A/B split test confidence

[F8thless]

How do I determine the confidence interval/level (preferably within Excel) if I have 2 percentages, and their samples sizes.

For example: I have 2 ads on Google, showing for the same keyword. Ad 1 has 1,421 impressions and 45 Clicks for a click-thru rate of 3.16%. Ad 2 has 1,452, 41, and 2.82% respectively.

Should I be using clicks as the sample size, or impressions?

What is the test that this would use? - I'm thinking 2 sample t-test, but that seems to need a standard devation of the mean, which I don't think applies here given a %, but in any case, I only have this data.

Thanks for your help, and if you know how to solve this with Excel, that would be the prefered method.