Results from a Systematic Survey of X-Ray Emission from Hydrogen-poor Superluminous SNe

Superluminous supernovae (SLSNe) are among the most luminous known stellar explosions in the universe. Only recently recognized as a class, in 2009 (Chomiuk et al. 2011; Quimby et al. 2011c), SLSNe reach optical-UV luminosities L > $7\times {10}^{43}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$, $\sim 10\mbox{--}100$ more luminous than common SNe, and are likely associated with the death of massive stars. The source of energy powering their exceptional energy release is still debated (e.g., Gal-Yam 2012). The proposed energy sources include: (i) radioactive decay of large amounts of freshly synthesized ⁵⁶Ni (M_Ni ≳ $5\,{M}_{\odot }$), a signature of pair-instability explosions (as proposed for SN 2007bi, Gal-Yam et al. 2009); (ii) SN shock interaction with dense material in the environment (e.g., Smith & McCray 2007; Chevalier & Irwin 2011); and (iii) a magnetar central engine (e.g., Kasen & Bildsten 2010; Woosley 2010; Nicholl et al. 2013). The narrow features ($v\lesssim 100\,\mathrm{km}\,{{\rm{s}}}^{-1}$) in the spectra of hydrogen-rich SLSNe like SN 2006gy clearly indicate that the interaction of the SN blast wave with the medium plays a role (e.g., Ofek et al. 2007; Smith & McCray 2007), while it is unclear if a single mechanism can power hydrogen-stripped SLSNe (i.e., SLSNe-I). Indeed, SLSN-I iPTF 13ehe has been interpreted as the combination of energy extracted from a magnetar central engine coupled with radiation from the radioactive decay of $\sim 2.5\,{M}_{\odot }$ of ⁵⁶Ni, and a late-time interaction of the SN shock with the medium (Yan et al. 2015; Wang et al. 2016b).

A number of independent lines of evidence support the idea that SLSNe-I might harbor an engine. Observations of SLSN-I host galaxies indicate a preference for low-metallicity environments, which inspired a connection with long gamma-ray bursts (GRBs, Lunnan et al. 2014; Leloudas et al. 2015b; Chen et al. 2017; Perley et al. 2016; see, however, Angus et al. 2016). Along the same line, Greiner et al. (2015) reported the detection of SN 2011kl associated with GRB 111209A (Kann et al. 2016) with color and luminosity properties that are reminiscent of SLSNe, and suggested that a magnetar central engine powered both the initial burst of γ-rays and the later optical/UV SN emission (Metzger et al. 2015). Milisavljevic et al. (2013) found links between the late-time emission properties of a subset of energetic, slow-evolving supernovae, and the superluminous SN 2007bi. They suggested that a single, possibly jetted, explosion mechanism may unify all of these events that span $-21\leqslant {M}_{B}\leqslant -17$ mag.¹¹ Additionally, nebular spectroscopic studies by Nicholl et al. (2016b) revealed similarities between the SLSN-I 2015bn and SN 1998bw, associated with GRB 980425, suggesting that the cores of their massive progenitors shared a similar structure at the time of collapse. In another source, SLSN-I Gaia 16apd, Nicholl et al. (2017) further demonstrated that the luminous excess of UV emission originates from a central source of energy, rather than reduced UV absorption or shock interaction with a thick medium (see, however, Yan et al. 2017). Finally, luminous X-ray emission has been detected at the location of the SLSN-I SCP 06F6 (Gänsicke et al. 2009) with a luminosity ${L}_{x}\sim {10}^{45}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ ∼ 70 days (rest-frame) after the explosion (Levan et al. 2013). At this epoch, SCP 06F6 even outshines GRBs by a large factor, suggesting the presence of a still-active central engine that manifests itself through very luminous and long-lasting X-ray emission (Levan et al. 2013; Metzger et al. 2015). Before our efforts, SCP 06F6 was the only SLSN-I for which an X-ray source was detected at a location consistent with the optical transient.

These observational results suggest a connection between SLSNe-I and engine-driven SNe. However, it is not yet known how the properties of the engines (successful jet? relativistic ejecta? collimated or spherical central-engine-powered outflow?), progenitor stars, and circumstellar environments would compare. Here, we present the results from a systematic search for X-ray emission from SLSNe-I both at early and at late times, which directly depends on the properties of the immediate environment and central engine (if any). The direct detection of the stellar progenitors of SLSNe-I in pre-explosion optical images is not possible due to their large distances ($z\geqslant 0.1$). Sampling the circumstellar density profile in the closest environment is thus the most direct probe of their progenitors and their recent mass-loss history before stellar death.

The data set that we present here includes the deepest X-ray observations of SLSNe-I with Swift, XMM, and the Chandra X-ray Observatory (CXO), extending from the time of discovery until ∼2000 days (rest-frame) after explosion, and led to the discovery of X-ray emission at the location of the slowly evolving SLSN-I PTF 12dam. These observations, described in Section 2, indicate that superluminous X-ray emission similar to what was observed in association with SCP 06F6 is not common in SLSNe-I (Section 3) and allow us to place meaningful constraints on the environment density at the SLSN site (Section 4). We constrain the properties of central engines in SLSNe-I in Section 5 by investigating the presence of late-time X-ray re-brightenings that can either be due to emission from off-axis collimated relativistic outflows similar to GRBs, or to the ionization breakouts from magnetar central engines (Metzger et al. 2014). Conclusions are drawn in Section 6.

Since 2011, we routinely followed up all publicly announced nearby ($z\lesssim 0.5$) SLSNe-I with Swift-XRT with a series of observations acquired between the time of discovery and ∼360 days (rest-frame) after explosion. For a subset of events we acquired deep X-ray observations with dedicated programs on the CXO and XMM-Newton. Additionally, we searched the Swift-XRT, CXO, and XMM archives for serendipitous or unpublished observations of SLSNe-I discovered before 2016 May. Our final sample consists of 26 SLSNe-I discovered between 2006 and 2016 May. The data set covers the time range between ∼days after explosion until ∼2000 days (rest-frame) and comprises ∼700 hr of observations. We update the X-ray observations of the sample of 11 SLSNe-I from Levan et al. (2013) with the most recent data¹² and we add 15 new SLSNe-I. Inserra et al. (2017) presented a selection of Swift-XRT observations of three SLSNe. The much longer temporal baseline and better sensitivity of the X-ray data set presented here allow us to constrain the environments and the properties of central engines possibly powering the SLSN emission.

We divide our sample into three groups: the "gold sample" (Table 1) contains 4 SLSNe-I with X-ray detections or well-sampled optical bolometric light curve and deep X-ray limits obtained with XMM or the CXO. The "bronze sample" contains 12 SLSNe-I with sparser optical data but with good Swift-XRT X-ray coverage (Table 2), while the "iron sample" comprises 10 SLSNe-I with very sparse optical and X-ray data (Table 3). Given the peculiar nature of ASASSN-15lh (Dai et al. 2015; Metzger et al. 2015; Bersten et al. 2016; Chatzopoulos et al. 2016; Dong et al. 2016; Godoy-Rivera et al. 2017; Kozyreva et al. 2016; Leloudas et al. 2016; Sukhbold & Woosley 2016; van Putten & Della Valle 2017; Margutti et al. 2017a), this transient is not part of the sample of bona fide SLSNe-I analyzed here. However, we discuss and compare the X-ray properties of ASASSN-15lh in the context of SLSNe-I in Section 4, 5.1 and 5.2.

Swift-XRT data have been analyzed using HEASOFT (v6.18) and corresponding calibration files, following standard procedures (see Margutti et al. 2013b for details). For each SLSN-I we provide stacked flux limits (for visualization purposes only), and flux limits derived from individual observations (see Table 5 in the Appendix). CXO data have been analyzed with the CIAO software package (v4.9) and corresponding calibration files. Standard ACIS data filtering has been applied. XMM data have been analyzed with SAS (v15.0). For the non-detections, we perform a flux calibration adopting a power-law spectral model with index Γ = 2 corrected for the Galactic neutral-hydrogen absorption along the line of sight (Tables 1–3), as inferred from Kalberla et al. (2005). The details of the X-ray observations of specific SLSNe-I are provided in Section 2.1 for the gold sample, and in the Appendix for all the other SLSNe-I. Data tables can also be found in the Appendix. Figure 1 shows the complete sample of X-ray observations of SLSNe-I.

Zoom In Zoom Out Reset image size

2.1. Gold Sample

There are four objects in the gold sample: SCP 06F6, PTF 12dam, PS1-14bj, and SN 2015bn (Table 1). In this section we describe the discovery and properties of each object in the gold sample.

2.1.1. SCP 06F6

X-ray emission at the location of the type-I SLSN SCP 06F6 was first reported by Gänsicke et al. (2009) from XMM observations obtained 162 days after the initial detection of SCP 06F6 (PI Shartel, ID 0410580301). Levan et al. (2013) derived an X-ray flux ${F}_{x}\sim {10}^{-13}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ (0.3–10 keV) on 2006 August 2 (MJD 53949, ∼80 days rest-frame since explosion). Given the importance of the claim, we independently reanalyzed the XMM observations of SCP 06F6. We confirm the presence of severe contamination by soft proton flares and we confirm the detection of a point-like X-ray source in the 0.2–2 keV energy range in MOS1 and MOS2. The source is detected at the $\sim 5\sigma$ confidence level at coordinates R.A. = 14^h32^m27586, decl. = 33°32'2533 (J2000), consistent within the 4'' position error with the optical position of SCP 06F6. The 0.3–10 keV flux inferred from the count rate $(12.8\pm 2.3)\times {10}^{-3}\,\mathrm{cts}\,{{\rm{s}}}^{-1}$ is in agreement with the findings from Levan et al. (2013).

Follow-up observations of SCP 06F6 with the CXO obtained on 2006 November 4 (MJD 54043, ∼126 days rest-frame since explosion, PI Murray, ID 7010) led to a non-detection. The corresponding flux limit is ${F}_{x}\lt 1.4\times {10}^{-14}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ (Levan et al. 2013). We adopt these flux values here and the correct redshift for this event, which is z = 1.189 (Quimby et al. 2011c).

2.1.2. X-Ray Emission at the Location of PTF 12dam

PTF 12dam (Nicholl et al. 2013; Chen et al. 2015; Inserra et al. 2017; Vreeswijk et al. 2017) belongs to the small subset of SLSNe-I with slowly evolving optical light curves. At the time of writing, this group includes SN 2007bi, PTF 12dam, iPTF 13ehe, SN 2015bn, PS1-14bj, and LSQ 14an. The slow evolution of these transients, and of SN 2007bi in particular, inspired a connection with pair-instability explosions (Gal-Yam et al. 2009) that was later debated by Nicholl et al. (2013). X-ray observations of the SLSN-I PTF 12dam have been obtained with Swift-XRT and the CXO. Swift-XRT observations span the time range of $\sim 43\mbox{--}900$ days rest-frame since explosion and revealed no detection down to a flux limit ${F}_{x}\sim 5\times {10}^{-14}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ (Figure 2, Table 5).

Zoom In Zoom Out Reset image size

A set of three deep CXO observations was acquired between 2012 June 11 and 19 (δt ∼ 60–68 days rest-frame since explosion; observations IDs 13772, 14444 and 14446, PI Pooley). An X-ray source with a soft spectrum is clearly detected in the merged event file (total exposure of 99.9 ks) at the location of PTF 12dam with significance of $4.8\sigma$ in the 0.5–2 keV energy range. The measured net count rate is $(7.1\pm 2.8)\times {10}^{-5}\,{\rm{c}}\,{{\rm{s}}}^{-1}$ (0.5–2 keV), which corresponds to an unabsorbed flux of $(7.3\pm 2.9)\times {10}^{-16}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ (0.3–10 keV) assuming a power-law spectrum with photon index Γ = 2. For a thermal bremsstrahlung spectrum with $T=0.24\,\mathrm{keV}$ (see below) the corresponding unabsorbed flux is $(8.9\pm 3.5)\times {10}^{-16}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ (0.3–10 keV), and $(3.5\pm 1.4)\times {10}^{-16}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ (0.5–2 keV). In both cases, the inferred X-ray luminosity is ${L}_{x}\sim 2\times {10}^{40}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ in the 0.3–10 keV (Figure 2).

PTF 12dam exploded in a compact dwarf galaxy with fairly large star formation rate $\mathrm{SFR}\sim 5\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ (Lunnan et al. 2014; Chen et al. 2015; Leloudas et al. 2015b; Thöne et al. 2015; Perley et al. 2016) Following Mineo et al. (2012), the expected apparent diffuse X-ray emission associated with star formation is ${L}_{x}/\mathrm{SFR}\approx 8.3\times {10}^{38}\,\mathrm{erg}\,{{\rm{s}}}^{-1}{({M}_{\odot }{\mathrm{yr}}^{-1})}^{-1}$, which translates into ${L}_{x}\approx 4.2\times {10}^{39}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ (0.5–2 keV) for $\mathrm{SFR}\sim 5\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$. As a comparison, for a thermal bremsstrahlung spectrum with $T=0.24\,\mathrm{keV}$ (average temperature of the unresolved X-ray component in galaxies, Mineo et al. 2012), for PTF 12dam we calculate ${L}_{x}\approx (1.0\pm 0.4)\times {10}^{40}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ (0.5–2 keV). We therefore conclude that star formation in the host galaxy of PTF 12dam is likely contributing to at least some of the X-ray luminosity that we detected at the location of the transient. In the following analysis sections we treat our measurements as upper limits to the X-ray emission from PTF 12dam. These observations provide the deepest limits to the X-ray emission from a SLSN-I to date. Future observations will constrain the late-time behavior of the X-ray emission at the location of PTF 12dam and will clarify its association to the optical transient.

2.1.3. PS1-14bj

Two epochs of deep X-ray observations of the type-I SLSN PS1-14bj (Lunnan et al. 2016) have been obtained with XMM (PI Margutti, IDs 0743110301, 0743110701) on 2014 June 9 (δt ∼ 135 days rest-frame since explosion, exposure time of 47.6 ks), and 2014 November 7 (δt ∼ 235 days rest-frame since explosion, exposure of 36.0 ks). The net exposure times after removing data with high background contamination are 3.6 ks and 29.8 ks, respectively (EPIC-pn data). We do not find evidence for significant X-ray emission at the location of PS1-14bj in either observation and derive a 3σ 0.3–10 keV count-rate upper limit of $9.4\times {10}^{-3}\,{\rm{c}}\,{{\rm{s}}}^{-1}$ ($1.5\times {10}^{-3}\,{\rm{c}}\,{{\rm{s}}}^{-1}$) for the first (second) epoch, which translates into an unabsorbed flux of $1.9\times {10}^{-14}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ ($3.3\times {10}^{-15}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$). The corresponding luminosity limits are shown in Figure 3 and reported in Table 5. Analogous with the type of analysis performed on SCP 06F6, we also inspected the 0.2–2 keV images (for both EPIC-pn and MOS) without filtering for the flares. However, for PS1-14bj we confirm the non-detection in both epochs.

Zoom In Zoom Out Reset image size

2.1.4. SN 2015bn

X-ray observations of the SLSN-I 2015bn (Nicholl et al. 2016a, 2016b) have been obtained with Swift-XRT and XMM (PI Margutti, IDs 0770380201, 0770380401). A first set of observations was presented in Nicholl et al. (2016a), while Inserra et al. (2017) included in their analysis five Swift-XRT pointings. Here, we present the complete data set. Swift-XRT started observing SN 2015bn on 2015 February 19 until 2016 July 23, covering the time period $\sim 44\mbox{--}522$ days since explosion rest-frame. No statistically significant X-ray emission is blindly detected at the location of the transient (Figure 4).¹³

Zoom In Zoom Out Reset image size

Two epochs of XMM observations have been obtained on 2015 June 1 (δt ∼ 145 days rest-frame since explosion) and 2015 December 18 (δt ∼ 325 days rest-frame since explosion) with exposure times of 28.0 ks and 25.1 ks, respectively (EPIC-pn data). After excluding time intervals heavily affected by proton flaring, the net exposure times are 7.3 and 18.8 ks. No X-ray source is detected at the location of the SLSN-I 2015bn. We derive a 3σ 0.3–10 keV count-rate upper limit of $4.2\times {10}^{-3}\,{\rm{c}}\,{{\rm{s}}}^{-1}$ ($2.3\times {10}^{-3}\,{\rm{c}}\,{{\rm{s}}}^{-1}$) for the first (second) epoch, which translates into an unabsorbed flux of $9.8\times {10}^{-15}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ ($5.3\times {10}^{-15}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$). The results from our X-ray campaign are listed in Table 5 and displayed in Figure 4.

In this section we derive constraints on the possible presence of superluminous X-ray emission in SLSNe-I that was not detected because of the discontinuous observational coverage. No assumption is made about the physical nature of the emission. SLSNe-I are treated here as different realizations of the same stochastic process (which is the underlying assumption behind any sample analysis).

The hypothesis we test is that superluminous X-ray emission of the kind detected at the location of SCP 06F6 (i.e., isotropic L_x ∼ 10⁴⁵ erg s⁻¹) is ubiquitous in SLSNe-I. We furthermore assume that the superluminous X-ray emission is "on" for a total time ${\rm{\Delta }}{t}_{\mathrm{active}}$ (not necessarily continuous) and shuts on and off instantaneously, and that the probability of turning on is uniformly distributed during the interval of time of investigation. Our sample of observations comprises 253 spacecraft pointings, for a total observing time of ∼30 days at t < 2000 days (rest-frame). Out of 253 trials, observations only led to one success (i.e., in the case of SCP 06F6). We can use simple binomial probability arguments to constrain the maximum and minimum ${\rm{\Delta }}{t}_{\mathrm{active}}$ that would be statistically consistent at the 3σ c.l. with $\leqslant 1$ successes (for ${\rm{\Delta }}{t}_{\mathrm{active},\max }$) and $\geqslant 1$ successes (for ${\rm{\Delta }}{t}_{\mathrm{active},\min }$) out of N trials, where $N\equiv N(t)$. In this context a trial consists of an observation that is deep enough to be sensitive to L_x ∼ 10⁴⁵ erg s⁻¹. For each trial, the probability of success is $p={\rm{\Delta }}{t}_{\mathrm{active}}/{\rm{\Delta }}{t}_{\mathrm{total}}$, where ${\rm{\Delta }}{t}_{\mathrm{total}}$ is the entire range of times during which we conduct our search for superluminous X-ray emission (for the entire sample, ${\rm{\Delta }}{t}_{\mathrm{total}}\sim 2000$ days). For t < 100 days, ${\rm{\Delta }}{t}_{\mathrm{total}}\sim 90$ days (i.e., our search starts at $t\sim 10$ days), N = 110. With these parameters we find $200\,{\rm{s}}\leqslant {\rm{\Delta }}{t}_{\mathrm{active}}\leqslant 5\times {10}^{5}\,{\rm{s}}$. For shorter ${\rm{\Delta }}{t}_{\mathrm{active}}$ the probability of having one success out of N trials is below the 3σ c.l., while for longer ${\rm{\Delta }}{t}_{\mathrm{active}}$ we would have expected to have more successes in our search at 3σ c.l. If we consider the entire sample of observations ${\rm{\Delta }}{t}_{\mathrm{total}}\sim 2000$ days, N = 253 and we find $2000\,{\rm{s}}\leqslant {\rm{\Delta }}{t}_{\mathrm{active}}\leqslant 5\times {10}^{6}\,{\rm{s}}$. As a cross-check, under the same assumptions, but adopting the Bayesian technique of Romano et al. (2014) we obtain for ${\rm{\Delta }}{t}_{\mathrm{active}}$ similar upper limits.

We conclude that superluminous X-ray emission is not a common trait of SLSNe-I. If present, the superluminous X-ray emission requires peculiar physical circumstances to manifest and its duration is ≤2 months at t < 2000 days and $\leqslant$ few days at t < 100 days.

Inverse Compton (IC) emission is a well-known source of X-rays in young stellar explosions (Björnsson & Fransson 2004; Chevalier & Fransson 2006). X-ray emission originates from the up-scattering of optical photons from the SN photosphere by a population of relativistic electrons accelerated at the shock front. While always present, IC is the dominant emission mechanism at early times (t ≤ optical peak) for SNe propagating into low-density media. In the case of strong SN shock interaction with the medium, the dominant X-ray emission mechanism is instead bremsstrahlung (Björnsson & Fransson 2004; Chevalier & Fransson 2006), as it is indeed observed in type-IIn SNe and as recently confirmed by the first broadband X-ray spectra of strongly interacting SNe (Ofek et al. 2014; Margutti et al. 2017a). The analysis of the optical emission from SLSNe-I in the context of the interaction model (e.g., Nicholl et al. 2014, 2016b) suggests that if SLSNe-I are powered by interaction, then the shock breaks out around the time of optical maximum light and the medium consists of a thick shell confined to small radii ($R\sim 5\times {10}^{15}\,\mathrm{cm}$ for SN 2015bn) surrounded by a lower density region. This conclusion is consistent with the lack of observed narrow lines in the optical spectra of SLSNe-I (in sharp contrast to ordinary and superluminous type-IIn SNe): the presence of an extended unshocked region of dense circumstellar medium (CSM) would likely imprint low-velocity features that are not observed in SLSNe-I (see also Chevalier & Irwin 2011). The X-ray observations that we will use in this section have been obtained at the time of maximum light or later, which is after the shock has broken out from the thick shell of material if a shell is there. In the following, we thus constrain the density around SLSNe-I under the conservative assumption that IC is the only source of X-ray radiation. Since we sample the time range $t\gt {t}_{\mathrm{peak}}$, our density limits apply to the region $R\gtrsim {10}^{16}\,\mathrm{cm}$.

The X-ray emission from IC depends on (i) the density structure of the SN ejecta and of the CSM, (ii) the details of the electron distribution responsible for the up-scattering, (iii) the explosion parameters (ejecta mass M_ej and kinetic energy E_k), and (iv) the availability of seed optical photons (${L}_{x,\mathrm{IC}}\propto {L}_{\mathrm{bol}}$, where L_bol is the bolometric optical luminosity). We employ the formalism of Margutti et al. (2012) modified to reflect the stellar structure of massive stars as in Margutti et al. (2014). We further assume a wind-like medium with ${\rho }_{\mathrm{CSM}}\propto {R}^{-2}$ as appropriate for massive stars, a power-law electron distribution ${n}_{e}(\gamma )={n}_{0}{\gamma }^{-p}$ with $p\sim 3$ as indicated by radio observations of H-stripped core-collapse SNe (Chevalier & Fransson 2006) and a fraction of post-shock energy into relativistic electrons _e = 0.1 (e.g., Chevalier & Fransson 2006). Since ${L}_{x,\mathrm{IC}}\propto {L}_{\mathrm{bol}}$, it is clear that the tightest constraints on ${\rho }_{\mathrm{CSM}}$ will be derived from the most nearby SLSNe-I, which have very bright optical emission and deep X-ray limits (i.e., they have the largest flux ratio ${F}_{\mathrm{opt}}/{F}_{x}$ constrained by observations). To this end, we analyze below the SLSNe-I 2015bn and PTF 12dam. We also provide constraints for the peculiar transient ASASSN-15lh.

For SN 2015bn we follow Nicholl et al. (2016a) and Nicholl et al. (2016b) and adopt a range of ejecta masses ${M}_{\mathrm{ej}}=7\mbox{--}15\,{M}_{\odot }$ (Table 4). With these parameters and the optical bolometric light curve from Nicholl et al. (2016b; Figure 4), our X-ray non-detections constrain the pre-explosion mass-loss rate from the stellar progenitor of SN 2015bn to $\dot{M}\lt {10}^{-2}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ ($\dot{M}\lt {10}^{-1}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$) for ${E}_{k}={10}^{52}\,\mathrm{erg}$ (${E}_{k}={10}^{51}\,\mathrm{erg}$) and wind velocity ${v}_{w}=1000\,\mathrm{km}\,{{\rm{s}}}^{-1}$, (Figures 5, 6), which is $\dot{M}\lt {10}^{-4}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ ($\dot{M}\lt {10}^{-3}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$) for wind velocity ${v}_{w}=10\,\mathrm{km}\,{{\rm{s}}}^{-1}$. In this context, the analysis of the radio observations of SN 2015bn indicates $\dot{M}\lt {10}^{-2}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ for ${v}_{w}=10\,\mathrm{km}\,{{\rm{s}}}^{-1}$ at $R\gt {10}^{15}\,\mathrm{cm}$, while $\dot{M}\sim {10}^{-2}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ would be needed to explain the late-time optical light curve of the transient through continued ejecta-CSM interaction (Nicholl et al. 2016b). The X-ray analysis thus argues against the presence of an extended CSM region if ${E}_{k}\gt {10}^{51}\,\mathrm{erg}$ (as is likely the case) and suggests that another source of energy is powering the light curve after the peak. This result is consistent with the conclusions by Nicholl et al. (2016a): based on the spectroscopic similarity of SN 2015bn with the GRB SN 1998bw in the nebular phase, Nicholl et al. (2016a) concluded that a central engine is driving the explosion.

Zoom In Zoom Out Reset image size

For PTF 12dam we use ${M}_{\mathrm{ej}}=7\,{M}_{\odot }$, as inferred by Nicholl et al. (2013) from the modeling of the optical bolometric emission (Table 4). We detect an X-ray source at the location of PTF 12dam with ${L}_{x}\sim 2\times {10}^{40}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$. We treat this value as an upper limit to the X-ray luminosity from the transient to account for possible contamination from the host galaxy. For these values of the explosion parameters and the measured L_x, the inferred mass-loss rate is $\dot{M}\lt 2\times {10}^{-5}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ ($\dot{M}\lt 4\times {10}^{-6}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$) for ${E}_{k}={10}^{51}\,\mathrm{erg}$ (${E}_{k}={10}^{52}\,\mathrm{erg}$) and ${v}_{w}=1000\,\mathrm{km}\,{{\rm{s}}}^{-1}$ (Figures 5, 6). These are the tightest constraints to the pre-explosion mass-loss history of SLSNe-I progenitors.

Zoom In Zoom Out Reset image size

If the peculiar transient ASASSN-15lh is associated with an ${E}_{k}={10}^{52}\,\mathrm{erg}$ explosion with ejecta mass ${M}_{\mathrm{ej}}=5\mbox{--}10\,{M}_{\odot }$ (Metzger et al. 2015; Bersten et al. 2016; Chatzopoulos et al. 2016; Dai et al. 2016; Dong et al. 2016; Kozyreva et al. 2016; Sukhbold & Woosley 2016), the X-ray observations from Margutti et al. (2017b) imply $\dot{M}\lt 5\times {10}^{-6}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ ($\dot{M}\lt 5\times {10}^{-5}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$) for ${v}_{w}=1000\,\mathrm{km}\,{{\rm{s}}}^{-1}$ for a thermal (non-thermal) X-ray spectrum. These values are typical of mass-loss rates from H-stripped compact massive stars (Figures 5 and 6).

From our analysis, we conclude that for PTF 12dam the inferred limits rule out the densest environments that characterize type-IIn SNe (Figure 5), indicating that the strong SN shock interaction with an extended medium is unlikely to be the primary source of energy sustaining the very luminous display. Our data, however, do not constrain the presence of a dense medium confined to d < 10¹⁶ cm, which might be the result of massive shell ejections by the stellar progenitor in the ∼10 yr before collapse (for a shell ejection velocity of $1000\,\mathrm{km}\,{{\rm{s}}}^{-1}$). Interestingly, a low-density environment with $\dot{M}\sim 4\times {10}^{-6}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ was also inferred from radio and X-ray observations of the type-Ib SN 2012au (Kamble et al. 2014), which showed spectroscopic similarities with SLSN-I (Milisavljevic et al. 2013). The tight constraints obtained for PTF 12dam point to a clean environment, and argue against the dense CSM typical of extended progenitors like RSG stars. This result suggests that at least some SLSN-I progenitors are likely to be compact stars surrounded by a low-density medium at $d\gt {10}^{16}\,\mathrm{cm}$ at the time of stellar death.

5.1. Constraints on On-axis and Off-axis Collimated and Non-collimated Relativistic Outflows

The search for off-axis and on-axis relativistic GRB-like jets in SLSNe-I is motivated by two recent observational findings: (i) the association of SN 2011kl with GRB 111209A. SN 2011kl bridges the luminosity gap between GRB-SNe and SLSNe-I, and shows similarities to SLSNe-I (Greiner et al. 2015); and (ii) nebular spectroscopy of the SLSN-I 2015bn revealed close similarities to the engine-driven SN 1998bw, associated with GRB 980425, suggesting that the core of engine-driven SNe and SLSNe-I share some key physical properties and structure (Nicholl et al. 2016a).

Early-time X-ray observations of SLSNe-I acquired at $t\lesssim 40$ days generally rule out on-axis collimated ultra-relativistic outflows of the type associated with energetic long GRBs (Figure 7, cloud of filled gray squares). We constrain the presence of off-axis relativistic outflows by generating a grid of off-axis GRB X-ray afterglows with the broadband afterglow numerical code Boxfit v2 (van Eerten et al. 2012).¹⁴ The observed X-ray emission depends on the kinetic energy E_k of the outflow, the density of the medium ${\rho }_{\mathrm{CSM}}$ (we explore both an ISM-like medium ${n}_{\mathrm{CSM}}=\mathrm{const}$ and a wind-like medium with ${\rho }_{\mathrm{CSM}}\propto {R}^{-2}$), the microphysical shock parameters ${\epsilon }_{B}$ and ${\epsilon }_{e}$ (post-shock energy fraction in magnetic field and electrons, respectively), the jet opening angle ${\theta }_{j}$ and the angle of the jet with respect to the line of sight ${\theta }_{\mathrm{obs}}$. We explore the predicted X-ray signatures of collimated θ_j = 5° outflows with _e = 0.1 and _B = 0.01 (as derived from first-principle simulations of relativistic shocks, e.g., Sironi et al. 2015), isotropic kinetic energy in the range ${E}_{k,\mathrm{iso}}={10}^{52}\mbox{--}{10}^{55}\,\mathrm{erg}$, environment density in the range $n={10}^{-3}\mbox{--}10\,{\mathrm{cm}}^{-3}$ (ISM) or mass-loss rate $\dot{M}={10}^{-7}\mbox{--}{10}^{-3}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ (wind) and observed angles θ_obs ≤ 90°. These values are representative of the parameters inferred from accurate modeling of broadband afterglows of GRBs.

Zoom In Zoom Out Reset image size

Based on these simulations and the X-ray observations from the entire sample of SLSNe-I, we find that relativistic collimated outflows with ${E}_{k}\gt {10}^{51}\,\mathrm{erg}$, $n\gt {10}^{-3}\,{\mathrm{cm}}^{-3}$ and ${\theta }_{\mathrm{obs}}\lt 2{\theta }_{j}$ are ruled out. Powerful jets with ${E}_{k}\gt {10}^{52}\,\mathrm{erg}$ expanding in a thick medium with $n\geqslant 10\,{\mathrm{cm}}^{-3}$ and θ_obs ≤ 30° are also ruled out. For a wind environment, our observations are not consistent with jets with ${E}_{k}\gt {10}^{50.5}\,\mathrm{erg}$ expanding in a medium enriched with $\dot{M}\geqslant {10}^{-7}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ and ${\theta }_{\mathrm{obs}}\lt 2{\theta }_{j}$. At higher kinetic energies, observations rule out jets with ${E}_{k}\gt {10}^{51.5}\,\mathrm{erg}$, $\dot{M}\geqslant {10}^{-4}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$, and θ_obs ≤ 30° or ${E}_{k}\gt {10}^{52}\,\mathrm{erg}$, $\dot{M}\geqslant {10}^{-3}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$, and θ_obs ≤ 45°.We are not sensitive to jets viewed at θ_obs > 30° for the ISM, and θ_obs > 45° for the wind medium.

In the case of PTF 12dam, observations argue against jets with ${E}_{k}\gt {10}^{51}\,\mathrm{erg}$ propagating into a medium with $n\gt {10}^{-3}\,{\mathrm{cm}}^{-3}$ or $\dot{M}\gt {10}^{-7}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ and ${\theta }_{\mathrm{obs}}\lt 2{\theta }_{j}$. The portion of the parameter space associated with ${E}_{k}={10}^{50.5}\,\mathrm{erg}$, $\dot{M}\sim {10}^{-6\mbox{--}7}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ and ${\theta }_{\mathrm{obs}}\lt 2{\theta }_{j}$ is also ruled out. Dense environments with $n\,\gt 10\,{\mathrm{cm}}^{-3}$ or $\dot{M}\gt {10}^{-4}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ would also produce X-ray emission in excess of what we observed for outflows with ${E}_{k}\gt {10}^{51}\,\mathrm{erg}$ viewed at θ_obs < 30°.

For the SLSN-I 2015bn observations rule out systems with ${E}_{k}\gt {10}^{52}\,\mathrm{erg}$, $n\gt {10}^{-3}\,{\mathrm{cm}}^{-3}$, or $\dot{M}\gt {10}^{-7}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ for ${\theta }_{\mathrm{obs}}\lt 2{\theta }_{j}$. Even the most energetic outflows in our simulations with ${E}_{k}\gt {10}^{52}\,\mathrm{erg}$ would fall below our detection threshold for θ_obs > 30° and the range of densities considered. These observations complement the results from deep radio non-detections of SN 2015bn (Nicholl et al. 2016b), which argue against powerful on-axis or off-axis jets with ${E}_{k}=2\times {10}^{51}\,\mathrm{erg}$ propagating into an ISM-like medium with density $n=1\,{\mathrm{cm}}^{-3}$.¹⁵

Finally, we consider the observable X-ray signatures of non-collimated mildly relativistic outflows. X-ray observations of the majority of SLSNe-I in our sample are not sensitive to the faint X-ray emission of mildly relativistic non-collimated outflows typical of low-energy GRBs like 980425, 031203, 060218, and 100316D (e.g., Pian et al. 2000; Kouveliotou et al. 2004; Watson et al. 2004; Soderberg et al. 2006a; Margutti et al. 2013a). However, our deepest X-ray limits obtained with the CXO and XMM are sensitive enough to probe the parameter space populated by the weakest GRB-SNe. For the SLSN-I 2015bn, our XMM observations probe and rule out luminosities ${L}_{x}\gt 2\times {10}^{41}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ at $t\,\sim 100\mbox{--}300$ days, which are comparable to the detected X-ray emission of GRBs 980425 and 031203 at a similar epoch, ${L}_{x}\sim {10}^{41}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ (Figure 7). Remarkably, in the case of PTF 12dam, CXO observations acquired at the time of optical peak rule out even the faintest non-collimated X-ray emission ever detected from a low-energy GRB (Figure 7), indicating that if PTF 12dam is an engine-driven stellar explosion, the jet never successfully broke out from the stellar envelope, in close analogy to the picture recently suggested for the relativistic SNe 2009bb and 2012ap (Margutti et al. 2014 and references therein).

To conclude, the analysis of our deep X-ray limits in the context of GRB afterglow simulations and the recent finding of similarity in the nebular emission from SN 2015bn with engine-driven SNe, suggest that if SLSNe-I are jet-driven explosions, then either SLSNe-I are powered by very energetic collimated GRB-like outflows that were pointing far away from our line of sight (θ_obs > 30°), or that SLSNe-I harbor failed jets that do not successfully break through the stellar envelope and are associated with weak X-ray emission. Late-time radio observations of SN 2015bn (Nicholl et al. 2016a) argue against the off-axis relativistic jet scenario in the case of energetic jets (see also Coppejans et al. 2018; Margalit et al. 2018). However, the association of the overluminous SN 2011kl with GRB 111209A suggests that some SLSNe-I might harbor relativistic jets. We therefore propose that, in strict analogy to H-stripped core-collapse SNe of ordinary luminosity (e.g., Mazzali et al. 2008; Xu et al. 2008; Lazzati et al. 2012; Margutti et al. 2014), SLSNe-I are also characterized by a continuum of jet strengths and lifetimes of the central engine.

5.2. Constraints on Magnetar Central Engines: The Ionization Breakout

We compute the ionization breakout time and the X-ray luminosity at breakout following Metzger et al. (2014) and Metzger & Piro (2014). We consider a central engine with an UV/X-ray luminosity L that releases an energy L × t in ionizing radiation on a timescale t. The radiation ionizes its way through the ejecta on a timescale

$$\begin{eqnarray}&&{t}_{\mathrm{ion}}\approx \left\{\begin{array}{l}120\,\mathrm{days}\,{M}_{3}^{3/4}{v}_{9}^{-5/4}{T}_{5}^{-0.2}{\left(\displaystyle \frac{{X}_{A}}{0.1}\right)}^{1/4}{\left(\displaystyle \frac{{Lt}}{{10}^{52}\mathrm{erg}}\right)}^{-1/4}{Z}_{8}^{3/4},({\eta }_{\mathrm{thr}}\ll 1)\\ 110\,\mathrm{days}\,{M}_{3}{v}_{9}^{-3/2}{T}_{5}^{-0.4}{\left(\displaystyle \frac{{X}_{A}}{0.1}\right)}^{1/2}{\left(\displaystyle \frac{{Lt}}{{10}^{52}\mathrm{erg}}\right)}^{-1/2}{Z}_{8}^{3/2},({\eta }_{\mathrm{thr}}\gg 1),\end{array}\right.\end{eqnarray} \tag{ 1 }$$

where ${M}_{3}\equiv {M}_{\mathrm{ej}}/(3{M}_{\odot })$, ${T}_{5}=T/{10}^{5}$ K is the temperature of electrons in the recombination layer, ${v}_{9}\equiv v/{10}^{9}\,\mathrm{cm}\,{{\rm{s}}}^{-1}$, X_Z is the mass fraction X_Z of elements with atomic number Z = 8Z₈ in the ejecta and

$$\begin{eqnarray}&&{\eta }_{\mathrm{thr}}\approx 0.7{\left(\displaystyle \frac{{Lt}}{{10}^{52}\mathrm{erg}}\right)}^{-1}{M}_{3}{v}_{9}^{-1}\left(\displaystyle \frac{{X}_{A}}{0.1}\right){T}_{5}^{-0.8}{Z}_{8}^{3}\end{eqnarray} \tag{ 2 }$$

is the ratio of absorptive and scattering opacity in the ejecta (Metzger et al. 2014). The spin-down timescale ${t}_{\mathrm{sd}}$ of a magnetar central engine is given by

$$\begin{eqnarray}&&{t}_{\mathrm{sd}}\sim 4.7{\rm{d}}{B}_{14}^{-2}{P}_{\mathrm{ms}}^{2}\end{eqnarray} \tag{ 3 }$$

and the spin-down luminosity is given by

$$\begin{eqnarray}{L}_{\mathrm{sd}} & = & 5\times {10}^{46}{B}_{14}^{2}{P}_{\mathrm{ms}}^{-4}{\left(1+\displaystyle \frac{t}{{t}_{\mathrm{sd}}}\right)}^{-2}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\\ & \approx & \,1.1\times {10}^{48}{B}_{14}^{-2}{t}_{{\rm{d}}}^{-2}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\end{eqnarray} \tag{ 4 }$$

for $t\gg {t}_{\mathrm{sd}}$, ${B}_{14}\equiv B/{10}^{14}\,{\rm{G}}$, ${P}_{\mathrm{ms}}\equiv P/\mathrm{ms}$, ${t}_{{\rm{d}}}\equiv t/\mathrm{days}$ and we adopted the vacuum dipole spin-down convention employed by Kasen & Bildsten (2010). For $L={L}_{\mathrm{sd}}$ and $t\gg {t}_{\mathrm{sd}}$ the ionization timescale of Equation (1) can be written as

$$\begin{eqnarray}&&{t}_{\mathrm{ion}}\approx \left\{\begin{array}{l}280\,\mathrm{days}\,{M}_{3}{v}_{9}^{-5/3}{T}_{5}^{-4/15}{\left(\displaystyle \frac{{X}_{A}}{0.1}\right)}^{1/3}{B}_{14}^{2/3}{Z}_{8},({\eta }_{\mathrm{thr}}\ll 1)\\ 1273\,\mathrm{days}\,{M}_{3}^{2}{v}_{9}^{-3}{T}_{5}^{-4/5}\left(\displaystyle \frac{{X}_{A}}{0.1}\right){B}_{14}^{2}{Z}_{8}^{3},({\eta }_{\mathrm{thr}}\gg 1).\end{array}\right.\end{eqnarray} \tag{ 5 }$$

The X-ray luminosity at ionization breakout is ${L}_{x}\approx {L}_{\mathrm{sd}}({t}_{\mathrm{ion}})/14$. In the following we assume that oxygen (Z₈ = 1) dominates the bound-free opacity in the ∼keV X-ray band, we use and electron temperature T = 10⁵ K, X_A = 0.1, velocity of the order of 10⁹ cm s⁻¹ and we compute the ionization timescale and the X-ray luminosity at breakout and compare to our X-ray limits and measurements.

For SLSNe-I with well-constrained optical bolometric emission, we use the magnetar parameters M_ej, P, and B that best fit the optical bolometric luminosity to estimate t_ion and ${L}_{x}({t}_{\mathrm{ion}})$ (Table 4). From Table 4 and Figures 2–4 it is clear that most of the ${L}_{x}({t}_{\mathrm{ion}})$ are too faint to be detected and that t_ion is usually much larger than the ∼2000 days that we cover with our observations. However, it is also clear that t_ion and ${L}_{x}({t}_{\mathrm{ion}})$ are very sensitive to the magnetar parameters and qualify as excellent probes of central engines in SLSNe-I. Magnetar central engines that would produce very similar optical bolometric outputs that cannot be distinguished with current optical-UV photometry are instead clearly differentiated in their ${L}_{x}({t}_{\mathrm{ion}})$ − t_ion properties. As an example, for the best-fitting magnetar parameters of SN 2015bn in Figure 4, ${L}_{x}({t}_{\mathrm{ion}})$ spans ∼5 orders of magnitude and t_ion ranges from 0.6 to 88 years. For this SLSN-I, our deep X-ray limits obtained with XMM and the combined limit from Swift-XRT favor models with $P\gt 2$ ms. The fastest spinning magnetar model with P = 1.5 ms and relatively small ejecta mass ${M}_{\mathrm{ej}}=7.4\,{M}_{\odot }$ from Nicholl et al. (2016a) predicts ${L}_{x}({t}_{\mathrm{ion}})\gt {10}^{43}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ at ${t}_{\mathrm{ion}}\sim 0.6\,\mathrm{years}$ and it is therefore disfavored by our X-ray observations (Figure 4).

Figure 8 shows how the X-ray observations from our sample compare to the predictions from the magnetar ionization breakout model. We investigate a wide range of central engine parameters P = 1–7 ms and B₁₄ = 0.2–10 G for ejecta masses ${M}_{\mathrm{ej}}=1\mbox{--}20\,{M}_{\odot }$. Current X-ray observations are not sensitive to magnetars with ${B}_{14}\geqslant 2$ (Figure 8). For ${B}_{14}\lt 2$, observations favor models with larger ejecta mass: for B₁₄ = 0.2, 0.5, 1.0 G the allowed parameter space is ${M}_{\mathrm{ej}}\gt 20,7,3\,{M}_{\odot }$. X-ray observations indicate that if a magnetar central engine powers the emission from SLSNe-I then it has to be either associated with a large magnetic field or with a large explosion ejecta mass. These results are independent from (but in agreement with) the values inferred from the modeling of the optical emission from SLSNe-I (Table 4, Inserra et al. 2013; Nicholl et al. 2013).

Zoom In Zoom Out Reset image size

We end by noting that our analytic treatment of ionization breakout from the supernova ejecta requires confirmation by a more detailed photoionization calculation in future work, as well as a more accurate model for the spectral energy distribution of the young pulsar wind nebula. Also, by adopting a relatively low mass fraction of ${X}_{Z}\approx 0.1$ of CNO elements (which contribute most of the bound-free opacity in the keV range) we may be underestimating the ionization breakout time and thus overestimating the associated X-ray luminosity if the true mass fraction is higher. On the other hand, asphericity in the ejecta (e.g., along the rotation axis) would reduce the breakout time along directions of lower than average density and introduce a viewing angle dependence to the emission. An extension from an analytical 1D model (used here) to detailed multi-D formulations is indeed necessary to fully characterize the expected X-ray signature from ionization breakout, and possibly solve the current tension between the anticipated versus observed spectral features and their evolution in the magnetar model (e.g., Liu et al. 2017).

We present the results from an extensive systematic survey of X-ray emission from 26 hydrogen-stripped SLSNe in the local universe with Swift, Chandra, and XMM. These data cover the SLSNe-I evolution from ∼days until 2000 days (rest-frame) since explosion, reaching L_x ∼ 10⁴⁰ erg s⁻¹. The unprecedented depth of these observations provided the deepest limits on the X-ray emission from SLSNe-I to date and enabled the detection of X-ray emission at the location of the slowly evolving SLSN-I PTF 12dam. The major results from our investigation can be summarized as follows:

• 1.  
  Superluminous X-ray emission L_x ∼ 10⁴⁵ erg s⁻¹ of the kind detected at the location of SCP 06F6 is not a common trait of SLSNe-I. Superluminous X-ray emission requires peculiar physical conditions that are likely not shared by the majority of SLSNe-I. If present, its duration is ≤2 months at t < 2000 days and $\leqslant$ few days at earlier epochs t < 100 days.
• 2.  
  We place sensitive limits on the sub-parsec environments of the SLSNe-I with the deepest observations and constrain the pre-explosion mass-loss histories of the stellar progenitors of SLSNe-I (Figures 5–6). The most sensitive X-ray observations in our sample rule out the densest environments typical of LBV eruptions and type-IIn SNe at distances $d\gt {10}^{16}\,\mathrm{cm}$. For PTF 12dam, the observations point to a clean environment similar to engine-driven SNe and argue against the extended dense CSM typical of stellar progenitors like RSG. Observations indicate $\dot{M}\lt 2\times {10}^{-5}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$. This result suggests that sustained CSM interaction is unlikely to play a key role in the process that powers the luminous display in some SLSNe-I and that at least some SLSN-I progenitors end their lives as compact stars surrounded by a low-density medium at distances $d\gt {10}^{16}\,\mathrm{cm}$. Our data do not constrain the properties of the nearby d < 10¹⁶ cm environment, sculpted by the mass-loss history of the progenitor star in the last ∼10 yr before core-collapse.
• 3.  
  We do not find compelling observational evidence for relativistic outflows in SLSNe-I. SLSNe-I might either be powered by energetic relativistic GRB-like outflows that we did not detect because they were positioned far away from our line of sight (θ_obs > 30°), or might harbor failed jets that do not successfully pierce through the stellar envelope. Deep X-ray observations of PTF 12dam rule out even the weakest emission from uncollimated GRB outflows (Figure 7), suggesting that if PTF 12dam is a jet-driven explosion, then the jet never successfully broke out from the surface (in close similarity to the relativistic SNe 2009bb and 2012ap). However, the luminous SN 2011kl was found in association with the fully relativistic, fully successful jet of GRB 111209A, suggesting that jet-driven explosions can give rise to SNe more luminous than the average H-stripped core-collapse SNe. We thus propose that, just like H-stripped core-collapse SN, SLSNe-I might also be characterized by a continuum of jet power and central-engine lifetimes.
• 4.  
  The X-ray ionization breakout is a very sensitive probe of the properties of a hidden magnetar central engine in SLSNe-I. Magnetar central engines that would produce very similar optical/UV displays are instead clearly differentiated in terms of X-ray luminosities and timescales of the ionization breakout (Figures 3–4). Current X-ray observations indicate that if a magnetar central engine powers SLSNe-I, then it has to be either associated with a large magnetic field ${B}_{14}\gt 2$ G or large ejecta mass (${M}_{\mathrm{ej}}\gt 20,7,3\,{M}_{\odot }$ for B₁₄ = 0.2, 0.5, 1.0 G).

This X-ray campaign provided constraints on the sub-parsec environment and properties of central engines in SLSNe-I. To further advance our knowledge and understanding of SLSNe-I it is necessary to systematically explore the region of the parameter space with ${L}_{x}\lt {10}^{41}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ both at very early ($t\lt 30$ days, rest-frame) and late times ($t\gt 1000$ days, rest-frame), where the emission from an off-axis relativistic jet, weak uncollimated relativistic outflow or magnetar ionization breakout might be found. This parameter space is almost an entirely uncharted territory of exploration and holds promise for future discoveries.

We thank the referee for constructive criticism and suggestions that helped improve the quality of the paper and the clarity of presentation of our results. We thank A. Kann, K. Murase, and N. Soker for their comments and detailed reading of the manuscript that we posted on the archive. This research has made use of the XRT Data Analysis Software (XRTDAS) developed under the responsibility of the ASI Science Data Center (ASDC), Italy. We acknowledge the use of public data from the Swift data archive. This work is partially based on data acquired with the Swift GO program 1114109 (PI Margutti). The scientific results reported in this article are partially based on observations made by the Chandra X-ray Observatory under program GO6-17052A (PI Margutti), observation IDs 17879, 17880, 17881, 17882, and IDs 13772, 14444, 14446 for PI Pooley. This work is partially based on observations by XMM-Newton, IDs 0743110301, 0743110701, 0770380201, 0770380401 (PI Margutti, proposal 74311). D.C. and R.M. acknowledge partial support from programs No. NNX16AT51G and NNX16AT81G provided by NASA through Swift Guest Investigator Programs. C.G. acknowledges University of Ferrara for use of the local HPC facility co-funded by the "Large-Scale Facilities 2010" project (grant 7746/2011). Development of the Boxfit code was supported in part by NASA through grant NNX10AF62G issued through the Astrophysics Theory Program and by the NSF through grant AST-1009863. Simulations for BOXFIT version 2 have been carried out in part on the computing facilities of the Computational Center for Particle and Astrophysics (C2PAP) of the research cooperation "Excellence Cluster Universe" in Garching, Germany. G.M. acknowledges the financial support from the UnivEarthS Labex program of Sorbonne Paris Cité (ANR10LABX0023 and ANR11IDEX000502).

Here we provide the details of the X-ray observations of SLSNe-I in the "bronze" and "iron" samples. Table 5 reports the measured fluxes for the entire sample of 26 SLSNe-I analyzed in this paper. For the non-detections we assume a non-thermal power-law spectrum with a photon index Γ = 2 and Galactic absorption.

Table 5.  X-Ray Observations of SLSNe-I

SN	tSTART	tSTOP	Unabsorbed Flux (0.3–10 keV)	Instrument
	(MJD)	(MJD)	(${10}^{-14}\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$)
SCP 06F6	53949	53949	13.	XMMa
	54043	54043	<1.40	Chandra
PTF 09atu	55062.188	55062.328	<14.50	Swift-XRT
	57667.055	57667.055	<89.37	Swift-XRT
PTF 09cnd	55061.883	55062.000	<12.02	Swift-XRT
	55065.828	55065.945	<12.35	Swift-XRT
	55069.016	55069.289	<14.47	Swift-XRT
	55073.637	55073.777	<12.43	Swift-XRT
	55077.312	55077.594	<17.12	Swift-XRT
	55084.012	55084.887	<13.73	Swift-XRT
	55097.812	55097.945	<25.94	Swift-XRT
	55107.375	55107.453	<22.40	Swift-XRT
	57422.016	57422.766	<57.30	Swift-XRT

Note.

^aFrom Levan et al. (2013).

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

A.1. SN 2009jh/PTF 09cwl

A.2. PTF 09atu

A.3. PTF 09cnd

A.4. PTF 10aagc

A.5. SN 2010md/PTF 10hgi

A.6. SN 2010gx/CSS100313/PTF 10cwr

A.7. SN 2010kd

A.8. SN 2011ke/CSS110406/PTF 11dij

A.9. PS1-11bdn

A.10. PTF 11rks/SN 2011kg

A.11. SN 2012il/PS1-12fo/CSS120121

A.12. DES15C3hav

A.13. iPTF 13ehe

A.14. LSQ 14an

A.15. LSQ 14fxj

A.16. LSQ 14mo

A.17. CSS140925-005854

A.18. DES15S2nr

A.19. OGLE15qz

A.20. OGLE15sd

A.21. PS16aqv

A.22. PS16op