.../...
I don't know where I read this, but I believe it was in one of the thousands of posts.
True, EagleWorks melted the nylon screws holding in the dielectric. The reason was the threads on the nylon screws didn't bind well in the treads. With the small gaps left the threads the microwave actions around the threads heated up the nylon screw and melted it. They used Teflon tape to wrap the screws and make a more secure fit and eliminated the problem of melting.
Shell
My understanding was that the Nylon screws were holding (through) a PTFE disk, the thick slab of dielectric, not that any thin Teflon tape was involved in wrapping things.
I have the same understanding
Upon further review of this subject:
1) I have not found anything on my notes, or on the EM Drive threads (using Google Advanced Search) concerning NASA using Teflon tape to solve this Nylon screw "melting" problem.
Perhaps Shell's memory about Teflon is related to one or both of these things:
1A) One of the dielectric materials that NASA used, (besides the now standard HDPE dielectric) was PTFE, and as we know, DuPont's tradename for PTFE is Teflon. An event described as the Nylon screws "melting" occurred when they were holding a PTFE dielectric, as one can notice in the following photograph:

1B) NASA considered (or used ?) a center bolt made out of PTFE ("Teflon") (*), since the center bolt did not have an issue with "melting" because of its center position (recall that the issue with "melting" screws was mode-dependent and therefore location-dependent as the "melting" was due to induction-heating of the end plate, and not just thermal contact between the induction heated metal end plate but also dielectric heating due to any gap thereof), much lower tan delta than Nylon (PTFE "Teflon" tan δ =0.00028 @ 3 GHz vs. PA "Nylon" tan δ =0.0800 @ 1 MHz ), and higher melting temperature than Nylon (PTFE "Teflon" Tm = 620.3°F vs PA "Nylon" Tm= 428 °F)
1C) aero made MEEP computer runs showing the electromagnetic field being able to escape through thread gaps
2) Review of the material properties makes it clear that the problem with the PA "Nylon" screws was due entirely to its huge value of tan δ (tan δ =0.0800 @ 1 MHz,
(tan δ > 0.1 @ 10 GHz see Krupka) which is known to be
hundreds of times higher than the tan δ for HDPE or PTFE. Hence
dielectric heating of Nylon is hundreds of times greater (for the same electromagnetic power) than for materials like HDPE or PTFE.3) I'm sure that NASA must have solved the PA "Nylon" screw issue purely by addressing the huge value of tan δ, and substituting with a polymer screw of similar tan δ as the HDPE or PTFE dielectrics used by NASA.(*)
4) Also upon reviewing this subject it is obvious that the HDPE dielectric never melted in any of NASA's tests. Not only it was inspected after tests and found to have preserved its shape, but due to its very low tan δ, the HDPE dielectric experiences very low dielectric heating at these levels of power input used in EM Drive tests (as one can readily verify by runnig calculations, using COMSOL, FEKO or ANSYS FEA or BEM), the heating experienced by the dielectric HDPE being mainly due to any thermal contact it may have with the induction-heated metal end.
BOTTOM LINE: PA "Nylon" screw "melting" was due to Nylon's huge value of (imaginary permittivity/(real permittivity)) tan δ ~ 0.1000 at 2 GHz, hundreds of times higher than the tan δ for HDPE or PTFE . This huge value of tan δ for Nylon is related to Nylon's time-temperature dielectric transition (as shown in Krupka's dielectrometer measurements).
The Nylon screws were going through the metal end plate to hold the dielectric, and hence the Nylon screws were subjected to dielectric heating (even if there would not be any electromagnetic fields escaping through thread gaps).
_____________
(*) Looking at readily available (McMaster-Carr), relatively-inexpensive, polymer cap screws, it looks like Polypropylene cap screws would be a great choice as cap screws for this application to hold the HDPE dielectric, because of its higher modulus of elasticity than PTFE (three times greater, see https://en.wikipedia.org/wiki/Young%27s_modulus, 1.5–2 GPA for Polypropylene vs 0.5 GPA for PTFE) while also having a low value of tan δ (see the classic masterpiece by Krupka et.al:
http://tinyurl.com/j899zq4and this nice chart
http://tinyurl.com/z464crm for comparison of PP, PTFE and HDPE and PA)
(mod edit - tinyurl inserted to prevent page break on long urls)