Below is a table of elements used, ordered by boiling point. Beryllium has the lowest boiling point at 2742K. Zirconium has the highest melting point at 2125K. This means the furnace temperature was greater than 2125K.<p>TIG temperature can easily reach 3000C (close enough to 3000K) [1], more than the boiling point of beryllium. The temperature of the metal slug would rise until it reaches thermal equilibrium through heat loss, the energy loss likely being due to the vapourisation of the element with the lowest boiling point: beryllium.<p>That smoke was probably beryllium vapour.<p>I'm no expert, but I would assume the beryllium vapour would not have made it out of the apparatus. By luck, it would have condensed either before reaching the HEPA filter or on its way through the HEPA filter, even though the HEPA filter was not designed to stop a vapour. Nile Red also had the sense to use a fume cupboard. Despite this, the inside of the apparatus could well have been coated with friable condensed beryllium, which would probably not have been flushed by the argon. If doing this, I'd take additional precautions against condensing beryllium vapour containing the equipment (I don't know what they might be).<p>Maybe it's a case of controlling arc temperature to keep it below the boiling point of beryllium? Maybe a small amount of impurity with a lower boiling point can be added to the mix, so it boils off first and avoids the beryllium boiling? This might also prevent loss of beryllium from altering the composition of the glass?<p>How would the pros handle this risk?<p>I gather there are similar issues with the refining of silicon, in which silicon vapour can cause silicosis.<p><pre><code> Element Melt Boil
Beryllium 1560K 2742K
Copper 1357K 2835K
Nickel 1728K 3003K
Titanium 1491K 3560K
Zirconium 2125K 4650K
</code></pre>
[1] <a href="https://hypertextbook.com/facts/2007/AnthonyHo.shtml" rel="nofollow">https://hypertextbook.com/facts/2007/AnthonyHo.shtml</a>