I think style is more important than topic. A talk with different viewpoints (math - physics - experiment - connections between topics) and different levels of abstraction and complexity can be really great. I find nobody is offended if you spend some time on the big picture as long as you don't dumb down the entire talk. People will forgive you if they become confused when you spend some time trying to be clear about a really complex issue. People tend not to be offended if they are bored or confused by some fraction of the talk (as long as its not the whole thing). And people tend to like things that have big connections or give glimpses to complicated stuff that they would like to learn more about. People like it if you take the time to explain some tricky stuff (even if they forget it two seconds later). And the glimpse to the future or to big (and important) unsolved problems can be a hook for your audience.
Another style issue is to consider your talk as a `story'. People like certain types of stories, ..e.g., I tried this but it failed, I was misled by this idea but then understood why and then I tried this and it worked! (Obstacles overcome by protagonist). If you set up a paradox and then explain it. If you present something as hopelessly complex and then outline a route to breaking it down to understandable pieces. If you suggest a simple experiment and then describe why it gave wrong and misleading results.... I try to remember that as a lecturer I am essentially an entertainer. I am not naturally a joker so the story line is key. I would recommend you choose a topic you really care about (and let that come through in your story) rather than suggest a topic.