I recently switched our index rebuild scripts to use tempdb for the sort runs instead of the database containing the index. I had the same issue where the transaction log was ballooning to about half the size of the database during the nightly rebuild task. With a 500 GB database and a Neverfail implementation all that excess log file I/O was definitely becoming a headache.

Switching to tempdb resolved the excessive log growth issue. The amount of time it takes to perform the reindexing also seems to have improved. The only catch is that you have to make sure have sufficient space on your tempdb disk set to accommodate the amount of index data it will store during the sorting. With the size of your database now you really don't have to worry but I would definitely recommend using this (or at least testing the performance) as you manage larger DBs.

Hope this helps.