This is not an easy question to answer.
"I think the memory property is gained by getting the right percentage of closed cell honeycomb like structure thoughout at the right density."
I do not believe this statement to be true. On a pillow or bed the closed cells would be crushed open, then you would not have memory, according to that statement.
I have made memory foam. It is based on PU foam chemistry. You need a special polyol. I used an ether based polyol, trifunctional. I believe the MW was around 3300. The polyol was of high ethylene oxide content. Standard carpet cushion polyol is 3000-4000 MW, mainly propylene oxide. PO is cheaper than EO.
Polyol blend = 70-80% EO polyol and 20-30% styrene/acrlynitrile graft polyol
Then you will need some isocyanate. MDI/TDI blend. Large amounts of EO polyol and TDI can cause tightness, shrinkage will occur. TDI is not so nice to breathe.
Water will control density. Low density = more water. 3.0% water should give a nice foam
A surfactant, 0.1%. This helps with bubble stability.
Amine catalyst, I used ZF-123 from Huntsman. This initiates the blow reaction, (water + isocyanate). Loading level will be determined by polyol temp and MDI/TDI ratio. MDI blows faster than TDI, in my experience.
Tin catylst, this initiated gel reaction (polyol + isocyanate)
Stannous Octoate, loading will be determined on reaction speed. I would start with 0.15%
Calculate polyol number for sum of water and polyol, add isocyanate mixture to 80% of complete reaction.
My understanding is hydrogen bonding of the high EO polyol gives the memory like foam.