Why do ice cubes come out easier from top trays? This is my "hey, I've noticed that too!" question for the week. If you stack two plastic ice cube trays with water in them in a freezer, the resulting ice cubes in the top tray will usually come out cleanly and easily, while the those in the bottom tray will stick and crack. (Try it!) So... why?

Summarized answer, 2013-01-12
Ice expands as it freezes. So, if freezing starts at the top surface of the water, the liquid water below it allows the cube to edge gently upwards as it freezes, without much stress on it. This same upwards edging effect also keeps the ice from every gaining a solid hold on the entire surface of the tray. Result: The top surface of the ice cube stays flat, and the cube comes out of the tray easily without cracking.
Conversely, if freezing starts from below at the tray surface, the freezing water has the time and opportunity to lock itself tightly to all parts of that surface. But as the ice continues to grow inward, it tries to expand from all sides at once into the internal space. This creates a lot of stress in the ice, and causes the cube to form a peak at the top due to the internally expanding ice having no place else to go. So, since the ice has both had ample opportunity to freeze tightly to all of the tray surface, and because it was stressed internally as the freezing progressed, the ice cube becomes a disaster when you try to get it out. Much of it sticks to the tray, and the parts that do come out are often cracked.
How is all of this connected to whether a tray is stacked on top or bottom?
It's because stacked trays almost always impedes the cold air of the freezer from getting between the trays. That causes the stacked trays to behave largely like a single unit, with freezing occurring from the outside of that two-tray unit. The top tray therefore freezes mostly from the top down, producing ice cubes that have flat surfaces and come out easily. The bottom tray freezes mostly from the bottom up through the surface of the tray, and so ends up producing peaked ice cubes that stick to the tray and crack when you try to remove them.
(See the next addendum for some actual experiments you can do to try to confirm this.)
(Ron Maimon's was the first one to figure this out; see his accepted answer for details.)

Addendum 2012-12-04:
Here's an experimental prediction based on Ron Maimon's answer:


*

*If the ice cubes have flat tops, they will come out easily. Conversely:

*If the ice cubes have central peaks, they will crack badly when you remove them.
See the answer summary above for why.
Based on the answer, this experiment should make ice cubes come out easier: Tape an empty tray to the bottom of the tray with water in it. This will create an air-tight insulating air pocket below the filled tray, and so should result in flat-topped ice that comes out of the top tray easily.
You can also try for he opposite result by taping a tray upside down over the top of the tray with water in it. When it is done freezing, the ice cubes should be strongly peaked and should stick and crack when you try to remove them.
So, anyone interested: Get crackin'!
 A: Why the top tray seems to freeze differently and crack loose so easily is the question. When and if I ever find a plausible answer I will consider my personal compendium of everyday science to be completed. I have wondered about this behavior since the first time I ever cracked and refilled trays. My folks had the old aluminum or galvanized metal trays with a mechanical lever to break the cubes free but it only worked on the top tray in the stack. My theory is that the top tray is able to warm up slightly when the freezer cycles off. Trays 2 and beyond always have ice above and below. The tray and ice expand at a different rate when warmed slightly during defrost causing the ice to break free and ready to dump. I also expect automatic ice makers dump right after the defrost cycle and some even heat the metal rack slightly so the ice come out cleanly/easily.
A: I am losing confidence in my first answer, it was too impulsive. I'll leave it up, but I think it's wrong. I don't think sublimation is important over short time scales, like an overnight test. Sublimation should be at the top, and it is wishful thinking to believe that it will physically detach the ice from the sides and certainly not the bottom.
I think what is going on is the different mechanism of heat transfer in the two trays lead to a different method of crystallization. The bottom tray is releasing heat primarily through the walls to the fridge walls, so the freezing starts by attaching water solid to the sides, where the temperature first falls to freezing, and then the ice spreads inward. By contrast, the top tray is releasing heat either radiatively, evaporatively, or through heat diffusion to air, which will be primarily from the top of the water, leading to freezing from the top down. The top down freezing crystals might make a bad contact with the sides when the freezing reaches the sides, since the water is attaching to the center-out growing crystal, and has no reason to make a molecular scale contact with the wall.
To test this, you can open the fridge at intermediate freezing stages, and see if it is freezing outside in on the bottom, and inside out on the top. Further, you can place the lower tray on a thermally insulating material (put a few layers of foam/carpet below the lower tray, and see if this makes the bottom tray easy removal.
A: I'm not a scientific person at all, I was just searching for the answer to something that has always bothered me.  I don't understand sublimation, so I can't speak to that discussion.  Based on my minimal experiments, Ron Maimon's explanation regarding two method's of crystallization makes most sense.  I've only ever had one freezer, and I've only ever had plastic trays.  I have swapped the trays to see if one tray was the culprit and debunked that theory immediately.  I always wondered if the problem would exist with metal trays, now I know.  I think whatever is causing the sticking has to do with the way the ice freezes, not the conditions after it's frozen.  To avoid the sticky bottom tray, I freeze my two trays individually in separate areas of my freezer, then stack them until I'm ready to dump them.  The bottom tray never sticks even after weeks of being stacked, leading me to believe the freezing process is the issue.  I know I haven't provided a lot of scientific data, but I think this offers proof in the direction of Ron's second theory.
A: I think this is entirely due to the fact that the ice on the top tray sublimates and shrinks freely, evaporating some of itself, and detaching from the sides as it contracts. The bottom tray is covered, and does not sublimate as well, as there is less air flow and the air is more vapor-saturated.
You can test this by putting saran-wrap on the top tray and bottom tray. This should eliminate the difference entirely if this is the cause.
A: Not an answer but the results from a few more experiments...
In a freezer with a wire rack shelf with space made for 2 stacks of 2 rigid plastic trays.  Several inches of open space on all sides for the right stack, left surrounded closely (1/2") on sides by unused icemaker, filled ice drawer, frozen foods below. 
Bottom tray was more difficult to break ice no matter what permutation. (room temp trays, cold trays, trays 1 thru 4 rotated in all positions)
Bottom tray in left (enclosed) area tended to be the most difficult (to the point of having to run water over the back to release or shattering the ice upon release)
However, unstacked trays released easily from either left or right. 
Does that help with any conclusions?
A: I have tried to find answer for this problem but i cant find any rational answer for this question but i have found some interesting details and links about ice. May be these links help you to find an answer for this query
http://www.lsbu.ac.uk/water/ice1c.html
http://www.pnas.org/content/109/4/1041.full
http://www.sciencedaily.com/releases/2010/06/100616090128.htm
http://en.wikipedia.org/wiki/Ice_Ih
http://www.answerbag.com/q_view/41749
A: I believe this is caused by the slight displacement of the water bringing it closer to surfaces of the ice trays causing a more complete bonding between the ice and the tray, as there is less displacement in the upper tray as the weight from the first tray ever so slightly compresses the bottom tray. 
To test this theory pre-freeze the trays without any water then add cool water to the trays and quickly get them in the freezer.
A: I have discovered that if you place an already frozen tray on top of an unfrozen tray, the covered tray produces ice cubes that come out almost as cleanly as a top tray. You can actually alternate frozen and unfrozen trays and even the bottom most tray will freeze fairly cleanly.
