Sciencemadness Discussion Board - HMX via Bachmann route

HMX via Bachmann route

markx - 5-8-2014 at 04:42

Continuation of the experiments with 70% nitric acid in Bachmann process as discussed previously in the RDX synthesis thread:

http://www.sciencemadness.org/talk/viewthread.php?tid=250&am...

To push things a little further I thought to try for HMX synthesis with the low 70% nitric acid. If RDX is well possible, then HMX should also work acceptably well. And if it does, then it might just deserve its own thread.

As a basis for selecting the suitable synthesis conditions and methods were the following documents:

http://www.google.com/patents/US2983725
http://www.google.com/patents/US2798870
http://www.google.com/patents/US2678927
http://nopr.niscair.res.in/bitstream/123456789/7060/1/IJCT%2013(4)%20404-410.pdf
http://www.chemikinternational.com/pdf/2012/01_2012/CHEMIK_2...

Especially the US2983725 patent seems to offer a rather convenient method for small scale amateur synthesis of HMX without the separation of intermediate precursors (DPT). As from the viewpoint of industrial scale production, the method described in the patent is not very suitable, since a large amount of waste acetic acid is generated. It is hard to utililize and regenerate, hence the large scale manufacture tends to opt for other routes that generate less waste.
For my small scale experiment it poses no obctructions and I happily opted for the synthesis procedure described in US2983725 with the amount of acetic anhydride corrected to compensate for the extra water in the 70% nitric acid that I use.

Experimental part

Into the thermostated reactor at 44C were added as heel:
15,7g Glacial acetic acid (GAA)
0,26g Ac₂O
0,34g Paraformaline

Magnetic stirring was applied.
The following liquid feeds were prepared:

10,29g Ac₂O
4,66g Ammonium nitrate (AN) + 6,17g HNO₃ 70%
2,02g Hexamine + 3,3g GAA

The liquid feeds were added to reactor in 4 portions during 15min at 44C.
1,33g Hexamine in GAA + 1,93g Ac₂O at a time. After all the hexamine solution had been added, another portion of 2,56g Ac₂O was added in bulk.
DPT was formed gradually as a white precipitate:

The reaction mixture was aged for another 15 min at 44C.

As the second stage (conversion of DPT to HMX) the following liquid feeds were added to the reactor at 44C:

9,6g Ac₂O + 5,42g AN in HNO₃

The addition was as follows:

Gradual and proportional addition of 6,4g of Ac₂O and 5,42g AN in HNO₃ in approximately 15minutes, followed by 3,2g of Ac₂O in bulk.

The DPT layer was dissolved quite rapidly and almost completely during the first additions of the liquid feeds:

After about 20 minutes from the start of the second liquid feed stage a gradual appearance of another white precipitate was observed (HMX):

The reaction mixture was let age for 60 min this time, at the end of which the reactor was filled with a thick slurry of white matter (supposedly HMX).

7g of water was added to the reactor in bulk and the temperature brought up to 98C for 30min. The digestion stage will purify the product by hydrolysing some side products.
After the purification, the reactor contents were crashed into 500ml of cold water and the formation of a thick voluminous precipitate was observed:

[Edited on 6-8-2014 by markx]

markx - 5-8-2014 at 04:54

The precipitate was filtered, neutralised with sodium bicarbonate solution and washed with coupious amounts of water:

After being squeezed dry with paper dowels and dryed at room temperature for 24 hours, the final yield was 2,8g of HMX. That translates to 66,5% yield if my calculations have not failed on me...

Bert - 5-8-2014 at 05:45

Looking forward to watching this thread develop!

As the most difficult reagent for an amateur to acquire is often acetic anhydride, my curiosity caused a quick scan of the procedure & yield. ROUGHLY 10 grams of anhydride was used per gram of HMX produced... If product was pure HMX.

Praxichys - 5-8-2014 at 07:53

Your yield might not be a whole lot of HMX.

HMX is produced by the nitration of hexamine with ammonium nitrate and nitric acid in an acetic acid/acetic anhydride solvent at 44°C. The raw materials are mixed in a two-step process and the product is purified by recrystallization. This is a modification of the Bachmann Process used to produce RDX, another explosive. The yield of HMX is about 55-60%, with RDX as an impurity. RDX produced by the Bachmann Process usually contains about 8-12% HMX as an acceptable byproduct.

Do you have a means of analysis? Melting point depression could tell you how much RDX is in your HMX. Density might also be another option.

markx - 5-8-2014 at 08:10

Unfortunately I have no reliable means for an in depth analysis at the time to comment on the purity or composition of the product. But constructing a melting point measuring apparatus could be a good starting point. I will also try to revive my old contacts in the university, perhaps I can gain access to machinery that allows for a more reliable analysis of the topic.

Dany - 5-8-2014 at 08:32

For Markx and Praxichys,

there is a fast solution to determine if at least there is one or a mixture of compounds. Do a TLC plate. For this, dissolve the compound in hand in acetone and spotted on silica gel plate. The plate are developed with acetone-benzene eluant. The HMX and RDX are separated and detected as dark spots under 254 nm UV light. From the number of spots and the intensity under UV light you can at least estimate how pure is your sample. The procedure described here can be found in the following document page 3. This procedure is specific for RDX and HMX mixture.

P.S. if you don't have benzene than try toluene.

www.dtic.mil/dtic/tr/fulltext/u2/a155983.pdf

Dany.

[Edited on 5-8-2014 by Dany]

DubaiAmateurRocketry - 5-8-2014 at 08:50

You can also find its density? Assuming only RDX and HMX exists in the final product, measuring the density shall tell you everything

Dany - 5-8-2014 at 09:56

I don't advise the density method. This method is only applicable if one is sure that the sample contain only RDX and HMX. The presence of impurities will lead to erroneous determination of the composition. Also, this method will not give accurate value if one compound is very minor relative to the other. The first thing a synthetic chemist do is to check his réaction média or the isolated products with a TLC plate. NMR or other analytical studies should be performed latter.

Dany.

DubaiAmateurRocketry - 5-8-2014 at 10:04

Yeah I agree that finding the density might not be a good determination. I wonder what is the smallest cheapest NMR or Spectrophotometer we can purchase?

roXefeller - 6-8-2014 at 02:08

Have you considered starting from hexamine dinitrate? If it helps with RDX, maybe it can help with HMX.

markx - 6-8-2014 at 03:13

Quote: Originally posted by roXefeller

Have you considered starting from hexamine dinitrate? If it helps with RDX, maybe it can help with HMX.

For the sake of scientific curiosity it can be easily tried out, but for practical purposes I personally see little gain in introducing 2 nitro groups into the setting via a hygroscopic, realatively unstable and poisonous substance (HDN), that adds a whole new synthesis chapter to an already rather complicated system. Aside from that I do not recall having seen an HMX synthesis based on HDN...might be wrong though.

roXefeller - 6-8-2014 at 13:24

Winkler posits that the Bachmann process actually forms hexamine mononitrate as an intermediate, while the direct nitrolysis forms the dinitrate intermediate.

In direct nitrolysis, substituting HDN for hexamine is intended to help the yields in the absence of optimum quantity or quality of reagents, like how you are substituting 70% HNO3. There is less demand for concentration when two moles of -NO3 are present.

Edit: I've included a paper by Winkler when he wrote up his results for differential hydrolysis to measure percent HMX in RDX.

[Edited on 7-8-2014 by roXefeller]

Attachment: differentialhydrolysis.pdf (112kB)
This file has been downloaded 990 times

markx - 7-8-2014 at 02:34

Quote: Originally posted by roXefeller

Winkler posits that the Bachmann process actually forms hexamine mononitrate as an intermediate, while the direct nitrolysis forms the dinitrate intermediate.

In direct nitrolysis, substituting HDN for hexamine is intended to help the yields in the absence of optimum quantity or quality of reagents, like how you are substituting 70% HNO3. There is less demand for concentration when two moles of -NO3 are present.

Edit: I've included a paper by Winkler when he wrote up his results for differential hydrolysis to measure percent HMX in RDX.

[Edited on 7-8-2014 by roXefeller]

Yes, of course you are right about the lesser need for dehydration agent when using the NO3- incorporated into the HDN structure, but what I meant under "practical purposes" is the fact that this decrease for does not come for free. Ultimately the NO3- has to come from nitric acid, albeit not a highly concentrated one and you need to separate and dry the HDN (probably under vaccuum or dessicant to separate the water completely) whilst dealing with its poisonous nature. All that adds up to a lot of extra work which can be avoided with a little extra Ac2O. But that is a matter of choice, taste, curiosity and scale....in principle I do agree with you on the concept

BTW...does anyone have an opinion on the feasibility to try the TLC method with filter paper as the column medium? I do not have the original silica coated plates available, but perhaps a strong porous paper can serve as the separation medium as well.... The retention constants will be different though, so the direct comparison with official results could be problematic, but at least we should get an idea of the number and extent of components in the product?

[Edited on 7-8-2014 by markx]

markx - 10-8-2014 at 08:54

Well gentlemen....I constructed a makeshift melting point determination apparatus and ran the whole lot of substances that I had in my toolbox through it. The result were following:

1) ETN, recristallised from ethanol: 62-63C
2) PETN, mixed acid synthesis, recristalised from acetone: 141-142C
3) RDX, Bachmann route, full fraction of product. not recristallised: 197C
4) RDX, Bachmann route, first fraction obtained by cooling of synthesis mixture, not recristallised: 203-205C
5) HMX, Bachmann route, recristallised from acetone: 282-283C

There seems to be a systematical bias of few degrees upwards...not a surprise with an uncalibrated K-thermocouple. But all in all the values add up and the HMX seems to be quite pure too

Some photographical evidence:

BTW, I will have to repeat the HMX synthesis and measure the mp before and after recristallisation. Right now I have only a recristallised sample left from the first synthesis. I can't comment on how much product was lost by the recristallisation process because I messed it up. I always hot filter the recristallisation solution to remove any insoluble particles from it before the actual recristallisation occurs via cooling or dilution. But with HMX this proved to be near impossible....the acetone solution immediately cloggéd the filter paper and not a single drop came through. I tried several types of papers, but all I accomplished was a considerable loss of solution that was soaked into the filter paper and became unredeemable. Hence I crashed the remaining solution directly into water and saved what I could. I did not even bother to weigh the result...it would not have told me anything reasonable in terms of process yields.

I will try the TLC plate method too, but first I must perform another synthesis to get a representing sample of product. I will recristallise most of it and leave a small sample for analysis from the raw alpha HMX directly from synthesis. That way we can get a nice comparable result on the effect of recristallisation and the yields and of course the number and rough estimation of content of products formed.

[Edited on 11-8-2014 by markx]

markx - 14-8-2014 at 12:28

I did try the TLC method using filter paper as the column media and 50/50 by volume mixture of acetone/toluene as the eluent...it's a no go. The filter paper is absorbing the eluent at a relativistic speed and everything is rushed up by the solvent front...no reasonable separation is achieved. I rigged up a UV-c germicidal tube to light up any information on the TLC plate, but I could not make out anything of useful purpose. I guess I'll have to order some proper TLC plates from ebay or try to make some DIY ones from microscope slides.
Is there any other reasonably realised methods for developing a TLC plate with nitramines besides the UV radiation? I did find some methods on the web, but none of them seem to be very suitable for the TLC plate.

Praxichys - 15-8-2014 at 04:09

This is pure speculation, but perhaps if the plates were slowly heated, varying decomposition temperatures of some compounds might be evident?

markx - 17-8-2014 at 23:14

Quote: Originally posted by Praxichys

This is pure speculation, but perhaps if the plates were slowly heated, varying decomposition temperatures of some compounds might be evident?

I think if a paper substrate is used, the decomposition of cellulose will appear before any of the nitramines show signs of alteration. Anyways, the fisrt step should be the aquisition of proper TLC plates. Constructing the needed tools for any job will always draw attention away from the objective and make the whole process very tedious. Therefore I have begun to accept the philosophy of buying the tools instead of making them (at least the simple ones)

As an example of the opposite I did construct a drop tester:

A simple design with a quick release dropping a 2kg hammer from the preset height onto a sample packed in Al foil. Works like a charm.
The results:
ETN: 30cm
PETN: 40cm
RDX Bachmann route mp 197C: 55cm
RDX Bachmann route mp 203-205C : 65cm
HMX Bachmann route mp 260C: 78cm (a single partial detonation at maximum height of coulmn, I could not replicate the result)
HMX Bachmann route mp 282C*: - no go at maximum height of 78cm with 2kg weight

* I must note a correction to the melting point of recristallised HMX sample from first synthesis. The corrected mp is 270C and there is clearly another component(s) in the product (supposedly RDX). The measuring technique makes all the difference in the final result. At first go I supplied the sample of HMX onto a relatively cool (150C) measuring apparatus and raised the temperature slowly until it reached the 282C mark and the sample melted abrubtly. But that was all wrong, because what I actually did, was to evaporate and decompose all the impurities in the sample as I raised the temperature slowly and was left with the highest melting component in the end...HMX. A simple distillation if you will. Since the sample was small I did not notice the other components leaving it. After realising my error, I heated up the apparatus to the expected mp and then placed the sample on to the machine. This revealed the corrected values for HMX samples.
Not recristallised alpha form of HMX, directly from syntesis: 260C
The sample softened at 255C and melted at 260C after which the impurities quickly boiled off and the sample resolidified until 282,7C was reached.

Recristallised sample of beta HMX from first synth: 270C
Sample softened at 260C and partly melted at 270C (crystals became transparent and a liquid phase was formed). Again the impurities boiled off quickly and sample resolidified until 282,7C was reached.

I also remeasured the mp of the rest of substances mentioned in previous posts ( RDX, PETN, ETN)...their mp-s remained the same.

Jimbo Jones - 17-8-2014 at 23:37

Excellent work. Thanks for sharing.

Vpatent357 - 18-8-2014 at 01:22

i like it, can you make a hammer test drop with SA.DS for show, pm me

Thanks

markx - 18-8-2014 at 04:55

Quote: Originally posted by Vpatent357

i like it, can you make a hammer test drop with SA.DS for show, pm me

Thanks

It will be a short fall and a loud bang...to gain any systematic knowledge about the matter of SADS a series of experiments is in order. Different synthesis conditions vs impact- friction- and ED sensitivity. Perhaps even some mixtures of SADS with energetic or OB balancing additives, even though the mixture part fills me with doubt and fear. Perhaps incorporating the additives as a solution would make reasonable sense. Anyhow, I have it planned for the future, but I am in the midst of constructing the friction sensitivity measuring rig right now. Anybody have any clever ideas for constructing an ED sensitivity measuring machine?
A set of HV capacitors with varying values, coupled to a low power switch mode dc HV generator would perhaps be the simplest option. The combination of capacitance and varied spark gap distance can be recalculated into energy, assuming the dielectric properties of air are constant over different experiments. In reality it can be subject to change (relative humidity). The reliable determination of the energy of the discharge seems to be a tough part, but without that the results are pretty much worthless.

Hennig Brand - 18-8-2014 at 08:13

Good job. Building a drop hammer device has been on my to-do list for a while now.

roXefeller - 18-8-2014 at 17:07

Quote: Originally posted by markx

In reality it can be subject to change (relative humidity). The reliable determination of the energy of the discharge seems to be a tough part, but without that the results are pretty much worthless.

Can you purge and fill a bell jar-enclosed tester with argon weld gas?

markx - 20-8-2014 at 04:10

Quote: Originally posted by roXefeller

Quote: Originally posted by markx

In reality it can be subject to change (relative humidity). The reliable determination of the energy of the discharge seems to be a tough part, but without that the results are pretty much worthless.

Can you purge and fill a bell jar-enclosed tester with argon weld gas?

Can be done....actually only the electrodes and the sample need to be enclosed in a purge chamber. The rest of the tester can be situated in normal atmosphere. Although the chamber has to be able to withstand quite a shock.

Freeze frame of a 2-3mg PETN sample going off in the drop tester:

As can be seen from the size of the fireball, the term "energetic material" is well justified.

markx - 24-8-2014 at 07:51

A few frames of the melting point measuring apparatus....I built a more civilised setup with a USB camera looking into the sample well:

I must admit the heat transfer block from a welder rectifier diodes heat sink is not the best of choices, but it was the only one I had laying around. I probably will exchange it for a piece of rectangular aluminium stock if I get the chance, the heat loss at higher temperatures is just too big (it is a heat sink after all). The setup is powered by 220V 100W mould heater. A very compact and powerful heating element that is compatible with my trusted PID controller.

Hennig Brand - 6-12-2014 at 10:16

You have built some very neat things. I really like the drop hammer/impact sensitivity tester. For the melting point apparatus I would suggest a block, or cylinder would even be better, of aluminum with no fins and maybe even some added insulation would be a good idea. What works best is a large thermal mass. A large thermal mass resists sudden changes in temperature (the thermal flywheel effect), which is what works best when trying to get an accurate value for melting point. Fins are designed to allow the heat sink to transfer heat/energy to the surroundings as quickly as possible. I see that you did already say that it wasn't the best choice of heat sink, but I thought I would comment anyway.

markx - 6-12-2014 at 13:06

Quote: Originally posted by Hennig Brand

You have built some very neat things. I really like the drop hammer/impact sensitivity tester. For the melting point apparatus I would suggest a block, or cylinder would even be better, of aluminum with no fins and maybe even some added insulation would be a good idea. What works best is a large thermal mass. A large thermal mass resists sudden changes in temperature (the thermal flywheel effect), which is what works best when trying to get an accurate value for melting point. Fins are designed to allow the heat sink to transfer heat/energy to the surroundings as quickly as possible. I see that you did already say that it wasn't the best choice of heat sink, but I thought I would comment anyway.

You are absolutely right....hence I did reconstruct the contraption to a more suitable form for the task at hand:

http://www.sciencemadness.org/talk/viewthread.php?tid=250&am...

I cut off the fins and added an insulating layer of wood (again perhaps not the best of choices for temperatures upwards of 280C) around the block of remaining aluminium. Works like a charm now

Hennig Brand - 6-12-2014 at 13:27

Nice looking setup you got there. I have been using a three or four hundred gram block of aluminum heated with a low flame on a propane torch and a thermocouple probe to measure temperature. Not very high tech, but with a good size piece of aluminum (fairly large thermal flywheel) and the thermocouple probe calibrated against a decent mercury thermometer the setup gets me reasonably close. Your setup looks quite accurate and sophisticated, nice work.

markx - 6-12-2014 at 14:04

Quote: Originally posted by Hennig Brand

Nice looking setup you got there. I have been using a three or four hundred gram block of aluminum heated with a low flame on a propane torch and a thermocouple probe to measure temperature. Not very high tech, but with a good size piece of aluminum (fairly large thermal flywheel) and the thermocouple probe calibrated against a decent mercury thermometer the setup gets me reasonably close. Your setup looks quite accurate and sophisticated, nice work.

The PID controller+SSR combo that I constructed (I really can not remember for what purpose exactly) has so far proven to be the most versatile piece of equipment ever. I can use it to control basically any equipment that works off the 220V line and can be PWM-d. Starting from soldering iron and ending up with the HMX reactor or a melting point apparatus

But the key is to properly calibrate the sensor against a reliable standard if one wishes to get precise results. The k type thermocouples can be off several degrees, especially in the low end spectrum (say below 100C ). Always calibrate against the desired setpoint...the fact that the sensor is true at 350C does not mean that it will be true at 70C.

Hennig Brand - 6-12-2014 at 17:51

I recently calibrated one of my k type thermocouple probes and meters against a good quality fisher scientific mercury thermometer at about 80C in hot water. I was aiming for about the melting point of TNT. It was off (low) somewhere between 2 and 2.5C. I have tested the mercury thermometer in boiling water too just to make sure it showed 100C. I have some cheap alcohol filled thermometers that read high by as much as 5C in boiling water. I don't mean to imply that only mercury thermometers are accurate however. I was going to go with an electrical resistance heater too, because of the great control which can be obtained, but I never got around to it yet.

PHILOU Zrealone - 8-12-2014 at 09:06

This temperature problem is why in lab we usually do mp testing in parallel inside glass capilarity tube.
Pure sample (or other compound with identical or close mp) vs the unknown/to test sample.

nzyzaw - 15-10-2018 at 09:16

please, why crude HMX and RDX mixture dissolved in hot water and don't crystallization

nzyzaw - 25-10-2018 at 07:38

[Edited on 25-10-2018 by nzyzaw]