I've used Ed Summers' twarc to collect #critlib tweets. twarc generates a JSON file, which I then convert to CSV using twarc's json2csv script. Files are named with the date when I grabbed the tweets; twarc grabs a week's worth of content at a time.
March 27, 2015: I've added in the JSON and CSV for the #critlib15 unconference held at ACRL 2015 in Portland, OR. I ran twarc around 10pm on Thursday, March 26. twarc had some problems and was repeatedly capturing a single tweet hundreds of times, so I've edited the JSON and the CSV files to remove the 400 or so repeated tweets.
March 31, 2015: It's worth reiterating that twarc searches about 1 week's worth of tweets, so the mar312105 files will have material from last week (which turns out to be a good thing because it looks like some folks were using #critlib rather than #critlib15 during the unconference).