Slack space is a form of internal fragmentation, i.e. wasted space, on a hard disk. When a file is written to disk it’s stored at the “beginning” of the cluster. A cluster is defined as a collection of logically contiguous sectors and the smallest amount of disk space that can be allocated to hold a file.
Rarely will there be an even match between the space available in a cluster (or collection of clusters for longer files) and the number of bytes in the file. Left over bytes in the cluster are unused, hence the name slack space.
For more information, see "E-Discovery Q & A for Data Warehouse Administrators."