Dwarf Satellites of High-z Lyman Break Galaxies: A Free Lunch for JWST

The James Webb Space Telescope (JWST; Gardner et al. 2006) will provide unprecedented observations of the most distant galaxies, revolutionizing our knowledge of the high-redshift Universe. It will be able to detect remarkably faint objects during the Epoch of Reionization (EoR; z ≳ 6) reaching ultraviolet (UV) absolute magnitudes of M_UV ≃ − 15 at redshift z ≃ 6 (e.g., Finkelstein 2016). The wavelength coverage of instruments like NIRCam (0.6–5 μm, Horner & Rieke 2004) will allow us to study their rest-frame optical and UV properties for the first time.

Among the faint objects expected to be present in large numbers at high-z are dwarf galaxies (M_⋆ ≲ 10⁹ M_⊙). Indeed, according to the hierarchical ΛCDM model, dwarf galaxies are the most abundant systems at all cosmic times, being the first galaxies to form and the basic building blocks of the massive galaxies that we see today.

The physical and observational properties of isolated dwarf galaxies located in the field have been widely studied through cosmological simulations (Wise et al. 2014; Zackrisson et al. 2017; Jeon & Bromm 2019). However, we also expect the presence of an additional category of dwarf galaxies: the satellites orbiting around more massive galaxies in the densest regions of the cosmic web (e.g., Conroy et al. 2006). In particular, Gelli et al. (2020) showed that a typical massive z ≃ 6 Lyman break galaxy (LBG) is surrounded by dwarf satellites, living inside its virial halo. These satellites represent an interesting complementary population with respect to isolated field dwarf galaxies, since their properties and evolution are affected by the peculiar environment in which they dwell.

JWST is expected to detect many LBGs: the emission of dozens of them will most likely appear already within the deep fields of high-priority surveys like the JWST Advanced Deep Extragalactic Survey (JADES; Rieke et al. 2019), that will reach exposure times of ∼20 hr for imaging with NIRCam. This provides the remarkable opportunity to exploit planned JWST surveys to detect for the first time and for free their dwarf satellite population.

Will JWST be able to effectively detect dwarf satellites of massive LBGs through imaging and photometry? Will such observations shed light on the star formation rates, stellar metallicity, and presence of young/metal-poor stars in these systems? We address these issues by using high-resolution, zoom-in cosmological simulations (Pallottini et al. 2017) reproducing the evolution of a typical z ∼ 6 LBG. In Gelli et al. (2020) we studied the stellar populations of the dwarf satellites. Here, we make a further step and model their emission, focusing on the stellar continuum.

2.1. Simulating High-z LBGs and Their Dwarf Satellites

2.2. Spectral Energy Distribution (SED) Building and Synthetic Images

To derive the intrinsic SEDs of the dwarf satellites based on their simulated stellar populations, we use STARBURST99 (Leitherer et al. 1999), also consistent with the simulation prescriptions. We proceed as follows: considering the total SFH of the selected galaxy (up to z ∼ 6, final snapshot of the simulation), we sample it in timesteps of Δt = 5 Myr, which assure a good convergence for the spectra. At each timestep we consider a single burst of star formation: the stellar mass, metallicity, and formation time are given as an input to STARBURST99. The final output consists in the rest-frame intrinsic spectrum of the galaxy at all times from its formation up to z ∼ 6.

The intrinsic spectrum can be attenuated by dust, for which we adopt the SMC synthetic extinction curve by Weingartner & Draine (2001). This is the natural choice for dwarf galaxies and a good approximation for high-z sources at λ > 0.3 μm (Gallerani et al. 2010). We compute the optical depth as: τ = C_ext × N_H[ cm⁻²] × Z[ Z_⊙], where C_ext is the tabulated cross section per H nucleon, and N_H and Z are the gas hydrogen column density and metallicity averaged on different line of sights. The latter two are retrieved directly from the simulation, considering the gas composition in the selected satellite galaxy at the snapshots available (i.e., every 18 Myr), and interpolating at intermediate times. The final SEDs (Figure 1) are obtained by reprocessing the intrinsic stellar spectra with the derived dust transmission curves.

Zoom In Zoom Out Reset image size

To enable a detailed spatial/morphological analysis, we have produced synthetic images of the central LBG and its dwarf satellites at z ∼ 6 (Figure 2). The images are built from stellar density maps of the simulated, face-on LBG disk (e.g., see Figure 2 in Gelli et al. 2020), and using the proportionality between stellar mass and observed magnitude in the selected bands. We simulate the instrument effects by convolving the high-resolution images with a Gaussian with the FWHM of the selected NIRCam filter, and then rebinning to match the instrument pixel size. No sky background has been simulated when creating the images.

Zoom In Zoom Out Reset image size

The resulting SEDs of the satellites and the central LBG Althæaare shown in Figure 1, in the NIRCam wavelength range, 0.6–5 μm, and for two different redshifts: z ≃ 6.5 (left panel) and z ≃ 6 (right panel). We adopt the same notation as in Gelli et al. (2020) to indicate the satellites, i.e., they are named from S1 to S6 with increasing stellar mass. S1 is the proto-globular cluster: its flux is more than 2 orders of magnitude below that of the other satellites, hence invisible even for JWST deep surveys. We notice that the flux of the central LBG (gray dashed curve) always exceeds that of the satellites by more than an order of magnitude at λ > 2 μm, because of the larger stellar mass. The two displayed redshifts pinpoint two relevant evolutionary stages of the three smallest satellites (S2, S3, S4, M_⋆ ≲ 5 × 10⁸ M_⊙):

• 1.  
  Bursty phase (z ≃ 6.5, left panel): in this epoch the least massive satellites are all experiencing the brief ( ≲ 50 Myr) burst of star formation in which they form all of their stars. At this stage they are characterized by relatively high SFRs, up to SFR ∼ 14 M_⊙ yr⁻¹. Hence, their stellar populations are dominated by young newly formed stars, resulting in spectra with an intense rest-frame UV flux at λ ≲ 2.5 μm.
• 2.  
  Passive phase (z ≃ 6, right panel): this stage takes places ∼ 100 Myr later, when the star formation in the least massive satellites has been completely suppressed by SN feedback. The older stars and lack of ongoing star formation result in redder spectra. Note that at λ ≲ 2.5 μm the flux has decreased by a factor ∼10.

SEDs of the two most massive satellites (S5 and S6, M_⋆ ≳ 5 × 10⁸ M_⊙) are dominated by old stellar populations in both epochs, since they are older than 300 Myr. The SFH of these systems is characterized by a low-level, continuous activity fueled by merger events that produce peaks with SFR < 5 M_⊙ yr⁻¹ (e.g., see Figure 7 from Gelli et al. 2020). By z ∼ 6.5 these high-mass systems can be considered as "stabilized" since the bulk of their stellar population is already in place. As a result, their SED is almost unchanged between the two stages. Also, in the first stage they are fainter than S2, S3, S4 due to the lower SFR, and to their more dusty nature; as lower systems fade, massive systems are the brightest ones in the passive phase. ⁴

To highlight dust effects on the final spectra, we compare the intrinsic stellar emission for S3, S4 and S6. As the dust content broadly scales with galaxy stellar mass, massive systems suffer a larger attenuation. For instance, even though S3 and S4 have similar intrinsic fluxes at wavelengths λ ∼ 1 μm during the bursty phase, the more massive S4 is more attenuated. However, low-mass systems are particularly affected by dust extinction during their starburst phase, which is characterized by higher gas column density in the star-forming regions. Interestingly, the predicted fluxes are in most cases well above the sensitivity of NIRCam in different filters, shown for a S/N = 5 in ∼3 hr by horizontal black lines. However, these estimates are valid for point sources. To draw solid conclusions on their observability, we need to assess whether the satellites can be spatially resolved by JWST.

Figure 2 shows the synthetic images of the LBG system at z ∼ 6 over an area of 25 × 25 in two wide filters (F200W and F356W). The colored pixels within the images show the flux at S/N∼3, with exposure times of ∼20 hr (top) and ∼3 hr (bottom). The white contours indicate ⁵ the areas with S/N ∼10. The image shows the system at z ∼ 6, i.e., the last snapshot available during the passive phase, when the three low-mass satellites are at their faintest level. This conservative choice reveals how, even in this worst case of detectability conditions during their evolution, many satellites do exceed the sensitivity threshold. Namely, the fluxes of S5 and S6 reach S/N = 10 in both filters and they are detectable in just 3 hr of observations. The emission of all five dwarf galaxy satellites exceeds the S/N = 3 sensitivity threshold in the wide filter F356W in 20 hr. In F200W, S2 and S3 remain undetected at z ∼ 6, but they are detectable in ∼17 hr at previous stages when their flux is higher (6.5 ≲ z ≲ 6.2). Figure 2 shows that to distinguish the dwarf galaxy satellites from the central LBG, a separation of ≳025 is required. Noticeably, this requirement is achieved for all dwarf satellites with the exception of S5, which is located in the inner regions (1.2 kpc from the center of Althæa) and whose emission is likely confused with that of the central LBG.

In conclusion, dwarf satellites of high-z LBGs can be actually observed with JWST within the planned deep surveys exceeding ∼20 hr of exposure time. What information can be inferred from their photometry, in particular through color–magnitude diagrams (CMDs)? While retrieving colors and magnitudes, we checked the possible contamination by nebular emission lines (e.g., H_α , H_β , [O III]λ5007 and [O II]λ3727), post-processing the simulation outputs with cloudy (Ferland et al. 2017), with the caveat of not having a consistent modeling of the interstellar radiation field with radiative transfer (see Pallottini et al. 2019). Our estimates show that emission lines are expected to produce negligible variation in both color and magnitude ( < 0.1 dex) for the three satellites farther from Althæa (S3, S4, S6), while the two located in its proximity ( ≲ 04) suffer heavier line contamination due to the nearby LBG disk. Figure 3 shows the CMDs from z ∼ 6.5 (t_Uni ∼ 850 Myr) to z ∼ 6 (t_Uni ∼ 950 Myr) of the three best-target satellites, which (i) are located farther from the central LBG, and (ii) can be identified in only 7 hr observations in F356W at z ∼ 6 (the faintest evolutionary stage).

Zoom In Zoom Out Reset image size

In the upper four panels (a-b-c-d) we plot the color F200W–F356W as a function of the AB magnitude in the filter F200W, and the symbols are color-coded according to SFR, stellar metallicity, Z_⋆, fraction of young, f_⋆(Age < 5 Myr), and metal-poor f_⋆(Z_⋆ < 10^−0.5 Z_⊙) stars. The path of each satellite in the CMD helps clarify how its SED evolves with time. We notice two trends: (i) a complex one characterizing "active" low-mass satellites (S3 and S4), which first (bursty phase) become brighter (from m_AB ≈ 29 to m_AB ≈ 27) with roughly constant color (F200W–F356W ≈ −0.5), and then (passive phase) become progressively dimmer and redder as they age; (ii) the simpler one of the high-mass S6, which by z ≈ 6.5 is already "stabilized", i.e., old/red (F200W–F356W ≳ 0.5), and therefore only shows a modest variation (≲1 dex) in both magnitude and color.

Since the high-mass satellite forms only ∼6% of its stellar mass during the time considered, no significant variation takes place in terms of both Z_⋆ (always super-solar, panel b) and f_⋆(Z_⋆ < 10^−0.5 Z_⊙) (≲15%, panel d). Conversely, its SFR varies due to a short bursts of star formation (SFR ∼ 5 M_⊙ yr⁻¹, panel a), which, however, is too weak to significantly affect the fraction of young stars (panel c). On the other hand, low-mass satellites, during the first ∼ 10 Myr displayed, experience the peak of their bursty star formation period in which they form the bulk of their stellar mass. The SFR reaches its peak ( ≳ 10 M_⊙ yr⁻¹, panel a), and the SED is mostly determined by newly formed young (≳50%, panel c), and metal-poor stars (≳80%, panel d). The stellar metallicity shows a rapid increase during the bursty phase (Z_⋆ ≳ 10⁻¹ Z_⊙, panel b). Over the ∼ 100 Myr shown, these low-mass systems spend only ∼ 20 Myr in the initial bursty phase, implying that in principle, when observing a high-z low-mass satellite, it will be more likely to detect it during the longer passive phase when their SFR has decreased and Z_⋆, f_⋆(Age < 5 Myr), and f_⋆(Z_⋆ < 10^−0.5 Z_⊙) settle to their final, constant values.

The two bottom panels (e–f) show the color F090W–F150W as a function of the F150W magnitude, and are color-coded according to the V-band dust optical depth, τ_V, and N_H. The SED shape in this spectral range is determined by two competing factors: young stellar populations, enhancing the flux at short wavelengths, and dust drastically reducing it. The trends in the CMD show that in general more massive satellites, which are dustier and denser, have higher F090W–F150W. The smallest object displayed (S3) shows a constant trend of the F090W–F150W, reflecting its simple SFH in which it settles to constant values of metallicity and column density. More massive satellites show a more complex behavior, in which the color becomes bluer at each small burst episode. The trends of the optical depth (e) and column density (f) are associated with the color: for instance, objects with F090W–F150W ≳ 2.5 have τ_V ≳ 1 and N_H ≳ 10²²cm⁻².

Based on our high-resolution cosmological simulations, we show that JWST will be able to detect for the first time faint dwarfs satellite galaxies of typical LBGs at the end of the EoR. Remarkably, this should happen already during the first planned observations. The JADES survey, for instance, will reach exposure times of ∼20 hr, sufficient for observing most satellites even during the faintest stages of their evolution. The survey will detect ∼50 LBGs with stellar mass M_⋆ ∼ 10¹⁰ M_⊙ (Williams et al. 2018), i.e., similar to Althæa. According to our simulations, we then expect to detect a total of ∼100–250 companion dwarf satellites, ∼40 of which will be most likely star bursting and dominated by young and metal-poor stars. In a future work (V. Gelli et al. 2021, in preparation) we will investigate the expected statistics of high-z dwarf satellite galaxies and probability of observing them in different phases, using a larger sample of LBG targets from the SERRA suite (A. Pallottini et al. 2021, in preparation). The simulation used in this work (Pallottini et al. 2017) is ideal to study this kind of sourced as it reproduces key observed properties of known LBGs, from L_{[C II]} luminosities (Carniani et al. 2018) to the Schmidt–Kennicutt relation (Krumholz et al. 2009).

Once candidate dwarf satellites are individuated based on the proximity to a massive LBGs, a measurement of their photometric redshift is needed to infer if they do belong to the LBG system, or if we are rather dealing with an interloper. NIRCam will be able to retrieve z ∼ 6 galaxies redshifts with a percentage of misidentifications as lower z galaxies of 9% (Bisigello et al. 2016), hence allowing a successful identification of the target LBG satellites.

An additional interesting finding of the simulation analysis (see Gelli et al. 2020) is the presence of a proto-globular cluster among the satellite population of the LBG at z ∼ 6. This is characterized by a stellar mass M_⋆ ∼ 10⁶ M_⊙, formed in an single burst of star formation ( < 15 Myr), and completely lacking both gas and dark matter. In this study, we find that such objects, despite their predicted existence in the vicinity of LBGs, are not bright enough to be observed with JWST/NIRCam: more than ∼1000 hr would be required to detect their light at S/N ∼ 3. Since within the JADES survey we do not expect the presence of gravitationally lensed sources, the hundreds of expected observed satellites around LBGs will be indeed dwarf galaxies, and not globular clusters (e.g., Vanzella et al. 2017). Among them, some will be most likely detected while experiencing their burst of star formation. We showed how photometric observations will be most relevant to shed the first light on high-z satellite galaxies and, provided that many of these sources will be discovered in the surroundings of LBGs, we suggested that the F200W–F356W color is a powerful diagnostic tool to understand their properties. For example, F200W–F356W ≲ −0.25 characterizes star-bursting (SFR ∼ 5 M_⊙ yr⁻¹), low-mass (M_⋆ ≲ 5 × 10⁸ M_⊙), low stellar metallicity systems, with ∼80% of their stellar population composed by young and metal-poor $[\mathrm{log}({Z}_{\star }/{Z}_{\odot })\lt -0.5]$ stars.

We thank the referee for the careful reading of the Letter and the insightful comments. We acknowledge support from PRIN-MIUR2017, prot. n.2017T4ARJ5, ERC Starting Grant NEFERTITI H2020/808240 and ERC Advanced Grant INTERSTELLAR H2020/740120.