It is not correct to compare sorted tree (B-tree) and simple binary tree, they are not equal. So i suppose that you mean binary search tree.
B-Tree was designed to be efficient when data are stored on relatively slow storage. For example when you load or save data from file system with cluster size 4kb it doesn't matter how much of data in this range 0..4kb you need, it will take same time to read 1 byte or 4kb and it will really take time. B-tree keeps in mind that fact and use it. So in all normal/general usage scenarios it will be more efficient to use B-tree (from point of view used space and performance).