First Release of High-Redshift Superluminous Supernovae from the Subaru HIgh-Z SUpernova CAmpaign (SHIZUCA). I. Photometric Properties

HSC16adga (SN 2016jhm) was discovered shortly after the beginning of the survey at (R.A., decl.) = (10:02:20.12, +02:48:43.3). It appeared in a galaxy with the COSMOS photo-z of ${2.26}_{-0.30}^{+0.25}$ and the MIZUKI photo-z peaking at 2.19 (Figure 2). The MIZUKI photo-z PDF has the second major peak at z ≃ 2.45. The SN spectroscopic redshift is z = 2.399 ± 0.004 (C19), which is consistent with the broad photometric redshift probability distribution, especially with the second peak in the PDF in Figure 2. The SN location is 036 ± 010 (2.9 kpc at z = 2.399) away from the host galaxy center. We estimate the host galaxy properties by fixing the redshift (z = 2.399) using MIZUKI. It is estimated to be a star-forming galaxy with the stellar mass of ${9.3}_{-1.3}^{+0.9}\times {10}^{9}\,{M}_{\odot }$ and the SFR of ${13.1}_{-2.7}^{+0.3}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$.

Figure 4(a) presents the observed light curves (LCs) of HSC16adga. The original data are summarized in Table 2. Figure 5 is the rest-frame spectral energy distribution (SED) obtained with the broadband photometry at the selected epochs shown in Figure 4(a). We do not take any host galaxy extinction into account in this paper. Because at early epochs the photometric temperatures are higher than ≃20,000 K, the peak of the SEDs is not constrained by our optical photometry. Afterward, the photospheric temperatures gradually cool to ≃15,000 K (20 days) and ≃10,000 K (30 days). The temperature evolution is consistent with that of luminous SNe IIn (e.g., Fassia et al. 2000).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Using the results of the blackbody SED fitted to the broadband photometry, we estimate the rest-frame ultraviolet (the uvm2 and uvw1 filters from Swift) and optical (the u band filter from the Sloan Digital Sky Survey and the g band from HSC) LCs as shown in Figure 4(b). The u band magnitude of HSC16adga is similar to that of SN 2010jl, although the LC of HSC16adga is probably still rising. The g band magnitude is getting more luminous, and it is about to reach −21 mag. Although HSC16adga does not reach exactly −21 mag in optical, which was used to define SLSNe under one criterion (Gal-Yam 2012), it is currently known that many SLSNe, which are spectroscopically classified nowadays, do not reach −21 mag in optical (e.g., De Cia et al. 2018; Quimby et al. 2018). These facts, as well as the proximity to −21 mag, lead us to claim that HSC16adga is likely a SLSN.

HSC17auzg (SN 2016jhn) was first detected on 2016 December 23 at (R.A., decl.) = (09:59:00.42, +02:14:20.8) in the z band (Figure 3). It appeared in a galaxy with the COSMOS photo-z of ${1.65}_{-0.08}^{+0.06}$ and the MIZUKI photo-z centering at 1.78 (Figure 2). The spectroscopic follow-up observations confirmed the redshift of 1.965 ± 0.004 (C19). The spectroscopic redshift is at the upper limit allowed by the photometric redshift PDF. The SN is at 078 ± 005 (6.5 kpc at z = 1.965) away from the host galaxy center. The host galaxy is estimated to have the stellar mass of ${3.0}_{-0.3}^{+0.1}\times {10}^{10}\,{M}_{\odot }$ and the SFR of ${33.8}_{-0.8}^{+8.1}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ by MIZUKI using the spectroscopically confirmed redshift.

The LCs of HSC17auzg are shown in Figure 6 (see Table 3 for the data). After the first detection, its brightness continued to increase for about 3 months in the z band, while the rise times in the bluer bands are shorter (Figure 6(a)). The SED evolution obtained by the broadband photometry is presented in Figure 7. The blackbody temperatures evolve from above ≃21,000 K to ≃11,000 K in about 40 days. The corresponding blackbody radii are ≃7 × 10¹⁴ cm to ≃2 × 10¹⁵ cm.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 6(b) shows the rest-frame ultraviolet and optical LCs of HSC17auzg, which are obtained by using the blackbody SEDs in Figure 7. The LCs are compared with those of SLSN IIn 2006gy (e.g., Smith et al. 2007) and Type IIn SLSN LSQ15abl (Brown et al. 2014). Although the rest-frame ultraviolet LCs of SN 2006gy do not exist, its optical LC rising rate is found to be consistent with HSC17auzg. The LC declining rate of HSC17auzg is consistent with that of LSQ15abl. The peak absolute magnitudes in the wavelength range presented in the figure do not reach −21 mag, which classically separates SLSNe from the other SNe by one definition (Gal-Yam 2012). However, it is currently known that SLSNe do not necessarily reach −21 mag (e.g., De Cia et al. 2018; Quimby et al. 2018). The LCs in the bluer bands are found to start declining earlier, and the LCs in the redder optical bands may actually keep rising in HSC17auzg (Figure 6(b)). Therefore, we believe that HSC17auzg is likely a SLSN.

The third spectroscopically confirmed high-redshift SN is HSC17dbpf (SN 2017fei) at (R.A., decl.) = (09:58:33.42, +01:59:29.7). It appeared in a galaxy with the COSMOS photo-z of ${2.25}_{-0.53}^{+0.08}$ and the MIZUKI photo-z centering at 1.58 (Figure 2). The redshift of HSC17dbpf is confirmed to be 1.851 ± 0.004 with the spectroscopic follow-up observation (C19). Our spectroscopic redshift is higher than that estimated by the MIZUKI code, but it is in the 1σ error range of the photometric redshift estimated by COSMOS. The SN location is 058 ± 008 (4.9 kpc at z = 1.851) away from the host galaxy center. MIZUKI estimates the host galaxy stellar mass of ${9.2}_{-3.1}^{+0.2}\times {10}^{9}\,{M}_{\odot }$ and the SFR of ${40.0}_{-1.1}^{+5.1}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$.

The LCs of HSC17dbpf are reported in Figure 8. The original data are in Table 4. HSC17dbpf has the LC that evolves more rapidly than the other two high-redshift SNe we found. The rest-frame SED evolution of HSC17dbpf is presented in Figure 9. The blackbody temperature in the first epoch is not well constrained, but the steep rise in the SED indicates a high temperature above 30,000 K. The blackbody temperature evolves to ≃14,000 K on the second epoch and then goes down to ≃11,000 K on the third epoch. The early high temperature is consistent with that found in SN 2003ma (Rest et al. 2011) and Type I SLSNe (e.g., Nicholl & Smartt 2016). However, Type I SLSN spectra below around 3000 Å often deviate from a blackbody (e.g., Yan et al. 2017, 2018).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 8(b) presents the ultraviolet and optical LCs of HSC17dbpf in the rest frame. The rest-frame LCs are obtained by using the blackbody SEDs shown in Figure 9. The peak absolute magnitudes are consistent within the uncertainty to −21 mag. The ultraviolet magnitudes of HSC17dbpf are similar to those of some Type I SLSNe, and it is also likely that the LCs in the redder optical bands are still rising. Therefore, we claim that HSC17dbpf is a SLSN. The rapid LC rise is consistent with that of Type IIn SLSN 2003ma, although our rest-frame absolute magnitude estimates have large errors at the beginning. Such a rapid rise is not found in optical LCs of Type I SLSNe, but the rest-ultraviolet LCs of Type I SLSNe to compare are lacking. It is also interesting to note that the rapid evolution is consistent with the rapidly evolving transients from the Dark Energy Survey recently reported by Pursiainen et al. (2018), as well as those found by our previous HSC survey (Tanaka et al. 2016). The recent discovery of the Type I SLSN 2017egm at z = 0.031 (e.g., Nicholl et al. 2017a; Bose et al. 2018) provided an opportunity to obtain the rest-ultraviolet photometry of the Type I SLSN, but the rise is slower than that of HSC17dbpf (Bose et al. 2018). The rapid rise is also found in some SNe in the luminosity range between SNe and SLSNe (Arcavi et al. 2016), but HSC17dbpf is brighter (Figure 8(b)). The rapid decline of HSC17dbpf is consistent with that of Type I SLSN 2016els, although the rise of this SN was not observed (Brown et al. 2014).

We have reported spectroscopically confirmed z ≃ 2 SNe so far. As presented at the beginning of this section, the z ∼ 2 candidates are selected based on the photometric redshift of the transients' host galaxies. We obtained about 10 SLSN candidates at 1.5 ≲ z ≲ 2.5 with this method, but we often did not have opportunities to conduct spectroscopic follow-up observations for them because of weather conditions and telescope availability. The priorities for spectroscopic follow-up observations on a given night, which is not target of opportunity but classically scheduled in the case of the Keck follow-up observations, are decided by the LC information, the apparent magnitudes of the candidates, and the uncertainties in the photometric redshift estimates. We first select SN candidates with at least a few epochs of LC information to see if they are consistent with being SNe. Then, we sort the candidates with host galaxy photometric redshifts. The higher-redshift SN candidates with small redshift uncertainties had higher priority for the spectroscopic follow-up. Then, we made the final decision based on the apparent magnitudes, which should be brighter than ∼25.5 mag for the Keck/LRIS follow-up or ∼24.5 mag for Gemini/GMOS-S follow-up to take a spectrum in a few hours. For example, HSC16apuo had the highest priority to follow for a while, but bad weather prevented us from taking its spectrum when it was bright enough. The remaining candidates and their LCs will be reported elsewhere after the completion of the survey.

HSC16apuo (AT 2016jho) was discovered at (R.A., decl.) = (10:01:29.42, +02:28:33.8) on 2016 December 23 in a galaxy with the COSMOS photo-z of ${2.82}_{-0.70}^{+0.47}$ (Figure 10). It appeared 056 ± 005 away from the host galaxy center, which is located in the southwest of HSC16apuo. The corresponding physical distance is at z = 3 is 4.3 kpc. The PDF of the photometric redshift of the host galaxy estimated by MIZUKI is shown in Figure 11. The most probable redshift is 3.22, but the PDF has the relatively extended distribution ranging from z ≃ 2.8 to 3.5. The SED of the host galaxy is presented in Figure 12. The relatively large photometric uncertainties lead to the uncertainty in the photometric redshift, but its faintness and the small flux below ∼4000 Å support the photometric redshift of around 3 and reject the low redshift possibilities. The estimated stellar mass and SFR of the host galaxy are ${4.5}_{-1.3}^{+2.1}\times {10}^{9}\,{M}_{\odot }$ and the SFR of ${6.0}_{-1.7}^{+2.5}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$, respectively. The best-fit SED model presented in Figure 12 has χ²/dof = 26.80/30. Assuming z = 3, the SN is brighter than −21 mag at the peak (Figure 13; see Table 5 for photometry data).

Zoom In Zoom Out Reset image size

Although HSC16apuo appears to be at the edge of this galaxy, there is another galaxy at 19 toward the northwest. The photometric redshift of this galaxy is degenerate, and it could be either z ∼ 0.5 or 3.

HSC17dsid (AT 2017fej) is the most distant SN candidate found in the survey so far. It was discovered at (R.A., decl.) = (10:02:58.12, +02:13:04.1) on 2017 April 26 in a galaxy with the COSMOS photo-z of ${4.1974}_{-0.13}^{+0.09}$ (Figure 10). The SN location is 016 ± 007 away (1.1 kpc at z = 4) from the host galaxy center. The PDF of the photometric redshift of the host galaxy estimated by MIZUKI is shown in Figure 11. The most probable redshift is 4.16. The PDF has a narrow distribution and HSC17dsid is likely at z ∼ 4. The host galaxy SED presented in Figure 12 clearly shows the signature of the Lyman break at ∼5000 Å, which strongly supports the host redshift of z ∼ 4. The stellar mass and SFR of the host galaxy are estimated to be ${3.4}_{-2.0}^{+4.8}\times {10}^{9}\,{M}_{\odot }$ and ${24.2}_{-24.2}^{+15.1}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$, respectively. The best-fit SED model presented in Figure 11 has χ²/dof = 22.39/30.

The LC of HSC17dsid is shown in Figure 14 (see also Figure 16). HSC17dsid was discovered at the end of our transient survey period. However, additional photometry was obtained by Gemini/GMOS-S on 2017 May 31 (UT) with the i band (these data are reduced following standard procedures; open diamond in Figure 14). Because HSC17dsid was much brighter than the host galaxy (24.8 mag in the i band) when it was observed with Gemini, we did not perform the image subtraction for the Gemini photometry, and we use the aperture photometry. We also had an opportunity to take another z band image with HSC on 2017 June 20 (UT). Assuming z = 4, HSC17dsid reaches −22.7 mag at 1600 Å in about 10 days. It is brighter than the ultraviolet-brightest SLSN currently known (Gaia16apd, Nicholl et al. 2017b; Yan et al. 2017), which could be due to a smaller metallicity and thus lower extinction from the ejecta.

Zoom In Zoom Out Reset image size

Two SLSNe have been reported in the literature that are at the similar redshifts to our SLSNe: DES15E2mlf at z = 1.861 (Pan et al. 2017) and DES16C2nm at z = 1.998 (Smith et al. 2018). Their peak optical magnitudes are at around 23 mag, which is brighter than the peak magnitudes of our SLSNe (∼24 mag). They are found in the shallower transient survey, and HSC allowed us to find fainter SLSNe.

Our high-redshift SLSNe are fainter in the rest-frame ultraviolet wavelengths that we directly observe, as well as in the rest-frame optical bands whose luminosity is estimated in the previous section. While DES15E2mlf and DES16C2nm are brighter than −22 mag in optical, our SLSNe basically remain fainter than −22 mag in the rest-frame optical. The rest-frame optical peak magnitudes of SLSNe are usually between −19 mag and −22 mag (e.g., Nicholl et al. 2017a; De Cia et al. 2018; Lunnan et al. 2018). Therefore, our survey is capable of finding SLSNe with peak magnitudes fainter than −21 mag, while the previous surveys could only observe the most luminous end of the high-redshift SLSN populations.

We roughly estimate the event rates of SLSNe at high redshifts based on the three spectroscopically confirmed SNe and SN candidates we present here. Because the complete selection of high-redshift SN candidates from the survey has not yet been finished, our rate estimates are approximate. We have at least a few more good SN candidates at z ∼ 2 whose redshifts are not spectroscopically confirmed. The actual number depends on the choice of criteria, which include detection sigma and number of filters exceeding the threshold, LC evolution, color evolution, interpolation over missing epochs and missing filter information, etc. In addition, our high-redshift SN candidates are selected based on the photometric redshifts of the host galaxies, and we have not completed our investigation into how our selection methods affect our discoveries. For example, we may have missed SLSNe without apparent host galaxies or with faint ones (see also Section 5.3 for related discussion). These remaining candidates and final rate estimates will be presented elsewhere.

The event rate can be expressed as ${\sum }_{i}(1+{z}_{i})/{\varepsilon }_{i}{VT},$ where T is the survey period, V is the comoving volume, z_i is the redshift of an event, and ε_i is the detection efficiency of an event (e.g., Prajs et al. 2017). We assume ε_i ∼ 1 here for simplicity. The actual efficiency is clearly less than 1 because we could not confirm the redshifts of many high-redshift SLSN candidates, and this makes our rates only approximate. The survey period is set as T ∼ 0.5 yr. Assuming a typical limiting magnitude of 26.5 mag, SNe evolving with absolute magnitudes brighter than ≃−20 mag can be detected up to z ∼ 4.5.

We estimate the event rates in the redshift ranges 1.5 ≲ z ≲2.5, 2.5 ≲ z ≲ 3.5, and 3.5 ≲ z ≲ 4.5. We confirmed three SLSNe at 1.5 ≲ z ≲ 2.5. Summing up the three events, we find the approximate event rate of ∼900 ± 520 Gpc⁻³ yr⁻¹ using Poisson error. This rate is consistent with the expected total SLSN rate at 1.5 ≲ z ≲ 2.5 obtained by extrapolating the total SLSN rate at z ∼ 0.2 based on the cosmic star formation history (Figure 15).

Zoom In Zoom Out Reset image size

We obtained at least one SLSN candidate at 2.5 ≲ z ≲ 3.5 and another SLSN candidate at 3.5 ≲ z ≲ 4.5. The approximate total SLSN rate based on these candidates are ∼400 ±400 Gpc⁻³ yr⁻¹ (2.5 ≲ z ≲ 3.5) and ∼500 ± 500 Gpc⁻³ yr⁻¹ (3.5 ≲ z ≲ 4.5) including the Poisson error. The approximate rates are already consistent with the high-redshift SLSN rate estimated by Cooke et al. (2012) (Figure 15). We note that Cooke et al. (2012) argue that their rate estimate is a lower limit.

The approximate event rates estimated in this section are summarized in Figure 15. We repeat that the estimates for the event rates presented here are still very approximate. A complete study of the high-redshift SN rates from the HSC-SSP transient survey will be presented elsewhere after the completion of the survey.

The host galaxies of the high-redshift SNe and SN candidates reported here have ∼−21 mag in the rest ultraviolet and are relatively massive galaxies (∼10¹⁰ M_⊙). Meanwhile, high-redshift (z > 1) SLSNe have been classified as Type I SLSNe and are typically found in less massive galaxies (∼10^9–8 M_⊙) (e.g., Schulze et al. 2018). The SFRs of our host galaxies (∼10 M_⊙ yr⁻¹) are also higher than those of the other SLSN host galaxies (∼1 M_⊙ yr⁻¹), but their specific SFRs (∼10⁻⁹ yr⁻¹) are similar to those found in the other SLSN host galaxies (e.g., Schulze et al. 2018). The MIZUKI code is shown to give galaxy property estimates that are consistent with the other methods, and this difference is not likely to originate from systematic uncertainties (Tanaka 2015).

Previous high-redshift SLSN discoveries are limited to Type I SLSNe, which are found to preferentially appear in low-metallicity environments (e.g., Howell et al. 2013; Pan et al. 2017; Smith et al. 2018). However, the galaxy mass–metallicity relation at z ≃ 2 shows that galaxies with ∼10¹⁰ M_⊙ at z ≃ 2 are as metal-poor as the ∼10⁸–10⁹ M_⊙ hosts in the local universe (e.g., Erb et al. 2006; Steidel et al. 2014; Sanders et al. 2015), where Type I SLSNe are found (e.g., Chen et al. 2017). Indeed, high-redshift Type I SLSNe are sometimes found in such massive galaxies with ∼10¹⁰ M_⊙ (Cooke et al. 2012; Perley et al. 2016; Schulze et al. 2018). Type II SLSN host galaxies have large diversities (e.g., Neill et al. 2011), and our high-redshift SLSN host properties are consistent with being both Type II and Type I host galaxies.

We note, however, that our preference to find SLSNe in relatively massive galaxies could be partly due to our method of selecting high-redshift SN candidates, i.e., using photometric redshifts of the host galaxies. Brighter galaxies would have better estimates of photometric redshifts, and we tend to follow high-redshift SN candidates with better photometric redshifts to secure our high-redshift SN discoveries. Further systematic investigations are required to conclude the host galaxy properties of the high-redshift SLSNe.