Sciencemadness Discussion Board - Preparation of elemental phosphorus

Quote: Originally posted by Myfanwy

could phosphine and hydrogen peroxide yield some P4?

2PH3 + 3H2O2 -> 6H2O + 1/2 P4

Possibly but why go to the trouble of making a highly toxic, highly inflammable gas when you can make P from Calgon tablets, some fine sand and some Al powder?

Sorry to bring up an old post, but many people seem to make phosphine inadvertently, as a result of accidental production of phosphide sideproducts.

Yes I know phosphine is notorious, but what if we tried to make a phosphide on purpose and then convert the phosphine into P4?

Does phosphorus react with H2O2?

Rogeryermaw - 24-5-2012 at 08:11

@ magpie: beautiful work sir! glad you were able to recover the fitting. i wonder that using a gas for positive pressure may have contributed to the early cooling of the P4 gas inside the glassware? is it possible to inject the gas from the other side of the heat source so that your yield stays hot until it reaches the water? the general form of a tube furnace suggests much difficulty in the proposition... what alterations do you have in mind for future runs?

@bromic acid: your effort in compiling and efficient pruning of this thread will be a great help to anyone new who may wish to take up the torch and run. it is visually appealing as well. excellent work!

Magpie - 24-5-2012 at 13:01

@ magpie: beautiful work sir! glad you were able to recover the fitting. i wonder that using a gas for positive pressure may have contributed to the early cooling of the P4 gas inside the glassware? is it possible to inject the gas from the other side of the heat source so that your yield stays hot until it reaches the water? the general form of a tube furnace suggests much difficulty in the proposition... what alterations do you have in mind for future runs?

Thank you. Injecting the argon from the back end would be possible but would take some extra work and expense. When the retort is in the green stage a mullite tube, such as a thermocouple sheath, could conceivably be attached to the back end. But I'm skeptical that this would be of much help. Also the argon would then have to move through or around the molten slag. This might or might not be a problem. Although from my picture it seems as though there is a lot of P in the tailpipe, it is just a sublimed film so I don't think it amounts to much P.

I'm working on a prototype of a 12" air condenser that would extend down to the water receiver. I'm expecting that the P will condense in the tube and freeze. But the tube is vertical so I can keep it moving down to the receiver using a heat gun.

I spent some time reviewing your excellent work, specifically looking for indications of yield. The numbers I found indicate that your typical yield was about 6.5g. This was on a charge of 102g of NaPO3 which would give a theoretical amount of P = 31g. So your yield was ~ 21%. Please correct me if my figures are wrong.

I'm hopeful that as we research this method that we can find ways to increase the yield as there is obviously a lot of P not being captured.

Rogeryermaw - 24-5-2012 at 19:11

your figure seems about correct. i have been thinking a lot about this lately and i wonder how much the yield of this sort of method can really be increased. is it really possible to stop or even reduce the formation of PH3? when i have done this at night i have also noticed (the setup has a glass receiver with a loose fitting lid) a bluish-white glow around the area where the pipe passes the lid which indicates a loss of P4 to inefficient cooling caused by fear of suck-back related catastrophe (shallow immersion giving the extremely hot P4 gas little time to cool to the condensation point). there are numerous possibilities to consider . all in all, considering the state of the world, i'm happy to be able to experiment with phosphorus at all, even if procuring it is, as yet, inefficient.

AndersHoveland - 24-5-2012 at 19:15

The yield of white P in the water receiver was very small, ie, less than 1 g, and this had to be melted with a heat gun to get it to drain into the receiver.

That could be because phosphorous, in a finely divided state, might actually react with water.
http://www.sciencemadness.org/talk/viewthread.php?tid=20064

This could potentially explain your low yields passing hot phosphorous vapor to condense in water.

[Edited on 25-5-2012 by AndersHoveland]

Magpie - 24-5-2012 at 20:33

...is it really possible to stop or even reduce the formation of PH3?

The formation of PH3 requires the presence of hydrogen. None of the reactants contain this. So if we can get the bulk of the P4 condensed to a liquid before it hits the water there should be little phosphine produced.

...all in all, considering the state of the world, i'm happy to be able to experiment with phosphorus at all, even if procuring it is, as yet, inefficient.

Amen.

AndersHoveland - 24-5-2012 at 23:34

I really do not know, but I do not think that elemental phosphorous readily reacts nitrogen directly. There seems to be very little information about this available.

http://www.sciencemadness.org/talk/viewthread.php?tid=20208#...

[Edited on 25-5-2012 by AndersHoveland]

Rogeryermaw - 26-5-2012 at 12:37

yes specific information is hard to find but there are some syntheses using elemental phosphorus that recommend nitrogen as an inert atmosphere so i rather doubt they react directly.

Zan Divine - 28-5-2012 at 07:13

We know, from existing literature, that P & N probably don't react at these low temperatures. We also know that leading hot P4 vapors into H2O is not generally problematic in that it's been done that way for a long time and still is.

I'll be the first to admit that when all factors are considered, it's way, way easier said than done, but the mass change of the reactor/tailpiece (once P is melted out) could be hugely indicative of the nature of the problem. For instance, you may have to pre-bake the reactor, reactants, & luting to get a reliable starting weight.

I'm not sure I'd automatically assume that you released all the missing P as PH3 or as P that oxidized to P4O10. You'd have really noticed a lot of smoke and/or stench. I'm more inclined to worry about incorporation into the slag as red P or possible incomplete reaction due to some reactant segregation.

A long standing practice in leading hot vapors into water is to use a wide tube to avoid suck-back of water into the reactor. You want the tube to be a significant volume compared to the volume of the reactor. Your new vertical tailpiece may be very appropriate if wide enough.

I've been handling a lot of molten P recently, and the thing that amazes me is the simplicity of operations and ease of handling.
I was a little surprised to find that 1 or 2 gram samples of commercial white P can be placed on a room temperature surface outside and they just sit there, emiting an occasional wisp of smoke and liquify and after 5 minutes will not ignite.

It's time to end some of the widely held ideas that P is a fearsome thing too dangerous to play with. It's really this simple: Keep it underwater, work in a ventilated area. You can easily move it through air with forcepts to its next watery location. No fire. I was amazed to find that this dragon was, in reality, a docile pet when just the simplest precautions are taken.

This doesn't take away from the inherent level of risk you faced with P vapor @ >500 C. That may well be the dragon.

When I lit it, it burned fiercely with mountains of the oxide billowing away. I'd guess that a gram of P being liberated as this was would totally fill a room with thick smoke. Your aerosol size many vary, of course.

You really took some time with your work & presentation, Magpie, and your work has all the hallmarks of a career scientist.

[Edited on 28-5-2012 by Zan Divine]

Magpie - 28-5-2012 at 09:15

... We also know that leading hot P4 vapors into H2O is not generally problematic in that it's been done that way for a long time and still is.

Agreed.

I'll be the first to admit that when all factors are considered, it's way, way easier said than done, but the mass change of the reactor/tailpiece (once P is melted out) could be hugely indicative of the nature of the problem. For instance, you may have to pre-bake the reactor, reactants, & luting to get a reliable starting weight.

Agreed. I plan to do just that.

I'm not sure I'd automatically assume that you released all the missing P as PH3 or as P that oxidized to P4O10. You'd have really noticed a lot of smoke and/or stench. I'm more inclined to worry about incorporation into the slag as red P or possible incomplete reaction due to some reactant segregation.

Agreed. In reviewing some of Rogeryermaw's work it is apparent that a relatively small amount (I'm assuming <6g) of burning P emits a huge amount of white smoke. See his video where he burns P on top of a wood post on his patio.

A long standing practice in leading hot vapors into water is to use a wide tube to avoid suck-back of water into the reactor. You want the tube to be a significant volume compared to the volume of the reactor. Your new vertical tailpiece may be very appropriate if wide enough.

It's not very wide and I'm hoping that this will not become a problem. It may be an Achilles heel, unfortunately - not from a suckback standpoint as I will be using an argon purge, but from a plugging standpoint.

I've been handling a lot of molten P recently, and the thing that amazes me is the simplicity of operations and ease of handling.
I was a little surprised to find that 1 or 2 gram samples of commercial white P can be placed on a room temperature surface outside and they just sit there, emiting an occasional wisp of smoke and liquify and after 5 minutes will not ignite.

My observation, also.

It's time to end some of the widely held ideas that P is a fearsome thing too dangerous to play with. It's really this simple: Keep it underwater, work in a ventilated area. You can easily move it through air with forcepts to its next watery location. No fire. I was amazed to find that this dragon was, in reality, a docile pet when just the simplest precautions are taken.

Agreed.

This doesn't take away from the inherent level of risk when you faced with P vapor @ >500 C. That may well be the dragon.

Most assuredly.

Yes - see Rogermyermaw's video.

You really took some time with your work & presentation, Magpie, and your work has all the hallmarks of a career scientist.

Thank you. Unfortunately the career scientist thing is all behind me now.

It's good to have this body of work critiqued by a fresh set of experienced eyes.

[Edited on 28-5-2012 by Magpie]

[Edited on 29-5-2012 by Magpie]

Rogeryermaw - 28-5-2012 at 20:31

many of your observations are spot on. a couple of points if i remember right...the molar mass of phosphorus is close to 31 grams. i based my experiments on 1 mole of sodium hexametaphosphate. when i measured, i had a rough starting mass of 217 grams with all reactants combined (after drying in oven) and a post run weight of 199.8 grams. even with a conservative estimate (owing some loss to the brittle nature of the slag) the reactant mass lost in the neighborhood of 15-17 grams. with the collection of 6+ grams of P4, that is an anus load of product to account for. the weight is not there so we can rule out P4 being trapped in the reaction mass. i know for a fact that some burns away from performing this experiment at night. there is no mistaking the ghostly glow of phosphorus vapor. it is possible that we don't see as much of the P4O10 as we should think; perhaps it is absorbed into the receiver water forming phosphoric acid. a post run pH test of the water may confirm this but, even if positive, this does not explain all the losses. the thought of the mass containing some red phosphorus had not occurred to me and certainly bears investigation but i have a feeling that the answer to this will only be solved through trial and error. we will chip away little by little at the imperfections of this process until the yields increase and then there will be some sore foreheads after the mass *facepalm* of realization.

only one thing bugs me here. please do not get comfortable with the safety of phosphorus. she is a wicked, backstabbing bitch that will only bite you when you stop looking. the reason that it sometimes fails to auto-ignite is that even the slightest breeze can carry enough heat away from it to slow the process. if it is sunny out and the wind stops, it will go eventually. don't handle it with your hands too much, even in gloves. not only can your body heat do the job, a residue can rub off on your hands that may contaminate other things. the lowest known lethal dose of P4 is in the neighborhood of 15mg (i know the LD50 is closer to 50 mg). mostly, however, you are correct. with the safeties in place, phosphorus has been a pussycat to handle. i just don't want the young bucks to get it in their heads that the dangers are imaginary.

for those interested, some health and safety info can be found here: www.atsdr.cdc.gov/toxprofiles/tp103-c2.pdf

[Edited on 29-5-2012 by Rogeryermaw]

White Yeti - 29-5-2012 at 09:25

Has anyone tried powdered iron as a reducing agent? Perhaps the lower melting point of its oxide as compared to those of aluminium and magnesium may allow for better mixing and a more complete reaction.

Rogeryermaw - 29-5-2012 at 13:17

iron readily forms phosphides with P4 in high heat.

that is actually involved in one of the industrial methods to produce P4 (well sort of. the iron is from impurities and undesirable). iron in the crucible will form iron phosphide from which PH3 is produced and then burned away to yield P4. however, the handling of that much PH3 is too scary for me.

[Edited on 29-5-2012 by Rogeryermaw]

garage chemist - 30-5-2012 at 11:03

Excellent work, Magpie.
I also have a selfmade ceramic retort at home, but haven't been willing to use it for this synthesis because I knew that it would be a single-use item, with no possibility of reuse because of the hard glassy slag that cannot be removed from the retort.
With the amount of work and time that goes into making a ceramic retort, I don't think of this as a good way to make phosphorus, though at least it DOES work, which sets it apart from so many of the theoretical methods posted here.
I also haven't been able to attach a glass tube to my retort, which you were able to do.

[Edited on 30-5-2012 by garage chemist]

Magpie - 30-5-2012 at 12:26

Quote: Originally posted by garage chemist

Excellent work, Magpie... With the amount of work and time that goes into making a ceramic retort, I don't think of this as a good way to make phosphorus, though at least it DOES work, which sets it apart from so many of the theoretical methods posted here.

Thank you. Yes, it does work, but still needs a lot of development.

The tube materials are very cheap (<$1) but the time required is considerable. Also, one must have a 1300C furnace for firing, as you know. It takes me about a week, working on and off, to make one. At least now I have a sure-fire procedure and don't get any failures. The last tube I made was #21.

Quote: Originally posted by garage chemist

I also haven't been able to attach a glass tube to my retort, which you were able to do.

Yes, I tried a lot of formulations before finding one that works. Here's the formula:

*10g pottery grade ZrSiO4
*17 drops "as received" pottery grade waterglass
*add just enough water to get good fluidity

I make mine in a small beaker using a stirring rod.

The bond can then be destroyed with near boiling 10% NaOH.

I have used it twice now, both times with good results.

[Edited on 30-5-2012 by Magpie]

[Edited on 30-5-2012 by Magpie]

Zan Divine - 30-5-2012 at 15:37

many of your observations are spot on. a couple of points if i remember right...the molar mass of phosphorus is close to 31 grams. i based my experiments on 1 mole of sodium hexametaphosphate. when i measured, i had a rough starting mass of 217 grams with all reactants combined (after drying in oven) and a post run weight of 199.8 grams. even with a conservative estimate (owing some loss to the brittle nature of the slag) the reactant mass lost in the neighborhood of 15-17 grams. with the collection of 6+ grams of P4, that is an anus load of product to account for. the weight is not there so we can rule out P4 being trapped in the reaction mass.

[Edited on 29-5-2012 by Rogeryermaw]

Forgive me if you've already said so, but was the phosphate really dry? Unless you dry it by fusion, or to constant weight in a drying pistol at high temp with P2O5 in the other end, you can be sure that some of the weight loss was water. You really have to work to get it dry. I melted it in an electric furnace and got a hard glassy solid that is real work to powder.

[Edited on 30-5-2012 by Zan Divine]

Strepta - 31-5-2012 at 08:36

Reduction of sodium hexametaphosphate with a sufficiently active metal such as pyrotechnic aluminum or powdered magnesium and either silica or boria is relatively straight forward. The reactants, when thoroughly mixed and sufficiently heated, will combust to provide the required reaction heat. If the reaction tube is long enough, you should have no difficulty segregating the free P from the reaction slag using heat.
Pure magnesium is a bit too fast ( I used 100 mesh powder in my one experience with it) as it blew out the stopper of the reaction tube with a loud bang although it did not destroy the test tube reactor. Perhaps a powdered magnalium (Mg/Al, 50:50) would work a bit better. In the US at least these materials are readily available, particularly at this time of year.
The best I could do in efficiency was about 50%, however. It would indeed represent some advancement if someone could positively identify where and in what form the other 50% of theoretical P was.

Magpie - 31-5-2012 at 09:06

Reduction of sodium hexametaphosphate with a sufficiently active metal such as pyrotechnic aluminum or powdered magnesium and either silica or boria is relatively straight forward.

What differences did you notice when boria was substituted for silica?

Strepta - 31-5-2012 at 09:24

When heating the reaction mix with boria, it began to noticeably shrink and curl before combustion onset as boria fuses at a lower temp than silica. Also, I obtained by best yield with boria (52%) although it was not significantly greater than with silica (48-50%). Does the lower melting point of boria mean it provides a better flux?

Magpie - 31-5-2012 at 09:49

I'm not sure of the mechanics of boria vs silica. I just know that boria has a much lower mp (450C or 510C) than silica (1600-1725C). This should mean that it is a liquid at the reaction temperature, where silica might not be. Boria should promote a more efficient reaction, as all reactants would then be liquid.

We are already fighting a lack of any stirring or agitation.

It is known by potters that alumina increases the viscosity of a melt. So I assume this is working against us. Alumina, however, is a by-product in both cases when aluminum is used as the reductant.

If boria produces a less viscous less voluminous slag this might also be an advantage.

[Edited on 31-5-2012 by Magpie]

Rogeryermaw - 31-5-2012 at 12:10

as dry as i could get it by conventional means. about 2+ hours in the oven at over 350f. it was partially fused and had to be broken up before loading the vessel.

When heating the reaction mix with boria, it began to noticeably shrink and curl before combustion onset as boria fuses at a lower temp than silica. Also, I obtained by best yield with boria (52%) although it was not significantly greater than with silica (48-50%). Does the lower melting point of boria mean it provides a better flux?

if it swells less then a larger charge could be used. this doesn't change efficiency but would allow one to produce more P4 per run thereby requiring less runs to produce the product needed. definitely an advantage.

[Edited on 31-5-2012 by Rogeryermaw]

White Yeti - 2-6-2012 at 12:25

I'm not sure if this has already been discussed, but is trisodium phosphate a suitable replacement to sodium hexametaphosphate?

Rogeryermaw - 3-6-2012 at 20:33

Quote: Originally posted by White Yeti

I'm not sure if this has already been discussed, but is trisodium phosphate a suitable replacement to sodium hexametaphosphate?

i don't see why not. there is more sodium to account for and sometimes it comes as the hydrated form which you absolutely must dry (don't want any hydrogen source in your reactants) but otherwise, it should be as feasible as 6(NaPO3).

learningChem - 4-7-2012 at 17:20

Success!!

experiment :

I mixed 1.7g sodium hexametaphosphate, 0.7g Al and 0.5g of something that is sold as "quartz". All three substances where already 'fine' powders when I bought them. I didn't grind anything. The powdered Al is not pyrotechnic grade as far as I can tell - I'm assuming pyrotechnic Al is black? This one would make your fingers silvery if you touched it.

I put half of the mix (~1.5g) in a 15cm test tube, plugged it with that yellow stuff which is supposed to be glass wool (but yellow? why?) and wrapped a bit of wet cloth at the top.

Heated the stuff for something like 10-15 seconds. The mixture started to glow red and expanded somehwat. It took 3 or 4 more seconds for the reaction to complete and the test tube to almost melt, as can be seen in the picture.

I put the test tube in a dark room and could see it faintly glow. Now a few hours have passed and the glow is almost invisible.

Not sure what to do next...

Strepta - 5-7-2012 at 08:32

Congratulations learningChem. What to do next? Try heating the bottom of the test tube and forcing the free P toward the cooler end where it should condense if the wet cloth is kept cool. The P which accumulates at the open end can be removed with a flat spatula and stored under water. Since you are not using an inert atmosphere but a plug of glass wool, oxygen will diffuse through and consume your hard-won product!

Words of caution: this is not an experiment which should be done indoors without proper ventilation (preferably with an efficient fume hood). Also, you are handling the test tube after the experiment without appropriate protection for your hands--be advised that white/yellow phorphous burns are particularly nasty and painful.

learningChem - 5-7-2012 at 10:38

Hi Strepta! Thanks for your message.

I did the reaction outdoors and I kept/keep the tube outside as well. Hopefully none of my cats will try to eat it =P

I saw in your paper how you drive out the P. Quite neat.

What I'm wondering is, if I do that, what happens then when I take the plug out? Will the phosphorous catch fire at once? Or will I have time enough to put it under water?

As to protection, I have some nitrile gloves. Are they OK for P?

Strepta - 6-7-2012 at 02:59

What happens when you take the plug out seems to mostly depend on the temperature of the P. If it's warmer than ambient, it is more likely to inflame. Why not remove the plug under water and eliminate the concern?

As regards gloves, remember the hazard is twofold with P--tendency to inflame and toxicity. Any gloves you have should not be tight fitting in case you need to remove them quickly. A bucket of water at hand is a must--you can always immerse them in water. Tidiness is also important--clean up your work area after experimenting to avoid having any missed particles inflame after you've left. I used to wave a torch over my ceramic work surface to "clean up" any missed P.

learningChem - 6-7-2012 at 10:53

>Why not remove the plug under water and eliminate the concern?

Yes. That's probably the best idea. I wasn't keen on doing that because manipulation seems a bit harder under water, yet it's probably the best solution.

As to gloves, I didn't feel at ease using them because as you noted they can't be taken off quickly.

Thanks for all the info! It's much appreciated. Soo, I'm off to the 'lab'...

Magpie - 16-7-2012 at 10:09

Here's the results of my testing yesterday using a retort made from 3/4" EMT (steel electrical conduit). The retort is shown below.

EMT retort

The closed end is about 14" long and the remainder is about 7" long. I first soaked the conduit in 6N HCl in an attempt to remove the zinc coating. I don't think it all came off. I'm not sure why. The tube was bent to a 90 degree angle using a tubing bender. The long end was closed in a vice then welded shut. A piece of 1/4" ss tubing was then welded to the short end.

Stoichiometric quantities of NaPO3, Al, and SiO2 were then ground together in a mortar. Basis was 6.4g of P. This powder was poured into the retort. The retort was then placed in a tube furnace. The receiver beaker contained water which was maintained at a temperature of 50-60C using a hotplate. Submergence of the retort outlet was 1.5". Argon was led to the retort using the 1/4" ss tube, as shown below.

EMT retort in furnace

Run log:

t=0 min: Set furnace current I= 6.5a.
t= 32 min: Steady bubble generation due to heatup.
t= 51 min: Turned on argon bubbler. T= 419C. Turned on hood fan. Set I = 7a.
t= 67 min: T=539C.
t= 72 min: Smoke coming off receiver now. T=572C.
t=74 min: T= 597C. Smoke and phosphorus vapor burning at receiver water surface. No pops heard.
t= 75 min: phosphorus drops falling into receiver now. Only about 7 drops came over, basically all at once.
t=102 min: Still just smoky puffs. T= 772C. I=8a.
t= 142 min: T= 1055C. Managed to tease a few more drops out by introducing the argon very slowly.
t= 147 min: Turned off furnace.

The drops of white phosphorus can be seen in the picture below.

P in water receiver.

Today the receiver was heated to 45C, melting the drops of P. They were easily consolidated by pushing them together with the end of a thermometer. The product weight was 1.2g for a yield = (1.2/6.4)100 = 19%.

Discussion & Analysis
The boiling point of white phosphorus is 280.5C and its freezing point is 44.2C. Based on my earlier tests with a ceramic retort and glass condenser it seemed that the condenser was too cold as most of the P was subliming onto its internal surface. Therefore, I theorized that a metallic retort might be an improvement in that the condenser temperature would likely be suitable for condensation but above the freezing point. Some dry run testing with an empty retort confirmed this. Temperatures above 44.2C could further be assured by keeping the water in the receiver in the range of 50-60C. Taking a lesson from Rogermeryaw I constructed the above described steel retort using EMT.

When the drops of P came raining over I first thought that I had finally found a good system. However, this high production spike was short-lived. After that the P4 vapor just continued to puff out and burn at the receiver water surface. When the hood light was turned out a green glow was visible in the puffs.

My hypothesis for why the puffs of P4 are not condensing is as follows: the P4 vapor is simply not concentrated enough in the argon gas to reach saturation & thereby condense. This is exactly analogous to why rain does not form in damp air until the water concentration (humidity) is high enough. The P4 is carried by argon to the surface of the water in the receiver where it now contacts oxygen and burns giving the green glow and white smoke.

So, this method for producing phosphorus seems to have the serious limitation of only producing yields in the vicinity of 20%. This doesn't seem to have near as much to do with condenser design as it does with the chemical reaction itself. That is, we are using a dry powder with no agitation that produces a viscous slag. I believe that this inherently gives poor yields and that we would find that most of the phosphorus remains locked in the slag.

Please provide any questions, comments, or suggestions that you may have.

[Edited on 16-7-2012 by Magpie]

[Edited on 16-7-2012 by Magpie]

[Edited on 16-7-2012 by Magpie]

[Edited on 16-7-2012 by Magpie]

[Edited on 17-7-2012 by Magpie]

blogfast25 - 17-7-2012 at 04:16

It would be interesting to digest the slag (I know: it's a very hard material to grind down) to check for phosporus.

Are you sure Ar is really needed? If the free volume in the reactor is small enough the oxygen present should be 'mopped up' by the first P. Even 24 L of air (STP) is only about 0.2 mol of O2...

Edit:

But I’m not convinced that the viscosity of the slag really is a problem, based on a single experiment of mine described way up above.

I filled a steel shot measuring cup (for measuring 1 fl.oz. of spirits in bars and restaurants) with the right mixture of NaPO3, Al and SiO2 and simply heated it on the smallest flame of my lab gas cooker. When the bottom of the crucible began to glow dull red/brown, the reaction started and jets of P2O5 (burning P) streamed out. I shut down the gas and the reaction seemed to continue, self-sustainingly. My IMPRESSION was that the reaction ran to completion.

[Edited on 17-7-2012 by blogfast25]

Magpie - 17-7-2012 at 08:27

It would be interesting to digest the slag (I know: it's a very hard material to grind down) to check for phosporus.

Yes, that would be very enlightening. I'm not sure how that could be done.

Are you sure Ar is really needed? If the free volume in the reactor is small enough the oxygen present should be 'mopped up' by the first P. Even 24 L of air (STP) is only about 0.2 mol of O2...

I agree that the Ar is likely not needed to prevent consumption of phosphorus. I had it cut way back to just barely provide a bubble. It's mostly for my peace of mind that I'm not going to get a suckback.

But I’m not convinced that the viscosity of the slag really is a problem, based on a single experiment of mine described way up above.

I filled a steel shot measuring cup (for measuring 1 fl.oz. of spirits in bars and restaurants) with the right mixture of NaPO3, Al and SiO2 and simply heated it on the smallest flame of my lab gas cooker. When the bottom of the crucible began to glow dull red/brown, the reaction started and jets of P2O5 (burning P) streamed out. I shut down the gas and the reaction seemed to continue, self-sustainingly. My IMPRESSION was that the reaction ran to completion.

Those are interesting results. Perhaps a vertical retort would be an improvement on my horizontal retort.

I feel that more such "test tube" experiments are needed to find better retort designs and perhaps better reactants.

[Edited on 17-7-2012 by Magpie]

blogfast25 - 17-7-2012 at 12:35

Prolonged boiling in conc. H2SO4 of the ground up slag should get any remaining phosphates in the watery phase. But grinding that slag, going by what I saw, would be punishing. Then neutralise with NaOH and filter. Demonstrate phosphate with ammonium orthomolybdate.

Magpie - 17-7-2012 at 13:19

I also find a test for phosphate using a "magnesia mixture" (MgCl2, NH4Cl, and aqueous ammonia). This solution, added to a slightly acidic solution of the unknown, will form a white precipitate of crystalline magnesium ammonium phosphate.

But this is a qualitative test and I'm not sure of what value that would be.

Endimion17 - 27-7-2012 at 04:15

I'm trying to purify my batch. I made a dilute solution of chromic acid and melted phosphorus in it, in a beaker to use a larger surface area. As noted, it remained molten at room temperature, but after several hours the surface got yellow precipitate, and the orange solution right above the melt is darker, which is obviously reduced chromium.
I will leave it in the dark for several days, molten.

BTW, if you're making WP from RP, be sure to mix it with some fine quartz sand. It helps with the bumping a lot.

edit: I've set aside one smaller batch (roughly 3.9 g) and as I write this, it's being stirred molten (<60 °C), under 15% nitric acid. It's constantly being dispersed into small beads and I think it's getting less and less orange, and going towards yellow. I'll let it go for an hour or two.
Meanwhile, there's progress with the batch being purified with chromic acid. The precipitate is half gone.

[Edited on 27-7-2012 by Endimion17]

Endimion17 - 27-7-2012 at 10:47

Success.

This is only a small part of the pure batch. Here you can see it resting in benzene.

The melt was completely transparent and I had quite a bit of trouble solidifying it. It took some ice and poking to finally do it. It looks like a piece of pale amber and compared to the impure samples, the glow is even more powerful.

There's still this orange gunk sediment (probably a variety of red allotrope). However, the chromic acid method seems to be slowly eating it in the mentioned, stil molten batch, so I suppose that if the chromic acid method preceeds the nitric acid swirling bath, one might obtain a nearly transparent block of slightly yellow allotrope.
I'm quite excited, I've never made so much pure WP. The glow is enchanting and I'm trying to find a way to capture it properly.

The final purification is vacuum distillation in the dark, which should produce transparent and colorless WP, its true form. Unfortunatelly, I don't have the equipment.

[Edited on 27-7-2012 by Endimion17]

learningChem - 1-8-2012 at 22:53

My experiment was a failure of sorts after all - I couldn't extract much P from reaction mix - I'm not even sure if the orange stuff that coated the walls of the tube was P? Anyway, here are some pictures of the slag. There doesn't seem to be any P in it?

Magpie - 25-8-2012 at 14:03

I'm going to try a new retort design for making phosphorus. Initially at least I will use the now popular Al/SiO2/NaPO3 recipe firing to about 700C. The retort will be fired in a muffle furnace from which I have removed the door. The door is now replaced with a shaped insulating firebrick as shown in the picture below.

The retort consists of a 1/2 pint (250ml) paint can and a section of curved 1/2" EMT (steel) pipe. The special connection of the pipe to the can lid allows easy removal of the lid. Hopefully this will allow the pipe piece to be reused for many firings.

I just finished pre-firing the assembly to 700C to burn off the can's tin coating. The fired pieces are shown in the other picture below.

A concern I have is the cleaning of the pipe piece following its use. I believe that it will have some internal phosphorus contamination, both white and red. What I want to be able to do is to clean the pipe so that it can be safely handled, with gloves, during reassembly for reuse.

I'm assuming that any white phosphorus will be quickly converted to P2O5 upon contact with oxygen. This then should dissolve in water and be pretty much harmless. The red phosphorous will likely remain unchanged, however.

Ideally the pipe piece could be soaked in a solution that would convert the red phosphorus to harmless products such as a phosphate. Any recommendations for how to do this are greatly appreciated.

[Edited on 25-8-2012 by Magpie]

Rogeryermaw - 28-8-2012 at 00:00

you may also consider insulation/ torch/ heating wire to keep your downpipe at a temp sufficient to keep your product in the vapor phase. this should help to minimize losses and dangerous residue in the pipe. elemental phosphorus scavenges a good deal of oxygen, so i have concerns that, if a residue were to build up, it may not all oxidize to P2O5. love your design though. looks like it may be able to produce much larger quantities in a single run than we have seen up till now.

i will say that in my experiments, i have not had the issue of a buildup in any of the surfaces of the reactor or piping. YMMV. can't wait to see how it goes. best of luck magpie!

of course, for cleaning, you know CS2 is about the best solvent for phosphorus. it is only very slightly soluble in benzene and may be helped along with a long bottle brush or carboy brush.

[Edited on 28-8-2012 by Rogeryermaw]

Magpie - 28-8-2012 at 07:36

Thanks Roger for your helpful comments.

I had not considered keeping the P in the vapor state. My goal has been to keep it liquid. The thermal conductivity of the steel pipe seems good for that based on my previous run.

My intent is to not permit any unsafe buildup of P on the internal (or external) surfaces of the pipe piece. I agree that it would be nice to give it a good soak/scrub with CS2 but I have none.

The best approach I have come up with so far is to soak it in a caustic bath (say 10% NaOH), scrub with a ss wire brush, follow with a caustic rinse, and finally do a water rinse. I will be trying my best to keep liquid waste at a minimum.

The 10% aqueous NaOH should convert the P to PH3. By doing this in my hood I should be able to safely vent off the PH3.

Reuse of the pipe piece is essential to the economics of this method. This piece is expensive as it was fabricated by SCWIM (skilled craftsman who isn't me).

Further suggestions and comments are welcomed.

watson.fawkes - 28-8-2012 at 09:11

I had not considered keeping the P in the vapor state. My goal has been to keep it liquid. The thermal conductivity of the steel pipe seems good for that based on my previous run.

My intent is to not permit any unsafe buildup of P on the internal (or external) surfaces of the pipe piece. I agree that it would be nice to give it a good soak/scrub with CS2 but I have none.

I've been thinking over issues involved in high-temperature condensers, and while you don't have an explicit condenser here, you do have a de facto one. My original motivating example was a miniature CS2 plant, where you've got the issue that generally there's going to be some sulfur that you want to be under reflux and not escaping with the product, condensing in the product pipe and eventually clogging it. Similar concerns apply to other (semi-crazy) ideas such as mercury purification stills and phosphorus production. Clearly these won't apply to your present effort, since it's already fabricated. Therefore, FWIW:

You want the wall temperature of the condensing chamber above the condensation point and the condenser to be below it. This allows you to control where the condensate goes. In practice, this means at minimum measuring these two temperatures, and possibly actively controlling them.
For the condenser itself, say, a coil in the shape of a cold finger, you'd measure the output temperature of coolant. For example, when doing sulfur condensation, mineral oil or a glycol would make adequate coolants, since you need something with a higher boiling point than water. You don't want to overcool the condensate and risk solidification, either. For full control, you 'd also measure the input temperature and control the pumping rate. This leads to something like a flow calorimeter operated "in reverse", where you're measuring the heat extracted with the condenser, and thus estimating the condensation rate.
For the wall of the chamber, you'd use temperature sensors. If insulation is insufficient to keep the operating temperature high enough, you'd use a heater and a thermostat to keep it there.
The condenser should be demountable for cleaning. If the condensate is isolated on the condenser itself, it means you've got an alleviated issue with cleaning the chamber, which has more surface area and is physically larger. The condenser could be dunked into a cleaning tank; not so feasible for the chamber as a whole.

Magpie - 6-9-2012 at 15:09

I made my first P production run with the new paint can retort configuration today. The run was a failure due to the reactant charge being way too large. However, the retort per se performed well.

As has been my practice heretofore I filled the retort (250mL can) half full of the reactants (NaPO3/Al/SiO2). The theoretical yield for this 149g charge was 25g of P.

After drying the charge at 120C for 1/2 hr the EMT tube w/lid was pressed onto the retort body and the assembly inserted into the furnace. The insulating brick door was installed and the outlet of the EMT tube was submerged about 1/2" in a beaker of water. The argon supply system was made ready. The thermostat was set for 700C and heating was commenced.

At about 400C bubbling (air expansion) was regular and there was some smoke coming out along with the bubbles. Shortly thereafter quite a few drops of phosphorus fell into the receiver, but concurrently there was a breach of containment inside the furnace! A voluminous quantity of smoke came pouring out along the sides and top of the firebrick door. I quickly pulled the plug on the furnace, then backed well away from the hood. My strong hood flow was just able to keep up with the generation of smoke. I could see it pouring out of the fan duct outlet. Fortunately there was a slight wind. In a few minutes this had abated and no more smoke could be seen emanating from the louvered outlet at the peak of my garage.

I took four pictures following this run. The first shows the system in the operating configuration.

P retort in operation

The second picture shows the furnace with the firebrick door removed. The white material on the slag is a thin layer of de-laminated firebrick. You can see smears of red P on the top of the firebrick.

P retort with inuslating brick removed.JPG - 127kB

P retort with insulating brick removed

The 3rd picture shows the retort with slag, removed from the furnace.

P retort following run

The final picture shows the slag on the furnace floor. This was removed fairly easily by scraping with a large screwdriver.

P slag on furnace floor

Disscussion

I admit that I should not have loaded the retort so full on the first run for this new retort design. But greed overcame better judgement.

Unfortunately this slag foam is a serious drawback for this formulation of reactants and significantly cuts into yield potential.

I am pleased with the paint can/EMT tube retort design, which incorporates compression seals, allowing reuse of the expensive tube. Perhaps it can be improved to allow for a larger yield.

Although I don't think it had anything to do with causing too much foam I note that I did grind my NaPO3 in a coffee grinder prior to mixing it with the other ingredients. This is the first time I have done this. This grinder does a fine job of pulverizing the NaPO3 to a fine powder.

As always, comments, suggestions, and questions are welcomed.

[Edited on 6-9-2012 by Magpie]

watson.fawkes - 6-9-2012 at 16:13

Although I don't think it had anything to do with causing too much foam I note that I did grind my NaPO3 in a coffee grinder prior to mixing it with the other ingredients. This is the first time I have done this. This grinder does a fine job of pulverizing the NaPO3 to a fine powder.

As always, comments, suggestions, and questions are welcomed.

The fine grind might have contributed. The reaction mixture, partly liquid, had some characteristic viscosity. Gaseous products (e.g. phosphorus) would have some characteristic bubble-up time to reach the surface and escape. If the rate of gas production exceeded the time for a bubble to escape, you'd expect to see volumetric expansion and creation of a foam. Finely ground reactants could well have played into this by increasing the gas production rate beyond some threshold.

Suggestions. (1) (Not too helpful) Use a tall, thin reaction chamber and expect foam. That would mean a new oven and chamber, but it would allow the foam to form without much consequence. (2) Use an inert viscosity modifier to thin the reaction mixture. CaF is what immediately comes to mind, but I don't know if it would apply here. You could test multiple compositions in one furnace run with 1 g batches and little reaction thimbles, packed into your reaction can. I'd suggest a simple rack also, with holes for the thimbles and a couple of bent sides for legs.

Magpie - 6-9-2012 at 17:46

Yes - good thought. Some small scale thimble testing should determine the effect of NaPO3 particle size.

Suggestions. (1) (Not too helpful) Use a tall, thin reaction chamber and expect foam. That would mean a new oven and chamber, but it would allow the foam to form without much consequence. (2) Use an inert viscosity modifier to thin the reaction mixture. CaF is what immediately comes to mind, but I don't know if it would apply here. You could test multiple compositions in one furnace run with 1 g batches and little reaction thimbles, packed into your reaction can. I'd suggest a simple rack also, with holes for the thimbles and a couple of bent sides for legs.

(1) It is unlikely that I will be buying another furnace. But I do have an old homemade furnace that I nearly threw away that might suffice. The tricky part is finding a new form of cheap throwaway retort body that will seal to my EMT downspout.

(2) Rogeryermaw has done some runs using NaCl as a viscosity reducer. I'm not sure how much difference this made for him.

I will do some thimble testing with NaCl, CaF2, and a control. What % add of these reagents would you suggest?

[Edited on 7-9-2012 by Magpie]

blogfast25 - 7-9-2012 at 05:31

Amazing array of niggling little problems, you have to wonder if Hennig Brand with his more 'conventional' formulation also suffered these.

As regards fluxes, CaF2 is probably too high melting. Best might be a NaCl/KCl eutectic (or other inert, low melting eutectic) for maximum impact on melt viscosity. Start at 10 w% of total charge, I would say.

[Edited on 7-9-2012 by blogfast25]

watson.fawkes - 7-9-2012 at 06:04

The tricky part is finding a new form of cheap throwaway retort body that will seal to my EMT downspout.
[...]
I will do some thimble testing with NaCl, CaF2, and a control. What % add of these reagents would you suggest?

I don't know how much "throwaway" you need, but since you're using paint cans, you could consider modifying them. The machined lip is the most important part, since it's the mating surface and difficult to duplicate. So, one possibility is to cut a paint can in half across the middle of the cylinder. Then extend the height of the can with a tube. The problem is that this isn't exactly throwaway, given the effort to make a gas-tight retort this way. But, you could take the same idea and use the top half of a can to make adapters for other kinds of containers, such as, oh, gallon-sized paint cans. I should mention, although this might be in the realm of future plans, that you could split the retort into two parts: (1) an inner reactant crucible and (2) a gas-tight retort. As long as part (1) is large enough to contain any foam, part (2) should need minimal cleaning per run. Then only part (1) is your consumable. You might also use (3) an overflow tray, a semi-consumable.

As for percentages for flux reducers, I have no well-informed proposal about what percentages to start with. The idea came from flux modifiers, such as those that reduce viscosity or flocculate slag particles. So you might start by looking at what they use. I have doubts about direct application, though. My fallback position would be to look at the whole range first, from 0% to 100% in 10% increments, the endpoints being controls. This is one extra furnace run over just guessing what might work and would give some confidence that you were in the right ballpark when you get to smaller variations.

EDIT: copy editing.

[Edited on 7-9-2012 by watson.fawkes]

learningChem - 9-9-2012 at 18:12

Found a recipe in this book

http://books.google.com/books/about/One_thousand_processes_i...

page 61

They start with burnt bones - they add sulphuric acid to get rid of the lime as calcium sulphate - apparently the product is a solution of phosphoric acid?

Next step, they add a solution of lead acetate to the phosphoric acid solution and "a white powder falls to the bottom"

But they don't say what this 'white powder' is - some kind of lead phosphate maybe??

Then they mix the white powder (dried) with 1/6 of charcoal, put the mix in a retort and heat it to "red hot"

So, I'm wondering two things

1) is the white powder lead phosphate?

2) is this a relatively 'low' temperature reaction?

Magpie - 11-9-2012 at 18:44

As suggested by Watson I did some "thimble" testing today to determine what NaPO3/Al/SiO2 based formulation(s) might minimize slag formation. The results are somewhat surprising.

Procedure
Rather than steel thimbles I used the bottoms of 100mm test tubes cut in half with a Dremel tool. I filled 5 such tubes with 1g of base (stoichiometric) formulation modified as follows:

1. base formula - ground ingredients in a mortar per usual (the control).
2. pre-grind the NaPO3 in a coffee grinder; mix lightly in a mortar with the other 2 ingredients.
3. base formula with 10% NaCl add; grind all ingredients together in a mortar.
4. pre-grind NaPO3 + 10%NaCl in a coffee grinder; mix lightly with the other 2 ingredients in a mortar.
5. base formula w/10% add of 50 mole%KCl/50 mole%NaCl; grind together in a mortar.

The tubes were loaded with their respective powders and the height of the powder measured in mm. The tubes were then placed in a 50mL beaker in the retort. The retort was fired to 700C and kept at that temperature for 1/2 hr.

Results
When cool the retort was opened and the tubes removed. They are shown in the picture below.

P slag volume comparison

In the order tubes 1 through 5, the unfired powder heights were 9, 10, 13, 13, and 11 mm, respectively.

Likewise, the fired slag heights are 34, 22, 44, 34, and 34mm, respectively.

The computed volume proportionalities, fired:unfired, are:
3.8, 2.0, 3.4, 2.6, and 3.1, respectively.

Conclusions
Therefore, these results show that the NaCl and NaCl/KCl adds gave mixed results. The best apparent result was with the pre-ground (by coffee grinder) NaPO3.

The results bring up more questions. Like, what difference can be attributed to thorough grinding together in a mortar vs a light grinding in a mortar. The better performers might just have a less complete reaction. It is impossible to say what is best without knowing the yield of P for each formulation. This suggests that a different line of further testing is needed.

Comments, questions, and suggestions are welcomed.

[Edited on 12-9-2012 by Magpie] conclusions revised

[Edited on 12-9-2012 by Magpie]

watson.fawkes - 11-9-2012 at 20:34

The results bring up more questions. Like, what difference can be attributed to thorough grinding together in a mortar vs a light grinding in a mortar. The better performers might just have a less complete reaction. It is impossible to say what is best without knowing the yield of P for each formulation. This suggests that a different line of further testing is needed.

Really interesting photo there. Good call on transparent thimbles; it's much easier to see what's happening. There are culture tubes that come already in about the size you got by modifying; they seem to come in boxes of 200. You might also consider a ceramic marking pencil that will fire on permanently at temperature.

The first thing that strikes me is the relative difference between 1 vs. 2 and 3 vs. 4. Start with color differences, particularly in the 3-4 pair. Based only on color it's possible that #3 reacted the most, converting dark Al to light Al₂O₃. The coarser formulations have obvious pores and entrained volumes. The idea that this might be due to reaction times and degree of completion seems worth investigating.

One of the ways of testing for completeness of reaction would be to grind and lixiviate the samples with water and see how much soluble content remains in them. You'd have to distinguish between phosphate and chloride in the leachate. Perhaps then follow with something to preferentially dissolve unreacted Al.

As an alternate way of testing the amount of P formation, perhaps use a phosphorus getter above a glass wool plug. Use something that reacts completely with gas-phase P, maybe forming a phosphide, but scavenging all the P formed. Mass difference gives a direct reading of yield.

Seeing these samples has spurred some new ideas about additives. Principal are the constituents of soda-lime glass, oxides and carbonates of sodium, calcium, and magnesium. Looking at the samples leads me to think that the sooner you get into a liquid-phase reaction, the better off the reaction will be. Lead oxide would lower melting points, but it would reduce along with the phosphorus. I doubt you'd need bulk liquid, but some interfacial liquid on particle surfaces might be enough.

The other "additive" that comes to mind is a candidate reagent: calcium carbide. It seems to me it would be an excellent oxygen scavenger in this environment, lower melting points, and provide some mechanical sparging by CO₂ production. Wouldn't need much, I'd guess, to be active at the beginning of the reaction.

Magpie - 12-9-2012 at 08:12

There are culture tubes that come already in about the size you got by modifying; they seem to come in boxes of 200. You might also consider a ceramic marking pencil that will fire on permanently at temperature.

Readymade tubes would be convenient if I end up doing much more of this. Marking the tubes was a challenge. I made score marks with a file to number them.

The first thing that strikes me is the relative difference between 1 vs. 2 and 3 vs. 4. Start with color differences, particularly in the 3-4 pair. Based only on color it's possible that #3 reacted the most, converting dark Al to light Al₂O₃. The coarser formulations have obvious pores and entrained volumes. The idea that this might be due to reaction times and degree of completion seems worth investigating.

Yes, those color and pore volume differences are striking. There's a story there for sure. At first I thought the light pinkish color of #3 was due to red P. But I think your explanation is more plausible.

I took the picture below this morning by turning off the hood fluorescent light and using a flash:

P slags 2

One of the ways of testing for completeness of reaction would be to grind and lixiviate the samples with water and see how much soluble content remains in them. You'd have to distinguish between phosphate and chloride in the leachate. Perhaps then follow with something to preferentially dissolve unreacted Al.

I wished I had weighed the tubes+powder before firing. Then I could weigh them now and determine a weight loss. This could give an indication of completeness of reaction. But it still does not give a direct indication of useable product, ie, drops of P in the receiver. I'm afraid this will only be possible with a full scale test.

During this very small scale screening test no drops of P fell into the receiver. The only product I saw was smoke. This is telling me that a significant loss of yield may be due to aerosol formation (smoke) that does not coalesce in the EMT condenser.

As an alternate way of testing the amount of P formation, perhaps use a phosphorus getter above a glass wool plug. Use something that reacts completely with gas-phase P, maybe forming a phosphide, but scavenging all the P formed. Mass difference gives a direct reading of yield.

Seeing these samples has spurred some new ideas about additives. Principal are the constituents of soda-lime glass, oxides and carbonates of sodium, calcium, and magnesium. Looking at the samples leads me to think that the sooner you get into a liquid-phase reaction, the better off the reaction will be. Lead oxide would lower melting points, but it would reduce along with the phosphorus. I doubt you'd need bulk liquid, but some interfacial liquid on particle surfaces might be enough.

The other "additive" that comes to mind is a candidate reagent: calcium carbide. It seems to me it would be an excellent oxygen scavenger in this environment, lower melting points, and provide some mechanical sparging by CO₂ production. Wouldn't need much, I'd guess, to be active at the beginning of the reaction.

Lots of good ideas here. Thanks.

watson.fawkes - 12-9-2012 at 09:22

Marking the tubes was a challenge. I made score marks with a file to number them.
[...]
During this very small scale screening test no drops of P fell into the receiver. The only product I saw was smoke. This is telling me that a significant loss of yield may be due to aerosol formation (smoke) that does not coalesce in the EMT condenser.

Here's the marking device I had seen for glass:
Ti-Pen at manufacturer Aspen Glass
Ti-Pen at Wale Apparatus
I don't know if it's just a rod of Ti, sharpened and mounted, or whether it's some Ti alloy.

Searching for "borosilicate culture tubes", I found this page from VWR LabShop. They sell the 5 ml tubes, 12 x 75 mm, for about USD 31 for a package of 1000. I put the link here because there are number of other people working on process parameters who could benefit from some combinatorial testing. The point is that little glass tubes are comparable to cost of the reagents and can be considered expendable.

If you're getting smoke as a product, and you're purging with argon first, the only sources of oxygen to make P₂O₅ smoke are going to be (1) interstitial atmosphere left over within the powder of the charge, or (2) from within the charge itself, either the phosphate or the silica. There's just not much oxygen (in molar terms) in residual atmosphere to see very much smoke. This leads me to think that an overly high temperature might first reduce and liberate phosphorus and then reoxidize it in a gas-solid process. Two thoughts on how to deal with this. (a) Lower the temperature. (b) Add a bit of carbon. Some of the carbon will oxidize to CO. As a gas, will would react away any oxygen available at solid surfaces. The Al in the charge would would reduce CO and CO2 thus formed. (See the great Al + dry ice combustion video.) In this way carbon is acting in a catalytic cycle, with the advantage that it's gaseous for part of the cycle.

Magpie - 12-9-2012 at 10:07

If you're getting smoke as a product, and you're purging with argon first...

I did not purge with argon first...possibly a mistake. I did use argon for the bubbler as usual for preventing suckback.

Rogeryermaw - 12-9-2012 at 10:27

just an observation but, with your given conditions (meaning the retort of course) i would think that mix #2 would be where i would start. i really am not sure how NaCl affected my experimentation, but when i used it, i did not achieve that pinkish color. i wonder if there is a dissociation of the chloride ion at your temps that could allow for a reaction causing impure phosphorus.

with a boiling point of over 1400C dissociation is not the cause but perhaps the phosphorus vapor and chloride undergo some sort of replacement reaction leaving the sodium to react with??? oxygen from the phosphate? it is more reactive than aluminum. hmmm. something to think about. perhaps the some of the phosphorus would react with the sodium were that the case. there are several known compounds of phosphorus and sodium.

[Edited on 12-9-2012 by Rogeryermaw]

Magpie - 12-9-2012 at 15:09

I have added a snorkel to my condenser design. This will allow a greater fill and thus a greater yield. It is being "burnt-in" as I write.

[Edited on 12-9-2012 by Magpie]

Magpie - 14-9-2012 at 12:22

Yesterday I made a full scale production run using the basic (stoichiometric) NaPO3/Al/SiO2 formulation. Ingredients were thoroughly ground together in a mortar then dried for an hour in an oven. The charge, which weighed 54.4g, was placed in the retort and the condenser installed, the whole system having been backfilled with argon. The assembly was weighed at 368.8g. The retort void cavity had been measured at 225mL. The assembly was placed in the oven and the thermostat set for 700C.

At ~500C some smoke, burning phosphine, and drops of white P (very clean and white) were emitted from the condenser. Some white smoke was emitted from the crack above the furnace door indicating a small breach of containment of the retort. Heating was continued until the furnace temperature reached 700C. A picture of the P in the receiver is shown below:

base formulation P (54.4g charge).JPG - 109kB

base formulation P (54.4g charge)

When cooled the retort was removed from the furnace. Some white powder, P2O5 I presume, could be seen at the can rim in 2 places, and red P in 4 places. Upon opening the can the slag formation was observed as shown below:

base formulation slag (54.4g charge).JPG - 125kB

base formulation slag (54.4g charge)

Weight of the fired retort assembly was 367.6g indicating a loss in weight of 368.8-367.6 = 1.2g. The volume of slag was measured by backfilling with sand to be 166mL. Therefore the slag volume/g of charge = 166mL/54.4g = 3.1mL/g, which exactly matched that of the 1g test tube measurement!

The P produced was a disappointing 0.9g for a %yield of only 9.5%. I expected 19-20%. This weight was close to that shown in the assembly weight loss of 1.2g computed above, however. Therefore, it seems that the overwhelming contribution to poor yield is just the nature of the reaction itself.

Comments, questions, recommendations?

learningChem - 16-9-2012 at 16:42

I got a question...

Isn't the reaction self-sustaining? Shouldn't the ingredients react on their own, once 'ignited'? It should be possible to only heat a small portion of the retort, to kick start the reaction, and then wait a few seconds for it to complete?

Magpie - 16-9-2012 at 17:22

I got a question...

Isn't the reaction self-sustaining? Shouldn't the ingredients react on their own, once 'ignited'? It should be possible to only heat a small portion of the retort, to kick start the reaction, and then wait a few seconds for it to complete?

I think that is a possibility. In my experience it seems to fire off at about 500C. The P shows up over a period of a few seconds. It might be possible to just fire off the reactants using a Bunsen or Meker burner.

BromicAcid - 16-9-2012 at 19:53

Great work Magpie. Although the numbers don't look exciting I think your work is very valuable. It never seemed like the yields were quite where you would expect and it's great to finally see a true mass difference and percent yield along with information on the expansion of the reaction mixture. Great work and very concise!

Still, a disappointment from a preparative route. I would have never thought it was that low and you have given the reaction every benefit you could think of. Oh well, back to the drawing board, right?

Rogeryermaw - 16-9-2012 at 20:04

that would depend on the viscosity of the melt. it is quite possible that the reactor may need to be held at a temperature to keep the melt fluid so any phosphorus produced can be released from it.

Magpie - 17-9-2012 at 08:40

Thank you Bromic. Yes, I hope that by modifying the formulation and/or technique that the yield can be significantly increased.

watson.fawkes - 20-9-2012 at 07:45

Yes, I hope that by modifying the formulation and/or technique that the yield can be significantly increased.

I went and read the Kirk-Othmer article on white P production. It's mostly about primary extraction from phosphate ores, principally fluorapatite, so there's CaF₂ in the product. The reaction mixture also includes silica and carbon. Some of the silica comes from the ore, but some is added. The reasons listed are two: (1) thermodynamic favorability from the formation of calcium metasilicate CaSiO₃, and (2) lowering of the slag melting temperature. The temperatures given are as follows: the reaction starts at 1100 °C, the sweet spot is 1400 °C - 1600 °C, and unwanted side reactions (such as reduction of silica) become problematic at 1700 °C. Phosphorus comes off as the dimer P₂, which is favored above 800 °C gas temperature, condensing to P₄ below that temperature. (This is significant because this gas-phase condensation induces ΔP in the exhaust path.) Carbon monoxide is the carbon product, which is unreactive with P₂ and P₄ in the exhaust gas.

You, however, are not doing primary extraction. Starting with calcium phosphate should be considered a secondary extraction process. Sodium phosphate, though, looks to be a better starting material, or perhaps a good additional starting material, at least on the grounds of its lower-melting mixtures. It seems that the use of Al as a reducer is one of the problems, since alumina forms high-melting slags (stovetop glass cookware is aluminosilicate glass). Silica seems worth keeping, since it's a good slag-forming material. The liquidus temperature of soda-lime glasses are in the 1000 °C - 1100 °C range, though. I should also mention that we don't really care whether the slag forms a glass in its solid phase, though many considered here do.

I found a downloadable spreadsheet for estimating liquidus temperatures on this page on glass properties. A bit of dinking around with it found an approximate minimum at the mixture SiO₂ 75 wt% + Na₂O 25 wt%, yielding a liquidus temperature of 785 °C. (The temperature rises very quickly with decreasing Na content, much less so in the other direction.) Small additions of the oxides of Mg, K, Al, and Ca all raised the temperature. From what I can tell, this spreadsheet is interpolating between measurement data, and I can't vouch for its accuracy.

I don't have handy data for the heat of formation of the various metasilicates, and liquidus point is not the only factor to consider. Nevertheless, this 3:1 oxide-weight ratio seems like a decent place to start at the thimble scale.

Also, I've realized that chloride flux materials are probably not a good idea, since they may well form PCl₃ as a byproduct, which in this case would be a contaminant.

Strepta - 20-9-2012 at 08:19

Magpie: "The P produced was a disappointing 0.9g for a %yield of only 9.5%. I expected 19-20%. This weight was close to that shown in the assembly weight loss of 1.2g computed above, however. Therefore, it seems that the overwhelming contribution to poor yield is just the nature of the reaction itself. "

I agree and more specifically the likely result of reacting materials in the solid phase. Under the best of circumstances, the proximity and stoichiometry of the reactants in the neighborhood of the flame front is poor at or anywhere near the molecular level. As with black powder, better incorporation (mixing of reactants) may be the shortest path to higher reaction yield.

Also, have you experimented with boron trioxide? Since it begins to flow short of the reaction onset temperature, it's possible that holding the charge at its melting point for an extended time prior to raising the temp to the reaction point may be beneficial. Perhaps use a stoichiometric excess of B2O3 to facilitate flow?

[/rquote]

[Edited on 20-9-2012 by Strepta]

Magpie - 20-9-2012 at 08:30

Thank you watson for this research - much food for thought here. I have been doing some reading on steel making slag, which also is loaded with potential insights. I will post links and discuss it later.

Run #4
Another run was completed recently using the same NaPO3/Al/SiO2 formulation as the previous run except that the NaPO3 was pre-ground for 1 minute, a teaspoon at a time, in a coffee grinder. The ingredients were then mixed, without grinding, in a mortar.

Although the 1g test tube run (TT#2) indicated a slag volume of 2.1mL/g of charge, I was suspicious that this was too low. So I conservatively chose a charge size of 60.4g.

As before the charge fired off at about 510C and was complete in a minute or two. It was slower than the previous run. The yield was 1.7g of P for a % yield of 16.1%. The slag volume was 2.9mL/g, confirming my suspicion. This left a free volume of only 48mL in the retort. So it seems that 60.4g is a practical maximum charge size.

Weight loss of the assembly due to firing = 375.1g – 373.1g = 2.0g

Discussion
The reaction takes off when reaching sufficient temperature (~500C). The yield might be improved by slowing down the reaction. The obvious way to do this would be to add an inert substance. If that substance were also a viscosity reducer, all the better, it would seem. A less active reductant than aluminum might help also.

If this run can be taken as typical then the P reports as follows:

a. drops of white P in the receiver: 16.1%
b. PH3 & P2O5 exiting the condenser: 2.8%
c. tied up in the slag: 81.1%

Magpie - 20-9-2012 at 10:19

Thanks for your comment Strepta. Because we posted simultaneously I almost missed it.

[...
As with black powder, better incorporation (mixing of reactants) may be the shortest path to higher reaction yield.]

Even though the reaction is over with in at most a couple minutes, I understand that we are dealing with possibly a solid/solid/solid reaction, without benefit of mixing, and that this hinders the reaction at the molecular level. I can possibly improve this somewhat by mixing all ingredients in the coffee grinder, if it doesn't catch on fire!

Also, have you experimented with boron trioxide? Since it begins to flow short of the reaction onset temperature, it's possible that holding the charge at its melting point for an extended time prior to raising the temp to the reaction point may be beneficial. Perhaps use a stoichiometric excess of B2O3 to facilitate flow?

I have been thinking some of substituting B2O3 for SiO2 myself. I think it would be well worth trying.

[Edited on 20-9-2012 by Magpie]

learningChem - 20-9-2012 at 11:20

Quote:

tied up in the slag: 81.1%

So, in theory there is elemental P in the slag? If you break off bits of it, you should see white fumes for instance? Does that actually happen?

Assuming there is P in there, shouldn't it be possible to extract it somehow?

watson.fawkes - 20-9-2012 at 11:44

Assuming there is P in there, shouldn't it be possible to extract it somehow?

At least in some runs in the past, there doesn't seem to have been. But your comment did spark an idea.

It's not like you should expect any free P given the reaction mixture. Si, Al, and P all form oxide polymers, heavily cross-linked in the manner of ceramics. If I had to guess, but with some evidence from Magpie's pictures, I'd guess that the residuum is a mixed suboxide foam. So here's the idea, making more (potential) work for Magpie. Roasting the slag in an excess of oxygen, after grinding, might show a mass increase as the ordinary oxides form. Now this is a gas-solid diffusion reaction, so it would require a very fine grind in the powder, high temperature, a reasonable amount of time, some flow of atmosphere to provide makeup oxygen, and a refractory boat preferably made of an oxygen-saturated ceramic.

It's something still of a mystery as to where all the P ends up, but such an assay might allow an estimate of trapped P. To make the estimate, you'd assume that all the Al oxidized and that none of the oxygen became volatile. These assumption probably aren't perfect, but it might indicate something.

Magpie - 20-9-2012 at 12:43

I concur with watson's assesment that the P is most likely chemically bound as one or more oxides. Just where the oxygen originates is a bit of a mystery as the retort was backfilled with argon and kept under a slight argon overpressure.

When the slag is cracked open there is no smoking, which would be evidence of free white P. Also there is not much evidence of free red P except some obvious spots here and there. The beige coloration may be due to finely divided red P, however.

Shown below are two pictures that I just took of the slag from Run#4:

learningChem - 20-9-2012 at 13:05

Quote:

I concur with watson's assesment that the P is most likely chemically bound as one or more oxides.

Does that mean that the reaction doesn't work as advertised? Or that (most of) the P is first liberated from the phosphate and then it reacts again and forms a new P compound?

Also, according to Strepta

"From 3 g of the above mentioned mix, .5 g of P4 is available. The resultant P4 (after re-subliming) weighed .25g or just 50% of theory. "

What's the main difference between your setup and Strepta's? The scale? The shape of the container?

Magpie - 20-9-2012 at 13:51

Does that mean that the reaction doesn't work as advertised?

It obviously works: white P is produced. It's just not produced in high yield.

Or that (most of) the P is first liberated from the phosphate and then it reacts again and forms a new P compound?

These mechanics are not elucidated by our experiments so far. I think it is possible that much of the NaPO3 is left unreacted, probably dissolved in the glassy silicate.

Also, according to Strepta

"From 3 g of the above mentioned mix, .5 g of P4 is available. The resultant P4 (after re-subliming) weighed .25g or just 50% of theory. "

What's the main difference between your setup and Strepta's? The scale? The shape of the container?

Those, and there are differences in reactant quality and particle size, surely.

watson.fawkes - 20-9-2012 at 15:00

I concur with watson's assesment that the P is most likely chemically bound as one or more oxides. Just where the oxygen originates is a bit of a mystery

I didn't emphasize it, but I suspect that the oxides are more properly suboxides. More specifically, we have full oxides Al₂O₃, SiO₂, and P₄O₁₀, with oxidation states Al³⁺, Si⁴⁺, and P⁵⁺. In this hypothetical suboxide matrix, the average oxidation states for each of these elements are simply smaller than these values. As for the oxygen source, it was the silica and phosphate. There are two suboxides of aluminum Al₂O and AlO, with oxidation state Al⁺ and Al²⁺.SiO exists in reduced ceramic phases, with oxidation state Si²⁺. The phosphite ion has oxidation state P³⁺. Any or all of these could be present in some combination, and all of these ions polymerize. It's basically a partially reduced cement.

The reaction mixture itself, in a sealed can with no external atmosphere, makes a reducing environment. The metallic aluminum ensures that; it acts as the reducer. The goal is reduce the phosphate, liberating free phosphorus; that happens to a limited extent. But an incomplete reaction means that the phosphorus only partially reduces, that the silica might reduce, and that the aluminum might not fully oxidize. This seems like it may be the key to carbothermic reduction of phosphate: the oxidation product of carbon as a gas, its monoxide, does not hinder the formation of a liquid-phase reaction medium as alumina would.

In any case, this observation is the source of my suggestion to calcine the product slag in moving atmosphere. If it's a suboxide, it should gain mass. As a qualitative test, also, free P₄O₁₀, that is, free of polymeric links to Si and Al, should also form, which could be detected as a pH change when dissolved in water, as compared to the pH of the slag immediately after the initial reaction.

Strepta - 21-9-2012 at 08:43

A very interesting thought. Would it be fair to assume that since the complete oxides have larger reaction enthalpies, that the existence of suboxides implies a shortage of oxygen in the reaction plasma? Does this mean that better mixing may at least, in part, lead to higher yield of the desired P?
@Magpie: I would not be too concerned about the mix igniting in the coffee grinder. It will ignite and burn slowly after being heated to 500C, but as Blogfast noted several pages up-thread, the reaction enthalpy is just barely positive.

[Edited on 21-9-2012 by Strepta]

watson.fawkes - 21-9-2012 at 13:42

Would it be fair to assume that since the complete oxides have larger reaction enthalpies, that the existence of suboxides implies a shortage of oxygen in the reaction plasma? Does this mean that better mixing may at least, in part, lead to higher yield of the desired P?

Better mixing might lead to a better reaction. The problem is that means reducing particle size, and then you are also simultaneously removing voids through which gaseous product phosphorus can travel. It seems that the result we're seeing is that firing this reaction mixture leads principally to suboxide foam unless the P can escape quickly. In this model for what's happening, I'd guess the voids seen in the test run were initially formed by gaseous P₄, which then reacted back into the foam matrix when it couldn't escape.

For the first issue, there's manifestly an oxygen deficit. That's the whole point of the reaction mixture, to have an oxygen transfer from phosphate to a reducer. This is not a triviality, since it leads immediately to a quantitative issue with the size of the oxygen deficit in the reagent charge. If aluminum is forming suboxides, it's effective power as a reducer is lowered. One way to deal with that is to triple the amount of Al. That way, even if it only forms Al₂O, with oxidation state 1+, it will scavenge as much oxygen as if the original amount oxidized completely to a 3+ oxidation state. This seems like an easy thing to try, since it doesn't require any new reagents. If the silica is also being reduced, even more Al would be required.

Magpie - 21-9-2012 at 14:05

Keep the good ideas coming.

I just finished making some B2O3 from H3BO3. Thank science for coffee grinders. That B2O3 glass is hard as hell.

My next run will be a 50g charge based on the following assumed stoichiometry:

10Al + 6NaPO3 + 6B2O3 ---> 3Na2B4O7 + 5Al2O3 + 6P

My choice of Na2B4O7 (borax) as the end product is somewhat arbitrary. I'm assuming that since it is commonly found in nature it is a stable ground state.

Note that this requires twice as many moles of B2O3 as SiO2 based on the standard stoichiometry:

10Al + 6NaPO3 + 3SiO2 ----> 3Na2SiO3 + 5Al2O3 + 6P

Strepta - 21-9-2012 at 14:37

When I was actively working on this some years ago, I did a number of experiments with varying stoichiometry. My best results (~50%) yield were with a slight excess (15%) of Al. What I recall vividly was that the mix had the rheology of a light weight oil--it would sway like a liquid when the test tube was jostled and spurt up when the tube was rapped on the work table. This was the result of the use of pyro grade Al and extensive mixing (with a coffee grinder) and drying of the mix prior to use.

Yes, the boria is a pain to powder--supposedly best due by vacuum heating of the boric acid. This results in a matrix of B2O3 with voids where the H2O has vacated, and this is easily powdered.

watson.fawkes - 21-9-2012 at 15:35

10Al + 6NaPO3 + 6B2O3 ---> 3Na2B4O7 + 5Al2O3 + 6P

My choice of Na2B4O7 (borax) as the end product is somewhat arbitrary. I'm assuming that since it is commonly found in nature it is a stable ground state.

There seems to be only one simple oxyanion of boron, the borate, with boron oxidation state 3+. Borax has this same oxidation state, as do all the common polymeric borate relatives. A hypothetical borite ion, in oxidation state B¹⁺, might only exist in an excited state and transiently. So one potential advantage of using borate is that it's less likely to partially reduce. And it seems thermodynamically unfavorable to reduce to B instead of P. There's still the issue of the Al not completely oxidizing. But if you're lucky, the borate might further oxidize to perborate with oxidation state B⁵⁺.

Another advantage is that the melting point of borax is 743 °C. That might be a eutectic temperature for soda-boria. I haven't tried looking up the data. Regardless, the product alumina will surely raise the melting point.

White Yeti - 22-9-2012 at 11:02

I know electrolysis is a pain in the ass to get to work, but has anyone contemplated the electrolysis of molten phosphoric acid? The idea is that phosphate is electrochemically reduced sequentially from phosphate to phosphite to hypophosphite and finally to phosphorus, the allotrope remains to be determined.

H3PO4(aq) + 2 H+ + 2 e− ---> H3PO3(aq) + H2O −0.276V
H3PO3(aq) + 2 H+ + 2 e− ---> H3PO2(aq) + H2O −0.499V
H3PO2(aq) + H+ + e− ---> P + 2 H2O −0.508V
O2(g) + 2 H2O + 4 e− ---> 4 OH−(aq) +0.40V

Hypophosphite disproportionates at around 250C to phosphine and phosphate, but if the temperature is kept around 60C this problem shouldn't arise.

If 85% phosphoric acid is used, would the water be electrolysed first instead of the phosphate? If so, is pure phosphoric acid conductive enough for the electrolysis to proceed?

Due to phosphorus's tendency to form polymeric anions such as pyrophosphate, phosphorus pentoxide or escape as phosphine, I honestly have no idea what the exact outcome of this electrolysis would be. Obviously due to the reactive nature of phosphorus, some method should be employed to ensure that the phosphorus and oxygen are not allowed to mix.

What do you think? Is it worth a shot?

plante1999 - 22-9-2012 at 14:49

Quote: Originally posted by garage chemist

Just some other experiment I found: a demonstration experiment to show that bones contain phosphorus:

Cleaned, boiled and dried chicken bones are burned with a bunsen burner on a fireproof surface and directly heated with the flame until they have turned into white ash.

2g of this bone ash are mixed with 0,5g magnesium powder and 0,5g kieselgur.
The mix is heated in a test tube which is plugged with a glasswool plug. After the reaction has finished, it is left to cool and the glasswool plug is removed in a darkened room and observed closely.
A glow is visible on the glasswool.
When the residue in the test tube is mixed with water, gas bubbles are evolved which self-ignite on contact with air. They are phosphine.

Reactions are on the site that I posted.

The important feature here is the use of magnesium instead of the often- used aluminium. Mg reacts at a much lower temperature than Al.
The SiO2 must be finely dispersed in the mix, hence the use of kieselgur. Quartz sand is not fine enough, even after good grinding.

I already made some P4 a few days ago with aluminium, NaPO3 and B2O3. I decided to try this experiment changing the Mg for a smaller amount of Al and I was able to get very small amount of phosphorus at very high temperature for a long time, so it seam that with calcium phosphate Mg is better suited.

I will try to find Mg then I will report my result here. As for the reagents, the bone ash is from chicken bones that I calcined to the white ash, and the kielsegur is from insect killer which is suppose to be 91% SiO2 and ''chemical free'' which is obviously false since there is something in the bottle.

bbartlog - 22-9-2012 at 17:07

Quote: Originally posted by White Yeti

I know electrolysis is a pain in the ass to get to work, but has anyone contemplated the electrolysis of molten phosphoric acid? ...
What do you think? Is it worth a shot?

I would expect it to be too poor a conductor. The dissociation constant is low; while there is water present I think the main products would be hydrogen and oxygen, and once you get to the point where you have a glassy mass of 90something% phosphoric acid I don't think you'll have a lot of moving ions. And even if you did I think you might find that P2O5 would be the end product.

Magpie - 22-9-2012 at 20:05

In preparing for a phosphorus run using B2O3 I became aware of an increase in weight of at least 3.2g on 50.0g of charge. I believe this is due to the pickup of water by the hygroscopic B2O3 and to a lesser extent NaPO3. I believe I will have to reject this charge and start over. Based on the following reaction every gram of water will kill 1.8g of P:

16P + 15H20 ---> 3P2O5 + 10PH3

(1g/18)(16/15)(31) = 1.8g

Since the theoretical yield of this charge is only 7.1g of P, this would be an unacceptable loss.

blogfast25 - 23-9-2012 at 11:33

In preparing for a phosphorus run using B2O3 I became aware of an increase in weight of at least 3.2g on 50.0g of charge. I believe this is due to the pickup of water by the hygroscopic B2O3 and to a lesser extent NaPO3. I believe I will have to reject this charge and start over. Based on the following reaction every gram of water will kill 1.8g of P:

16P + 15H20 ---> 3P2O5 + 10PH3

(1g/18)(16/15)(31) = 1.8g

Since the theoretical yield of this charge is only 7.1g of P, this would be an unacceptable loss.

B2O3? Boric oxide? To what desired effect?

Magpie - 23-9-2012 at 11:52

B2O3? Boric oxide? To what desired effect?

Have you read upthread for background. The B2O3 serves as a sort of "sink" to prevent the formation of AlP which would otherwise consume half the P. Same purpose as using SiO2 in the standard formulation...just a variant of this.

---------------------------------------------------------
I decided not to reject my prepared Al/NaPO3/B2O3 charge but to heat it to 350C in an attempt to dehydrate it prior to use.

blogfast25 - 23-9-2012 at 12:03

Sorry, read most but must have missed that bit. My bad.

White Yeti - 24-9-2012 at 12:31

Quote: Originally posted by bbartlog

I feared those potential problems, but would adding a small amount of sodium phosphate increase the conductivity of the solution? If the sodium salt is soluble in molten phosphoric acid, the conductivity would increase dramatically.

On the other hand, I highly doubt that the phosphoric acid could be dehydrated all the way to phosphorus pentoxide via an electrolysis alone.

Magpie - 24-9-2012 at 14:54

I just finished cleaning up after my phosphorus run (#5) with boria substituted for silica according to the following assumed reaction:

10Al + 6NaPO3 + 6B2O3 ---> 3Na2B2O7 + 5Al2O3 + 6P

The charge was 50g.

I had noted that the charge had picked up 3.2g during grinding and handling due apparently to the aggressively hygroscopic nature of B2O3. I felt that this water had to be driven off before firing because of the potential for a yield killing reaction below:

16P + 15H2O ---> 3P2O5 + 10PH3

Therefore I heated it for 2 hours at 350C. This did drive off the water but also melted the boria forming the charge into a hard monolithic disk 3" in dia and 1/2" thick. Wrongly or rightly I felt this disk had to be pulverized so pounded it with a hammer in a plastic bag then ground it up quickly in the coffee grinder.

I backfilled the retort ass'y with argon, started the argon bubbler, and turned on the furnace. When the temperature reached 700C nothing yet had happened. So I set the thermostat to 800C. When the temperature reached 750C that's when the fun started.

The reaction took off. Some few drops of red and black P fell into the receiver as shown below:

first P discharged

(The weight P in the receiver is 0.5g.)

Then a large fire in the furnace occurred, indicating a breach of containment of the retort. I pulled the plug on the furnace and took a few steps back. The generation of P2O5 smoke was voluminous, challenging my powerful fume hood.

After about a half hour I ran out of argon. Now a big goober of burning phosphorus dropped into the receiver. Oh well, all the exposed P in the furnace had to burn off at some time.

You can perhaps see the added goober in the picture below:

after 2nd P discharge

After cooling overnight I open the furnace for inspection. Here's what was left of the retort:

run #5 (boria) retort

You can see the glassy slag in the bottom of the retort and around the inlet of the snorkel. A second picture shows the snorkel in more detail. The inlet is about 1/2 closed off with glass.

run #5 (boria) snorkel

Discussion
My current retort is clearly not adequate for containing this reaction. However, I think the reaction has potential for producing a good yield of P. All three reactants are liquid at the apparent firing temperature of 750C. And the slag has a very low volume: maybe 25% of the retort, and likely has a low viscosity as there are very few vacuoles. There was much burning of P in the furnace indicating a potentially good yield.

Questions, comments, and recommendations are welcomed.

[Edited on 25-9-2012 by Magpie]

Strepta - 24-9-2012 at 15:38

Excellent! As soon as you "harden" your retort, we should see some nice results. I never used a charge larger than 4 g, so I can only imagine the magnitude of that conflagration. It will be interesting to see your yield and, then, whether further improvement can be made.

watson.fawkes - 24-9-2012 at 16:53

Questions, comments, and recommendations are welcomed.

One thing I immediately noticed is how much higher the initiation point was for this reaction, particularly since everything is liquid and the earlier mixture was powder. The answer, it seems to me, is that there's a gaseous intermediary in all the powder reactions that's able to set off a chain reaction. A little digging, a candidate popped up: P₄O₆, with boiling point 173 °C. I'd guess it reacts with Al, reducing to P₄. In any case, since gas molecule are more mobile that liquid ones, it could explain why the mixture ignites later.

I would have to guess that you've got phase separation at 700 °C, with a puddle of molten Al at the bottom, covered by a boron-phosphorus flux on top. That would mean that at the time the reaction starts you've got a planar interface layer where the reaction occurs, at least before it gets violent.

It's suspicious that the ignition temperature you report is just a few degrees above the melting point of borax, one of the putative products. It looks like the aluminum, whether solid or liquid, might be passivated by borax, kinetically limiting the reaction below its liquid threshold.

What's the enthalpy of this reaction? I know aluminum burns hot, but that crucible is just totally wrecked. Notice how the very bottom of the can is the one part that stayed partly attached. My guess would be that you can see the height of the molten Al in the secant across the arc of metal that's left.

Any reaction that involve combustion of liquid Al is going to need something heavier than sheet metal to contain the heat. I guess it's more obvious when you state it that way. From the photograph, it seems that the lid of the retort stayed on, and that the wall of the retort failed first. Confirm? As a more engineering-oriented comment, there's a hot spot. You need adequate thermal mass in the wall of the container so that it doesn't rise above its melting point at the surface level of the Al (where the reaction is happening) before the heat can dissipate outward.

You mention that the slag has volume about 25% of the retort. How did you measure that?

My one suggestion is to use carbon as part of the initial charge. I don't know that it ought to be the entirety of the reducer, but because its oxidation product is a gas you might see a lowering of the ignition temperature (see first comment). If it lowers below the melting point of Al, then the hot-spot problem will be alleviated. With liquid Al as a reagent, you've got the melting enthalpy of Al to dissipate as heat. I'd guess that the Al powder might still settle at the bottom, but it would be an improvement over a puddle. The carbothermic industrial reaction I reported on is at much higher temperature, but it's also reducing a Ca compound, not an Na one. It might be useful to run a small carbon-only reduction just to estimate its ignition temperature. Also, with liquid salts as the other reagents, porous chunk charcoal might work better than fines that can settle more easily.

Magpie - 24-9-2012 at 17:42

Excellent! As soon as you "harden" your retort, we should see some nice results.

Thanks. Yes, it's back to the drawing board for a more robust retort. The beauty of the paint can is that it is a highly engineered item yet is mass produced. So it is cheap and therefore, expendable.

UnintentionalChaos - 24-9-2012 at 17:48

Excellent! As soon as you "harden" your retort, we should see some nice results.

I have used a beefy retort made from black iron pipe parts that should hold up. The body is a piece of 2" pipe nipple with an appropriate cap on the bottom. A few reducing adapters and an elbow later, I added a long piece of narrow iron tubing. I've used it for half a dozen benzene runs, though the retort volume (and yield per run) is thus low compared to paint can approaches. It also takes a -long- time to get to temperature being several pounds of steel.

I don't think that the entire thing ran me more than $20. If you can weld, sealing the bottom off properly (I did have some seepage out of the threads) would help.

[Edited on 9-25-12 by UnintentionalChaos]

Magpie - 24-9-2012 at 18:17

My guess would be that you can see the height of the molten Al in the secant across the arc of metal that's left.

Sorry I can't easily confirm that as that can has been disposed. With some dumpster diving it would be retrievable, however.

From the photograph, it seems that the lid of the retort stayed on, and that the wall of the retort failed first. Confirm?

The lid and bottom retained integrity. The lid did stay on. The sidewall was trashed, however, as you can see.

You mention that the slag has volume about 25% of the retort. How did you measure that?

It's just an estimate by eye. It might be even less than 25%.

It might be useful to run a small carbon-only reduction just to estimate its ignition temperature.

The only experience I have with carbon is an attempt to run the industrial reaction with Ca3(PO4)2, C, and SiO2. This was in a ceramic tube in a tube furnace. There was no reaction. Maximum temperature was likely 1300C. I apparently did not record this experiment as I can't find it in my notebooks.

Magpie - 24-9-2012 at 18:29

Quote: Originally posted by UnintentionalChaos

I have used a beefy retort made from black iron pipe parts that should hold up.

Yes, I will probably have to go to some metal pipe for the boria reaction. That may be practical as the product is pretty much water soluble and should allow for recovery of the retort with a reasonable amount of effort/risk.

Quote: Originally posted by UnintentionalChaos

If you can weld, sealing the bottom off properly (I did have some seepage out of the threads) would help.

I can't weld (don' have the equipment) but know a SCWIM who can for a reasonable price.
--------------------------------------------------------------
Another possibility is to use a ceramic tube/tube furnace with the boria formulation. Because of the vastly reduced slag the charge could be considerably higher than when using silica, and thereby a decent yield could be achieved. I suppose there is a risk that the boria/borax would flux the mullite ceramic retort and eat through the wall, however.

[Edited on 25-9-2012 by Magpie]

[Edited on 25-9-2012 by Magpie]

Dave Angel - 25-9-2012 at 13:04

Hey Magpie - are you using something like this for your argon?:

I make use of these cylinders and I've just (very carefully after ensuring it was empty!) opened my last one up (for the Downs cell). Perhaps this could be the beginnings of the heavy duty container you seek?

Oh, and nice work so far by the way! I've been silently following this as WP is a little way down the line for me, but I'll look forward to making use of your learnings when I get to it.

Magpie - 25-9-2012 at 15:31

Quote: Originally posted by Dave Angel

Hey Magpie - are you using something like this for your argon?:

No, those don't seem to be available in the US. What I have is a much larger 40 cubic foot cylinder that I get refilled at my gas dealer. I have thought of using an empty disposable propane cylinder that is about that size, however.

I just got back from a trip to my local scrap dealer where I purchased some 0.125" plate, and 3" tube, wall thickness 0.065", for $8.50. Both are stainless steel. I will be visiting my SCWIM tomorrow to get these pieces fabricated into a new, much more beefy, retort.

Magpie - 27-9-2012 at 14:24

... I will be visiting my SCWIM tomorrow to get these pieces fabricated into a new, much more beefy, retort.

I have put a hold on fabrication of a new, ss retort. It's design was predicated on being reuseable. Because of the hardness and insolubility of some slag that hardened on the floor of my furnace during the previous run with boria that assumption may not be valid.

It is a dark, glassy slag. Dark because of unreacted aluminum I believe. The slag forms small bubbles in both HCl and NaOH but is otherwise unaffected. It is difficult to chip using a chisel and hammer. It will grind difficultly under attack with a file. My Dremel with diamond wheel even has trouble cutting it.

So, I have retrieved the slag mass from the dumpster. If I can't find a way to dissolve or otherwise break up this slag I will be forced to revert to an expendable retort body for the boria formulation.

Any suggestions on how to deal with that slag will be appreciated.

watson.fawkes - 27-9-2012 at 16:18

So, I have retrieved the slag mass from the dumpster. If I can't find a way to dissolve or otherwise break up this slag I will be forced to revert to an expendable retort body for the boria formulation.

Any suggestions on how to deal with that slag will be appreciated.

Vycor glass starts life as a kind of alkali-borosilicate glass. It's heat-treated to cause a phase separation into a high-silica (96%) phase and everything else. Then it's leached in acid, which dissolves out the non-silica phase. According to this document, it's hot sulfuric acid 90 °C, concentration not specified. The result glass is micron-sized porous, and it either used porous or heated again to shrink and seal the pores. So my guess is that the heat is a necessary driver to get the acid reagent into the reaction zone by using thermal motion. You don't have a silica phase, but an alumina one. If you're lucky, your slag will leach out similarly.

Magpie - 27-9-2012 at 17:27

Thanks watson for the document. I will do some experimenting with the boria slag in hot sulfuric acid.

Another approach to this problem, which I like even better, is to cheaply harden the paint can. I have had some ideas on how to do this:

1. Using a removable mandrel cast a layer of plaster-of-paris internal to the can. Idea thrown out because water would be driven off during the firing from CaSO4*2H2O.

2. Cast the internal surface with a clay based ceramic. Idea thrown out because of the 10-14% shrinkage expected.

3. Line with furnace cement.

4. Line with kaowool using waterglass as adhesive.

Other suggestions are welcomed.

watson.fawkes - 27-9-2012 at 18:56

Other suggestions are welcomed.

Line with sand and fireclay. Perhaps some grog, or small amounts of bentonite or ball clay. That's what I recall off the top of my head, without research, into one kind of shop-made crucible for small foundry work. The principle is having sand as a body largely eliminates shrinkage. The fireclay sinters enough to hold everything together. Additives such as grog, bentonite, and ball clay are used as shrinkage modifiers and cements (but you don't want very much cement at all in a fire clay mixture).

As I recall, the standard fire clay available on the US West Coast is Lincoln.

Magpie - 27-9-2012 at 19:49

The fireclay sinters enough to hold everything together.

This sounds like the right idea. What temperature is needed to sinter the fireclay? Wiki indicates 1600C for firing. My furnace will only reach 1000C.

[Edited on 28-9-2012 by Magpie]

-----------------------------------------------------------
According to this link, a ceramic of 90%talc/10%clay, fired at 1040C, shrinks very little if at all:

http://www.pottery-magic.com/pottery/clay/low-shrink.htm#.UG...

[Edited on 28-9-2012 by Magpie]

watson.fawkes - 28-9-2012 at 06:45