Modern CPUs are pipelined, meaning that the clock signal doesn't have to propagate across the entire chip each tick. Instead the subsections act like a bucket brigade, with each one handing the results of the partially completed work to the next stage.
There is a limit to how small these subsections can be practically made due to pipeline length and the formation of "bubbles" on instruction branching that take time to clear. Eventually the cost of these bubbles outweighs the gains made from more pipeline stages.
So, if a 6502 or a Z80 were smaller than a single pipeline stage in a modern processor it could potentially have an even higher clockrate, albeit while doing massively less work per clock cycle. Though thermals might be a bottleneck before clock propagation is. Very small but extremely hot spots can be a problem in modern CPU design, with parts of e.g. the ALUs rising to unacceptable temperatures even as all the silicon around them is relatively cool. IIRC some Intel CPUs actually have instructions that are only able to be executed a limited number of times per second for this reason.
So an extremely small, extremely fast/hot 6502 might not have a much faster clockrate than a modern chip.