I’m starting this thread with the intention of trying to answer some of the questions that I’ve been asking myself about why and how crackle works.
As far as possible, I aim to base my answers on established chemical fact and/or the results of my own research. However, much of the evidence that I rely on is likely to be fairly indirect, so I will almost certainly have to resort to somewhat imaginative interpretations and – in cases where definitive evidence is lacking – to outright speculation.
Please feel free to respond, regardless of whether you do or don’t agree with my conclusions. In fact, I would consider it to be an ideal outcome if someone were to think ‘that’s not right’ and come up with a better answer.
I’d also be happy for people to ask their own questions and I’ll do my best to answer – even if all I can say is ‘I don’t know’.
Crackle Q & A
Moderators: richardh08, Boophoenix, Lloyd
-
richardh08
- Site Admin
- Posts: 2228
- Joined: Wed Aug 07, 2013 1:54 pm
- Location: Bedfordshire
Crackle Q & A
Even when I'm wrong, I'm convincing.
-
richardh08
- Site Admin
- Posts: 2228
- Joined: Wed Aug 07, 2013 1:54 pm
- Location: Bedfordshire
Re: Crackle Q & A
1. What is so special about copper, bismuth and lead oxides?
I’ve chosen a really big question to start with. It’s also one that my own research doesn’t (yet) address, so my answer will have to depend on a bit of chemistry and a lot of speculation.
Perhaps the first thing to note is the amount of energy that is released when the oxides react with magnesium, aluminium or a 50:50 alloy of the two. The values listed in Table 1 represent the energy, in kilojoules, that is released when one gram of each metal or alloy completely reacts with the relevant oxide. The table includes data for the oxides of antimony, manganese and tin, which have also been suggested as viable constituents of crackle. Table 1
It is clear that, of all these oxides, those of copper, lead and bismuth are the ones that are most capable of providing the energy to drive an explosive reaction.
But, for the reaction to start, the oxides have to release at least some of their oxygen so that it can react with the metal. It is therefore worth looking at how much energy is required to decompose the oxides. Table 2 lists the energy, again in kilojoules, that is required to release one gram of oxygen from the total decomposition of each of the oxides. It also contains their melting points, in kelvin. Table 2
Again, the oxides of copper, lead and bismuth head the list, in terms of how easily they release their oxygen (a fact not entirely unrelated to the values in Table 1). However, copper and lead oxides can release part of their oxygen even more easily. At a temperature of 773K, Pb3O4 partially decomposes:
2Pb3O4 -> 6PbO + O2,
which makes it a good low temperature source of oxygen. CuO also has the possibility of partially decomposing, by means of the following reaction:
4CuO -> 2Cu2O + O2.
Although this reaction needs a high temperature to run to completion, it is reported to take place to a noticeable extent at temperatures as low as around 600K. It is probably this ability to release oxygen at relatively low temperatures, combined with the high heat output from their thermite reactions that makes it possible for each of these oxides to be used as the only oxidiser, in combination with magnalium, aluminium or a combination of the two. The fact that the rate of propagation of the reaction in CuO-Al thermite is documented as being exceptionally fast probably helps in that case.
Lead oxide has one further property that makes it particularly suitable; after partially decomposing, the resulting PbO melts at a temperature of 1161K. Once the temperature passes this value, all the reactants will be liquid, enabling the reaction to progress much more rapidly until an explosion occurs. I suspect that this double advantage may well be the reason why the first crackle compositions to be discovered were based on lead oxide.
As far as I am aware, bismuth oxide is not susceptible to such a partial decomposition and is therefore unlikely to provide an effective low temperature source of oxygen. It is perhaps for this reason that it is not used as the only oxidiser, but is always combined with a proportion of copper oxide. However, its melting point, at 1090K, is very close to that of PbO and it seems likely that it will also be effective in enhancing the reaction rate once that temperature has been reached and all the reactants are in the liquid state.
What about the other oxides listed in Tables 1 and 2? I know of crackle formulations that contain antimony oxide, but none for which it is the only oxide present. In my experience, many of them appear to produce a softer-sounding report than is obtainable with copper, lead or bismuth oxides. I have seen MnO2 and SnO2 mentioned in patent documents but I am not aware of any specific formulation that includes either of them. I have tried both of them in crackle-style compositions and have found that they tend to burn without producing an explosion.
Although Sb2O3 is harder to decompose and doesn’t generate as much energy, its low melting point ought to make it, like bismuth oxide, an enhancer of the reaction rate at high temperatures. But maybe the melting point is just a bit too low. A general rule of thumb is that the rate of a reaction approximately doubles for every ten degrees rise in temperature. With a melting point some 160 degrees lower than bismuth trioxide, the reaction rate could be 2**16 = 65536 times slower when the antimony oxide melts. Perhaps, at this temperature, the reaction is proceeding so slowly that, even with the enhancement produced by melting the oxide, it doesn’t quite manage to reach the rate needed to produce an explosion. I accept that, as stated, this argument is extremely naïve, but I suspect that it may contain an element of truth.
Compared with antimony oxide, manganese dioxide has a smaller energy output and an even lower melting point so, provided the above argument is correct, it is likely to be even less effective in crackle compositions. Tin oxide generates even less energy and, with a melting point that is higher than that of copper oxide, is unlikely to have any comparable enhancing effect on the reaction rate.
I’ve chosen a really big question to start with. It’s also one that my own research doesn’t (yet) address, so my answer will have to depend on a bit of chemistry and a lot of speculation.
Perhaps the first thing to note is the amount of energy that is released when the oxides react with magnesium, aluminium or a 50:50 alloy of the two. The values listed in Table 1 represent the energy, in kilojoules, that is released when one gram of each metal or alloy completely reacts with the relevant oxide. The table includes data for the oxides of antimony, manganese and tin, which have also been suggested as viable constituents of crackle. Table 1
It is clear that, of all these oxides, those of copper, lead and bismuth are the ones that are most capable of providing the energy to drive an explosive reaction.
But, for the reaction to start, the oxides have to release at least some of their oxygen so that it can react with the metal. It is therefore worth looking at how much energy is required to decompose the oxides. Table 2 lists the energy, again in kilojoules, that is required to release one gram of oxygen from the total decomposition of each of the oxides. It also contains their melting points, in kelvin. Table 2
Again, the oxides of copper, lead and bismuth head the list, in terms of how easily they release their oxygen (a fact not entirely unrelated to the values in Table 1). However, copper and lead oxides can release part of their oxygen even more easily. At a temperature of 773K, Pb3O4 partially decomposes:
2Pb3O4 -> 6PbO + O2,
which makes it a good low temperature source of oxygen. CuO also has the possibility of partially decomposing, by means of the following reaction:
4CuO -> 2Cu2O + O2.
Although this reaction needs a high temperature to run to completion, it is reported to take place to a noticeable extent at temperatures as low as around 600K. It is probably this ability to release oxygen at relatively low temperatures, combined with the high heat output from their thermite reactions that makes it possible for each of these oxides to be used as the only oxidiser, in combination with magnalium, aluminium or a combination of the two. The fact that the rate of propagation of the reaction in CuO-Al thermite is documented as being exceptionally fast probably helps in that case.
Lead oxide has one further property that makes it particularly suitable; after partially decomposing, the resulting PbO melts at a temperature of 1161K. Once the temperature passes this value, all the reactants will be liquid, enabling the reaction to progress much more rapidly until an explosion occurs. I suspect that this double advantage may well be the reason why the first crackle compositions to be discovered were based on lead oxide.
As far as I am aware, bismuth oxide is not susceptible to such a partial decomposition and is therefore unlikely to provide an effective low temperature source of oxygen. It is perhaps for this reason that it is not used as the only oxidiser, but is always combined with a proportion of copper oxide. However, its melting point, at 1090K, is very close to that of PbO and it seems likely that it will also be effective in enhancing the reaction rate once that temperature has been reached and all the reactants are in the liquid state.
What about the other oxides listed in Tables 1 and 2? I know of crackle formulations that contain antimony oxide, but none for which it is the only oxide present. In my experience, many of them appear to produce a softer-sounding report than is obtainable with copper, lead or bismuth oxides. I have seen MnO2 and SnO2 mentioned in patent documents but I am not aware of any specific formulation that includes either of them. I have tried both of them in crackle-style compositions and have found that they tend to burn without producing an explosion.
Although Sb2O3 is harder to decompose and doesn’t generate as much energy, its low melting point ought to make it, like bismuth oxide, an enhancer of the reaction rate at high temperatures. But maybe the melting point is just a bit too low. A general rule of thumb is that the rate of a reaction approximately doubles for every ten degrees rise in temperature. With a melting point some 160 degrees lower than bismuth trioxide, the reaction rate could be 2**16 = 65536 times slower when the antimony oxide melts. Perhaps, at this temperature, the reaction is proceeding so slowly that, even with the enhancement produced by melting the oxide, it doesn’t quite manage to reach the rate needed to produce an explosion. I accept that, as stated, this argument is extremely naïve, but I suspect that it may contain an element of truth.
Compared with antimony oxide, manganese dioxide has a smaller energy output and an even lower melting point so, provided the above argument is correct, it is likely to be even less effective in crackle compositions. Tin oxide generates even less energy and, with a melting point that is higher than that of copper oxide, is unlikely to have any comparable enhancing effect on the reaction rate.
You do not have the required permissions to view the files attached to this post.
Even when I'm wrong, I'm convincing.
-
richardh08
- Site Admin
- Posts: 2228
- Joined: Wed Aug 07, 2013 1:54 pm
- Location: Bedfordshire
Re: Crackle Q & A
2. Why must magnesium be alloyed with aluminium?
Regardless of whether you do or don’t believe that crackle’s ‘dark’ reaction is due to the preferential oxidation of magnesium, the fact remains that an alloy of magnesium and aluminium produces the crackle effect, whereas a straight mixture of the two metals – even if in the same proportions – does not. When ignited, such a mixture just burns and never reaches an explosive phase. I suspect that the reason why this is the case might be revealed by careful consideration of the properties of the two elements.
Magnesium and aluminium are both highly reactive metals. Such are the wonders of chemistry that, although the reaction of one gram of aluminium will generate more heat than the equivalent reaction with one gram of magnesium, magnesium is the more reactive of the two. But the difference is not great; the reaction between magnesium and aluminium oxide – to yield magnesium oxide and metallic aluminium takes place relatively slowly – and produces less than 8% of the heat that is generated by either of them when they react with copper oxide.
As an aside, I might point out that these facts support the notion that preferential oxidation of magnesium is not a dominant mechanism. At any given temperature, the reactions between either magnesium or aluminium and, say, copper oxide would be expected to progress at a much greater rate than the reaction between magnesium and aluminium oxide. In my opinion, in the short time between ignition and explosion of a crackle grain, it is unlikely that the latter reaction would have the chance to take place to any significant degree. This contention might be able to be proved or disproved by someone who knows more about reaction kinetics than I do.
Both metals are sufficiently reactive to be oxidised by atmospheric oxygen. Indeed, if magnesium is exposed to moist air it will fairly quickly be covered with a layer of magnesium oxide. If left long enough, the reaction will continue until all the magnesium is consumed. However, if the same experiment is performed with aluminium, it appears to be unaffected. The difference in behaviour is due to the unique properties of aluminium oxide; the aluminium is initially attacked, but the molecules of aluminium oxide that are formed have exactly the right size and shape to form a thin, transparent, but continuous layer over the surface of the metal, protecting it from further attack. If the layer of oxide is disrupted (one way is to rub a little mercury onto the surface) the oxidation continues at a rapid rate. In contrast, magnesium oxide does not form such a continuous, protective layer.
I don’t know what happens at the surface of a magnesium-aluminium alloy, but it does appear to be a fact that such an alloy is less susceptible to surface oxidation than is pure magnesium. It might be that a surface layer of aluminium oxide is able to provide some measure of protection to both metals, or that a mixture of magnesium and aluminium oxides has some of aluminium oxide’s protective properties. In any case, it seems likely that the formation of aluminium oxide is a significant factor.
It is tempting to propose that the ‘dark’ phase of the crackle reaction is mitigated by the presence of aluminium oxide. In the initial stages, when all the constituents are in the solid state, the reaction can only occur at particle interfaces, with partial decomposition of the oxide(s) allowing oxygen to diffuse towards the metal(s). A growing intervening layer of aluminium oxide would provide a barrier to such diffusion, preventing the reaction from progressing so easily. The same reasoning might apply to the increase in the amount of magnesium oxide, but the above discussion suggests that it is likely to be a less effective barrier.
This proposal is consistent with the ineffectiveness of a mixture of elemental magnesium and aluminium, compared with their alloy. If magnesium oxide is ineffective as a barrier, then the magnesium would react vigorously during the early stages and the conditions necessary to create an explosion might never occur.
Further supporting evidence for this proposal is provided by including cryolite (sodium hexafluoroaluminate, Na3AlF6) in a crackle composition. When added, even in very small amounts (say, around 1%) it totally destroys the crackle effect; the behaviour becomes very similar to what occurs when the mixture contains elemental magnesium. Molten cryolite is a very good solvent for aluminium oxide – hence its use in the electrolytic production of aluminium. The observed effect can be explained if the cryolite breaks down the aluminium oxide barrier, allowing both the magnesium and aluminium to react more freely at relatively low temperatures.
Regardless of whether you do or don’t believe that crackle’s ‘dark’ reaction is due to the preferential oxidation of magnesium, the fact remains that an alloy of magnesium and aluminium produces the crackle effect, whereas a straight mixture of the two metals – even if in the same proportions – does not. When ignited, such a mixture just burns and never reaches an explosive phase. I suspect that the reason why this is the case might be revealed by careful consideration of the properties of the two elements.
Magnesium and aluminium are both highly reactive metals. Such are the wonders of chemistry that, although the reaction of one gram of aluminium will generate more heat than the equivalent reaction with one gram of magnesium, magnesium is the more reactive of the two. But the difference is not great; the reaction between magnesium and aluminium oxide – to yield magnesium oxide and metallic aluminium takes place relatively slowly – and produces less than 8% of the heat that is generated by either of them when they react with copper oxide.
As an aside, I might point out that these facts support the notion that preferential oxidation of magnesium is not a dominant mechanism. At any given temperature, the reactions between either magnesium or aluminium and, say, copper oxide would be expected to progress at a much greater rate than the reaction between magnesium and aluminium oxide. In my opinion, in the short time between ignition and explosion of a crackle grain, it is unlikely that the latter reaction would have the chance to take place to any significant degree. This contention might be able to be proved or disproved by someone who knows more about reaction kinetics than I do.
Both metals are sufficiently reactive to be oxidised by atmospheric oxygen. Indeed, if magnesium is exposed to moist air it will fairly quickly be covered with a layer of magnesium oxide. If left long enough, the reaction will continue until all the magnesium is consumed. However, if the same experiment is performed with aluminium, it appears to be unaffected. The difference in behaviour is due to the unique properties of aluminium oxide; the aluminium is initially attacked, but the molecules of aluminium oxide that are formed have exactly the right size and shape to form a thin, transparent, but continuous layer over the surface of the metal, protecting it from further attack. If the layer of oxide is disrupted (one way is to rub a little mercury onto the surface) the oxidation continues at a rapid rate. In contrast, magnesium oxide does not form such a continuous, protective layer.
I don’t know what happens at the surface of a magnesium-aluminium alloy, but it does appear to be a fact that such an alloy is less susceptible to surface oxidation than is pure magnesium. It might be that a surface layer of aluminium oxide is able to provide some measure of protection to both metals, or that a mixture of magnesium and aluminium oxides has some of aluminium oxide’s protective properties. In any case, it seems likely that the formation of aluminium oxide is a significant factor.
It is tempting to propose that the ‘dark’ phase of the crackle reaction is mitigated by the presence of aluminium oxide. In the initial stages, when all the constituents are in the solid state, the reaction can only occur at particle interfaces, with partial decomposition of the oxide(s) allowing oxygen to diffuse towards the metal(s). A growing intervening layer of aluminium oxide would provide a barrier to such diffusion, preventing the reaction from progressing so easily. The same reasoning might apply to the increase in the amount of magnesium oxide, but the above discussion suggests that it is likely to be a less effective barrier.
This proposal is consistent with the ineffectiveness of a mixture of elemental magnesium and aluminium, compared with their alloy. If magnesium oxide is ineffective as a barrier, then the magnesium would react vigorously during the early stages and the conditions necessary to create an explosion might never occur.
Further supporting evidence for this proposal is provided by including cryolite (sodium hexafluoroaluminate, Na3AlF6) in a crackle composition. When added, even in very small amounts (say, around 1%) it totally destroys the crackle effect; the behaviour becomes very similar to what occurs when the mixture contains elemental magnesium. Molten cryolite is a very good solvent for aluminium oxide – hence its use in the electrolytic production of aluminium. The observed effect can be explained if the cryolite breaks down the aluminium oxide barrier, allowing both the magnesium and aluminium to react more freely at relatively low temperatures.
Even when I'm wrong, I'm convincing.
-
richardh08
- Site Admin
- Posts: 2228
- Joined: Wed Aug 07, 2013 1:54 pm
- Location: Bedfordshire
Re: Crackle Q & A
3. Is it essential that magnesium be present?
The short answer is no, but…
I have created a number of experimental formulations that produce the crackle effect, despite containing aluminium as the only metal. In addition, I have found that Lloyd’s crackle produces equally loud reports if the magnalium is replaced by 350 mesh atomised aluminium (so that the total metal content is 15% Al).
But … they all are significantly harder to ignite and exhibit much longer and more variable delays. Although they work, I can’t see many (if any) of them becoming viable alternatives to some of the more popular formulations. For me, the chief interest is what information such compositions give about how crackle works. As of this moment, that is still a work in progress.
The short answer is no, but…
I have created a number of experimental formulations that produce the crackle effect, despite containing aluminium as the only metal. In addition, I have found that Lloyd’s crackle produces equally loud reports if the magnalium is replaced by 350 mesh atomised aluminium (so that the total metal content is 15% Al).
But … they all are significantly harder to ignite and exhibit much longer and more variable delays. Although they work, I can’t see many (if any) of them becoming viable alternatives to some of the more popular formulations. For me, the chief interest is what information such compositions give about how crackle works. As of this moment, that is still a work in progress.
Even when I'm wrong, I'm convincing.
-
richardh08
- Site Admin
- Posts: 2228
- Joined: Wed Aug 07, 2013 1:54 pm
- Location: Bedfordshire
Re: Crackle Q & A
4. Why use a nitrocellulose binder?
The vast majority of published crackle compositions use nitrocellulose (or its close cousin, double base powder) as the binder. Most of the authors recommend the addition of around 10%, and several have expressed the opinion that the more you add, the better the effect.
On the other hand, Lloyd Sponenburgh put a lot of effort into engineering a composition that had a reliably short delay, produced a loud report and was as simple as possible to manufacture. His composition is probably the best that I know of, yet it only contains about 2.5% (double base) binder. As far as I am aware, Mike Swisher was the first person outside China to create a lead oxide-magnalium crackle composition which, according to Shimizu in his Pyrotechnica XIII article, could be bound with either nitrocellulose or glutinous rice starch.
In my own research, at the time when I was looking to simplify the composition as much as possible, I experimented with many different binders. I eventually settled on phenolic resin, and now use it as standard in a wide range of mixes.
It seems clear that nitrocellulose is not an essential ingredient in many crackle compositions, so why do people continue to use it?
One clue to its possible significance comes from my experiments with Lloyd’s crackle, when I used phenolic resin instead of nitrocellulose (here in the UK, without a licence to possess a gun, I don’t have access to double base powder). The resulting grains were very slightly harder to ignite, but noticeably different in performance, with a strong tendency to produce multiple reports instead of a single loud explosion. An obvious reason for this change in behaviour is that the reaction proceeds at different rates in different regions which, in turn, strongly suggests that there are significant temperature variations from point to point within the grain.
That is, perhaps, not surprising; Lloyd’s composition contains a high proportion (71%) of bismuth trioxide, which is a relatively poor conductor of heat. I have run a computer model of the physical conditions inside grains with a variety of compositions and find that there are significantly greater variations in temperature in bismuth-rich mixtures than in ones where the principal (or only) oxide is CuO – which is a much better conductor of heat. In this respect, I believe it is significant that I don’t see the same binder-dependent differences in behaviour in my own experimental compositions, which generally contain high percentages of copper oxide.
I’m led to the conclusion that, at least in Lloyd’s crackle, the key function of the NC-based binder is to ensure smooth ignition throughout the grain, thereby minimising the temperature differences from one point to another.
I used to think that nitrocellulose achieved this aim by speeding up the progress of the flame front through the grain. To test this hypothesis I made a series of compositions that were identical except for the amount of nitrocellulose. I formed each into long thin grains and videoed their burning, when ignited at one end. I was very surprised to find that the ignition front progressed more slowly through the grains containing a higher percentage of nitrocellulose.
The only explanation that I can think of for why this behaviour might help produce a single, loud report is that the decomposition of the NC prevents the temperature from rising too high in any one locality until the whole grain is ignited. That fits with a suspicion that is starting to form in my mind, based on the subtle details that might be emerging from my research, that a necessary condition for the production of a loud report could be the existence of one – or perhaps (even preferably?) more than one – mechanism that inhibits, or slows down, the progress of the reaction at low to intermediate temperatures. Maybe the NC is providing one of those mechanisms.
Perhaps that is why – at least for some compositions – the correct rule for how much NC to use might well be ‘more is better’.
The vast majority of published crackle compositions use nitrocellulose (or its close cousin, double base powder) as the binder. Most of the authors recommend the addition of around 10%, and several have expressed the opinion that the more you add, the better the effect.
On the other hand, Lloyd Sponenburgh put a lot of effort into engineering a composition that had a reliably short delay, produced a loud report and was as simple as possible to manufacture. His composition is probably the best that I know of, yet it only contains about 2.5% (double base) binder. As far as I am aware, Mike Swisher was the first person outside China to create a lead oxide-magnalium crackle composition which, according to Shimizu in his Pyrotechnica XIII article, could be bound with either nitrocellulose or glutinous rice starch.
In my own research, at the time when I was looking to simplify the composition as much as possible, I experimented with many different binders. I eventually settled on phenolic resin, and now use it as standard in a wide range of mixes.
It seems clear that nitrocellulose is not an essential ingredient in many crackle compositions, so why do people continue to use it?
One clue to its possible significance comes from my experiments with Lloyd’s crackle, when I used phenolic resin instead of nitrocellulose (here in the UK, without a licence to possess a gun, I don’t have access to double base powder). The resulting grains were very slightly harder to ignite, but noticeably different in performance, with a strong tendency to produce multiple reports instead of a single loud explosion. An obvious reason for this change in behaviour is that the reaction proceeds at different rates in different regions which, in turn, strongly suggests that there are significant temperature variations from point to point within the grain.
That is, perhaps, not surprising; Lloyd’s composition contains a high proportion (71%) of bismuth trioxide, which is a relatively poor conductor of heat. I have run a computer model of the physical conditions inside grains with a variety of compositions and find that there are significantly greater variations in temperature in bismuth-rich mixtures than in ones where the principal (or only) oxide is CuO – which is a much better conductor of heat. In this respect, I believe it is significant that I don’t see the same binder-dependent differences in behaviour in my own experimental compositions, which generally contain high percentages of copper oxide.
I’m led to the conclusion that, at least in Lloyd’s crackle, the key function of the NC-based binder is to ensure smooth ignition throughout the grain, thereby minimising the temperature differences from one point to another.
I used to think that nitrocellulose achieved this aim by speeding up the progress of the flame front through the grain. To test this hypothesis I made a series of compositions that were identical except for the amount of nitrocellulose. I formed each into long thin grains and videoed their burning, when ignited at one end. I was very surprised to find that the ignition front progressed more slowly through the grains containing a higher percentage of nitrocellulose.
The only explanation that I can think of for why this behaviour might help produce a single, loud report is that the decomposition of the NC prevents the temperature from rising too high in any one locality until the whole grain is ignited. That fits with a suspicion that is starting to form in my mind, based on the subtle details that might be emerging from my research, that a necessary condition for the production of a loud report could be the existence of one – or perhaps (even preferably?) more than one – mechanism that inhibits, or slows down, the progress of the reaction at low to intermediate temperatures. Maybe the NC is providing one of those mechanisms.
Perhaps that is why – at least for some compositions – the correct rule for how much NC to use might well be ‘more is better’.
Even when I'm wrong, I'm convincing.