[Enthalpy]

Hello nice people!

Many syntheses use a metal to bind two reactants, cyclize or bridge a reactant... This is usually done in a solvent, which introduces drawbacks like its own reactivity, and can take hours or days. So I wonder if a metal vapour  can do the task, when the reactant can be gaseous.

The vapourized metal being atomized, it becomes much more reactive, which shall address the speed concern  but the obvious consequence is that the metal loses selectivity, so the process isn't expected to fit every subtle reaction; I hope that the careful control of dilutions and residence times can keep some simple reactions under control.

To begin, here under are the vapour pressure versus temperature for some metals. Figures are from
http://en.wikipedia.org/wiki/Vapor_pressures_of_the_elements_(data_page)
which I just put in shape. More curves are in
http://yorkamo.phys.yorku.ca/general_stuff/2012/09/post.html
and some metals achieve a notable vapour pressure at temperatures bearable by organic compounds - by some simple ones.

Among monovalent metals, lithium looks too hot, but sodium and potassium could fit. Rubidium and caesium are more volatile but more bulky. (You may have to log in and click on the images for full size)

Among bivalents, magnesium seems difficult, zinc and cadmium better - since atomic metals must react very quickly, a low pressure can still be productive, and the compounds can leave the hot zone quickly. I've put the volatile mercury as well, though its two valences won't help. Other metals like calcium, strontium, barium are less volatile.

Trivalent metals are little volatile. Thallium is the least bad I saw.

Most metals will be liquid at the operating temperature.

Halides of these metals use to have a high boiling point: at least 1171°C at 1atm for I-VII salts. II-VII salts are less refractory, but oxides of most bivalent metals are. This means that the salt by-products will fall away from the reactants and products.

An example of a reactor is to come.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Here under is a sketched example of a reactor for metal vapour.

The metal can be introduced continuously or not, heated separately or, if any, by the carrier gas.

The carrier gas helps much if other reactants must exceed the metal amount. Because the individual metal atoms are expected to react snappily, they wouldn't have time to dilute in the other reactant, hence would be the major reactant locally where they're introduced. Pre-diluted in the carrier gas, the metal can be the minor reactant right from its introduction, surrounded by a concentrated reactant. The carrier gas, probably a rare gas under such conditions, also gives the flexibility of partial versus total pressure. In addition, it can keep the reaction at a reasonable temperature.

Some reactants can be introduced first, if they must react with the metal before meeting other reactants. I imagine things like CH₃Cl or CH₂Cl₂ reacting with Zn or Cd, then brought to an alkene. The carrier gas can bring these first reactants to the metal. More dynamic reactor designs can also inject the reactants in sequence.

The reactants introduced later don't need to be as hot as the metal vapour. If the metal reacts quickly and is diluted in the carrier gas and the other reactants, it won't have time to meet other metal atoms to condensate.

The sketch lacks, among others, some ways to evacuate the salts and other by-products, nor does it display the subsequent separation of the products, the reactants and the carrier gas.

Liquid metals are corrosive. The pressure vessel must resist at least the outer pressure. And expect the unexpected as in any design.

Marc Schaefer, aka Enthalpy

[Enthalpy]

To check roughly what plain radicals can be created, I've computed heats of reaction. Starting from halogenated methane, as bigger molecules ease the reactions. Just at 298K and producing solid salts, and without evaluating the entropies, so the contentious cases can't be decided from that, but some general conclusions are already clear.

Every monohalomethane can react with any alkaline element to make a methyl radical:
(g)A + CH₃X -> (s)AX + .CH₃X
from the heat criteria, where 300-400kJ are released. The interrogation is rather if a gaseous metal is advantageous, and whether it will damage the reactant.

Free carbene is hard to obtain, and a gaseous metal eases it, at least from the computed heat: see the attached table.

• Two alkaline atoms would easily abstract an oxygen from CH₂O or two halogens from CH₂X₂, but why should they proceed to carbene and stop there? I prefer a bivalent metal.
  
• Heavier halogens improve the reaction heat.
• Lighter metals improve the reaction heat but need a higher temperature to vaporize.
• Mercury must be impossible in every case.
• Magnesium widely outperforms zinc and cadmium. It's the only bivalent that abstracts oxgen from CH₂O.
  
• Zinc and cadmium can abstract only heavy halogens from CH₂X₂. Only an accurate heat of reaction, or better the entropy, can conclude.
  
• A heavier carbene would help.

Free carbene will recombine snappily, so using it demands a special reactor design. Later.

Bridging a dihalo cyclic compounds needs no free carbene and looks easier.

Some salts decompose at heat, like ZnI₂ at 1150°C, which helps recycling at the factory.

Marc Schaefer, aka Enthalpy

[Enthalpy]

At least gaseous potassium is already used and nobody tells me a word...

In the 2007 PhD thesis of Carles Ayats Rius:
Síntesi i reactivitat de derivats del triciclo[3.3.0.0(3,7)]octá (it's stellane)
on page 95 (Pdf p119 of 446) and scheme 3.8:
"la deshalogenació de l'1,4-diiodonorbornà 272 en fase gas, utilitzant potassi atòmic com a agent reductor"
(if you wonder, it's Catalan, more or less understandable to Spanish speakers)

which uses gaseous K to remove iodine from diiodonorbornane and obtain [2.1.1]propellane - single metal atoms bring no other radical, advantage over a liquid, and are more reactive, without any altered surface.

[Enthalpy]

Instead of spreading metal vapour in a big volume, we could evaporate the metal locally and let other reactants flow there. Concentrated laser power is a usual means to evaporate a solid, which doesn't melt substantially if the peak power suffices. For instance short pulses made by a diode-pumped solid laser, possibly with a Q-switch, and concentrated on a spot, are know for this purpose. The spot can be swept, or the target moved, or both; enough speed might enable a Cw laser.

Advantageously, the other reactants feel heat very briefly, only through the metal vapour if they don't absorb the light - lasers for peak power use to radiate infrared.

• Reactants and products less robust to heat get possible.
• Metals less volatile are enabled, for instance with valence >2 (Ga, In, Sb, Tl...).
• Besides the usual coupling metals, others too inert as a solid may become reactive as a vapour.
• The hotter target widens the choice of operating pressure.

The reactor design doesn't seem very constrained. Referring to the sketch:

• The input windows could be a lens.
• Here the metal reactant is a tube evaporated at its tip. With the other reactants (and optional carrier gas) flowing around the tip into the tube, unreacted metal has a chance to condense back on the tube for reuse. The light spot is also more easily swept.
• The proper flow of other reactants can limit the deposition of metal on the reactor walls, and especially on the window.
• The light's direction orients somewhat the vapour emission. A tube hit from inside is better than a wire.
• The tube is fed as it gets consumed. Or the optics could follow the tube's tip.
• The tube can be produced by extrusion within the reactor vessel for longer operation. Have a piston, a die and a coolant or chiller as usual - or even a pump with some soft and fusible metal injected liquid in the die. Here exceptionally, the die could have a hole at the middle for the products.
• More reactants can be fed downstream, within the reactor, and even within the tube.
• The products can be chilled within the reactor, and even within the tube.

It must just need a few trials and tunings...
Marc Schaefer, aka Enthalpy

[Enthalpy]

I've looked if Grignard-like reactants R-M-X can be synthesized using vapours of bivalent metals by trying to evaluate the enthalpy of reaction. The figures in the appended table are very imperfect because I found the enthalpy of formation only for R-Mg-Br and R-Mg-I and in solution, so I tinkered horribly:

• From ZnCl₂, I took the 149kJ/mol for fusion plus vaporization and applied it to the other MX₂...
  
• The Hf for solid MX₂ with Zn, Cd, Hg combined with F, Cl, Br, I come from the CRC Hdbk of Chem&Phys, all at 298K without correction.
  
• So do the Hf for gaseous M(CH₃)₂ and for gaseous metals. Though, CH₃- is known to bind with metals better than bigger hydrocarbyls do.
  
• I took the mean value of gaseous MX₂ and M(CH₃)₂ as an estimate of CH₃-M-X, porca miseria!
  
• But for CH₃-Mg-Br and CH₃-Mg-I I corrected by 31kJ/mol only the enthalpy of dissolution, so it's for the solid rather! CH₃-Mg-X must release 150kJ/mol less when still gaseous.
  

So take with due mistrust.

The result is that

• The heat brought by evaporating the metals makes the synthesis exothermic with Zn and Cd, not just Mg. 130 and 112kJ/mol make the difference. Hg remains marginal. Ca, Sr, Ba have a low vapour pressure, Mg isn't very volatile neither.
  
• F, Cl, Br, I make no big difference here while lighter metals are more reactive.
• The oxide layer at the metal surface is no worry more, traces of moisture and oxygen in the reactants neither. Metal vapours are perfect getters.
• The dangerous activated metal is replaced by metal vapour, which would react with air, but is prepared in minute amount like 1Pa and consumed immediately.

I imagine (wishfully?) the reaction to be immediate. With an excess of R-X it should produce R-R (which alkalis would too). To get R-M-X it seems better to introduce R-X slowly and at low pressure (diluted in a carrier gas?) in the metal vapour and try to remove the R-M-X quickly.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Since the boiling points differ a lot between

• a carrier gas like Ar or if possible N₂;
  
• my favourite reactants and products around C₅ and C₁₀;
  
• useful metals like K for couplings and, if willing to work, Cd for Grignard-like;
  
• the possible resulting salt,

a reactor could separate them autonomously.

In the appended sketch, the hot condensed metal is in direct contact with the products and possibly the gaseous reactants. Depending on the potential consequences, it may better be kept separate. Many varied valves are also needed but not displayed.

The carrier gas must dilute the reaction heat by >100: 100kJ/mol for Grignard-like Cd, 400kJ/mol for a hydrocarbyl and KCl prior to dimerization. Carles Ayats Rius used N₂ in his already mentioned thesis, and the heat didn't destroy the molecule's skeleton. Dilution shall also favor Grignard-likes over dimers when desired.

The reactant(s) is evaporated to the desired partial pressure, diluted and heated. Gaseous metal is supposed to make a quick reaction. K needs +200°C to reach 1Pa, Cd (sublimates) +260°C, ouch; Rb and Cs, since they're recycled? Possible salt is to gather at the bottom and be split for local reuse.

The gas exiting the reaction vessel is to contain little reactants left. Cooling it to an intermediate temperature lets the metal condense back to the reactor. Then, further cooling condenses the products, and the carrier gas is reused.

Heat from the coolers can boil and pre-heat the reactants, since the reaction produces heat and the products often have a higher boiling point.

In the lab, introducing slowly the reactants in an evacuated vessel with heated metal is simpler; Cd reaches 1Pa before melting (+321°C) and can be the vessel.

Opinions welcome of course!
Marc Schaefer, aka Enthalpy

[snorkack]

To present metals, and semimetals, of interest (boiling under 2300 C, to include Ga you mentioned) in a table visible in the posting body

Element symbol	Boils at 100 kPa, C	Vapour pressure 1 Pa, C
Li	1337	524
Na	880	280
Mg	1088	428 (s)
K	756	200
Ca	1482	591 (s)
Mn	2060	955 (s)
Zn	912	337 (s)
Ga	2245	1037
As	601 (s)	280 (s)
Se	685	227
Rb	685	160
Sr	1373	523 (s)
Ag	2160	1010
Cd	767	257 (s)
In	2067	923
Sb	1585	534 (s)
Te	992	data missing
Cs	667	144
Ba	1897	638 (s)
Sm	1788	728 (s)
Eu	1523	590 (s)
Tm	1944	844 (s)
Yb	1192	463 (s)
Hg	356	42
Tl	1485	609
Pb	1754	705
Bi	1562	668
Ra	1526	546 (s)

Also concerning flash heating: if metal vapour is distributed into inert carrier gas and rapidly diluted, then yes it may form supersaturated vapour or if it does condense, form a fine aerosol which is reactive compared to bulk surface.
But laser is not the only option for flash heating: seeing you deal with metals, which are electrically conductive, how about spark discharge?
This way, you can also minimize the heating of your organic reactant: place your spark gap and the metal wire you are evaporating into an outlet of gas, where rapid gas flow blows away the arc and prevents your organic reagent from entering the arc.

[Enthalpy]

Thanks for your interest and contribution!

What kind of temperature organics accept briefly is still unclear to me. Some monovalent and bivalent metals are possible (K is used), the others are doubtful. Maybe experiments bring a good surprise - or not.

I like laser for flash evaporation because it spares the organic reactant, and it has the best power density as short pulses, which produces vapour and nothing else, but lasers have a limited efficiency.

Diluting metal vapour first in the inert gas look like a good alternative as an electric discharge. It's a matter of figures as usual: what pressure, voltage and resulting spark nature, residence time, and so on - I had planned to check it. Paperwork may give first guidelines.

Metal droplets are reactive, yes, and would advantageously replace superfine metal powder, as droplets made in tiny amounts in the reactor are less dangerous. They don't bring the other advantage of vapour, which renders some reactions exothermal, especially Grignard-like with cadmium. Alternatives to droplets already exist but may be less convenient: within the reactor so they don't oxidize before use, extrude many thin wires, grind tiny chips...

If you really want droplets, a rocket injector achieves them, for instance by blowing the carrier gas on the liquid - here premolten metal. If the reaction with the organics is snappy enough, it makes a tiny reactor.

----------

Meanwhile I like less cadmium in the last sketched separating reactor, because it sublimates, so after condensation in the outlet pipes it wouldn't flow back to the melt: improvement needed. Also, the solid reacting with some residual halogenated compounds risks to develop a hermetic salt layer, end of the game.

So this reactor seems to fit monovalent metals better, and flash evaporation retains a strong interest for cadmium.

[snorkack]

Just at 298K and producing solid salts, and without evaluating the entropies, so the contentious cases can't be decided from that, but some general conclusions are already clear.

Every monohalomethane can react with any alkaline element to make a methyl radical:
(g)A + CH₃X -> (s)AX + .CH₃X
from the heat criteria, where 300-400kJ are released.

And of course the reaction actually forms gaseous AX molecules. With poor reaction enthalpy because AX tends to be high-boiling.
Also: when you are dealing with supersaturated vapour, you are apt to have stuff like dilithium.

[Enthalpy]

Precisely because alkali halides aren't very volatile, I expect them to be produced solid, and the reaction to release much heat. I see no good reason why the reactants in the gas phase would make gaseous products. Nucleation may take some induction time, but the reaction supposedly happens at the dust's surface. Just like gaseous HCl and NH₃ make solid NH₄Cl, without passing endothermally by a gaseous salt.

The reaction with gaseous potassium and organic iodides is already used.

If any necessary, the reaction can be started by a powder of seed crystals.

Do you see reasons against?

I don't consider lithium as a first choice, because it needs an excessive temperature. Potassium is a better choice (and already used), and if cheap enough or recycled, caesium and rubidium.

[snorkack]

Precisely because alkali halides aren't very volatile, I expect them to be produced solid, and the reaction to release much heat. I see no good reason why the reactants in the gas phase would make gaseous products. Nucleation may take some induction time, but the reaction supposedly happens at the dust's surface.

Then are you likely to have metal condense on the surface of halide dust?

[Enthalpy]

I can only try to imagine what happens... like the hydrocarbyl's halogen sticking for a limited time on the salt's surface and getting stabilized there if an alkaline metal sticks next to it, usually by rolling on the surface, and when one X and one M more pertain to the salt, the hydrocarbyl radical is released. Or the metal first and the halogen next, this makes no difference.

M-X needs an ionic bond to be part of the growing crystal, and once the halogen receives an electron from the metal it releases the hydrocarbyl.

How widely known is the reaction of gaseous potassium with organic iodides? Maybe the mechanism is already known.

[Enthalpy]

An other already known reaction with gaseous K+Na:
1,4-dichlorobutane to cyclobutane with 70% yield at 220°C and 0.1 torr,
as reported in chapter 48.3.1.2 of
"Science of Synthesis: Houben-Weyl Methods of Molecular Transformations"

Only Li amalgam and dibromobutane in dioxane would improve the yield to 80% but this needs 3h and mercury and dioxane. Gaseous K+Na is supposedly quick because the step to hydrocarbyl radical (even to momentarily gaseous KCl) needs half as much energy as in the bromination to t-Butyl. A smaller reactor reduces the amounts of active reactants and products.

Whether gaseous cadmium improves over potassium? The hydrocarbyl radical hurdle is too high here, but gaseous Grignard-like and CdCl₂ would form exothermically.

[discodermolide]

Toxicity of cadmium salts! Surely this makes this method very unattractive?