Ok, a lesson for you all. BTW, totally plagiarized from my pond forum.
Water physics
The temperature of 39 Deg F is CRITICAL to understand. 39 deg F (4 deg C) is that temperature where water is at its most dense. And here is where the understanding of freezing water gets bizarre. But let's be really accurate for a second (there is madness to my method here)... at 39.2 deg F (4.0 deg C), PURE water is its most dense. This would be distilled water with all of the impurities, minerals, etc removed. Just hold this thought for awhile.
When any substance is at its most dense, we typically assume that these substances diminish in size as the molecules contract to the point where they can't move anymore and so everything is tighter and smaller. This is a natural physical law. The exception is water (and some other liquids). At 39 deg F, water is at its densest and so in theory, the volume is less as the molecules contract. This is why cooler water sinks.
So... if water is at its most dense, how does it freeze?? Wouldn't freezing constitute a more dense configuration of water?? The answer is no. Ice (which is a scientific terms and it comes different stages (I, II.....IV) is what typically referred to as solid water. But it is not solid. Unique to water (and some other liquids) is a process called crystallization. Crystallization begins when the water temperature starts to drop below 39 deg. During crystallization, the molecular shape of the normally very orderly water compound changes with the re-alignment of the hydrogen and oxygen molecules. This re-alignment of molecules does a couple of things:
First, it permits added oxygen into the structure (the more it crystallizes, the more O2 added) until at 32 deg F, the structure is saturated with O2 and the whole thing "freezes" solid. You now have a crystal, a snowflake, a hail stone, an ice cube, etc. These are all super-saturated water molecules and all float. And, this is why ice floats because of the saturated O2 levels.
Next, the crystallization process gives the ice its structure. While we can say that no two snowflakes are alike, in fact, they are all structurally similar. If we take away any movement of the molecules (water movement, for instance) we will get a glass-like, highly similar ice crystal structure. Ice is just a series of interconnected crystals. As our ponds freeze, note that they do so from the outer edge to the center and from the top down. The outer edge to center process is easy to explain as the cold ground provides the reduced temperatures to allow crystals to form. The crystals use as moisture, the water vapor from the pond's surface (from evaporation) as its "water" to build themselves. In this manner, the crystallization process moves away from its cold source (the ground) over the water vapor source.
Any continued movement of the water disrupts the crystallization process and the water won't "freeze."
As the crystals expand to form ice, there is an appreciable air gap between the 32 deg F crystals and the water surface. At the surface the water may be below the 39 Deg F level as it gains O2 and becomes saturated but all the while it is evaporating and providing more water vapor for the crystallization process. And this is how the ice thickens from top to bottom.
In an ice cube (or very shallow pond), the crystallization process is happening in 360 deg mode, outside-in, where in our ponds, just on the surface. This is why not quite "done" ice cubes are still liquid in the middle.
Ken is correct that evaporation probably does not occur on the inside of an ice cube but the crystallization process is still working. And if you look at an ice cube you will that it is more dense toward the center than the edges.. Less O2 in the center.
Oh and the crystallization process is also why water is unique in that it expands as it gets cold and crystallizes (to crack our pipes and such). The expansion of ice is caused by the re-alignment of the molecules during crystallization with the added saturation of O2.
