Both sequence diagrams are correct from an UML perspective, but they show slightly different sequences.
In the first diagram, with the new activation on object2, the call from object2 to object6, with the new activation, is not the result of the asynchronous call from object4 to object2. Rather it is initiated independently by a trigger that is not shown in the sequence diagram.
In the second diagram, with one continuous activation on object2, the call from object2 to object6 does happen in response to the asynchronous call to object2.
Based on how you draw your diagrams, a tool like Sparx Enterprise Architect will make some assumptions regarding what you mean and create the diagrams according to those assumptions. Typically you can do things like extending an activation box to tell the tool that a second outgoing arrow belongs to the same activation.