.  If there is a vacuum in a surrounding tile, it uses its internal temperature in its place.  So, if you have a 1 tile contact point, the aerogel starts at 300 K and the other tile starts at 2000 K, then it will take 300 K as the three surrounding tiles and set itself to (2000+300*3)/4 = 725 K.  Then, it checks to see if it should melt, after which it sets the temperature back to the radiant pipe.  The equation does not appear to use this temporary temp to determine whether it should melt, but instead uses either the initial temp, which will be the temperature of the pipe.

My bet is that's part of how all thermal conductivity is calculated, except it has caps and limits to how much temperature change happens that micro-aerogel never hits.

Therefore a vertical wall is pretty safe, but inside corners are especially vulnerable.

Edited Tuesday at 10:58 PM by Zarquan