Understanding how a crystal formation occurs in volcanic activity is much like what happens in steel. Liquid magma dumped straight into the ocean instantly solidifies, locking in the natural state of the magma. If allowed to cool slowly over thousands of years the different elements begin to clump together forming crystals. Quartz, silver, gold. Cripple Creek Colorado is an excellent geological example of this. This same concept can be applied to heat treating. Instead of thousands of years as a time metric, you're dealing with seconds.
The normalizing process I find particularly interesting. If you do several normalization cycles each time lowering the temperature you take it to prior to air cooling you can produce a very fine micro grain structure.
Each time you normalize the crystalline structure has time to grow as it cools. The goal is to take your metal and make the entire piece uniform in its crystalline structure, with the smallest crystal growth possible.
For example, lets say you are working with CPM S110V (my favorite fillet knife material for my personal blades) stainless. This would be the correct heat treatment for the blade.
All steel manufacturers include data sheets for their products. It doesn't have to be a special alloy steel, you can find these data sheets for many common metals both stainless and non-stainless.
Crucible Industries makes S110V, the data sheet can be found here: http://www.crucible.com/PDFs/DataSheets2010/Datasheet CPM S110Vv12010.pdf
CPM S110V much be quenched at 2150F. That would be your minimum temperature for normalization cycles. This is the austenitization temperature. It should be non-magnetic, and the lowest temperature that loosens the bonds between the crystalline lattice in the material.
Set your soak temp for 2250F and soak for 1hr. This will homogenize - think of it as almost liquidizing the crystalline structures in place, they are still there roughly but are in a state where they will reform once cooled the bonds that hold them together are broken, but they are still there "thermal memory". You will never completely get rid of them without melting down the steel. The higher your temp is the longer it takes to cool. You want to start above your quench temperature, but without excessive heat. Go with the closest margin of error your heat treat furnace can achieve. Given the same rate of cooling in air a higher starting temperature will result in larger crystals. This is important to remember. Once you start your normalization cycles you want to make sure you never cycle at a higher temperature than your last cycle. If you do it will continue to improve uniformity, but not so much the grain size. By reducing the temperature a bit on every cycle you both create uniformity of crystalline structure AND force smaller crystals by reducing the time in which the crystals have to grow.
Set soak temp for 2225F and soak for 30mins, air cool. This slightly lower temperature will reduce the time it takes for the steel to cool below the austenitization temperature. The crystals that formed in the first cycle will be unable to reform to their original size and the micro-stresses of the crystal formation will force the bigger crystals (thermal memory here) from the previous cycle to break up. This is very good.
Set soak temp for 2200F and soak for 20mins, air cool. Again lowering the temperature a bit forces the thermal memory of the previous cycle to break up even smaller.
Set soak temp for 2175F and soak for 20mins, air cool.
At this point you have a very uniform crystalline structure throughout your steel and a very small grain structure. By stepping down your temperature each cycle you have forced less of the crystals to homogenize each time, creating an effect that fills in the gaps so to speak between the larger left over crystalline structures that formed during the first 2 or 3 cycles.
At this point you want to go for your quench at 2150F. - - Note this is below the normalization heats. If you go above your normalization heat temps you ruin the work you have done by the step down cycle. Make sure you never do this, never quench from a heat higher than your normalization heats. This is an extremely important concept - the longer your steel remains takes to cool the larger your grain structure will be. You want your steel conditioned perfectly by the normalization process so that when you go for a quench you can do it right at the austenitization temperature. If you skip the normalization structure you lock in whatever stresses and non-uniform crystal growth size is in the steel. This is why you see smiths on forged in fire quenching at cherry red end up with high performing hardness and those quenching at bright yellow end up with chipping and rolling. Both are hard, but only one is strong. This is also why quenching in water is very dangerous for steels with high rates of crystalline formation. Where pure iron may survive, high carbon alloy steel with very complex crystalline and even carbide micro structures just shatter like glass under that amount of stress.
After the quench the uniform/tiny crystal formations are set.
Set soak temp for 975F and soak for 2 hours. Make sure it completely cools to room temp each time. Repeat 2 more times.