# Distill Rare Earth Metals

## Recommended Posts

Hello everybody!

The chemical separation of rare earth metals is difficult and costs more than the extraction of the ore. So could distillation separate rare earth metals? (Of course I didn't check if it's already done).

Most 1kPa boiling points spread nicely. Exceptions are 66K for Tb/Pr (or 2.6%, comparable with 100°C and 110°C), 18K for Ce/La, 2K for Pr/Gd. For those pairs, the 1atm boiling points or the melting points spread, so a later step might separate them, distillation at a different pressure or a chemical process.

1kPa       1atm      mp          Element
K          K       K
==============================================
1047       1465    1097    70 Yb ytterbium
1234       1796    1099    63 Eu europium
1421       2061    1345    62 Sm samarium
1570       2217    1818    69 Tm thulium
1954       2831    1680    66 Dy dysprosium
2040       2964    1734    67 Ho holmium
2163       3132    1802    68 Er erbium
2296       3336    1297    60 Nd neodymium
2505       3491    1629    65 Tb terbium
2571       3779    1208    59 Pr praseodymium
2573       3535    1585    64 Gd gadolinium
2653       3663    1925    71 Lu lutetium
2754       3705    1068    58 Ce cerium
2772       3726    1193    57 La lanthanum
==============================================

The distillation tower's material is badly difficult. The best refractory metals have a very low vapour pressure at such temperature (notice the 1kPa, 1Pa and 13mPa), but I suppose they dissolve in the molten mischmetal to pollute the most refractory rare earth outlet and fail mechanically.

 1Pa                 mp          Element
K                  K
==============================================
2639               2128    40 Zr zirconium
2689               2506    72 Hf hafnium
2742               2896    42 Mo molybdenum
2942               2750    41 Nb niobium
3297               3290    73 Ta tantalum
3303               3459    75 Re rhenium
3477               3695    74  W tungsten
==============================================

Would plain ceramic, or a ceramic coating, resist molten mischmetal better? For the following oxides, I'd say no. I've taken only oxides of refractive metals because of MgO's bad example: liquid Al corrodes it as metallic Mg is volatile. The heat of formation per mole of oxygen atoms is -409 to -635kJ/mol while rare earth metals have -568 (Eu3O4) to -633 (Er2O3). I haven't checked the carbides nor borides. Graphite?

13mPa     dHf/O      mp          Compound
K     kJ/mol       K
==============================================
2200       -409    2145          Ta2O5
2300       -635    2683          Y2O3
-380                  Nb2O5
2500       -550    2973          ZrO2
2800       -572    3031          HfO2
==============================================
-568                  Eu3O4
-633                  Er2O3
==============================================

So maybe uncoated C, Ta, Re, W can make the tower to distill the more volatile rare earth metals, which might be useful.

Marc Schaefer, aka Enthalpy

##### Share on other sites

It's an interesting idea. Even if you just split them into a few groups- high boilers, medium and low, you  might make a second step so much easier as to be worthwhile.

I'm pretty sure they will react with carbon.

I suspect that you are near enough to the melting points of all but the most refractory metals that they wouldn't have the mechanical strength to build vacuum chambers from (unless you happened to be on the Moon).

##### Share on other sites

The more refactive lanthanides would distill under lower pressure to spare the column's materials. Beginning here with terbium, all data from Wiki:

10Pa   100Pa    1kPa      mp          Element
K       K       K       K
===================================================
1979    2201    2505    1629    65 Tb terbium
1973    2227    2571    1208    59 Pr praseodymium
2028    2267    2573    1585    64 Gd gadolinium
2103    2346    2653    1925    71 Lu lutetium
2194    2442    2754    1068    58 Ce cerium
2208    2458    2772    1193    57 La lanthanum
===================================================

Tb and Pr spread more at 1kPa, Pr and Gd spread less badly at 10Pa and 100Pa, Ce and La remain close. The process speed must be a limit.

Marc Schaefer, aka Enthalpy

27 minutes ago, John Cuthber said:

I suspect that you are near enough to the melting points of all but the most refractory metals that they wouldn't have the mechanical strength to build vacuum chambers from (unless you happened to be on the Moon).

Hi JC, thanks for your interest!

I have data about Nb, Ta, W, Re creep at heat and could have a look.

In the low-pressure distillation tower, the trays don't feel the atmospheric pressure. Several thin cylinders of refractory material can enclose the trays as a heat insulation that limits the condensation, enclosed in a strong cylinder that operates at a lower temperature. This may need an additional gas like argon.

Marc Schaefer, aka Enthalpy

##### Share on other sites

Ramblings about the solubility of metals in melts.

Ag and Au dissolve in molten Sn63Pb37 but Cu doesn't.

• The difficulty to separate atoms from the solid metal, represented by the vapour pressure, is one factor.
• The temperature for 1kPa is 1782K for Ag, 2281K for Au, 2089K for Cu, so there are other factors.
• The strength with which the melt incorporates foreign metal atoms is the other factor.
• Cu, Ag and Au have similar valences, so this isn't the reason.
• But the molar volumes can explain these metal solubilities. Nothing revolutionary to metallurgy.
     cm3
=========
Cu  7.11
Ag 10.27
Au 10.21
=========
Sn 16.29
Pb 18.26
=========

Sn and Pb being much bulkier, they dissolve Ag and Au a bit, and the smaller Cu less.

So to build a distillation tower, what metal would not dissolve in molten mischmetal?

Refractory metals are small and lanthanides rather big:

     cm3
=========
Re  8.86
W   9.47
Ta 10.85
=========
Er 18.46
...
La 22.39
Yb 24.84
Eu 28.97
=========

This gives some hope that they dissolve little (or slowly). If following that stretched logic, W would dissolve less than Ta, and expensive Re even less.

Marc Schaefer, aka Enthalpy

##### Share on other sites
1 hour ago, Enthalpy said:

Ag and Au dissolve in molten Sn63Pb37 but Cu doesn't.

Yes it does.
I have a dead soldering iron bit which says that it does...

Lithium / lead alloys (which have seriously different sized ions)  are being considered for use in nuclear reactors.

The only way to find out if lanthanides attack tungsten is to try it.

##### Share on other sites

Eu, Yb, La have a molar volume distinctly bigger than the other lanthanides. Could this serve to scoop them from the mischmetal?

     cm3
=========
Sr 33.94
Ba 38.16
K  45.94
=========
Er 18.46
...
Ce 20.69
La 22.39
Yb 24.84
Eu 28.97
=========

Ligands make complexes selectively with preferred ion sizes. I bet this is already done for lanthanides.

Alternately, I suggest the selective dissolution of Eu, then Yb, later La, in a molten metal, if there is such a selectivity. Still stretching the previous logic, the solvent would be a metal with huge molar volume: Sr, Ba, K and few others, at a temperature where lanthanides are little soluble. Then, Eu would dissolve preferentially from mischmetal powder. With more heat, Yb would be scooped, and even more heat would extract La.

Extraction could evaporate the solvent. Moving the mischmetal and the solvent in opposite directions would extract much of the target element but evaporate only well loaded solvent.

Maybe. And playing with matches, the big way.

Marc Schaefer, aka Enthalpy

##### Share on other sites
23 hours ago, John Cuthber said:

Yes it [Cu] does [dissolve in Sn-Pb].
I have a dead soldering iron bit which says that it does...

Most electrical engineers have seen soldering tips eroded over time. Up to now I believed it resulted from the acidity of the flux. If copper in SnPb improves that, fine! But why don't do they it all the time since Cu is cheaper than Sn, instead of selling "neutral flux"?

My point is, if you weld with Sn-Pb once on 5µm Au or Ag, the 5µm are gone immediately. If you weld 7 times on 35µm Cu, the Cu is still there. So Cu is observably less soluble in Sn-Pb, or less quickly, than Ag and Au.

23 hours ago, John Cuthber said:

Lithium / lead alloys (which have seriously different sized ions)  are being considered for use in nuclear reactors.

Yes, an attempt as tritium regenerator blankets for tokamaks. I rang the bell about the pollution by these blankets.

I guess solubility rules address only small solute concentrations. Around 1:1 mole, it's no more a matter of solvent and solute and more of intermetallics.

23 hours ago, John Cuthber said:

The only way to find out if lanthanides attack tungsten is to try it.

Yes!

The sole reason for my ramblings is that I don't have the means to experiment.

Edited by Enthalpy

##### Share on other sites

Even if I completely botched the explanation based on molar volumes, but some melt achieves to dissolve selectively some of the lanthanides from mischmetal powder, I'll be happy of course.

On 11/10/2019 at 8:55 PM, John Cuthber said:

Lithium / lead alloys (which have seriously different sized ions)  are being considered for use in nuclear reactors.

The molar volumes could be computed, but this is already done at
webelements.com
The ratios of molar volumes are

1.4 = 18.3/13.0 for Pb/Li, they mix
1.6 = 16.3/10.2 for Sn/Au and Sn/Ag, dissolve
2.3 = 16.3/7.1 for Sn/Cu, no or little dissolution

The situations aren't quite the same because Pb and Li have melting points (601K and 454K) closer to an other, while I wonder about "dissolution" for Au, Ag, Cu which are far below their melting point (around 1300K) in just molten Sn-Pb (456K).

##### Share on other sites

The effect of atom size matching between metal solute and solvent is long known. One example there (I access only the abstract)
sciencedirect.com
They measured the solubility of Ta and W in lanthanides at varied temperatures, and:
"The solubility of [small] Ta and W vary inversely with the atom size of the lanthanide solvent."
"W is a much better crucible material for the lanthanides than Ta"

There too I access only the abstract:
aip.scitation.org
"Mol/mol solubility of W in Ce is 2ppm at 800°C, increasing to 240ppm at 1540°C."

There on page 25:
ameslab.gov
"Sm, Eu, Yb and Tm can be melted in Ta crucible without contamination" (1815K for Tm)
Also, they use metallic Ca to reduce R from RCl3, and later Ca is evaporated away from R.

This gives me hope that, as I suggested previously

• Some refractory materials let a distillation apparatus separate the most volatile lanthanides.
• Liquid Ca or other can selectively dissolve the biggest (smallest too?) lanthanides and be later evaporated away.

##### Share on other sites

Some rare earth metals have a distinctly smaller molar volume, especially Sc obtained from the same ore, and a few more. Sc is reportedly difficult to separate. Would a metallic melt of small molar volume dissolve selectively these elements? Or the most easily molten or evaporated target elements, combined with the molar volume.

I've listed some small solvent metals, volatile to separate from Sc, Er, Ho. Some add drawbacks to the general explosion risk, Ag seems less dangerous. Alloyed solvents make eutectics.

     cm3
=========
Mn  7.35
Zn  9.16
Ag 10.27
Ga 11.80
Li 13.02
Mg 14.00
Hg 14.09
=========
Sc 15.00
Er 18.46
Ho 18.74
Dy 19.01
Tm 19.13
Tb 19.30
Sm 19.98
Y 19.88
Gd 19.90
...
Eu 28.97
=========

Marc Schaefer, aka Enthalpy

## Create an account

Register a new account