Sciencemadness Discussion Board

Sulfur trioxide & Oleum: the ferric sulfate method

garage chemist - 25-2-2008 at 03:13


Sulfur trioxide and Oleum are substances with many potential uses for the amateur chemist, yet, a simple method for their preparation is still to be found.
Since they are manufactured industrially on a very large scale, one could theoretically downscale the principal method of industrial production, the contact process. However, this process requires a good deal of special apparatus, such as a catalyst tube, the gas preparation system, and a washing column to absorb the product gas mixture in concentrated H2SO4 and is therefore difficult to set up.
Fleaker has done so and documented his work:
The vanadium(V)oxide- catalyzed method

I have prepared small amounts of SO3 myself from sodium peroxodisulfate, as documented here:
The persulfate method

Another method that is known to work well is the reaction of phosphorus pentoxide with concentrated sulfuric acid. Since P2O5 is one of the most powerful hygroscopic agents known, it can dehydrate H2SO4 to its anhydride.
This method has been used extensively by members of the german forum to prepare SO3 in quantities of 10-100g since P2O5 is quite available here.
P2O5 is dissolved in concentrated H2SO4, the mixture left to stand until mostly homogenous and then the SO3 distilled with strong heat.
It is, however, unclear what species of (poly?)phosphoric acid is formed in the process, and which ratio of P2O5 to conc. H2SO4 gives the best yield of SO3 in regard to the P2O5 used. Too much H2SO4 is certainly detrimental to the yield since it contains 2-4% water.
The old Oleum & SO3 thread
Engager used this method on page 5 and posted a picture of the solid SO3.

Finally, the method I am going to document here has been discussed in this thread:
SO3 from Iron Sulphate.
On the first page, S.C. Wack posted a document pertaining to the thermal decomposition of different metal sulfates. I based my work partially on this.

The production of both fuming and concentrated sulphuric acid by thermal decomposition of metal sulfates, most importantly iron sulfates (“vitriol burning”), is a very old method and has been known by Islamic scientists since the 8th or 9th century and in Europe since the 13th century.
It was an industrial method of production since the 17th century and remained the only method for production of fuming sulphuric acid until the 19th century.
Green vitriol (ferrous sulphate) was used as the raw material, which was calcined and air-oxidised to basic ferric sulphate before the distillation.


The ideal equation for the thermal decomposition of ferric sulphate is:
Fe2(SO4)3 -----> Fe2O3 + 3 SO3

The theoretical SO3 yield governed by this equation can not be realized in practice, since SO3 is thermally unstable and increasingly dissociates into SO2 and O2 with increasing temperature.
The extent of this dissociation is shown in this diagram:

While the temperature for the decomposition of ferric sulphate is 600- 700°C and therefore more than half of the SO3 should theoretically be obtained undissociated, I found it necessary in practice to heat to 750- 950°C in order to realize a sufficient heat stream through the retort walls and reactants to make the endothermic decomposition of ferric sulphate happen at a reasonable speed (1,5- 2 hours reaction time).
Therefore, considerable decomposition of SO3 is bound to take place. What kinds of SO3 yields can be obtained shall be a subject of this investigation.

Preparation of the starting material

Ferric sulphate can be prepared in a variety of ways.

One would be to add H2SO4 and H2O2 to a solution of ferrous sulphate in order to oxidise the ferrous to the ferric ion and provide the additional sulphuric acid necessary for the formation of neutral ferric sulphate. The solution would then have to be evaporated and the solid dried and calcined to remove water of hydration.

Another method would be to simply dehydrate and oxidise ferrous sulphate by aerial oxygen by heating in an oven. This gives basic ferric sulphate, e.g. ferric sulphate with an excess of Fe2O3 over the theoretical formula. This was the material used in the early preparation of sulphuric acid, since ferrous sulphate (“green vitriol”) occurs naturally.

The method I have chosen is the reaction of concentrated sulphuric acid with red ferric oxide.
Its advantage is that it directly gives a solid product with little water content which can be dehydrated over a Bunsen burner, making this method simple and fast.
The initial reaction is exothermic and will proceed by itself once it has been initiated at one point of the reaction mixture. If the bulk of the reaction mixture is at room temperature when the initiation is done, the reaction is harmless and easily controlled, but if one brings the whole reaction mix to near the initiation temperature before the initiation, it proceeds vigorously with hissing and eruption of steam. If one does larger batches than described, the latter condition must be avoided.
After the initial exothermic reaction, the reaction must be completed and the water removed by further external heating.

An old chemistry book I have describes using 1,5 parts per weight of concentrated sulphuric acid per part Fe2O3. This is only ca. 75% of the theoretically needed amount of H2SO4, since the stochiometry of the reaction
Fe2O3 + 3 H2SO4 -----> Fe2(SO4)3 + 3 H2O
requires ca. 2 parts of H2SO4 per part Fe2O3.
I have first tried using stochiometric amounts of H2SO4 and Fe2O3, but found that this brings the disadvantage that some concentrated H2SO4 (unreacted, apparently) distills off from the ferric sulphate before SO3 evolution starts. Due to the construction of the apparatus, this is highly inconvenient since any conc. H2SO4 that distills over before SO3 evolution starts must be collected separately or left to escape as vapor, otherwise the conc. H2SO4 would continuously condense and drip down into the glowing hot retort, causing steam explosions.
So it is indeed favourable to prepare basic ferric sulphate by using the described excess of Fe2O3 in its preparation.

30g red ferric oxide and 45g concentrated sulphuric acid were mixed thoroughly in a beaker.
Any lumps of Fe2O3 should be crushed beforehand by grinding in a mortar! Incomplete reaction of the lumps can lead to the dangerous unwanted evolution of H2SO4 vapor before SO3 production, as described.

Now the slurry is heated on one spot with a Bunsen burner.
The begin of the reaction shows itself by steaming and solidification of the mix.
Here you can see the reaction front very well, the upper liquid part being the unreacted mix:

After the reaction has subsided, the product is immediately broken up into lumps with a spatula or screwdriver (it becomes hard rapidly) and the granular product coarsely ground in a mortar (I failed to do this here, which turned out to be a mistake- do grind the raw product if you want to do it yourself).
It is then strongly heated over a free flame with constant stirring until no more steam is being given off and a dry powder results. This is how it looked:

This was, however, a bad batch since the (rock hard) granules contained considerable amounts of unreacted Fe2O3 which led to some H2SO4 evolution in the retort later, despite the excess Fe2O3 used. How to remedy a bad batch is described later.
The following picture is of a good, homogenous pulverized batch filled into the retort, it looks reddish due to the excess Fe2O3. This is how it should look like!

This retort is a quartz test tube with 14/23 ground glass joint, 20cm long without the joint and with 25mm outer diameter. The batch fits completely in there, as seen.
The wire construction seen on the joint serves to keep the test tube from sliding down into the inclined tube furnace.

Apparatus and Procedure

The tube furnace is inclined ca. 30 degrees against horizontal. An extension made of quartz is attached to the retort. A pipe adapter made of normal glass holds a bent glass pipe that leads down into the cylindrical receptacle, which is empty and directly condenses most of the SO3 (bp: 44°C). It is cooled with ice-water from the outside.
The joints are completely clean, no grease whatsoever is used. Graphite could maybe be used, I haven’t tried it.
Any organic grease would be turned into black goop by the SO3!

Here the receptacle, it is surrounded with ice water in operation.

The quartz-to-glass joint connection must stay cool (and did so in my setup) since the greater thermal expansion of the glass would otherwise lead to cracking of the outer joint. Keeping the heat away from such a connection is the reason why I had this quartz extension made by the glassblower.

The lower end of the furnace is tightly plugged with ceramic fibre to stop the “chimney effect” (hot air rising through the furnace) from happening and carrying away heat from the inside of the furnace .

After the retort has been filled and inserted into the furnace, it is slowly heated up (ca. 15-20°C/min) without the extension attached to drive out residual water.
The thermocouple is located at the lower third of the test tube.
If at ca. 400- 500°C lots of H2SO4 fumes appear because the ferric sulphate wasn’t homogenous (there ideally shouldn’t be any), the retort is removed from the furnace, left to cool a bit, and 2cm of ferric oxide are put on top of the ferric sulphate. On continuing the heating, the H2SO4 vapor will react with the ferric oxide and is bound up that way. Some more steam will appear as the reaction product.

The quartz extension and the receptacle are not attached before SO3 actually starts to be given off, which has always been the case for me at 750°C. This is the temperature in this picture,
dense white smoke which doesn’t condense in the cool extension is the sign of SO3 production starting.

The receptacle is now connected. The white fumes collect in there.

Now the furnace power is reduced so that the temperature increase is only about 2°C per minute. If the temperature is increased too rapidly, more SO3 will be decomposed and the yield will be smaller.

In the receptacle, a clear mobile liquid slowly collects. The gas in the receptacle also becomes clear since all aerial moisture is bound up in there.

From the top of the receptacle, smoke can be seen pouring out. This is the SO2/O2 gas with a small amount of SO3 that is not being condensed reacting with aerial moisture as it hits the air. A strong smell of SO2 is noticeable, which is why the receptacle is located under the fume hood.

The temperature was increased from 750°C to 950°C over the course of 90 minutes, giving about 2°C per minute, with the heating a bit faster starting at 920°C since the SO3 production was slowing down. At 950°C, it was essentially over.
The manual power regulation on my furnace allows such slow heating curves to be realized easily, opposed to an automatic thermostat that only gives power in single bursts.
A look down the furnace at ca. 800°C, also showing the wire construction:

The raw product:

It weighed 15,4g.

I wondered why I got a liquid product- SO3 should solidify rapidly after preparation, its melting point is 27°C in its lowest-melting modification.
Then I realized that what I got was a solution of SO2 in SO3- those two are miscible with each other, and SO2 will of course dissolve in SO3 to a large extent, depressing its melting point.
I expelled the SO2 by gentle heating and swirling, lifting the stopper slightly to allow the gas to escape.
After the weight of the product was down to 11,8g, the SO3 was then pure enough to solidify:

It seems that what I got was the asbestos-like, polymerized form of SO3.
The black discoloration came from residues of joint grease being turned into black goop by the SO3 and running down the walls of the receptacle.

With 11,8g SO3 obtained from 45g conc. H2SO4, my yield therefore is 33,4%.
Part of the reason for the not optimal yield is surely due to incomplete condensation of the SO3 at 0°C, as the fumes pouring out the top of the receptacle have shown.

A better method for collecting the SO3 out of a gas stream is to absorb it in H2SO4.
H2SO4 absorbs SO3 as disulfuric acid H2S2O7, which has a very low vapour pressure of SO3 over it. Also, the solubility of SO2 in H2SO4 is low.
I therefore did a second experiment with absorption of the product gas in H2SO4 as the method of workup.
The apparatus is simple as well, the glass pipe dips into a test tube with 24,5g conc. H2SO4 and the opening of the test tube is lightly plugged with ceramic fibre to prevent ingression of atmospheric moisture.

The disadvantage of this method is the water content of the conc. H2SO4, which reacts with some of the SO3. One should therefore not use too much H2SO4 in the bubbler.
However, after only 20 minutes of bubbling the gas, a small sample of the H2SO4 in the test tube already fumed lightly in air, showing that the water had already been neutralized!

The absorption of SO3 was very efficient, the exiting SO2/O2 mix did not fume at all upon mixing with air!

After the gas production had subsided at 950°C, the H2SO4 was weighed and its weight had increased by 12,7g. The yield of SO3 was therefore 36%.
If the 24,5g of H2SO4 contained 2% (0,5g) water, 2,2g SO3 have been lost to this.
I have therefore obtained 10,5g SO3 in 37,2g total mass of the oleum, e.g. oleum with 28% SO3.

This is the Fe2O3 residue in the retort after cooling down:

One can see that the material in the lowest part, which has been exposed to the highest temperatures, has shrunken very much and detached from the retort walls. The deep color, almost purple, suggests that crystallization to hematite has taken place since the color of a pigment becomes deeper as the particles become larger.
The material in the upper part, not exposed to the full end temperature of 950°C, has not shrunken, but is still the known Fe2O3. However, it has a core of unreacted material, suggesting that the retort should not be filled as full as I did, or maybe inserted deeper into the furnace.


It has been shown that the thermal decomposition of ferric sulphate is a viable method for the preparation of small amounts of SO3 and oleum in an amateur setting.
The absorption of SO3 in H2SO4, as opposed to the condensation of SO3 at 0°C, has been shown to give a slightly better yield and Oleum directly instead of pure SO3 which is difficult to handle due to its very high vapour pressure and tendency to polymerize into a form that cannot be melted at atmospheric pressure.
Oleum with less than 40% SO3 stays liquid at room temperature, has a low vapour pressure of SO3 and if needed pure SO3 can be distilled out of it.
The 100% H2SO4 that remains behind can be used again to absorb more SO3, leading to the elimination of SO3 loss from water in the conc. H2SO4.
My conclusion is that the absorption of SO3 in sulphuric acid is the better method of product recovery, despite the initial SO3 loss.

It has also been shown that thermal decomposition of SO3 is a big problem with this method, apparently destroying more than half of the SO3.
An idea to partially remedy this problem would be to put a catalyst above the ferric sulphate in the lower temperature zone, allowing some recombination of the SO2 and O2.

However, the low yield of the method must be weighted against the cheapness of the precursors. Since the Fe2O3 is regenerated in the pyrolysis, it is not used up and can be used over and over again to make fresh ferric sulphate. The only necessary precursor is therefore concentrated H2SO4.
If one litre of 98% H2SO4, weighing 1,84kg, is converted to ferric sulphate and decomposed to SO3 with a yield of 36%, 530g SO3 would be obtained.
A litre of 98% H2SO4 costs EUR 7,50 here.
So the method is still quite effective in terms of money.

[Edited on 25-2-2008 by garage chemist]

not_important - 25-2-2008 at 06:42

Nice write-up.

Do you have an feek for the rate of SO3 decomposition at differing temperatures? (yeah, I could chase it down but I'm being lazy)

Fleaker - 25-2-2008 at 17:00

Very nice. What is the maximum amount of ferric sulfate you could make and use in your quartz tube?

len1 - 25-2-2008 at 20:04

Thank you garage chemist, your posts are always worthwhile. I was just wondering, given you carried out the persulfate method before, why would one want to use the ferrous sulfate method given the former runs at a much lower temperature where SO3 is stable? - I understand the low yield you got for the former you think is due to small batch sizes - which seems reasonable. According to the diagram you posted the equilibrium of SO3 at 950C is about 20% so its a surprise indeed you got the yield you have. Plus the conversion of ferrous to ferric is annoying.

Given the 80% disproportionation of SO3 maybe its worthwhile to just heat the ferrous sulphate - the overall increase in SO2 will not be huge over what you already got - and I suspect

2FeSO4 -> Fe2O3 + SO2 + SO3

runs at a lower temperature which will compensate by lower decomposition of SO3, so its yield might actually be the same as from Fe2SO43. It can be absorbed in NaOH and make bisulfite.

[Edited on 26-2-2008 by len1]

garage chemist - 25-2-2008 at 20:08

The batch size of ferric sulfate I described is the amount that will fit into the retort. It's about 64g, if I remember correctly (I didn't weigh the last two batches, just filled them in completely). A little less would be better, as you can see that decomposition temperature is not reached completely in the top part of the retort.
It really depends on the bulk density of your ferric sulfate, the product obtained by dehydrating commercial material might be very different. If it is more dense, you can do larger batches in the same test tube.

not_important, what do you mean by "feek"?

len1, this experiment was done to explore this specific method of SO3 production. Sure, low- temperature methods are better in that they don't decompose the SO3. But I didn't know that I would have to heat so strongly here! The german paper in Saurons ferric sulfate thread specified 600-700°C as the temperature at which ferric sulfate is completely decomposed.
In a search for good methods of SO3 production, I will of course come across ones which don't work very well due to one reason or another.
The persulfate method is simple and doesn't require a furnace, but the yields were quite low as well due to incomplete reaction.

[Edited on 26-2-2008 by garage chemist]

len1 - 25-2-2008 at 20:10

I think he means 'do you have a feel'

I do think FeSO4 is worth a try.

garage chemist - 25-2-2008 at 20:20

Ferrous sulfate first decomposes into basic ferric sulfate with evolution of SO2 upon heating. After that it's the same as with my ferric sulfate: 600- 700°C in theory.
Perhaps I should try holding my ferric sulfate at 700°C for a really long time, so that the reaction takes place slowly and with better yield?

len1 - 25-2-2008 at 20:35

Hmm the book I have lists it decomposing at 550C and gives the reaction scheme as I posted. Perhaps thats like a lot of these books, talking about reactions the author hasnt first experience of.

Temperature has a huge effect on reaction rate (exponential for many reaction I believe) so if you really need 950C for a 'reasonable' rate, I cant imagine below say 850C youll get much at all. Ill have to investigate.

30% would be alright I suppose but I would start from FeSO4 seeing as its much cheaper than H2SO4 especially given the low yield.

The other consideration is the decomposition temperature of sulphates is generally related to the bacisity of the metal oxide. Thus Na2SO4 dont decompose on boiling, CaSO4 at 1200C, ZnSO4 at 900C. So less basic sulphates such as CuSO4 SnSO4 will decompose at even lower temperatures. The metal can be recycled so its cost is not a consideration

garage chemist - 25-2-2008 at 23:20

Well, the SO3 production starts at 750°C and is quite vigorous at 800°C. At 900°C most of the gas has already been expelled and the 950°C were really just to bring it to an end.
I'll experiment with a smaller batch size and with lower temperature for a longer reaction time.

len1 - 26-2-2008 at 00:06

Sorry, my prediction is it wont help, the yield might actually go down, because when you had rapid SO3 elimination you might actually have got SO3 out of the reaction mixture before the SO2 SO3 equilibrium is reached. The gain in equilibrium yield changing T by 50-100 degrees isnt that much to compensate.

But I can suggest the following. This would have to be in steel with a tube at one end which passes thru the furnace. H2SO4 dried air blown into the tube would increase the partial pressure of O2 and so shift equilibrium in the reaction

SO3 -> SO2 + 1/2O2

to the left. By constantly removing SO3 it will also shift the equilibrium

Fe2SO43 -> Fe2O3 + 3SO3

to the right. The thermodynamic effect is a yield of almost 80% as opposed to the 25% now, whether the kinetics bears it out is another matter.

[Edited on 26-2-2008 by len1]

chloric1 - 26-2-2008 at 03:31

Hmm if there was some anhydrous protonating compound that one could add to ferric sulfate then the temperature of SO3 evolution might be lowered. Phosphoric acid condensed by heating comes to mind if only I had a suitable container to generate metaphosphoric acid in.:(:(

not_important - 26-2-2008 at 07:21

'feel' was supposed to be 'feel'.

Len's going after what I was - shifting the equilibrium. I was also thinking of rapid heating and cooling in an attempt to freeze the gases before much decomposition had occurred.

Starting with ferrous sulfate might actually help, in that the SO2 formed could help push the equilibrium , preserving the SO3.

len1 - 26-2-2008 at 19:30

Agreed - and also actually push the SO3 along out of the reaction zone, in a literal sense.

That is it will reduce SO3 partial pressure pushing Fe2(SO4)3 decomposition to the right, while by increasing SO2 partial pressure it will push SO3 decomposition to the left, as well as flushing the SO3 out of the high-T zone.