Towards a consistent framework of determining active galactic nucleus contribution fraction and host galaxy properties

abstract · abstract

Decomposing active galactic nucleus (AGN) emission from host-galaxy light is essential for identifying AGN-dominated systems and accurately deriving host-galaxy physical properties. However, estimating AGN contributions from multi-wavelength photometry remains challenging due to inherent parameter degeneracies in spectral energy distribution (SED) fitting. In this work, we establish a unified framework for estimating AGN contribution fractions and host-galaxy properties by combining complementary diagnostics: SED decomposition with two independent fitting codes, CIGALE and GRAHSP, and deep-learning-based imaging decomposition. We apply this framework to galaxies in the COSMOS-Web field using multi-wavelength photometry from the ultraviolet to the far-infrared. We calculate the AGN contribution fraction in the JWST/NIRCam F150W filter and compare the SED-derived estimates with independent AGN fractions obtained from deep-learning image decomposition. Our results reveal significant degeneracies in current SED-fitting approaches based on empirical or theoretical AGN templates and demonstrate that incorporating independent morphological information can help break these degeneracies and improve the reliability of AGN and host-galaxy property estimates.

S5.SS1: 5.1 AGN-significant galaxies: CIGALE refitting with f AGN DL f_{\mathrm{AGN}}^{\mathrm{DL}} / block 1 · html

Figure 5 illustrates the differences in reduced χ 2 \chi^{2} and key physical parameters between the original fits (open symbols) and the refitted results (black-edged symbols), colour-coded by f AGN DL f_{\mathrm{AGN}}^{\mathrm{DL}} . Despite a maximum variation of 0.4 in f AGN DL f_{\mathrm{AGN}}^{\mathrm{DL}} , AGN fraction constraints produce generally only modest changes in derived physical parameters and goodness-of-fit: maximum variations are 1.27 dex (stellar mass), 0.69 dex (dust attenuation), 0.81 dex (SFR), and 0.52 dex (reduced χ 2 \chi^{2} ). While adopting f AGN DL f_{\mathrm{AGN}}^{\mathrm{DL}} leads to a slight increase in χ ν 2 \chi^{2}_{\nu} , the resulting fits remain physically plausible and consistent with the observed photometry. The corresponding refitting SED-fitting results and parameters are shown in Fig. 11 and listed in Table 5 .

S6: 6 Conclusions / block 1 · html

Strong agreement for AGN-negligible galaxies. The three AGN identification methods demonstrate remarkable consistency for low-AGN systems. Specifically, 82.4% of the entire sample is identified as AGN-negligible by both CIGALE and GRAHSP, with 86.4% of these cases also showing a minimal AGN contribution according to the DL imaging analysis. This high concordance suggests that SED fitting and imaging methods reliably identify pure galaxy systems.

S6: 6 Conclusions / block 1 · html

Discrepancies increase systematically for AGN-dominant scenarios. When comparing f AGN f_{\mathrm{AGN}} from the two SED codes, we find that 15.0% of the sources are inconsistent between CIGALE and GRAHSP. Strikingly, in four-fifths of these inconsistent cases, GRAHSP identifies a significant AGN component where CIGALE does not, indicating a systematic tendency for GRAHSP to favor higher AGN fractions. Only 2.7% of the galaxies are classified as AGN-significant by both SED codes. Within this subset, only 40% show tight quantitative agreement, while the remaining 60% exhibit substantial scatter, indicating uncertain AGN characterisation even when both codes agree on AGN presence.

S6: 6 Conclusions / block 1 · html

Agreement remains limited when incorporating the DL predictions. Among galaxies identified as AGN-significant by both SED fitting, only 7.6% are consistently classified as hosting a significant AGN by the DL method. Furthermore, within this small subset, only ten galaxies exhibits robust quantitative agreement across all three methods.

S6: 6 Conclusions / block 1 · html

Re-running CIGALE by fixing f AGN f_{\mathrm{AGN}} to the DL-estimated value for 20 AGN-dominant galaxies yielded statistically acceptable fits, with a maximum reduced χ 2 \chi^{2} shift of 0.52 dex. Fixing the f AGN f_{\mathrm{AGN}} value resulted in maximum parameter variations of 1.27 dex in stellar mass, 0.69 dex in attenuation, and 0.81 dex in SFR.

S6: 6 Conclusions / block 1 · html

For galaxies with negligible f AGN SED f_{\mathrm{AGN}}^{\mathrm{SED}} but significant f AGN DL f_{\mathrm{AGN}}^{\mathrm{DL}} , we find that ∼ \sim 30% fail to yield a physically plausible solution when refitted with f AGN C ​ _ ​ DL f_{\mathrm{AGN}}^{\mathrm{C\_DL}} , as evidenced by their elevated reduced χ 2 \chi^{2} . Furthermore, we observe no clear correlation between the success of these refits and the magnitude of the resulting shifts in physical parameters.

A4: Appendix D SED fitting results and galaxy imaging for the AGN-significant and -negligible samples / block 1 · html

Table 5 presents the SED-fitting parameters for 30 AGN-significant galaxies identified by CIGALE, GRAHSP, and DL. For each physical parameter, we report results from both SED methods. Values in brackets correspond to the CIGALE refits with f AGN f_{\mathrm{AGN}} fixed to the DL estimates.

A4: Appendix D SED fitting results and galaxy imaging for the AGN-significant and -negligible samples / block 1 · html

Figure 11 presents the four randomly selected example galaxies that are identified as AGN-significant by the DL prediction and both SED decompositions. Galaxies with IDs 963088, 283630, and 428464 exhibit relatively larger scatter in f AGN DL f_{\mathrm{AGN}}^{\mathrm{DL}} compared to f AGN SED f_{\mathrm{AGN}}^{\mathrm{SED}} , and therefore undergo refitting with a fixed f AGN DL f_{\mathrm{AGN}}^{\mathrm{DL}} value as described in Sect 5.1 . The change in reduced χ 2 \chi^{2} is indicated in the titles of each sub-plot, and the dashed lines denote the refitted SED results. Figure 12 shows four randomly selected galaxies from the subsample with negligible AGN contributions according to the DL prediction and both SED decompositions.

A4: Appendix D SED fitting results and galaxy imaging for the AGN-significant and -negligible samples / block 1 · html

Table 5: Galaxy properties of 30 AGN-significant galaxies identified by all three methods.

S5.SS2: V.2 Results from SED fitting / block 1 · html

When including an AGN component, both Synthesizer-AGN and Prospector consistently yield log 10 ⁡ ( M ⋆ / M ⊙ ) ≈ 9 \log_{10}(M_{\star}/M_{\odot})\approx 9 , although Bagpipes retrieves a significantly lower stellar mass by ≈ 1 \approx 1 dex: log 10 ⁡ ( M ⋆ / M ⊙ ) = 7.92 \log_{10}(M_{\star}/M_{\odot})=7.92 . Similarly, CIGALE finds log 10 ⁡ ( M ⋆ / M ⊙ ) = 8.50 \log_{10}(M_{\star}/M_{\odot})=8.50 . Notably, Bagpipes predicts a stellar age of t age = 400 − 260 + 260 t_{\mathrm{age}}=400_{-260}^{+260} Myr, consistent with both CIGALE and Prospector within the uncertainties ( t age ≈ 400 − 600 t_{\mathrm{age}}\approx 400-600 Myr). However, its mass-weighted age, from Bagpipes , is only 4 − 3 + 2 4_{-3}^{+2} Myr, indicating that a substantial fraction of the stellar mass formed recently, which is consistent with a very recent burst of star formation.

S5.SS2: V.2 Results from SED fitting / block 1 · html

Interestingly, both Synthesizer-AGN and Prospector , when invoking an AGN, predict A V ≈ 2 − 4 A_{V}\approx 2-4 . In particular, Synthesizer-AGN predicts A V ≈ 0.74 A_{V}\approx 0.74 mag for the young population ( t age ≈ 10 t_{\mathrm{age}}\approx 10 Myr), in line with our estimate from NIRSpec/PRISM data, and A V ≈ 2.40 A_{V}\approx 2.40 for the older population ( t age ≈ 32 t_{\mathrm{age}}\approx 32 Myr). Similarly, Bagpipes and CIGALE yield A V A_{V} values consistent with those inferred from Balmer lines tracing the young population. Across all fits, the metallicity remains consistent at 10 10 – 20 % , Z ⊙ 20\%,Z_{\odot} , in agreement with our spectroscopic constraints.

S5: 5 Discussion / block 2 · html

mprovement in constraining the AGN and host galaxy properties provided when incorporating four additional FIR filters. B24 showed that we can effectively recognise AGN by using the PRIMA coverage to measure the shape of the IR dust continuum. B24 simulated PRIMA observations of “normal” and active galaxies over a large range of AGN fractions, IR luminosities, PAH contribution, dust properties, and redshifts and performed SED-fitting to investigate PRIMA capabilities in recovering the source properties. In particular, exploiting PHI, PPI1, and PPI2 filters, B24 were able to retrieve the AGN fraction with a dispersion of 0.06, the fraction of dust mass in PAH with a precision of ∼ 10 % \sim 10\% , and the IR luminosity with a scatter of 0.1 dex.

S5: 5 Discussion / block 2 · html

rily imply detections in all bands; if the flux plus noise fell below the band sensitivity threshold, the corresponding photometric point was treated as a non-detection. We found that with PRIMA coverage we can recover (within 0.1 dex of the true value) the bolometric luminosity for 75 % 75\% of our sources and for 63 % 63\% of those at z ≥ 3 z\geq 3 , a significant improvement over the 18 − 20 % 18-20\% of the sources without mid-IR and far-IR coverage. We also expect PRIMA to double the number of sources with reliable SFR (within 0.1 dex of the true value), with the fraction increasing from ∼ 32 % \sim 32\% to 63 % 63\% . We run further simulations focusing on sources at 6 ≤ z ≤ 8 6\leq z\leq 8 . We found that using PRIMA photometric bands, we measure bolometric luminosity within 0.1 dex of the true value for ∼ 30 % \sim 30\% of the simulated objects. More importantly, for the majority of sources with log ⁡ ( L bolo / erg ​ s − 1 ) ≥ 45 \log{(L_{\text{{bolo}}}/\rm{erg\,s^{-1}})}\geq 45 , the measured L bolo L_{\text{{bolo}}} is within 0.3 dex of the true value. Specifically, for sources with 45 ≤ log ⁡ ( L bolo / erg ​ s − 1 ) < 46 45\leq\log{(L_{\text{{bolo}}}/\rm{erg\,s^{-1}})}<46 , we obtain an average offset of log ⁡ L bolo sed − log ⁡ L bolo true = 0.24 ± 0.30 \log{L_{\rm{bolo}}^{\rm{sed}}}-\log{L_{\rm{bolo}}^{\rm{true}}}=0.24\pm 0.30 , similar to the bayesian uncertainties associated to the measured values. Considering the last two redshift bins in Fig. 5 ( 6 ≤ z < 7 6\leq z<7 and 7 ≤ z < 8 7\leq z<8 ), we can see that all the bolometric luminosity bins that we are able to populate with PRIMA are at log ⁡ ( L bolo / L ⊙ ) ≥ 11.5 \log{(L_{\text{{bolo}}}/\rm{L_{\odot}})}\geq 11.5 , or equivalently to log ⁡ ( L bolo / erg ​ s − 1 ) ≥ 45 \log{(L_{\text{{bolo}}}/\rm{erg\,s^{-1}})}\geq 45 . Therefore, we are confident that PRIMA will not only be able to detect sources up to z ∼ 8 z\sim 8 , but also to measure their L bolo L_{\text{{bolo}}} with reasonable accuracy, enabling us to reconstruct the bolometric LFs and BHAD even at these high redshifts.

abstract · abstract

We present full 3-28 $\mathrm{μm}$ JWST MIRI/MRS and NIRSpec/IFU spectra of the western nucleus of Arp 220, the nearest ultraluminous infrared galaxy. This nucleus has long been suggested to possibly host an embedded Compton-thick AGN. Millimeter observations of the dust continuum suggest the presence of a distinct 20 pc core with a dust temperature of $T_\mathrm{d} \gtrsim 500~\mathrm{K}$, in addition to a 100 pc circumnuclear starburst disk. However, unambiguously identifying the nature of this core is challenging, due to the immense obscuration, the nuclear starburst activity, and the nearby eastern nucleus. With the JWST integral field spectrographs, we can, for the first time, separate the two nuclei across this full wavelength range, revealing a wealth of molecular absorption features towards the western nucleus. We analyse the rovibrational bands detected at 4-22 $\mathrm{μm}$, deriving column densities and rotational temperatures for 10 distinct species. Optically thick features of C$_2$H$_2$, HCN and HNC suggest that this molecular gas is hidden behind a curtain of cooler dust, and indicate that the column densities of C$_2$H$_2$ and HCN are an order of magnitude higher than previously derived from Spitzer observations. We identify a warm HCN component with rotational temperature $T_\mathrm{rot} = 330~\mathrm{K}$, which we associate with radiative excitation by the hot inner nucleus. We propose a geometry where the detected molecular gas is located in the inner regions of the starburst disk, directly surrounding the hot 20 $\mathrm{pc}$ core. The chemical footprint of the western nucleus is reminiscent of that of hot cores, with additional evidence for shocks. Despite the molecular material's close proximity to the central source, no evidence for the presence of an AGN in the form of X-ray-driven chemistry or extreme excitation is found.

S6: VI Discussion / block 1 · html

Figure 9 highlights three “Good Fits” of SDSS1420+5300A. All three fits do a reasonably good job of modeling the radial profile of the host galaxy; however, the models span a wide range of effective radii, Sérsic indices, and AGN-to-host flux ratios. For this example, we find that the R Host = 20 R_{\mathrm{Host}}=20 model tends to underestimate the surface brightness of the central region, spiral arms, and extended wings of the source, while the larger radii models tend to model the bar and spiral arms more accurately, but overestimate the central region and wings. While the specifics vary from source to source, it is consistently true that the most important regions for constraining the Sérsic profile are the central pixels and the extended wings. As previously discussed, the central region is significantly influenced by residual effects from the AGN subtraction, making constraining the profile difficult. The extended wings of the source contribute very little to the overall χ ν 2 \chi^{2}_{\nu} of the fit—in Figure 9 , which shows minor but clear under- and over-subtraction of the extended wings between models, the χ ν 2 \chi^{2}_{\nu} of the three models differs by < 1 % <1\% . Thus, the shape of the profile is dominated by the region outside of the FWHM of the PSF, nearer to the effective radius. The larger distribution of profiles shown in Figure 8 demonstrate significant overlap in this region, leading to a wide range of “valid” fit parameters.

S4: IV Fitting Process / block 1 · html

Now that we have identified our AGN candidates and constructed PSF models, we can proceed to fitting our sources. We apply both Galfit and AstroPhot in order to compare the results between two often used 2D photometric fitting software packages. We follow the same process for both; we use a two-component model, with a Sérsic profile representing the host galaxy and a point-source model representing the AGN. By fitting the Sérsic and point-source components simultaneously, we can effectively disentangle the light from each source with minimal contamination. If the light from the central AGN is left unaccounted for, light is misattributed to the bulge, thus artificially inflating the Sérsic index and contaminating the characterization of morphology along with any galaxy magnitude or size measurement. Two-component fitting such as this is standard and has been thoroughly tested across both 1D and 2D-fitting [ 29 , 45 ] .

abstract · abstract

Little Red Dots (LRDs) are compact, red sources discovered by JWST at high redshift ($z \gtrsim 4$), marked by distinctive "V-shaped" spectral energy distributions (SEDs) and often interpreted as rapidly accreting AGNs. Their evolution remains unclear, as identifying counterparts at lower redshifts is challenging. We present WISEA J123635.56+621424.2 (here dubbed {\it the Saguaro}), a $z=2.0145$ galaxy in GOODS-North, as a possible analog of high-redshift LRDs and a potential missing link in their evolutionary path toward lower-redshift systems. It features a compact LRD-like nucleus surrounded by a face-on spiral host. Its connection to LRDs includes that: (1) its nuclear spectrum shows a clear "V-shaped" SED; and (2) when redshifted to $z=7$, surface brightness dimming makes the host undetectable, thus mimicking an LRD. This suggests that high-redshift LRDs may be embedded in extended hosts. To test this, we stack rest-frame UV images of 99 photometrically selected LRDs, revealing faint, diffuse emission. Stacking in redshift bins reveals mild radial growth, consistent with the expected galaxy size evolution. A simple analytic model confirms that surface brightness dimming alone can explain their compact appearance. Lastly, we show that {\it the Saguaro} is not unique by describing similar objects from the literature at $z\lesssim3.5$. Taken together, our results support a scenario in which LRDs may not be a distinct population, but could be the visible nuclei of galaxies undergoing a short-lived, AGN-dominated evolutionary phase, with their compact, red appearance driven largely by observational biases.

S6: 6 Summary / block 1 · html

We present one of the largest spectroscopic samples of broad Balmer line sources at high redshift to date, including 80 robust detections at z > 3.1 z>3.1 , 28 (35%) of which lie at z > 6 z>6 . Our detections are enabled by the unite package, which combines multiple NIRSpec spectra, i.e. the PRISM and G395M spectra, simultaneously to model emission lines and continua. This approach leverages the signal-to-noise of the PRISM and the resolution of G395M to identify kinematic components of emission lines, overcoming limitations of single-disperser analyses.

S6: 6 Summary / block 1 · html

The majority of sources with (spectroscopic) v-shaped continua also show a dominant point source component at rest-optical wavelengths. Remarkably, we robustly detect broad Balmer emission lines for upwards of 80% of these sources, with the remainder being inconclusive primarily due to data shortcomings. Similarly, the majority of sources with v-shaped continua show broad Balmer emission lines. Of this population, 80% also show a prominent point-source component in rest-optical imaging.

abstract · abstract

There is evidence for significant evolution in the gaseous and dust properties of galaxies since the era of cosmic noon ($1\lesssim z\lesssim 2.5$). The well known co-evolution of supermassive black holes with their host galaxies suggests a constant connection between the small-scale (nuclear) and large-scale regions of galaxies. A fundamental component of Active Galactic Nuclei (AGN) is the "torus", a dense, dusty structure that acts as the interface between the accretion disc and the ISM of the host galaxy. The transitional nature of the torus makes it a prime subject to search for evolution since cosmic noon. We use high-resolution near- and mid-IR imaging from the JWST CEERS program to disentangle the emission from the torus in unprecedented detail for 88 X-ray selected AGN at $z\sim2$. We employ a novel SED fitting technique that combines archival low-resolution multi-band photometry at UV to FIR wavelengths with the new high-resolution JWST photometry to constrain essential AGN and torus parameters, such as accretion disc luminosity, torus opening angle, and inclination angle. We demonstrate that this SED fitting approach leads to better AGN characterisation and tighter constraints on AGN parameters. The population-level analysis finds that the Covering Fraction ($CF$) distribution peaks at $\approx0.25$ with a long tail towards higher $CF$. Despite the well-known evolution of the ISM and structural properties of AGN hosts to these redshifts, the $CF$ distribution of our sample does not show any strong statistical difference with that found in local AGN of equivalent luminosity, or with those at intermediate redshifts.

abstract · abstract

Active Galactic Nuclei (AGN) significantly influence galaxy evolution. Specific sources such as obscured AGNs, especially Type II quasars (QSO2), still remain understudied. We characterise 366 QSO2 candidates in the redshift desert (median z~1.1) identified via machine learning from SDSS/WISE photometry, analysing their spectral energy distributions (SEDs) and deriving their physical properties. Using CIGALE, we estimated star formation rate (SFR), stellar mass (M), AGN luminosity, and AGN fraction. We compared these with SPRITZ simulations and the literature, placing results in the galaxy evolution context. Our QSO2 candidates show diverse evolutionary stages. The SFR-M diagram reveals high-SFR sources above the main sequence, linking AGN activity to enhanced star formation. Quenched galaxies may indicate obscured star formation or AGN feedback. Additionally, the physical properties align with SPRITZ composite systems and AGN2, endorsing our obscured AGN classification. This study validates machine learning for identifying AGN-host galaxies, beyond traditional colour-colour selections. Diverse candidate properties highlight this method's ability to identify complex AGN systems. This advances our understanding of AGN-driven galaxy evolution with new target selection.