A good topic is one from which you can get a "take home message". Also, the bulk of the paper/talk must explain the take home message. Science explanation must work for non-science & science audience.
Consider "bar code scanners". Science here is how a scanner reads the code the bars represent, independent of bar code orientation. But how the computer, having read the bar code, looks up the UPC in a database is a discussion of databases and not bar codes.
A poor topic is quantum computing, if paper goes like:
"quantum computers will avoid the one-atom limit (false),
uses qubits for data storage and computation,
can do true multitasking (misleading) and
has applications in cryptography."
The paper/talk science concerns how quantum-specific algorithms work. This is difficult and yields an uninteresting talk.
Topics need to have some depth. Consider the topic "how fireworks produce different colors". If the content is limited to a brief discussion of emission lines (for the non-science audience) and a list of which chemicals produce which colors, the content will not be sufficient to interest your science colleagues during your talk.
Astronomy topics are difficult as science tends to be vague.
History of science works only if history illustrates the science.