|MadSci Network: Physics|
A look at the phase diagram for water shows that below about -20 Centigrade, all phases of water are solid. Therefore, the answer is, "Water can freeze without expanding."
The phase diagram shows what "state" or "phase" water will be in, at equilibrium, at a given temperature and pressure. We're all accustomed to operating at a constant pressure of 1 atmosphere, which corresponds in this diagram to a vertical line very near the left edge of the figure. The right edge of the figure corresponds to some enormous pressure. The article accompanying the phase diagram gives an overview of the various phases of ice.
So the freezing of a puddle corresponds to moving down the left edge of the diagram. At a temperature of 0 Centigrade, we pass from the region labelled "liquid water" into the region labelled "Ih", the name for the familiar form of ice. The other forms of ice (II, III, IV, V, et cetera) occur at much higher pressures, and have water molecules arranged differently from ice Ih.
The experiment you propose, namely sealing water in an infinitely rigid container and cooling it, differs from the familiar world of constant pressure (1 atmosphere) by imposing, instead, the requirement of constant density: 1 gram/milliliter. It would proceed like this: at 0 C and one atmosphere of pressure, water attempting to freeze would raise the pressure. This causes the "freezing point" to decline, corresponding to the downward slope of the line separating "liquid water" from "Ih". The system would probably follow in the neighborhood of that line until it reached the boundary of the Ice III region. Ice III is more dense than 1 gram/milliliter, so formation of Ice III would probably relieve pressure, permitting Ice Ih to form, so we would end up with a mixture of Ice III and Ice Ih. Given that the density of Ice Ih is 0.93 and that of Ice III 1.15, to get an average density of 1 would require a mixture of about 2/3 Ice Ih and 1/3 Ice III.
Return to the MadSci Network