The word fail could mean 2 things , fail as in the drive is useless , or fail as in the raid array has errors but both drives can still be used / and the raid array can be recreated.
I have never heard of raid arrays speeding up true failures that render the drive useless. Running SSD's in raid is rarely necessary for consumer requirements, and running them in raid often prevents trim commands from reaching the drive which help keep performance at its best.
Assuming good quality HDDs and SSDs, they both work just fine in RAID. TRIM is an issue with Hardware RAID controllers which does reduce performance and longevity, as the crap data which has been deleted will be assumed to still exist, and will be moved instead of simply overwritten.
The bigger issue comes in at the end of the SSD's lifetime though. In RAID with parity information, the number of writes across both the parity/mirror drive(s) and the other drive(s) will be very nearly equal in most circumstances. When one drive's write endurance limit is reached, and it stops functioning properly, the others are probably not far behind. This can in some cases lead to the all the drives in the array dying all at once, leaving no time for rebuilding to new drives.
You should really look at your motivation of putting your SSD's in a RAID. If RAID1 it is a great idea since you will have a bit more uptime protection. If you are thinking of RAID0 you will find that while certain benchmark numbers will show large increases (Sequential Transfers) your real-world usage will show negligible to no benefit while doubling your points of failure. You will not see the drastic speed benefits like you did with spinners.