Your AVE value is good indicator of convergent validity since it's above .5 (Fornell & Larcker, 1981). However, CR is higher than .9. So, you should consider dropping some items off the scale (Hair, Black, Babin, & Anderson, 2010).
You would remove items with small loadings (typically, .4 or .5). Alternatively, you'd argue that the AVE has been deemed by some scholars as somewhat a strict measure of validity and hence keep the items with low loadings as long as you have sufficient theory to keep them.