Posted by Dave Arnold
Last week I participated in a seminar at Tales of the Cocktail entitled, “The Science of Shaking.” The panel was put together by moderator Eben Klemm (head of bar programs for the B. R. Guest restaurant empire, well-known innovative cocktail guy, and former biologist) with Alex Day (famed bartender from Death & Company and Franklin Bar) and myself as panelists.
Before I post the results of the seminar, I wanted to do a short post on some basic cocktail science. I am not a scientist, so please feel free to correct me.
Question: If ice melts at 0° Celsius, and I start with 0° ice, how is it possible that shaken drinks can get down to minus 7° Celsius?
Answer: First of all, yes, shaken drinks can get down to minus 7° just by shaking with ice. Â
Well, we all know that alcohol freezes at a much lower temperature than water, but that still doesn’t answer the question: how can ice make something colder than 0°?
This question can be approached several ways (colligative properties, vapor pressure, etc.), but I think the most fundamental way is to see the problem as a balance of changes in enthalpy and entropy. In other words, molecules are lazy but they also like to be free.
In any reaction, a change in enthalpy is a measure of the heat absorbed or released during that reaction (assuming a constant pressure, yadda yadda). In general, all things being equal, things want to give off heat. By giving off heat they have a lower internal energy. Things want to go to a lower energy state. Things are lazy. It takes energy to break ice molecules free of the crystal lattice, so there is less energy stored in an ice cube than in water at the same temperature and pressure (cause I had to dump in heat to make it into liquid water).  This heat that has to be added to ice to make water is called the enthalpy of fusion (or the heat of fusion). The heat of fusion of water is about 80 calories per gram, meaning that the heat required to melt one gram of ice is sufficient to heat one gram of water all the way from 0° to 80° C! Remember: melting ice requires heat (the heat comes from your drink so your drink gets colder). Making ice gives off heat, so enthalpy favors water turning to ice.
Entropyis a different story. Entropy is often described as a measure of disorder. Greater entropy equals greater disorder. A better way to think about it is as a measure of how many different states something can be in (scientists call these microstates). Things want to increase in entropy. Things want to maximize the number of available microstates and then commence to occupy those microstates in a random way. Things want to be free. At any given temperature, there are more positions, speeds, etc.—microstates—in a liquid than in a solid. Water molecules, for instance, are free to spin around and find new neighbors, etc.  Ice molecules are locked in a crystal. Being a solid is more constrained than being a liquid, so entropy favors ice melting into water.
So who wins, enthalpy or entropy? It depends on temperature.  As the temperature goes up, entropy tends to dominate and ice melts. As the temperature goes down, the heat of fusion tends to dominate and water freezes. At high temperatures, entropy wins because there are more microstates available to the molecules in the liquid water than at lower temperatures (cause they are moving around more). Thus, there is more of an entropy win by turning to a liquid than at lower temperatures. The freezing point of water (0° C) is the point at which the entropy gain from ice melting to water is exactly balanced by the amount of heat given off by water freezing into ice. Water molecules are constantly freezing into ice and melting into water at the same rate—they are in equilibrium. If you lower the temperature, the entropy gain becomes puny and water wants to freeze. If you raise the temperature, the entropy win outstrips the enthalpy part and the ice wants to melt. Got it?
What happens when you add alcohol? For the purposes of this discussion, let’s assume that the ice crystals remain pure water (that’s pretty true). Ok. We are at 0° C, we have ice and water at equilibrium, and we add alcohol into the liquid water. The heat given off by water molecules freezing into ice is the same as it was before, because the ice hasn’t changed; but the entropy win of ice melting into the water/alcohol mix has gone up. Ice melting into the water/alcohol mix has more microstates available, more ways of being arranged than were available in the pure water, because there are more different ways of arranging x water molecules and y alcohol molecules than there are of arranging x+y water molecules. So what happens? The entropy gain of melting wins and the ice starts to melt. Melting ice absorbs heat. The only place the heat can come from is from the ice and water/alcohol mixture (oh yeah, I forgot to mention I am assuming a closed system), so the whole shebang cools down below 0°.
What’s really cool is that as the ice melts, the solution gets more diluted, which reduces the magnitude of the entropy win at the same time as the temperature goes down. This happens until a new equilibrium is reached, when the entropy and enthalpy become balanced again—that is the new freezing point of your drink. The theory is pretty straightforward but figuring out the final temperature and dilution of a drink from first principles is well, well beyond my ability. I encourage you to try and tell us how.  One note before you try, though: I’d say it’s probably a lot harder than you think. Even assuming a closed system (false), and no energy input from shaking (false), and no problems with surface area and speed of agitation and quantity of ice vs mixture (somewhat false), it’s a hard problem to solve exactly.
Next installation: Tales of the Cocktail Seminar: The Science of Shaking, does type of ice matter?
Are you positive that the ice is at o°C? Shoving a probe thermometer into a shaker of ice is probably measuring the liquid phase, rather than the solid. An IR thermometer measurement says the ice from my freezer is at -20°C.
Yeah, our ice is at 0. We don’t use freezers for ice. Our ice machines store their ice in insulated, but non-refrigerated ice bins. To be sure, when we were doing shaking experiments, we equilibrated all our ice in a holding cabinet set at 0 degrees for a couple of hours. Bar ice is usually wet ice and wet ice is 0. Ice out of the freezer will, of course provide some cooling power before it starts to melt, but the cooling power you get from those 20 degrees of supercooling is small compared to the heat of fusion of ice melting. The heat capacity of supercooled ice is only half that of water (0.5 calories/gram).
dave… i usually have no problems hanging with your in-depth knowledge, this has truly dumbfounded me… i mean i get it, but my brain hurts this time. you and “mr. i” still thinking about ice cubes?
We’re always thinking about ice cubes!
Interesting. This would seem to suggest that having more ice (greater thermal mass) is more important than having colder ice.
Excellent post. The new mixture of ice/water/alcohol will reach state of equilibrium at a temperature lower than the freezing point of ice/water mixture i.e. 0*C if you have enough ice in your shaker that is. Dave may your gospel reach all bartenders calling them selves professional. When you shake keep your shakers full of ice!!!
P.S. Can’t wait your next post on the topic.Good night
tnx so much guys for sharing
Max
http://maxologist.blogspot.com
Credit you championing details. It helped me in my task
Another, no-entropy-mentioned, explanation:
Hi energy water molecules break off the ice randomly and can’t re-attach because alcohol is in the solution. This lowers the ice temperature until the water can begin freezing onto the ice again. (Just as wet clothing in air gets colder as water evaporates.)
Great post, how long does the liquid stay below zero for once its in the glass?
Good question. It depends on the ambient temperature, the mass of the cocktail, amount of ice present, etc. I haven’t done the measurements in a while, but the answer is … a pretty long time if ice is present, not long if ice isn’t present.
But on balance this article is pretty good. Congrats for actually doing the experiment. For accuracy you have Judith Miller beat, hands down.
You probably don’t want someone with a PhD in physical chemistry to intrude here but:
“Things want to go to a lower energy state. Things are lazy.”
Sorry, things are not lazy, nor are they industrious. Things don’t care. The reason that a system will go to a state of lower energy is that when the energy from the system is dissipated to the surroundings, there is an entropy gain of the surroundings. Changes in entropy are the only driving force for anything.
“So who wins, enthalpy or entropy?”
It is always entropy, entropy is the only thing that matters, for driving a process. At low temperature the entropy of the surroundings wins. Ice freezes at low T because the entropy gain from dissipating heat (the enthalpy of freezing) to the surroundings beats the entropy loss in the system (from ordering the water molecules in the crystal). At high T the direction switches (ice melts) because the entropy gain from dissipating heat is less than at low T. It is all entropy, all the time. You just have to do the accounting.
I am following this idea, but there is one thing I do not understand. If melting ice absorbs heat (read: gets hotter), and ice melts at 0 degrees C, and your ice is also 0 degrees C, then how, upon absorbing heat, does it remain ice?
You mention that alcohol’s freezing point does not answer this question. What results are achieved using water instead of alcohol? I believe it is possible to cool water below 0, but not by much without adding anything. I understand the point of your post, but remain unconvinced that the alcohol not the primary reason for achieving these temperatures. I would like to see what temperature is achieved using water only in the shaker. I don’t think it would be quite so low, but am no expert.
Howdy Chris,
The alcohol is the reason the temperature goes lower than 0C. Water and ice by themselves will never go below 0 C (even with external chilling, because it would be very difficult to supercool pure water with ice crystals already present). The reason the cocktail system goes below 0C is that it is energetically favored for ice to melt into a water/ethanol mixture even at 0C. When that ice melts into the ethanol mixture, heat is required (it takes energy in the form of heat to melt the ice). That heat doesn’t make the ice hotter, it just changes the water from solid ice to liquid water. That required heat (the heat of fusion of water) needs to come from somewhere, and it isn’t from the environment (heat leaking from the environment into a cocktail shaker isn’t fast enough). The heat comes from the remaining ice and the drink itself. If you remove heat from something without changing its phase, it gets colder –so the drink gets colder.
I don’t understand this “it would be very difficult to supercool pure water with ice crystals already present” — in the Modernist Cuisine, they talk about supercooled water (which actually remains liquid below freezing temperature because it is pure). I haven’t had the time to conduct the supercool water experiment, but relying on the Modernist Cuisine, your water plus ice will just turn to ice.
What I am saying is you can’t supercool water with ice crystals present in the water because the water will freeze onto the crystals instead of supercooling. When you supercool water and introduce some sort of nucleation site, ice crystals immediately begin to form and the temperature of the water rises to the freezing point. Further supercooling will be impossible.
That is correct — but I thought the question was whether shaking water with ice will turn the resulting stuff into something colder than ice. We are not talking about supercooling water, which I thought was something different entirely (but I could be mistaken). In any case, I actually asked a professor in thermodynamics this question — and will try to post whatever he says — I wasted too much time trying to read up on the laws of thermodynamics on my own. I did learn a lot, but nothing that would help me actually answer this particular question. It may have to do with the way alcohol evaporates.
If you take pure ice at 0 C and any ethanol solution and shake them, the temperature of the whole system (including the ice) will drop below 0. Shaking pure water in pure ice is different, the system will stay at 0 C. The ethanol in the liquid changes the balance of the Gibbs equation making the melting of ice favorable at 0 C. The temperature of everything will drop until the heat required to melt the ice is matched by the entropy win of melting into an ethanol solution.
While I agree that it is the presence of alcohol that seems to be the explanation – I really could not understand your explanation on August 10. I do wonder if you were to just shake alcohol on its own, whether its temperature will lower. I think I’ll try this at home if my thermometer allows such a thing.
You need to shake an ethanol solution with ice –not on its own. The same phenomenon is at work here as when you add salt to ice to make ice cream. The temperature of the whole system drops below it’s initial temperature. It’s all about entropy.