The weird science of ice and how to make “Premium Ice” at home

beautiful, clear ice

Update 2/27/13: Like this post? Check out the 250-page, 65-recipe book that came from this research.

I was inspired to post it here by a reader named Andrew, who sent us an e-mail recently asking how to make fancy crystal-clear ice cubes at home. He’d tried everything from using distilled water to sonication, but never got the perfect cubes he envisioned.

Andrew’s not the only one – a whole bunch of folks have tried to make clear ice cubes at home that can match up to the “premium” ones available through companies like Kold-Draft, which are often used in high-end bars and restaurants. I’ve tried all the methods over the course of the last year and figured out what really works. No esoteric equipment required.

What are the benefits of clear ice?

Using clear ice for mixed drinks is not just about aesthetics.

Reason #1: Purity

If ice is clear, then by definition, it contains minimal impurities. Impurities, like chemicals used to treat tap water, minerals, and particles of food picked up from the freezer, become more soluble in water as it gets colder. But anyone who has ever marveled at a snowflake knows that water forms a crystal structure when frozen. Water’s crystal lattice structure forms most efficiently without impurities, so impurities are slowly forced out as water freezes.

Experiment: Try dissolving a small pinch of salt into a cup of water.  Taste the mixture, remember how salty it tastes, and then freeze it for a few hours. Open the freezer a few hours later, and you should find that a strange ice cube has formed. Taste it.  The outer wall of the cube will taste salty, but the remaining ice will be less salty and have a loose, snowcone-like texture. This occurs because the salt was forced to the outsides of the cube and into its center as the water froze. In fact, crack the cube open and a small puddle of salty water should leak out. The loose texture of the ice is caused by all the holes the salt leaves as it migrates out of the water.

Normal tap water doesn’t have nearly as much mineral content as salted water, but the same principles apply. The more slowly you freeze ice, the more impurities get forced out to surface of the cube. Takeaway? Rinse ice off before you use it to remove most of the impurities.

Reason #2: (sort of) texture and dilution

I can say anecdotally that clear ice has a harder “crunch” than normal ice and that it melts more consistently.

Kold-draft and others claim that clear ice melts more slowly than normal ice. However, I have yet to see a test that compares the melting rates of equal weights of clear versus cloudy ice and I haven’t gotten around to doing it myself.

What causes cloudiness in “normal” ice?

The transparency of liquid water is caused by the lack of interfaces where light can scatter. With liquid water, dissolved salts and minerals don’t affect transparency.

Ice becomes cloudy for a number of reasons:

Reason #1: Dissolved gases. 

As gases are forced out of solution by crystal formation, they are trapped as tiny bubbles (see image above)

Reason #2: Crystal formation.

“At its freezing point, water molecules are moving so slowly that they begin to gather in loose, undefined clusters, called seed crystals. As the temperature continues to drop, molecules line up around the seed crystals in increasingly rigid formation until all of the water crystallizes. The longer it takes for crystals to form, the larger they will be.” [Joachim and Schloss, The Science of Good Food, Pg. 253]

You know how you’re supposed to freeze food quickly so smaller ice crystals form? You want the exact opposite effect when making clear ice cubes.

Reason #3: Density changes

Water is most dense at 4C or 40F, but it’s only slightly less dense at freezing temp, 0C/32F. Ice at freezing temperature is 8.3% less dense than liquid water at the same temperature – a substantial difference.

If water cools quickly, a density change can actually occur after the ice has already frozen solid. That’s right - molecules within ice continue to shift even in its solid form. Don’t believe me? Take a look at this time-lapse video of ice freezing.

The video makes it really clear (har): notice how as early as 1:00 minute in, air bubbles are trapped in solid ice. But the ice continues to grow significantly well after the entire block turns solid.

How do professionals make perfectly clear ice?

Ice sculptors have long made large blocks of clear ice by slowly freezing water from the outside in. They agitate the water in the center of the ice that remains liquid to allow gases to escape. When the ice is almost frozen solid, the very center of the block is drained. High-end ice machines spray layers of water onto a heat sink that freezes the ice layer by layer. Dissolved gases are quickly forced out of solution.

How do you make perfectly clear ice at home?

Two ways work: Directional freezing, and high-temperature freezing.

Option #1: Directional freezing

Cocktail journalist Camper English, did a fascinating series of experiments in which he tried a number of oft-recommended tricks for making clear ice ( – everything from consecutive distilled water to boiling and refreezing, to using carbonated water. In the end, he found that without access to specialized equipment, all homemade ice will trap gas. The trick, then, is to force the gas (and therefore the cloudy part of the ice) all to one area, where it can be isolated and removed from the clear remainder.
directional freezing, Camper English style

To make his ice cubes, Camper used custom molds and an igloo cooler to force long sticks of ice to freeze from the top down. The cooler’s insulation prevented cold air from getting to the sides or bottom of the ice. All the cloudiness was trapped at the bottom of the sticks; a quick melt easily got rid of it.

Option #2: High-temperature freezing

close-up of really clear ice

While Camper’s method certainly works, especially for large quantities or unique shapes of ice, it can be a little inconvenient, not to mention all the equipment involved.

A commenter on Camper’s blog turned me on to high-temperature freezing – basically, allowing water to freeze extremely slowly at just below its freezing point, ideally around 30F or -1C.

The method combats all the reasons ice turns cloudy that I talked about above. The slow freezing forces gases out, creates large crystals, and allows density changes to occur with minimal stress.

Using this method at home turned out to be a lot easier than I expected: simply set a compact dorm-style refrigerator to the warmest setting and use a normal ice tray.

If you don’t want your fridge at a really warm temperature, I’ve found that the top shelf of my refrigerator sits at about 29.5F when I turn it to its coldest setting. See below.

mini cocktail fridge for freezing ice

Stuff that doesn’t work

Search the internet for directions on making clear ice and the most popular recommendations are to boil the water first, use distilled water, and insulate the ice while freezing.

Boiled and distilled water help rid water of impurities, but remember that oxygen dissolves readily in water at room temperature and atmospheric pressure, so any gas you manage to boil out rapidly finds its way back in. More importantly, simply getting dissolved gases out doesn’t help to address expansion or crystal formation.

Insulation works to an extent, but not if your freezer is really cold. My main freezer sits at just below 0F/-18C. I tried a commercial insulated ice cube product, as well as about a dozen variations of DIY-jobs using styrofoam, PVC, water baths, etc. None of these worked in my freezer – all left a cloudy area somewhere in the cube.

Insulation will probably work if you have a warmer freezer – somewhere in the 17-25F range, though I haven’t test this.

Finally, I’ve read reports that if you first freeze ice, then pour boiling hot water on it and refreeze, the whole block will come out clear. I think this does work – but only because the hot water raises the temperature of the freezer.

What would you use fancy ice cubes for?

About Kevin Liu

Kevin Liu is a hopeless food geek obsessed with bacon, kale, and cocktails (but not usually all at once). You can follow him at @kevinkliu. Kevin is an editor at Science Fare.

, , ,

72 Responses to The weird science of ice and how to make “Premium Ice” at home

  1. Richard July 13, 2012 at 1:58 pm #

    After much trial and error and Internet research I found the Polar Ice Tray, purchased it and have found it to make the clearest ice yet, still not as clear as professional ice, but a great improvement over other home techniques.

    • Kevin Liu July 14, 2012 at 12:51 pm #

      Hey Richard, I’ve actually tried one of their products – the new “bamboo” version. The idea definitely does work, but raising the temp on your freezer should improve your results even more.

      • Andrew Ho August 15, 2012 at 1:25 am #

        Heh, I’m the Andrew mentioned in the intro. Thanks for writing this article!

        I have tried both Polar Ice products. The original pebble-shaped one makes decently clear ice, and is what I use at home (the main problems are that there is a rough edge on one end of the ice, and you have to cut it into cubes yourself). The Bamboo is crap, I couldn’t get a single clear cube out of it, and, you have to freeze a single cube at a time.

        • Kevin Liu August 15, 2012 at 6:53 am #

          Andrew, I tried the bamboo product as well and had the exact same thoughts.

  2. Barbara | Creative Culinary July 14, 2012 at 12:00 pm #

    I make cocktails for a weekly blog post and just make it simple on myself. I buy ice. I love Sonic’s ice for a lot of uses and the ice from the liquor store near me works fine…it is clear but I still long for large beautiful blocks. Maybe I’ll have to experiment. Maybe I need a wine fridge. :)

    • Kevin Liu July 14, 2012 at 12:53 pm #

      Barbara, these dorm-style fridges are all over craigslist – perfect for wine, cocktails, or any other sort of experimentation. My next project will be to hook one up to a PID controller…

  3. Rush July 26, 2012 at 9:14 am #

    Are you able to freeze bigger blocks of ice using the high temp method?
    Can you post a pic of your high-temp frozen ice?

  4. Kevin Liu July 26, 2012 at 5:30 pm #

    Hi Rush:

    Yes, you can definitely use this technique with larger blocks, though personally if you are going much larger (say, 4″ on a face or larger) I would recommend freezing at a much lower temperature and then simply carving pieces off as you need them.

    Check out – Camper has many different ways to deal with large blocks of ice.

    The reason I recommend this is because freezing at a high temperature takes a very long time and will take exponentially (squared) longer as you increase mass.

    As far as pictures – the pictures I used in my post are examples of the clear ice. Yeah, I know, there are a few bubbles leftover – but trust me, these were much more clear than normal ice.

  5. Jen July 28, 2012 at 10:48 am #

    I guess I’m confused. You say “simply set a compact dorm-style refrigerator to the warmest setting” So this dorm fridge would be setting at a warmer temperature than my normal fridge? How will ice ever form in a warm fridge? Do you then put the ice tray in the freezer compartment? Or do you mean to turn the dorm fridge to the coldest temperature and put the ice in the fridge part?

    • Kevin Liu July 29, 2012 at 9:51 am #

      Hi Jen,

      Sorry for the confusion! I’ve been getting this one a lot.

      If you choose to set the fridge to the warmest setting, you’ll want to put the ice tray in the freezer portion of the mini-fridge. The freezer portion should be right around 30F and the rest of the fridge will be around 45-50F. *However* this is somewhat dangerous, because then the food in the refrigerator portion will be in the DANGER ZONE for bacterial growth. This actually happened to me and a lot of the stuff in my fridge spoiled.

      As a workaround to this problem, I found that the top shelf on my mini fridge happens to sit around 30F when I set the fridge to the coldest setting. That’ s when I would put the ice tray on the top shelf. The freezer portion would be at something like 0F in this situation.

      The real takeaway, though, is that it’s worthwhile to get a decent infrared thermometer to better understand the temperature zones in your refrigerator. I’m still experimenting – right now I’m getting good results after removing the door to the freezer section in the mini-fridge. This results in a much smoother temperature gradient across the various zones of the fridge.

      Does that help answer your question?

  6. Colin Gore August 6, 2012 at 12:30 am #

    I saw your nod from Camper. That’s quite a compliment! Are you familiar with homoepitaxial growth? I believe the techniques you endorse are based on the principles of homoepitaxy of ice from water. The direction of crystalline growth is constrained to 1D, and each crystalline layer is in more-or-less perfect registration with the seed layer beneath it (or above it in this case), leading to good optical clarity.


    When you do a low temp freeze, does the water still freeze from the top down?

    • Kevin Liu August 8, 2012 at 6:13 pm #


      So I wasn’t aware of homoepitaxy, but I did suspect that crystal seeding could play a role in opacity. I believe water behaves somewhat like cholate – a few crystals of a particular size or shape can multiply, given the right conditions. I think, however, that this phenomenon applies more to the formation of smaller crystals, rather than the large crystals that are desirable for clarity. Does that make sense?

      I don’t think water ever really freezes from the top down. Water is densest at 4C, so I think the freezing process generates currents so the whole shebang freezes at roughly the same time. If you break into an ice tray before the cubes go solid, you’ll find crystal structures and a matrix of ice and water that permeates well into the core of the cube.

      • Colin Gore September 1, 2012 at 11:51 pm #

        Camper English got a clear ice tip from people that cut ice blocks from ponds (as was common back in the days before electrical refrigeration). It certainly freezes top down there because the earth is warmer 0C, and the resulting ice is incredibly clear. Mimicking that in a cooler was his grand discovery and causes a similar directional freezing for sure. It’s also how a few commercial ice makers make clear ice.

        You’re correct that this might not be the case for the slow freeze method, though. It would be interesting to determine, since I had assumed clarity was generally a result of unidirectional freezing.

        • Colin Gore September 1, 2012 at 11:57 pm #

          I suppose the insulated tray is like seeding from the surface. Seeding from the center would perhaps be ideal, but how do you suppose you could get the crystals to stay there without floating?

          • Kevin Liu September 7, 2012 at 7:49 pm #


            If you take into account the density of ice at temperatures close to freezing, you’ll see that directional freezing probably doesn’t apply here. Instead, I think a current actually develops in the ice.

            As room-temp water on the surface of the ice tray cools, it gets denser and sinks. When the cold water on the bottom of the tray eventually begins to freeze, it starts to float, as solid 0C H2O is much less dense than liquid 0C H2O.

            As far as crystals go, I’m not sure I would assume that they float. The assumption of floating implies the assumption that the crystals are forming in solid ice. I’m not sure seed crystals form only in solid ice; I remember reading some papers that assert crystals can form in liquid water as well. I may be misinterpreting – would love it if anyone out there knows the science.

          • Colin Gore September 8, 2012 at 1:43 am #


            Forgive me if this is heavy handed or old news. There are a few things to sort out here. I wasn’t being judicious in my use of “seeding” vs. “nucleation” above, so I hope that wasn’t the heart of any confusion. When you cool a liquid, crystals will nucleate from the liquid and grow. The faster you cool below the melting temperature, the more nucleation sites there are. At -1 C, there are a scant few nucleation sites that form by the time the whole solution has solidified, so those crystals grow larger and have fewer interfaces where defects will concentrate and diminish optical clarity. The general rule is that higher temperature leads to faster individual crystal growth (related to faster diffusion) thus larger crystals, but faster cooling rate leads to a greater number of nucleation sites (due to something called undercooling, cooling faster than molecules can rearrange), thus smaller crystals. Seeding is different.

            If you were to add already frozen ice crystals to the -1C water, they will act as additional nucleation sites. It takes less energy for those seed crystals to grow than it takes to create new seed crystals, so they dominate the solidification. This is what occurs when you temper chocolate by melting the base and adding some already tempered chocolate flakes. Perhaps you’re well aware of this, but it’s still good to clarify. On paper, it seems like bringing water to 0C and adding a seed crystal in the center would cause the crystal to grow from the center outward, preventing any gases and strain from volume expansion from concentrating in between crystals and causing cloudiness. One problem, though, is that freezing is exothermic (due to the latent heat of fusion) and adds heat to the solution surrounding the ice crystals as they form. Ice is way more complicated that it seems at first, which is why I find it so intriquing. It’s an everyday material, so you would think that it’s exhaustively understood. Turns out you can still get a paper published in Nature in 2002 on the freezing of water….that’s a real testament to the complexity.

            I actually plan to cover some phase transition thermodynamics for a post this weekend. One that’s been in the works for inexcusably long seeing as how I have it written out on paper already. I anticipate it will be interesting to ice nerds. Your comments on my blog and your own ice post have helped catalyze my overcoming of my sloth in this matter. Thanks!

            Would you mind elaborating a bit on the unsuccessful insulation vessels you have tried? I’d be interested in seeing the prototypes.


          • Kevin Liu September 8, 2012 at 10:57 am #

            Colin, here are some pictures.

            I used pvc pipe as insulation, hoping to freeze the top portion clear and end up with a cloudy bottom portion. I also sprayed home insulation around this doohicky.

            These were used to insulate a standard size Tovolo ice cube tray. This method actually worked decently, but there was always the problem of fitting the pieces together perfectly. Oh, and duct tape does not stay sticky when frozen.

            I tried submerging the trays in a larger container of water to see if the outside water would insulate it. The holes were poked to allow the tray to stay submerged (or something like that – I forget, and it didn’t work)

  7. Francisco Terrazas August 16, 2012 at 1:14 pm #


    Thank you for your time devoted to this topic that seems to have a myriad of opinions associated with it without much actual experimentation to back them up. I am a professional bar manager and have been trying to think of ways to upgrade the ice program at my restaurant with a limited budget and space. I am hoping to install a small fridge to devote to ice in the near future. Just to be sure that I understand, you are advocating placing ice trays in the normal compartment of the fridge and turning that to the warmest setting, correct? How long do you think it would take that process to freeze a couple of sheet pans full of ice?

    • Frank Shann August 16, 2012 at 10:56 pm #

      Kevin’s reply to Jen on the 29th of July 2012 (above) answers this

      • Francisco Terrazas August 19, 2012 at 5:08 pm #

        yeah…. Sorry. for some reason when I was looking at this post on my laptop, it wasn’t showing that response but my desktop did.


        That Guy

  8. AJD August 16, 2012 at 3:14 pm #

    Check out the review of the Polar Ice Tray on Serious Eats

  9. AJD August 16, 2012 at 3:18 pm #

    The link didn’t show up in my previous comment, so here’s another attempt:

    • Kevin Liu August 17, 2012 at 3:36 am #


      thanks! I took a look – interesting review – always good to spark conversation about this stuff.

  10. Gael August 17, 2012 at 2:59 am #

    I found this through camper like other people. You said that you found out about high temp freezing through a commenter on alcademics. I remember making a suggestion about freezing ~30f but I never did the experiment myself or else my food would suffer. I even cited some scientific literature. I was just curious if the commentator was named ‘Gael’?

    • Kevin Liu September 7, 2012 at 7:44 pm #


      Sorry – thought I responded to your comment earlier!

      I got the information from someone named “Cory Hain”.

      Can you share your original comment? I’d love to see what you wrote.

      • Gael October 22, 2012 at 9:37 pm #

        I make some comments here ultimately concluding that you need a temp near the freezing point to get clear ice. I even manage to find an academic paper as my argument. I also see Cory Hain on the same page but it’s two years after my posts so I didn’t see it. On another page I make a suggestion to freeze at 30F

        I would have done the experiment myself but I didn’t have that much control over my fridge and even if I did I have a lot of food in it that I didn’t want to risk being too cold or not cold enough. But I was pretty comfortable with my research that freezing at a high temperature would have worked.

        In the end it doesn’t really matter. I’m glad someone did the experiment so I know definitely that this method works.

  11. Steffen December 2, 2012 at 6:41 pm #

    Hi Kevin, I’ve just got my hand on the tovolo sphere ice molds. Im wondering if you have any set up that i could try to make a clear ice ball or the mold itself doesnt allow clear ice to be made. check the link to the product here. Thanks in advance.

    • Kevin Liu December 2, 2012 at 6:46 pm #

      Hey Steffen,

      That mold looks very cool! Really testing my willpower not to buy one right now… I’ve only used this mold:

      and I haven’t had good luck with it. But that may just be because I’ve been too impatient and keep checking the mold before the ice freezes fully. In theory, ice spheres should work, as long as you freeze them close to 30F. Take a look halfway down the page on this post, and you’ll see an example of clear ice balls made with a mold similar to the one I use (I think-the poster doesn’t say).

      I’m going to go toss a few ice balls in my fridge now and will report back if it works. Please comment if your mold works!

      • Kevin Liu December 21, 2012 at 6:30 pm #


        Quick follow-up: I’ve been having very poor luck doing clear ice spheres. They won’t freeze at the temperatures I use for tovolo trays, so I’ve had to change the temperature on my temp controller. At the same time, I think my temperature probe is starting to lose it calibration.

        Basically, I still think it’s possible, but you’ll probably have to use a lower temperature to beat the insulation of the plastic – something like 27F instead of 29 or 30F.


  12. Kevin La Presle December 21, 2012 at 3:02 pm #

    Hi Kevin,
    I just stumbled upon your cocktail blog and am looking forward to delving deeper. I was wondering about using warm water? It seems like this would make ice slowly.

    • Kevin Liu December 21, 2012 at 6:29 pm #

      Hi Kevin,

      Using warm water helps, but not a whole lot. Where I’ve seen warm water used successfully has been when the water was warm enough and there was enough of it that it actually raised the temperature of the freezer.

      Ultimately, if you freezer is at -20F and you’re only freezing a handful of cubes, warm water won’t make a difference either way.


  13. Douglas Shuntich January 8, 2013 at 4:04 pm #

    a few bits of info that might help; 1) attempting to freeze water at the warmer temps (~30F) will likely result in supercooling if the water is mostly or completely covered and void of an exposed surface. Nucleation sites tend to present themselves at the exposed surface, which is why most ice cubes freeze from the top down. If your water is not freezing at 30 deg, it is supercooled – give it a seed crystal and it will begin to freeze – though barely visible at that temp. 2) the thermostat and subsequent compressor cycling in your mini-fridge are likely to be woefully imprecise, meaning you may be getting 5-10 degF temp swings for significant lengths of time (30 minutes or longer) during cycling. No fridge holds an “exact” temp – it’s constantly on a curve up or down in it’s cycle. Though your thermometer may show an average of 30F, the lengthy temp excursions of imprecise equipment matter when it comes to controlling the freezing process. Invest in some precision usb temp sensors and a temp controller for the mini-fridge. 3) Nucleation and Freezing of water is exothermic to 32F – from the first crystal. A previous poster touched on this but didn’t go far enough IMO. It is very important in considering the entire thermodynamics of what is going-on inside both the cubes and the fridge/freezer. For example, supercooling and subsequently nucleating 2Liters of water will produce a significant amount of heat energy at exactly 32 deg F for up to 4 hours which can cause thermodynamic chaos in a small mini-fridge. Enough energy to nearly “thaw” other items that may be freezing or frozen that are close by. Even in the small environment of an ice-cube tray, understanding the heat transfer and resulting fluid dynamics due to the exothermic nature of nucleation from the first crystal is key. 4) Nucleation is stochastic – governed by laws of probability.

  14. Kevin Liu January 9, 2013 at 9:21 am #

    wow – thanks, Douglas! Yes this is very eye-opening and I’ll update the content in the book I’m writing to reflect this.

    • Douglas Shuntich January 10, 2013 at 10:22 am #

      here’s another one….

      5) some people find that warm water or boiled water will freeze (i.e. Nucleate) faster than cooler water. This is sometimes true. Here is what is happening…

      Start with 2 cups of water placed in the same freezer at 0-degF. Initial conditions are: Cup A is hot at 120-degF, Cup B is cool at 60 deg-F. Both cups begin a rapid cooling process, but Cup A’s cooling curve is much steeper due to the larger temp differential.

      Cup B will go through a relatively mild thermo flipping at 40degF (max-d) due to it’s softer temp decline curve and thus will be more ‘stable’ through this transition than Cup A.

      Cup B will subsequently go through the supercooling phase at 32degF in a more stable configuration than Cup A, resulting in ‘deeper’ supercooling temps prior to nucleation.

      Finally, Cup A will likely nucleate, i.e. begin freezing, near 30-32 degF due to it’s relative instability caused by rapid transitions through the 2 key phases mentioned above, while Cup B remains in supercooled liquid form even at 25-degF or lower. The untrained eye will not understand that the unfrozen Cup B is actually colder than the frozen Cup A. Furthermore, once Cup A nucleates, it begins pumping ‘heat’ energy into the freezer at 32 deg F which can further delay Cup B from nucleating.

      These results won’t always happen this way in a home environment, in part due to the stochastic nature of nucleation and in part due to other nucleation sites being present for both Cups. In controlled lab conditions, the warmer water will freeze first 70% of the time or more depending on initial starting temps vs. freezer temps. Tweaking the conditions carefully can bring it to 90%+….part of the keys to the kingdom.

  15. Shannon @ TheDrinkBuzz January 14, 2013 at 8:35 pm #

    I love all the information here about producing a crystal clear ice cube. Hmmm… I’m going to have to get on it and try some of these suggestions or maybe try the Polar Ice Tray. I also have a drink blog and need to make drinks often too. Thanks :)

  16. juicer February 13, 2013 at 8:16 pm #

    Here is my method for getting big (4+ ounce per cube) clear cocktail ice at home. Basically a more convenient version of the Camper/Alcademics method using a cooler. Same principles as the Polar Ice Tray, but you make one use-able piece of cocktail ice at a time.

    Start with a foam insulation beer can cozy. Insert into the cozy a plastic cup that fits tightly inside the insulation. I use a 16 ounce HDPE Nalgene bottle that I sawed the top off of. Just the right size — about 2.75 inch in diameter. So now you have a conveniently small insulated cooler with an open top to work with.

    Take a 5 ounce Dixie cup and punch a bunch of holes in the bottom. Put the Dixie cup into the cozy cooler. Fill with water to just below the rim of the Dixie cup. Freeze.

    When it is solid, remove the ice cylinder. Chip away the outside ice and then peel off the Dixie cup. The ice inside the Dixie cup will be almost entirely clear. There will be a mass of cloudy ice stuck to the underside of the bottom of the Dixie cup. Break that off with a putty knife.

    Like pond ice, the Dixie cup slowly freezes from the top down (with the sides and bottom being insulated by the surrounding water and the beer cozy). The gas bubbles get pushed down to the bottom of the Dixie cup, and then get pushed out of the Dixie cup at the very end of the freeze.

    You can use just about any kind of ice cube mold shape so long as (i) it fits inside the cozy and (ii) you drill holes into the bottom of the mold. Cylinder shaped plastic jars/bottles work very well. For a big cube, I use the 4 ounce Rubbermaid food storage container. Some smaller ice sphere molds will fit as well. Put something into the bottom of the cozy cooler (like a pair of dice) to elevate the bottom of the ice mold so that the gaseous water has room to flow out. [Not necessary with the Dixie cup, which has a rim which elevates the bottom of the cup.]

    For larger molds (like the Tovolo sphere which has a 3.5 inch bottom diameter) replace the cozy cooler with a bigger diameter cooler — a plastic insulated soup thermos or big coffee mug.

    • Kevin Liu February 13, 2013 at 10:48 pm #


      I tried a very similar setup to this – (see the picture above of the PVC piping. at one point, it was also surrounded with spray foam insulation). However, I never got good results with mine. Would you be able to measure the temperature in your freezer and post a few pictures too?

      That would be amazing! :-)

      • juicer February 14, 2013 at 1:08 pm #

        My freezer is set at the middle setting — 18F degrees.

        Just a $1 foam rubber beer cozy with a can in it, then slice off the top of the can to make it flush with the top of a cozy. You could use a beer can if you weren’t worried about the sharp metal edge; I have a sawed off plastic Nalgene bottle instead of the metal can.

        I suspect that the beer cozy is just a better insulator than your sprayed pipe experiment. It is thick and fits very snug around the can/bottle. It probably insulates as well as Camper’s open top picnic cooler.

    • Charlie Mote August 12, 2013 at 7:29 pm #

      Just tried this with 2 Tovolo spherical ice molds inside a Planet Smoothie insulated plastic mug. I filled the Tovols to the brim and inverted in the filled mug. I drained the mug down to above the silicon of the top mold. It took all night and a work day (the bottom sphere wasn’t totally frozen) and a hot water bath to release the molds but the results were great – very clear. And it produces a disc from between the top and bottom mold. For $10 for the mold and the free mug a pretty inexpensive clear sphere solution for my fresh juice.

      • Kevin Liu August 12, 2013 at 7:48 pm #

        That’s awesome Charlie – Thanks!

        • Charlie Mote August 13, 2013 at 10:57 am #

          My pleasure Kevin – thanks for your blog – I’m recommending it to my local Science Tavern Meetup group right now. Also, The sphere from the lower mold clouded on putting it back in the fridge (as I wrote, it wasn’t completely frozen before I took it out to inspect it.) I suspect the bottom of the mug actually cooled through conduction faster than the top down cooling from the open mug top so I am going to put a void underneath it and make sure the bottom mold is suspended off the bottom of the mug (I think I had it pressed to the bottom.)

          After reading everything here, I decided the whole thing was worth a shot for fun just to see, but now I need to go back and try a real documented run.

          It may be overkill as the molds themselves seem to get good reviews for clarity all on their own, but the science would seem to suggest used as directed would have to have some imperfections.

          Goal being two mugs in the fridge at all times to have four possible spheres on-hand; I like my juice cold!

  17. Colin Fogarty February 18, 2013 at 9:56 pm #

    Quick question. If we wanted to freeze it slowly, wouldn’t it be a good idea to put it in the fridge for an hour or so, stir it to agitate the oxygen. Then put it in the Freezer for the final touches?

    • Kevin Liu February 19, 2013 at 12:09 am #


      After reading through Doug’s contents, above, I find it to answer any ice question with certainty, but here is a quick thought:

      - It takes ice several days to freeze at 31F, so a pre chill in the fridge, or stirring with oxygen, shouldn’t make a big difference in the long run.

      Does that make sense?

    • Doug Shuntich February 19, 2013 at 8:28 am #

      Colin, pre-chilling the water will actually help it supercool when placed in the freezer which is not a good thing for clear ice because the crystaline structure can penetrate large volumes of supercooled water quickly upon nucleation – resulting in trapped air. However, in many cases it won’t make much difference. The more important factor will be the temperature inside your freezer during the freezing process. A relatively high temp of 28 – 31 degrees will cause the water to freeze slowly (due to it’s exothermic nature) and allow the trapped air to escape toward the unfrozen water – producing clearer ice. Ideally there will be some amount of flow, which can happen naturally in a container due to the density properties of water below 40 (warmer is denser and therefore sinks) and the exothermic nature of ice formation (creates heat at 32). It may all sound complicated, but the more you understand the various processes involved the better ideas you will generate on how to make practical applications that work. There are probably ‘sweet-spots’ in terms of temp, time, and configuration that none of us have formulated yet that could result in a very practical and repeatable formulation.

  18. Alex May 7, 2013 at 3:58 pm #

    This article here explains how the commercial ice makers make the perfectly crystal clear ice cubes (calling them gourmet cubes.) The process is pretty unique, and also very slow going. It’s almost impossible to replicate without some technological assistance, but the information Kevin provided is excellent!

  19. Dan Arnold May 15, 2013 at 6:05 pm #

    I’ve been making blocks of clear ice successfully for a few years, and have to credit Camper English for some of my ideas. One of the issues making the clear ice as easily as possible. There are diminishing returns for more intricate methods. I use a method that is reflected here. I simply pour filtered water into a small ice chest and freeze it. The slower you freeze it the better, so you want to lower your freezer temperature. If you time it correctly the block is clear, with the remainder still unfrozen. If you wait too long, you simply melt the small portion at the bottom that is not clear. I’ve made my own ice picks to cleave the block into cubes, using a pick that is slightly screw driver shaped to give some directionality to the chopping.

  20. Auri July 7, 2013 at 10:54 am #

    Hi everyone. I just made a 28 litre ice block in a cool box. It turned out extremely clear, but it seems to melt very fast. The old method of just freezing a bucket produces only parts of clear ice, but doesn’t melt as quickly. Any suggestions?

  21. Shrikant Singh August 6, 2013 at 1:12 pm #

    is ice cube is a crystal

  22. Jeremy August 14, 2013 at 1:51 pm #

    I bought this cake mold to make some fun ice, but it floats upside down. How do I go about getting it to float with the flower facing up? I assumed I would need to do a two part freeze with the second half being more dense, if so, how do I do that without changing the flavor?

    • Dan August 14, 2013 at 2:18 pm #

      Looks like it is made from pliable silicon, but if it is it will not float unless there is air trapped in it. You actually want it upside down as shown, so the air in the water can be pushed down as it freezes. I assume you want clear ice, so you’ll have to put it in water in an insulated tank with an inch or two of water at the bottom, having fully immersed it before turning it over. The other method is to keep it open side up, but poke sufficient holes in the bottom so the impurities and gasses in the water can be pushed down and out.

    • Kevin Liu August 14, 2013 at 7:17 pm #

      I’m assuming you’d want the finished ice to be clear as well? I mean, maybe you could freeze a weight of some sort into the ice? Or maybe just try filling the mold part way and see if it still floats upside down. The decorated part is the top; not having the entire shape might not matter?

      This is a tricky one!

  23. Onyr August 29, 2013 at 4:05 pm #

    After a lot of experimentation with different sized insulated boxes, with differing thickness of insulation, and different materials for dividers, depth of water, I’m ready to market an inexpensive device for making perfectly clear cubes with a minimum of fuss. I’m wondering what people think of as the ideally sized clear ice cube. What are your preferences?

    I’ve been happiest with a cube about 45 mm, or about 1.75″ to 1.875″, but want to get others’ thoughts.

    Also, I’m thinking that most people would prefer using something smaller than an Igloo 9qt. for the freezer portion of their refrigerator. What dimensions do people prefer for both the cube and the overall size of the ‘tray?’

    • Wes December 6, 2013 at 3:03 pm #

      Onyr, I have an ice sphere “press” that came with octagonal cups that are about 3.5 inches wide and 4.5 inches deep (or so). I’ve been experimenting with ways to use the cups and achieve a clear “blank” to be used in the press without having to cut the block. I tried insulating them with spray foam and submerging the cups in a cooler full of water, with both methods creating some clear ice at the top half or 2/3 of the cup, but not allowing for the cloudy part to be cut off without making the blank too short to be used in the press.

      So my goal is to figure out either how to use the provided cups with a method to create a clear blank (perhaps by drilling holes in them and elevating them with a mesh cooling sheet in an insulated cooler), or to use the cooler I have with some dividers to created the right sized blocks that I need. I purchased some lucite this week and am cutting the pieces to then fit together in a grid, creating 3.5 inch squares. My theory is that I can get clear blanks this way by placing the grid into the insulated cooler and elevating it a bit off the bottom, then filling with water to just below the top of the grid dividers. By leaving the bottom of the grid “open” and elevated above the bottom of the cooler, I figure the gases should get pushed below the level of the bottom of the grid and help with clarity. I’m also going to use the warm temp freeze method with the cooler, which is 17 x12 and 8 inches deep. I’ll complete the experiment this weekend.

      How is your prototype coming along, and is it similar to what I described above?

  24. fishmonger May 4, 2014 at 4:13 pm #

    hi, i have been following this concept of making clear ice at home. I would like to know if there has been any success to date, especially in the spherical ice.

  25. don cooper June 28, 2014 at 3:19 pm #

    This clear ice is great for pouring pop over without withou fizzing over the top of the glass

    • Douglas Shuntich June 28, 2014 at 5:47 pm #

      Clear-ice, if stored at relatively warm freezing temperatures (28-30 deg.F) has the added benefit of significantly reduced de-carbonating of carbonated beverages, while providing essentially the same cooling effect as that of much colder ice. Due to the melting properties of fresh-water ice cubes, “warm ice” and “sub-zero freezer ice” provide approximately the same results in cooling soda (or other beverages) to just about 32 degrees F….but the warmer ice has the above mentioned benefit, as well as usually being much clearer and visually appealing. Unfortunately, most standard fridge/freezers don’t allow storage of anything at those temperatures. Our company, Supercooler Technologies, is working on a device that will fill this temperature storage gap, and many others.

  26. Codge July 22, 2014 at 12:31 pm #

    I stumbled on a possible solution to this problem while trying to solve a slightly different one. I had silicone ice cube trays, which made for fairly cloudy cubes when using city water, but they came out of the trays fairly easily. I then switched to a counter-top model five-stage filter system, which came with a testing wand. The city water tested at 50 ppm, while the filter filters to 0. I used the filtered water in the trays, and it was extremely cloudy, had expanded in strange ways, and was impossible to remove from the trays. I think this was due to more rapid freezing. I decided that I would try to add back some salt to the filtered water, which would slow down the freezing, but be more appealing than whatever unknown contaminants are in the city water. I started with 50 ppm and increased by 50 ppm until I achieved perfect ice cubes. I found the sweet spot at 100-150 ppm. 50 was still cloudy and kind of lumpy, 200 froze too slowly. At 100 or so, the ice was very clear and removed easily from the trays. I noticed no salty taste at that dilution (it was only at around 600 ppm or so that I noticed a palpably salty taste, which is about 1/8 tsp per quart for reference).


  1. How to Make Clear Ice at Home | Craft Cocktails at Home - July 16, 2012

    [...] clear ice cubes used by the best cocktail bars at home. Over at my other blog, Science Fare, I explain. Go check it [...]

  2. Ice – Foam Panels - Three Sheets Mfg. Co. - March 13, 2013

    [...] *Kevin Liu breaks down the science going on along with another technique using a modified dorm fridge here. [...]

  3. Buying and Holding Your Way to the Poor House | Trading Blog Trading Blog - March 31, 2013

    [...] ice maker to a cold water supply, when science says the hot water supply would probably be better for various reasons? Because it’s what they were taught to do. Because “common sense” sometimes doesn’t make [...]

  4. Cocktail Science: 5 Myths About Ice, Debunked | Dessert Recipe Wall - June 18, 2013

    [...] 2: Use a temperature controller. I’ve written about my personal favorite method. I hook up a sous-vide temperature controller to a mini fridge so I can guarantee my ice freezes at just below 0&degC. The temperature [...]

  5. Cocktail Science: 5 Myths About Ice, Debunked | Recipes Pinboard - June 18, 2013

    [...] 2: Use a temperature controller. I’ve written about my personal favorite method. I hook up a sous-vide temperature controller to a mini fridge so I can guarantee my ice freezes at just below 0&degC. The temperature [...]

  6. Ice, Ice Baby – Bon Vivant Spirits Agents - August 27, 2013

    […] Science Fare describes an alternative to Camper's directional freezing method, using what he calls "high temperature freezing", but it requires a bit of an investment in a dorm-room bar fridge. The article is worth reading to […]

  7. Ice, Ice Baby – Bon Vivant Spirits Agents - August 27, 2013

    […] Science Fare describes an alternative to Camper's directional freezing method, using what he calls "high temperature freezing", but it requires a bit of an investment in a dorm-room bar fridge. The article is worth reading to […]

  8. The Gin Monkey Guide to a G&T | Gin Monkey - November 26, 2013

    […] Some geeky articles about ice: […]

  9. How to make a G&T: The Gin Monkey Guide | Gin Monkey - November 27, 2013

    […] Some geeky articles about ice: […]

  10. Clearing up this ice talk | THE BITTERS - December 5, 2013

    […] we already have an in-depth look at the science of water freezing, courtesy Science Fare and Kevin Liu: “The weird science of ice and how to make “Premium Ice” at home.” Check it out, it’s […]

  11. Lester Cheese on the Lives of Eskimos | Really Weird Things - December 16, 2013

    […] are all of these Eskimos freezing their damned asses off, trying to find food in this sub-sub-zero ice covered, wind swept wasteland, day in and day out, struggling through this miserable, cold […]

  12. How to Chill a Cocktail Correctly (Yes, There’s a Correct Way) « Food Hacks - April 27, 2014

    […] The second method? Freezing water slowly at a high temperature. When water becomes ice at a slower rate, it “forces gases out, creates large crystals, and allows density changes to occur with minimal stress,” as explained by Kevin Liu over at Science Fare. […]

  13. Ice Stand Corrected | Burners.Me: Me, Burners and The Man - July 4, 2014

    […] fact, not only does hot water chill quicker, many people think it produces abetter quality ice cube. And while ice is an afterthought for most folks, people who are serious about their […]

  14. Why are Some Ice Cubes Clear and Others Cloudy? | Steph's Random Facts - July 9, 2014

    […] The weird science of ice and how to make “Premium Ice” at home […]

leave a comment and we'll give you a cookie! (well, the internet kind...)