Homemade Bismuth Crystals

This post is based on the work of Theodore Gray in his blog for Popular Science. Mr. Gray, in addition to designing the UI for Mathematica and creating a literal periodic table… table, also appears to be among the few people who can make perfect bismuth crystals the first time he tries it. See his tutorial here

The principle behind metal crystals is relatively simple: the metals found in most everyday objects were formed by pouring molten metal into an appropriate mold and then rapidly cooling the assembly. This means that the atoms that comprise the liquid metal do not have a lot of time to diffuse through the material and find a location that minimizes their electrical potential relative to the other atoms—they might have time to trade places with their neighbors and form small, ordered regions within the metal structure, but overall the metal solidifies before each atom has had time to try out every possible position in the structure and find the one with the least repulsion from the other charged particles. This means that most metal objects lack global order—under a microscope, small crystalline patches will be visible, but overall the structure is a hodgepodge of various crystal structures and orientations.

This rudimentary explanation suggests that cooling a molten metal slowly will allow larger crystals to form, which provides the logic behind Mr. Gary’s approach, in which molten bismuth is slowly cooled on a stovetop. The idea of metal atoms rearranging themselves also suggests why blacksmiths anneal steel knives by raising them to high temperatures for long periods—at high temperatures, it becomes easier for bonds to break and atoms to trade places and find more energetically favorable positions in the material, thus making the knife more crystalline and thus rigid.

Cubic Bismuth Crystals

Bismuth crystals after cooling.

Bismuth crystals

Bismuth Oxide

Colorful bismuth oxide patterns appear when bismuth is melted and re-cooled.

Bismuth Crystals in Pot

The remainder of the bismuth after the melting process.

Above are some of my efforts to implement Mr. Gray’s approach. The setup is exactly as Mr. Gray and other sources describe–—the bismuth melts at a very low temperature, much like its noxious cousin, lead, and so a steel pot and butane flame are all that are needed to get started. I checked to see how much of the bismuth had re-solidified by blowing on the surface—do not shake the pot, as this will disrupt the formation of larger cubic crystals. Once half or so of the pot has congealed, pour off the remaining liquid bismuth to reveal the crystal structures.

The gorgeous color comes from bismuth oxide, which forms on the surface of the metal almost instantly. I acquired my bismuth from United Nuclear, who have a variety of excellent reagents at reasonable prices. It comes in 5-10 gram pellets like this:

Bismuth pellets

Small pieces of elemental bismuth, purchased from a chemical supplier.

Bismuth can also be found in certain types of game shot (it is often used in lieu of lead)—for more information, see what Scitoys uses bismuth for.


Sodium Metal Isolation: Electrolysis of Sodium Hydroxide

The compound in lye, sodium hydroxide, is among the most easily-accessible corrosive chemicals— a wide range of products ranging from Drano to specialty batteries contain it, despite it being the sort of substance that will happily make short work of your fingertips if you are not careful.

The reason the sodium hydroxide is so unreasonably caustic is not because it is acidic, but rather because it is so anti-acidic, or basic. While acids do have their fair share of malevolent uses—most notably to melt Coke cans—they only represent half of the story. Acids corrode things like metal because they love to toss protons at things they encounter, allowing them to provoke strange reactions in a wide variety of compounds. But bases are guilty of just the opposite; they steal protons from anything they come in contact with. Different materials will be more susceptible to acids or bases depending on their properties: while a metal might readily dissolve when exposed to an acid(usually because the acid’s protons encourage the metal to oxidize extra quickly), it might remain relatively inert when a base is applied to it. Likewise, organic compounds like cellular membranes are particularly susceptible to bases, making bases the reagents of choice for products like drain cleaner, which are used to break up the greasy globs of proteins that inhabit your kitchen sink after dinner. The susceptibility of organic tissue to bases is part of the reason why bases can be so dangerous.

As demonstrated by the familiar vinegar and baking soda volcanoes that annually discolor the floors of school gymnasiums nationwide, reactions between acids and bases are particularly vigorous. This makes sense, since one reagent is eager to steal a proton and another is eager to relinquish one. The products of acid/base reactions tend to be inert gases and water, making them extremely useful in bodily metabolic processes like digesting food.

A chemistry teacher once told me that half of all chemistry is simply keeping track of electrical charges. Acid/base reactions certainly seem to fall in this half. Thus it would seem natural that there are certain types of reactions that will only occur in the presence of an electric field, as not all compounds have the proclivity to exchange protons and electrons as acids and bases. Electrolysis reactions represent those charge exchanges that will not occur spontaneously unless abetted by an external electric field. These reactions generally only occur between metals within a conducting solution(like saltwater), because metals and ionic solutions are generally the best at conducting the DC currents necessary to sustain such reactions. Electrolysis and its variants are used in industry to plate metals with other metals, to etch designs without the need for a laser, and occasionally to isolate pure elements that tend to not exist in pure forms in nature—because the electric field puts so much energy into the reaction vessel, it allows otherwise unstable elements to be isolated.

The latter usage is quite beneficial to the home chemist—there are many great tutorials out there for making pure hydrogen and oxygen gases by running a battery through salt water. The reason I mention this usage of electrolysis in the context of acid base chemistry is because, one fine day in high school, I decided that it would be a prudent idea to heat sodium hydroxide to its melting point and then perform electrolysis with a high-current power supply. I wanted to isolate pure sodium metal, which has a variety of useful properties, and for this reason I was oblivious to the risks of running high power electrical current through boiling hot caustic lye. Here is the result, complete with Hollywood production values:

Here is what was supposed to happen: By running a cheap tube from a cooler filled with dry ice, I could pipe cold carbon dioxide into the melting crucible and displace any oxygen inside the chamber. This would supposedly suppress any fires from starting and keep side reactions in check. Supposedly, if I had done this perfectly, the molten sodium hydroxide would have reacted happily with the DC current and liberated a small amount of molten elemental sodium at one of the electrodes, which would have formed a shiny metallic pool at the top of the molten base.

Sadly, I seem to have undervalued the awesome power of bases. The molten sodium hydroxide will react with almost anything it can find, and so it begins to react with the metal sides of my melting vessel to yield samples of an intricate group of molecules known as “complex ions.” These are the sickening black substances that accrue inside my crucible—the molten sodium hydroxide normally should have remained transparent. The carbon dioxide seems to have done its job, as I still have all of my fingers, but the reaction nonetheless failed to yield an appreciable amount of sodium due to the complex ions’ interloping.

You can hear some sodium metal hissing as I drop the cathode into the pool of water at the end of the video, leading me to believe that a small amount of sodium was indeed liberated (and yes, I did wait for the cathode to cool before I dropped it in). The hissing is the sound of the elemental sodium aggressively recombining with water to form sodium hydroxide and hydrogen gas, the latter of which loudly deflagrates due to the heat of the reaction.


A few weeks ago I tried to make nitrocellulose, with mixed results:

A large portion of common explosives— including TNT (trinitrotoluene) and dynamite (nitroglycerin)—have the key prefix “nitro” in them. This is not a coincidence; nitrogen groups, when added to benign organic compounds, tend to suddenly make them more reactive. The full chemistry behind this is complex, but a good thing to keep in mind is that nitrogen, when added to a compound consisting predominantly of carbon, has the net effect of destabilizing the electronic structure of the molecules. Carbon has four electrons, and so it likes to form four bonds with neighboring molecules—either in the form of four single bonds, two doubles, or a triple and a single. In the chemical theory of Lewis structures, this property is the basis of the “octet rule”: if we count each single bond as two electrons “paired together” with a covalent bond(with one electron coming from the carbon, and the other from whatever it bonds to), then carbon is content when it has eight electrons around it—four from itself, and four from the various things to which it bonds. Nitrogen, on the other hand, has five electrons when it is not bonded to any other atoms. But rather than always forming five bonds, nitrogen and most other atoms prefer to stick to the octet rule, and so two of nitrogen’s electrons form an internal bond within the molecule(called a lone pair), the members of which don’t bond to anyone outside the molecule. As a result, nitrogen usually ends up forming three single bonds or a single and a double bond, with its lone pair of electrons allowing it to still have eight total electrons. While there are some great exceptions to this covalent bonding scheme(much to the dismay of high schoolers everywhere), the Lewis octet rule for drawing covalent molecules provides great insight into the relative stability of chemical compounds.

While the Lewis structure rules give us excellent insight into how organic molecules bond, they don’t actually tell us what the molecule looks like at the atomic level. As quantum mechanics has predicted(and modern molecular imaging routines like NMR have verified), the structure of an organic molecule is heavily influenced by the addition strange atoms like nitrogen. Nitrogen has relatively high electronegativity, which means that it tends to attract electrons to itself. The derivation of why nitrogen and other gases(particularly diatomic gases like halogens, which tend to covalently bond in pairs to form compounds like O2 and N2) is rather complex(see Linus Pauling’s derivation, or consult Wikipedia’s summary of it), but it essentially involves the idea that certain atoms just accept electrons in bonds much more readily—the reasons involve nuclear shielding, dissociation energy, and a litany of other complex quantum effects. But the relevant result is that, even in large organic molecules the electrons tend to be slightly delocalized—they like to move into parts of the molecule beyond their bonding pair, where they traditionally would be expected to stay. The electronegative nitrogen atom tends to pull all of them very slightly towards itself, ensuring that every electron pair has slight asymmetry due to the pull of the nitrogen group. While this pull varies based in distance in the molecule, it has the net effect of destabilizing bonds throughout the molecule. This allows the organic molecule to fall apart much more easily, making the decomposition reaction explosively exothermic.

Thus nitration of organic compounds with nitric acid is a powerful way of making normal compounds become explosive. The reaction itself is often carried out in an ice bath with nitric acid, and it usually must be carefully monitored to prevent the nitrogen groups from substituting in for too many carbons in the compound(an effect known as “runaway nitration,” which either creates a more dangerous product or merely an inert product), as well as to ensure that the compound does not detonate prematurely. The above video shows the combustion of “guncotton,” which is the result of nitrating the cellulose polymers that comprise ordinary household cotton. For obvious reasons, this compound is known as “nitrocellulose.”

A key thing to notice is that the guncotton dissolves in acetone—ordinary cotton does not do this quite so readily. This is because the electronegative nitrogen succeeds in making parts of the cellulose molecules polar—because they attract some of the electrons towards themselves, the molecules are left with a positive and a negative end that was not present before they had nitrogen attachments. This polarity allows the highly polar acetone molecules to easily pick apart the polymer chains and dissolve the cotton. If the acetone is then allowed to evaporate, the polar molecules recombine into solid nitrocellulose in a manner similar to when sugar water evaporates and leaves behind rock candy.

Thermite Recipes

Copper Thermite

Copper(II) thermite. The spray is molten copper metal.

Here are my recipes for thermite, based upon trial and error and some actual science. Note that while these ratios are based on molar stoichiometry, I’ve made some adjustments in order to attain certain desirable qualities (ie, more aluminum makes a brighter and faster reaction, and often lowers the activation energy). The exotic thermites all have different cool properties, but generally I’ve found that iron and sulfer give the best reactions.

Amazing Rust has a lot of excellent information on this subject, particularly in regards to exotic thermites involving other metals.


I live near Siesta Key beach in Florida, which boasts some of the purest and finest quartz sand in the nation. I’d always hoped to take advantage of its purity to some day melt some into glass- a goal that complemented my experience with thermite. In 2009, while reading about soda glass, I stumbled upon a recipe that claimed to reduce the melting point of quartz by a significant amount. I knew that even the best thermite experimenters sometimes had trouble making glass, and so I figured treating the sand would be a valid way to guarantee fusion.

The recipe called for adding sodium carbonate to the sand and then mixing the two together very well(presumably until each particle of quartz was uniformly coated). I made the sodium carbonate by just heating up baking soda in an oven (Sodium Bicarbonate decomposes to sodium carbonate and carbon dioxide in heat). Once it was done, I mixed it in with the sand in about a 1:4 Carbonate:Silica ratio. The resulting mixture performed well, as evidenced by the excellent green lime glass affixed to the pot in the final photo of this video:

Iron(II,III) Oxide and Aluminum

Cupric Oxide:

Cuprous Oxide:

Stannic Oxide:

Silicon Dioxide:

This reaction is very, very cool. The idea is that you can take ordinary sand, mix it with aluminum and sulfer, and create a thermite reaction that liberates elemental silicon. My only qualm with the reaction is that the resulting silicon has nasty pockets of sulfer byproducts in it, and so it will create an ungodly smell unless you find some way to seal it off (it won’t dry out- humidity in the air reacts with the sulfer compounds to create smelly H2S and sulfuric acid, which corrodes the product. There are ways of making the thermite without using sulfur to boost the reaction, but I have never tried them and they sound generally unreliable. Nonetheless, the reaction is totally worth trying, especially for the guffaws of onlookers when the thermite flame turns bright blue:

Thermite reaction

One of my first projects in high school was creating thermite entirely from household ingredients. This started my interest in hands-on science, and prompted me to become interested in chemistry and physics instead of my then-intended career of teuthology.

Thermite is a specific case of a general class of aluminothermic reactions. The premise of the reaction is that finely ground iron oxide, in the presence of aluminum or magnesium powder, will undergo an exothermic single replacement reaction in which the iron is reduced and the aluminum is oxidized. Because aluminum oxide is far more stable than iron oxide, the reaction releases a lot of energy– enough that the iron produced is molten. The reaction has the general form:

Fe_2 O_3 + 2Al \rightarrow 2Fe + Al_2 O_3

The exact stoichiometry (ie, the correct coefficients and thus relative ratios of molecules) will vary depending on which oxide of iron is used.

There are many exciting (but altogether dubious) stories reported about applications of this reaction—people have reported using the reaction for all manners of vandalism, including melting through a car’s engine block (later proven possible on Mythbusters) to breaking through padlocks and deadbolts (later simulated in an early episode of Breaking Bad). I recall watching a clip of the British television programme “Brainiac” in which two somber men in lab coats gleefully use thermite to melt through a dilapidated car.

Besides household rust, high-purity iron oxide can be obtained by isolating naturally-occurring magnetite, which I was capable of acquiring at a beach near my house. At the suggestion of various internet sources (and to the bewilderment of the lifeguards) I started wandering the beach during the day, dragging a magnet behind me in the hopes of picking up chunks of magnetite from the sand. I gradually built up my supply of iron oxide over several months. In order to generate aluminum powder, in the evenings I would “borrow” a mortar and pestle and sit in front of the TV grinding aluminum foil.

By the fall of my freshman year, I had accrued enough of each reagent to fill a film canister with thermite (roughly 100 grams). At the time I lacked a proper analytical balance, and so I recall mixing the powders volumetrically using estimates of their powdered densities and packing fractions. Fortunately, thermite is fairly robust to imperfections in stoichiometry, and so the final mixture (roughly 2:1 by volume if I remember correctly) was viable.

Thermite is notoriously difficult to light—while the reaction is very exothermic, a very large initial heat (really, the activation energy) is required to convince the reactants to move into states in which they are willing to react. The two most common methods involve adding lit magnesium ribbon and covering a layer of potassium permanganate with glycerin. The latter method is favored by demonstrators because it creates the illusion of fire coming from water—however it is also riskier, since the reaction can have a delayed start, making it unclear whether a failed demonstration can be safely approached. I decided that the best bet was to buy some magnesium ribbon online, which led me to the amusing webpage of United Nuclear.

The first ten or so times I attempted the reaction in my driveway, it sputtered and fizzled. But I found more success when I cut shavings of Magnesium ribbon and placed a pile of them on top of the thermite, and then stuck a long Mg “fuse” into the mixture that extended down the full length of the reaction vessel (a clay flower pot). The idea was that, if the reaction failed to propagate down, the still-burning magnesium backbone would help re-ignite it (I later learned that this is similar to how some of those irritating no-blow birthday candles work—Magnesium powder in the wick prevents them from cooling below the flash point of the cotton).

I could tell the reaction had succeeded by the vicious hissing that began to emanate from the flower pot. Soon fire was starting to spit out over the rim, depositing small droplets of molten metal that slowly sunk into the asphalt of my driveway. By the time I returned, a large chunk of formerly-molten iron was glowing in the charred remnants of the pot.

I later dropped my souvenir piece of iron and found that a large bubble had formed inside it. Had the iron remained molten long enough for this bubble to reach this surface and pop, a substantial quantity of molten iron could have been thrown from the flower pot because the pot was undersized relative to the amount of thermite. Inside the chunk of iron was a nice set of crystals—a sign that the iron produced by the reaction was somewhat pure and that it had cooled slowly.

I later understood that it was a lucky decision to use magnetite for the reaction—rather than being a single oxide of iron, like red ferric(iron III) or black ferrous(iron II) oxide, magnetite is a crystalline complex of both, denoted by IUPAC as FeO FeO·Fe2O3 or iron(II,III) oxide. The crystal produces the hottest thermite of any type, yet it can be isolated from beach sand.

Some details of my reaction, as well as pictures and video, can be found here.

Paraffin Wax Fireball

Here’s an old video of water being poured on burning paraffin wax.

Paraffin wax is the wax used to make those cheap tea light candles you see at drug stores (the short, fat kind that sit in little aluminum holders). Tea lights are different from standard candles in that the entire body of the candle melts, and the wick simply serves to suck the liquid fuel up towards the fire until there is none left.

A huge advantage of this design is that there is no messy leftover wax—the wax itself is burnt as part of the candle. The reason for this is that paraffin is more chemically similar to gasoline than ordinary candle wax. Paraffin, in general, is what’s known as a saturated hydrocarbon. This essentially means that the molecules that comprise it consist of chains of conjoined carbon atoms with a huge amount of hydrogens bonded to the chain along the side. Carbon atoms like to form four bonds apiece, and so, as one can easily visualize, a given carbon in the middle of a chain will have two bonds to neighboring carbons, and two open bonds to which hydrogen atoms can attach. Carbon atoms on either end of the chain have three open bonds(because there is no “next” carbon in the chain), leaving there three sites to which hydrogen can attach.

Unlike hydrogen, however, carbon can actually form more than one bond to the same neighboring atom. In a hydrocarbon(again, a molecule consisting of just hydrogen and carbon), the carbons have the option of forming multiple bonds to each other rather than just one. For example, if a carbon bonds to a neighboring carbon in the chain twice, and then each member of the doubly bonded pair bonds once to another carbon in the chain, then each of the original carbons only has one open bonding site left for hydrogen to attach to. As you can imagine, there is a great number of different possible combinations of single, double, and even triple bonded chains of carbon atoms(with any carbon atom with fewer than four bonds to other members of the chain having the difference made up by hydrogen attachments). The ability of carbon to form a great number of similarly structured(yet chemically and qualitatively distinct) compounds is the basis of the field of organic chemistry.

Our paraffin wax belongs to a very special case of carbon compounds, however, known as alkanes. Alkanes have the maximum number of hydrogens per molecule, which a little intuition will tell you means that all carbons in the molecule are single-bonded to one another. In fact, if we recall that each of our singly-bonded carbons bonded to two neighbors(leaving two of its allotted 4 sites open to hydrogen atoms), and if we remember that the endcap carbons of our chain had three open slots for hydrogen, then it should be clear that the number of hydrogens in an alkane molecule with n carbon atoms is 2n + 2.

Many well-known fuels satisfy this property. For example, propane, a common gas for outdoor grills, has the molecular formula C3H8. In general, alkanes make excellent fuels because there is a lot of energy stored within their single bonds that can be released through burning:

Formally known as combustion, burning is the situation in which we cash in all the energy stored in these single bonds in order to get heat. Technically, we start with any hydrocarbon and oxygen, and after adding heat we get water and carbon dioxide as output molecules. The latter set of molecules has much weaker bonds than the original hydrocarbon, and so the bonds have less energy. The difference in energy is manifested in all the heat released when we burn hydrocarbons. However, among the various bonds in the hydrocarbon, the most energetic are those between carbon and hydrogen(rather than between two carbons), and so there is a definite advantage to burning fuels with a lot of hydrogen bonds(hence an alkane fuel would be preferable to one that contains double carbon bonds). This is why alkanes are referred to as “saturated”– they have as much hydrogen as will fit onto a chain of carbons, which also means that they will release the most energy when burnt of all possible carbon compounds with that many carbons. This use of “saturated” is the same as that you see on food labels as “saturated fats”– sat fats contain the most energy per mass that gets stored in your body as fat.

Our paraffin wax is an alkane hydrocarbon with something like 40 carbons, meaning that there is a lot of energy available to burn. But unlike smaller molecules, like propane, the awkwardly-big wax molecules very slightly attract each other(see ), causing the wax molecules to clump together into a solid instead of a looser gas at room temperature. Thus even though we may expect paraffin wax to burn more vigorously than, say, propane, we actually find that ridiculously long carbon chains run into structural problems that diminish their usefulness as fuels.

But when paraffin wax is allowed to boil, the fumes it gives off are still flammable hydrocarbons, but in a much more spry gaseous form. These fumes burn very fast, creating enough heat to cause the wax to continue to boil. If the process were allowed to continue indefinitely(as occurs in a standard tea light, albeit at a much lower temperature due to the catalyzing effect of the wick), then the container would just burn through all the wax.

But when water is thrown it, it flash boils because its boiling point is so much below that of the wax. The rapid gaseous expansion of the steam causes it to throw tiny amounts of molten wax everywhere, freeing the molecules from sticking together and slowing things down and instead creating a simulation of propane, which burns very quickly because its molecules are so free to move about. The resulting fireball is qualitatively similar to a gasoline or propane explosion.

N.B: An equivalent argument is that the water, by throwing the wax everywhere, rapidly increases its surface area and allows it to burn more easily. This is the same principle behind why it is easier to burn paper than logs of wood, despite their similar composition. Under this logic, a valid model of the fireball would be that of the suspended particulate explosion.