Timescale-dependent X-ray to UV time lags of NGC 4593 using high-intensity XMM-Newton observations with Swift and AstroSat

TL;DR

This study uses high-cadence XMM-Newton, Swift, and AstroSat observations to measure and analyze the timescale-dependent X-ray to UV lags in NGC 4593, revealing shorter lags than previously reported and supporting disc reprocessing models.

Contribution

It provides the first detailed measurement of UV/X-ray lags on short timescales with simultaneous multi-instrument data, refining previous lag estimates and confirming disc reprocessing models.

Findings

UVW1 lags behind X-rays by ~30 ks, shorter than previous measurements.

Revealed bimodal lag distribution in Swift data, with a dominant short lag after detrending.

Lag spectrum consistent with disc reprocessing models, including BLR contributions.

Abstract

We present a 140ks observation of NGC 4593 with XMM-Newton providing simultaneous and continuous PN X-ray and OM UV (UVW1 2910\AA) lightcurves which sample short-timescale variations better than previous observations. These observations were simultaneous with 22d of Swift X-ray and UV/optical monitoring, reported previously, and 4d of AstroSat X-ray (SXT), far (FUV 1541\AA), and near (NUV 2632\AA) UV allowing lag measurements between them and the highly-sampled XMM. From the XMM we find that UVW1 lags behind the X-rays by 29.5$\pm$1.3ks, $\sim$half the lag previously determined from the Swift monitoring. Re-examination of the \textit{Swift} data reveals a bimodal lag distribution, with evidence for both the long and short lags. However if we detrend the Swift lightcurves by LOWESS filtering with a 5d width, only the shorter lag (23.8$\pm$21.2ks) remains. The NUV observations, compared to PN and SXT, confirm the $\sim$30ks lag found by XMM and, after 4d filtering is applied to remove the long-timescale component, the FUV shows a lag of $\sim$23ks. The resultant new UVW1, FUV, and NUV lag spectrum extends to the X-ray band without requiring additional X-ray to UV lag offset, which if the UV arises from reprocessing of X-rays, implies direct illumination of the reprocessor. By referencing previous Swift and HST lag measurements, we obtain an X-ray to optical lag spectrum which agrees with a model using the KYNreverb disc-reprocessing code, assuming the accepted mass of $7.63\times10^{6}M_{\odot}$ and a spin approaching maximum. Previously noted lag contribution from the BLR in the Balmer and Paschen continua are still prominent.