Crowdsourcing the Physics of the Backdraft Cocktail

As may be apparent from my earlier posts, I'm interested in "dangerous" foods and drink and playing with fire. This week, I've come across this flaming cocktail idea called "The Backdraft." Here's a bloke from New Zealand or Australia (I think) drinking one.

The recipe, abridged from Wikipedia is as follows

  1. A saucer is placed on a counter or table.
  2. A shot glass is placed in the center of the saucer, filled with Sambuca
  3. A pint glass is filled with 1 - 2 shots of Cointreau. Swirl this in the glass to coat the sides
  4. The Cointreau is lit and allowed to burn until the sides of the glass become warm to the touch
  5. The lit Cointreau is poured into the shot glass, igniting the Sambuca.
  6. The pint glass is lowered over the flaming liquid. As the atmosphere cools inside the pint glass it will try to suck the alcohol on the outside back into the upside down pint glass. This backdraft effect is the origin of the drink's name.
  7. The glass is removed and a straw is used to suck up the alcohol from the saucer and shot glass.
That seems a relatively straightforward preparation of a drink only a bit less goofy than a Flaming Moe. But I'm trying to understand the physics of the Backdraft. What causes the vacuum in the upside down pint glass? The Wikipedia explanation follows, but I'm not sure it's correct.

Backdraft physics follows

Once a gas has been warmed up, it tends to expand to fill a volume. It will replace other gases and expand due to its heat content. If this is done in an enclosed sealed space, and then the heat is taken away, as the gases cool, they condense, and decrease in volume, and create a vacuum. Thus when the flaming alcohol in a backdraft is covered with a pint glass over a saucer, the air (a heavier gas) is replaced with warm alcohol vapour (lighter gas)and warm air. As the remaining oxygen is used up, the fire in the pint glass goes out, and the heat source goes away. The alcohol vapor/air mixture now in the glass cools and begins to create a vacuum. This vacuum is responsible for sucking any liquid at the outside bottom of the pint glass further inside (as the seal of the glass and the saucer is not perfect). Once the majority of the liquid is inside the upside down pint glass, sometimes further oxygen can be seen to bubble up into the glass. At some point an equilibrium will occur, where the internal vacuum will hold the liquid inside the glass. This can be great enough at times, that the glass can be lifted, and the saucer will remain stuck to its underside.

When the pint glass is removed, ice is immediately added, thus causing the condensation of the alcohol vapour, creating a white mist in the glass. By covering the glass with the hand, this vapour is trapped until it is extracted by the process of inhalation, usually through a straw.


  1. The Wikipedia explanation is certainly confusing but is mostly correct. I do a version of this as a demonstration when my physics class reaches fluid mechanics — pour a modest amount of water into a steel drum and place it over a flame until the water boils. Seal the lid and transport the drum to an ice bath. In a few seconds, the water vapor inside the drum condenses back into fluid. Now the air pressure outside is much greater than inside since there is less gas inside and the drum is crushed. You’ve probably seen something like this.

    Same idea. Less dramatic just because the seal is really bad and there probably isn’t that much unburned alcohol vapor in the glass in the first place. But as long as there is an opening to the room, the gas pressures inside and outside the glass (or my steel drum) are teh same because the system is at equilibrium. Then the inside is prevented from equilibrating with the outside. Then, some of the gas inside is removed by condensing it into fluid, thereby lowering the gas pressure. In my case, that collapses the drum because the seal is gas tight. In this case, it does not because there is a leak. So outside air pressure drives fluid inside just as it does with a mercury barometer. The fluid entering the glass compresses the remaining gas inside, increasing its pressure until equilibrium is once again reached.

    It seems, however, like an excessively complicated way to get what amounts to a microshot of Everclear.

    1. I have nothing at all to disagree with about your comment except that:

      It seems, however, like an excessively complicated way to get what amounts to a microshot of Everclear.

      That’s what makes it so fun. You’re perfectly right, of course, but senselessly complex and hazardous things are often very good fun =)

  2. The other possible cause of the low-pressure atmosphere in the pint glass is from the combustion of alcohol vapour.

    C2H5OH + 3O2 → 2CO2 + 3H2O

    If the water vapour condenses out on the glass, that would result in a lower gas volume.

    1. If the water vapour condenses out on the glass, that would result in a lower gas volume.

      Minor nit: the volume is the same (i.e. the volume of the glass).

      It’s the actual number of gas molecules (measured indirectly, for example, as mass) that is reduced when the water vapor condenses into liquid. I.e. the “n” in PV=nRT. V is constant, T (for any gas remaining gaseous) and n (for any gas condensing into liquid) are decreasing, so P must decrease too.

  3. accents don’t sound australian or New zealand @ all. I am confused as to why you think they do…

    from youtube i’ve managed to extract this info…

    “This drink is served up by Harley Manion, the bar manager of the Paper Moon, 24 Blackfriars Road, London SE1”

  4. Here’s a bloke from New Zealand or Australia (I think) drinking one.

    They both sound English to my (New Zealander) ears, granted that the guy doing the actual drinking doesn’t say a lot.

    1. Yeah, as an Australian born to New Zealander parents, who lives in Australia but just visited New Zealand last week, I can tell you that this accent definitely belongs to neither country. My best guess is the UK.

  5. This is very impressive. Looks like a good start to an evening, but definitely not something to try after you’re already drunk.

  6. Good grief; I know the general level of physics literacy demonstrated on BB is not exactly stellar but this is utterly basic stuff. Ignore the effect of the warmed glass for a moment and just consider trivial physical chemistry. The atmosphere is approx. 20% oxygen; burn something in a sealed space and as long as you have enough to use up all the oxygen you will have taken away 20% of the gas (unless your burning creates a similar amount of gas. Finding the amount of exhaust gas and whether it is as much volume of waste gas as the oxygen consumed is left as an exercise for the reader) and thus the outside atmospheric pressure will push the outer puddle of alcohol into the upturned glass.

    The effect of the warming of the gases inside the upturned glass can be simply calculated by application of Boyle’s Law and measurements of the relevant temperatures. It is possible that for a brief instant the expansion caused by the warming might counter the contraction caused by the removal of the oxygen fraction.

    Even simpler experiment that I did at home as kid –
    Fix a small candle in the middle of a cereal bowl
    fill bowl with an inch or so of water
    light candle
    place suitable tall glass over candle and into water. yes, upside-down you dummy.
    observe water rising into glass
    when candle has burned all available oxygen it will go out (please tell me you’re not surprised by that part)

    Alternative and slower version –
    place small bowl in larger bowl
    fill larger bowl with some water
    fill small bowl with mouse, rat, or if appropriate, small brother
    place large-ish glass over small bowl and trapped sibling
    as oxygen is burned in body of expiring brood-mate, water level will rise. note that breathing of mammal takes oxygen out of atmosphere and replaces it with carbon di-oxide. Student should explain the relationship between amounts of gases. Student may be awarded extra credit for successfully resuscitating experimental mammal.

    1. Good grief; I know the general level of physics literacy demonstrated on BB is not exactly stellar but this is utterly basic stuff. Ignore the effect of the warmed glass for a moment and just consider trivial physical chemistry. The atmosphere is approx. 20% oxygen; burn something in a sealed space and as long as you have enough to use up all the oxygen you will have taken away 20% of the gas

      Seems to me, before one starts climbing up on their high horse and denigrating the physics literacy of others, they ought to get their own house in order.

      Combustion doesn’t “use up all the oxygen”. It simply rearranges the molecules. The exact outcome depends of course on what you’re burning (kind of mentioned parenthetically in the referenced post…but leaving “as an exercise for the reader” the key point in the question is pointless), but there’s no way you’ll ever use O2 at 20% partial pressure and literally reduce the gas quantity by 20%. Not even close. You’d only get that if the entire product of combustion was liquid or solid, which just doesn’t happen.

      And a “physics literate” person also ought to be able to use the word “volume” correctly in a sentence. :p

    2. I just tried a variant of the candle experiment using heat sinks instead of an actual flame. (so as to eliminate the chemical reaction variable.)

      Genuine $0.01 heat sinks (USA Pennies) x3
      Lit candle
      Small bowl partly filled with water
      Small buoyant container (or SS PV=nRT) to hold pennies (in my case, the cap of a cocktail shaker)
      Small glass that will encompass the container, yet fit within the bowl.
      (and tweezers to hold pennies instead of using your fingers. (OWOWOWOWOW!))

      Place SS PV=nRT in the partially water-filled bowl.
      Grasp the pennies with the tweezers. Heat the pennies over the flame. After humming a peppy, jazzy version of the chorus of “When the Saints Go Marching In” twice (or waiting 30-40 seconds), place the (now hot) pennies in the SS PV=nRT.

      Quickly and carefully place the glass over the boat and allow it to rest on the bottom of the bowl.

      Wait. Since the method of heat exchange is conduction (and much slower convection than an open flame) it will be less dramatic.

      What should happen (assuming it follows the same case as it did for me.) is that level of water in the glass should initially be pushed down (by the increasingly warm air that requires an increasingly larger volume to maintain the same pressure). Soon, some air will bubble out of the bottom of the glass.

      Eventually (for me ~1-2 minutes) the pennies and remaining air within the glass will cool down, causing a decrease in pressure. Water will be drawn back up inside the glass. The final level of the water will depend on the heat of the pennies, the volume of the glass, the volume of gas that escaped as bubbles (and a bunch of other things).

      On a completely unrelated note…
      Tim, if people’s lack of “physics literacy” upsets you, writing the a post in a condescending and exasperated tone does not exactly promote a spirit of healthy curiosity and scientific inquiry. Encourage people’s efforts to expand their knowledge of science rather than chide them for their lack of experience/education. You’ll do a lot more for increasing the public’s understanding and appreciation for science. Plus, you’ll help eliminate the stereotype of the arrogant, snippy, know-it-all scientist with poor people and language skills.

  7. The accents in the video sound British to me (a Kiwi), but Backdrafts have been popular in the student bars in Dunedin, NZ, since at least 2001.

  8. Tim wins!

    And I remember the candle in the upturned glass experiment being something I got from a book. I wonder which one it was. There must have been some other good stuff in there!

  9. there is no back draft going on here. A back draft is when you have a fire going in a sealed room, which burns up all the O2 and appears to go out. When you open the door to the room 02 rushes in, reigniting the fire. It is the surge of oxygen into a high heat, high carbon atmosphere that causes combustion to restart. (Some firemen have had their turnout gear ignite while testing SCBA because of the rush of oxygen on their gear which is covered in carbon from fire)

    This here is just simple condensation. This is how liquor is distilled in the first place. you bring a low concentrate of alcohol to a boil, the alcohol evaporates into the coils on the top of the still and drips down the side, giving you a higher proof. Thats all thats happening here, while its on fire the bartender puts the glass over it, capturing some of the vapors, then when he puts in the ice he is causing the vapors to condense back to a purer higher proof shot, which he sucks up with the straw. Silly thing about this shot, you would get more wasted if you just drank both shots without the fire. It feels strong because you are sucking it up concentrated, but there is actually less overall alcohol (some gets burned off before he puts the cup over it) then if you had just drank them straight.

  10. Wot Tim said… Bueno.

    Now for the *real* question: Why on earth would you want to burn off the active ingredient in a cocktail?

    (Countdown to some barroom brainiac yarbling on about “When you heat up the alcohol it gets absorbed sooner and metabolized later, so you really get mo’ drunker, longer, cheaper”. To which I would reply: “No”.)

  11. Saturday Morning Science Experiment time kids!

    Be sure that you have adult supervision, because we’re playing with fire!

    We’ll need:
    -8 or so quarters
    -a cooking pan
    -an old glass jar
    -a candle
    -a lighter

    1.) Fill the cooking pan with about 1/2 and inch of water.
    2.) Place the quarters in 4 stacks of two, space them evenly so that they will hold up the lip of the jar.
    3.) Light the candle and place it in the middle of the pan.
    4.) Invert the glass jar over the candle. Place it down to where the quarters provide a gap between the jar and the bottom of the pan.

    You’ll observe that the candle will burn for a moment, exhausting the Oxygen from the air that’s needed to supply the fire (Combustion uses up O2). When the fire goes out, you’ll find that the water level suddenly rises within the jar.
    When you try to remove the jar, the water will attempt to remain in the jar due to the suction place on it. Essentially, atmospheric pressure is pushing the water up to fill the empty space that was previously held by the oxygen in the air. After raising far enough, the weight of the water will break the seal and pour out of the jar.

    In engineering you learn about this as an example of fluid work as a change in pressure. You can actually calculate how much oxygen was in the air by measuring the column of water, and utilizing the following:
    Pressure = (density * acceleration due to gravity * height of water column) = Force * Area of Jar opening

    The Force is the evenly distributed weight of water. (F=mass * gravity) You can then find the mass of the water.

  12. Ah. You’ve got me there. I was getting sloppy by focusing on the pressure work. We will also assume that the wax and wick are negligible.

    Correct, it will transfer the O2 for CO2 and Heat, and there’s bound to be a sizable amount of incomplete combustion. CO2 has a density of 0.00184212 g/cm^3 (grams per cubic centimeter) (at STP), Oxygen has 0.001429 g/cm^3 (grams per cubic centimeter) (at STP). As CO2 is more dense than O2, then the resulting gap in densities causes a pressure drop, being as more mass is being compacted into a smaller space.

    Perhaps the method I mentioned earlier when taken into account with the chemical composition of air could utilize a general change in pressure to demonstrate the percentage of combustion obtained in the system?

    1. It depends.
      If you take the ideal gas equation of PV=nRT
      P is pressure,
      V is Volume (of the gas)
      n=number of moles of the gas (or, number of molecules of gas)
      R=constant (basically ignore for our purposes)
      and T=Temperature of gas.

      Given your requirement of “less molecules” (smaller ‘n’)….and assuming nothing else on the right side (temperature) changes, the left hand side of the equation (the product of Pressure and Volume) must decrease. This can be accomplished by the pressure of the gas decreasing, or the volume it fills decreasing (think of a mostly deflated balloon deflating even more).

  13. I think part of the fun is that you are going to get some alcohol vapor (that’s seems to be what he’s sucking up out of the pint glass and why the bartender keeps his hand over the glass).

    Dry ice in a liqueur that has strong flavor will give a great hit from the vapors. We had a bunch at a Halloween party once and decided to make alcohol fog. Midori works very well.

  14. So you start with 3-4 shots of already very sugary alcohol, evaporate away some of the alcohol but breathe most of that, and you’re left with 2-3 shorts of extremely sugary alcohol :-) I’m not sure that anise and orange flavors mix all that well either…

    I once microwaved rum to pour over a plum pudding and flame. In the process, I ended up breathing the vapors as I carried the glass over to the pudding; really gets your attention

  15. Screw the physics explanations, this is a very simple explanation :

    Take some alcohol, burn most of it off. Furthermore, it is a fire hazard, and for some reason the people engaging in it think it’s somehow enjoyable to inhale the carcinogenic byproducts of said burnt off alcohol.

    I toss this into the scrapheap of dumb ideas, like with those who think you have to “light” absinthe.

    Just do a YT search for “burning drinks” and you’ll see the hilarity of second degree burns this mostly results in.

    By the way – they’re British, not Australian.


  16. Lost my fascination with flaming drinks years ago when a waiter spilled a flaming Sambuca on my wife, setting her dress on fire. Minor burns and a destroyed dress. They paid for a new dress, but I always think we should have sued, as they didn’t really care.

  17. Woolly explanation. A gas *always* expands to fill its containing volume, so they start with a red herring. That said, what they appear to be trying to say makes sense.

    What they mean is that PV/T is (roughly, in real life) constant, so a cooling gas in an fixed enclosed space undergoes a pressure drop.

  18. Boo. I was expecting an actual backdraft effect (i.e. re-ignition). That would have needed some wierd physics.
    All I got was a few London cityboys (yes, they have a lower-middle class London ‘estuary’ accent like my own) chestbeating about making some fire.

    The error in the wiki description is the idea that the weight of the gas drives the effect. That’s arguably OK, but only by messing with the pV=nRT gas law. The effect is due to raising and lowering temperature. Volume remains constant.
    Under a flame the T and p go up, therefore in order to equalise the pressure, the amount of air in the glass (n) goes down.
    When the glass cools, T and p go down. To equalise the pressure, warm alcohol is sucked in and vapourises increasing n.
    Adding ice condenses the alcohol, reducing n again, which in turn causes p to decrease.

    There is actually very little alcohol vapour left in the glass, but it doesn’t take much for you to feel an effect.

    As a departing fact, a shot of spirit contains about 5 litres of alcohol vapour, if evaporated to dryness. This is approximately the volume of a cube with sides equal to the length of an average erect adult human penis.
    It also contains about 20 L of water vapour, which is approximately the volume of a sphere with a radius equal to the length of an average erect adult human penis.
    My chemistry degree has no end of practical uses…

  19. Sincere thanks to everyone who took time to answer this question for me. (Also, I admit, I do suck at accent recognition.)

    This particular experiment has interested me for a while, but I felt the reason behind it was never as simple as the reasons given by commenter #12. His explanation doesn’t hold water (so to speak), but you see it all the time.

    Let me direct those who after all this, are still interested in the question, to this site:

  20. Don’t know about the physics, but common sense tells me that extinguishing burning alcohol by blowing on it isn’t really the best idea.

Comments are closed.