a) Based on the best-fit parameters from the stacked spectrum we derive luminosities for all the individual observations between 2000−2011. The individual data sets are highlighted and so the combined plot is not evenly spaced in time. Observation # are as in Table 1. b), c) and d) Ionisation parameter (ξ) for the individual observations derived using stacked spectrum. Compared to a) it seems ξ varies in response to changing source luminosity. The response time is related to the gas density so the component in phase 3 is possibly denser than the remaining two warm-absorbers.