The currently executing SQL id is often not the culprit. The sql_id in v$tempseg_usage can be unreliable. See RTSM via v$sql_plan_monitor. Also cross reference underlying ASH tables and, depending on version, increasing values in temp_space_allocated column. Look as well at your top n SQL for the period concerned and between these, it should be easier to track down. ASH and RTSM require diagnostic pack license.