Any handbook about the strength of materials has a chapter about creep a.k.a. cold flow.
The handbooks from immediately after WW2 already included such a chapter, but I believe that the first studies of this problem must be much older.
When any metallic structure is designed, it must be verified that it will not fail in any of the possible modes, including due to flow over the intended lifetime.
By the approximate rule mentioned there, zinc begins to have non-negligible creep already above minus thirty Celsius degrees, so at normal ambient temperatures you must always compute the creep of zinc for any structural design.
On the other hand, metals like iron or copper have negligible creep at room temperature, even when pure.
Aluminum and magnesium begin to have non-negligible creep at temperatures only a little above normal ambient temperatures.
Hard alloys can have much lower flowing speeds than the metals included in their composition.
In integrated circuits, the metal connections are affected by electromigration, which is the flowing of the metal due to electrical current instead of mechanical stress.
The electromigration properties and creep properties of a metal are closely related. In the beginning, the ICs used pure aluminum for interconnections, but when their size was reduced, the connections began to fail after a too short lifetime.
The first solution for this problem was the replacement of pure aluminum with harder aluminum alloys, including small quantities of copper and/or silicon.
When the ICs became even smaller, the aluminum alloys had to be replaced with a metal having a higher melting temperature, i.e. copper, which fortunately also has a lower resistivity.
Sometimes it’s worth pausing a moment when seeing a “It’s well known that…” to check the timeline because the construction of Arecibo might well have taken place (or planned) before it was known (let alone well known).
Edit: a bit more searching suggests that this was studied in 1947 (Andrade’s Creep Law and the Flow of Zinc Crystals, by AH Cottrell)
This feels like the facility was in operation for so long, that the people who knew about this potential problem all retired and with them went that knowledge.
Andrade is the namesake of Andrade creep due to his work in the early 1900s, but the existence of creep had been known for a long time by then. I'd imagine smiths have been aware of creep throughout the history of metallurgy.
The handbooks from immediately after WW2 already included such a chapter, but I believe that the first studies of this problem must be much older.
When any metallic structure is designed, it must be verified that it will not fail in any of the possible modes, including due to flow over the intended lifetime.
See in:
https://en.wikipedia.org/wiki/Creep_(deformation)
at "Temperature dependence".
By the approximate rule mentioned there, zinc begins to have non-negligible creep already above minus thirty Celsius degrees, so at normal ambient temperatures you must always compute the creep of zinc for any structural design.
On the other hand, metals like iron or copper have negligible creep at room temperature, even when pure.
Aluminum and magnesium begin to have non-negligible creep at temperatures only a little above normal ambient temperatures.
Hard alloys can have much lower flowing speeds than the metals included in their composition.
In integrated circuits, the metal connections are affected by electromigration, which is the flowing of the metal due to electrical current instead of mechanical stress.
The electromigration properties and creep properties of a metal are closely related. In the beginning, the ICs used pure aluminum for interconnections, but when their size was reduced, the connections began to fail after a too short lifetime.
The first solution for this problem was the replacement of pure aluminum with harder aluminum alloys, including small quantities of copper and/or silicon.
When the ICs became even smaller, the aluminum alloys had to be replaced with a metal having a higher melting temperature, i.e. copper, which fortunately also has a lower resistivity.