Professional Documents
Culture Documents
Stellar Populations of A Sample of far-IR AGN and non-AGN Green Valley Galaxies
Stellar Populations of A Sample of far-IR AGN and non-AGN Green Valley Galaxies
Ethiopian Space Science and Technology Institute (ESSTI), P.O. Box 33679, Addis Ababa, Ethiopia
4 Instituto de Astrofísica de Andalucía (IAA-CSIC), Glorieta de la Astronomía s/n, 18008 Granada, Spain
5 Physics Department, Faculty of Science, Mbarara University of Science and Technology, P.O. BOX: 1410 Mbarara, Uganda
6 Southern African Large Telescope, P.O. Box 9 Observatory, Cape Town, South Africa
7 MSPE Department, School of Education, College of Education, University of Rwanda, P.O. Box 5039, Kigali, Rwanda
8 National Research Foundation, Box 2600, Pretoria, 0001, South Africa
Accepted 2022 April 22. Received 2022 April 8; in original form 2021 August 16
ABSTRACT
We present a study on the stellar populations and stellar ages of a sub-sample of far-
infrared AGN and non-AGN green valley galaxies analysed in Mahoro et al. (2017, 2019)
at 0.6 < z < 1.0 using the data from the COSMOS field. We used long-slit spectroscopy
and derived stellar populations and stellar ages using the stellar population synthesis code
“STARLIGHT” and analysed the available Lick/IDS indices, such as Dn4000 and HδA . We
find that both FIR AGN and non-AGN green valley galaxies are dominated by intermediate
stellar populations 67% and 53%, respectively. The median stellar ages for AGN and non-
AGN are log t = 8.5 [yr] and log t = 8.4 [yr], respectively. We found that majority of our
sources (62% of AGN and 66% of non-AGN) could have experienced bursts and continuous
star formation. In addition, most of our FIR AGN (38%) compared to FIR non-AGN (27%)
might have experienced a burst of SF more than 0.1 Gyr ago. We also found that our FIR
AGN and non-AGN green valley galaxies have similar quenching time-scales of ∼ 70 Myr.
Therefore, the results obtained here are in line with our previous results where we do not find
that our sample of FIR AGN in the green valley shows signs of negative AGN feedback, as
has been suggested previously in optical studies.
Key words: galaxies: active; galaxies: evolution; galaxies: star formation; infrared: galaxies;
galaxies: high-redshift; galaxies: structure; galaxies: stellar content
Table 1. Comparison of photometric redshift, SFR, stellar mass, and magnitude properties between the M20 and M19 samples of FIR AGN and non-AGN
galaxies.
Table 2. Number of galaxies (and fraction) in morphological visual classi- (e.g., Cid Fernandes et al. 1998; Schmitt et al. 1999; Oio et al.
fication from Mahoro et al. (2019). 2019).
STARLIGHT fits an observed spectrum Oλ with a model Mλ
FIR AGNs FIR non-AGNs
that blends up to N? single stellar populations (SSPs) with distinct
M20 M19 M20 M19
ages and metallicities from different stellar population synthesis
Class 1 3 (23%) 26 (25 %) 24 (22%) 452 (17%)
Class 2 5 (38% ) 27 (26 %) 50 (45%) 1204 (46%) models. The code works out on the following equation for a model
Class 3 - 2 (2%) - 87 (3%) spectrum Mλ :
Class 4 3 (23% ) 39 (38 %) 27 (25%) 494 (19%) N
?
X
Class 5 2 (15 %) 9 (9% ) 9 (8%) 372 (14%) Mλ = Mλ0 x j b j,λ rλ ⊗ G(v? , σ? ), (1)
j=1
2.1.2 Comparison of AGN and non-AGN in M20 sample. where b j,λ rλ is the reddened spectrum of the j-th SSP normalised
In this section we compared normalised distributions of AGN at λ0 ; rλ = 10−0.4(Aλ −Aλ0 ) is the reddening term; Mλ0 is the synthetic
and non-AGN green valley galaxies of the M20 sample analysed in flux at the normalisation wavelength; and ~x is the population vec-
this paper (see Figure 1, the right-hand column). First, we made a tor. The sign ⊗ indicates the convolution operator and G(v? , σ? ) is
comparison based on morphological classification done in Mahoro the Gaussian distribution used to model the line-of-sight stellar mo-
et al. (2019). We found a slight difference in class 2, where 38% tions. It is centered at velocity v? with dispersion σ? . The addopted
of AGN have spiral morphologies compared to 45% of non-AGN normalisation wavelength range of 4010 Å ≤ λ0 ≤ 4160 Å for ob-
(see Table 2). Secondly, we made a comparison of the photometric served spectra and at the base of λ0 = 4020 Å. The best fit model is
redshift, and found that AGN are at a slightly higher redshift, with determined by minimising χ2 (through a simulated annealing plus
50% of the sample being in the range of 0.69 and 0.93, while 50% Metropolis scheme and Markov Chain Monte Carlo techniques):
of non-AGN green valley galaxies have redshifts in the range of 0.7 X
χ2 = [(Oλ − Mλ )wλ ]2 , (2)
and 0.86. Thirdly, we compared the SFRs of AGN and non-AGN
λ
and found that AGN show higher SFRs, with 50% of the sample
being in a range of 63 [M /yr] to 131.82 [M /yr], while 50% of where emission lines, bad pixels and unknown features are masked
non-AGN have SFRs between 33.88 [M /yr] and 81.28 [M /yr]. by assigning wλ = 0 to those regions. We first moved to the rest-
Fourthly, we compared stellar mass (in log scale) of AGN and non- frame all observed spectra and corrected them for foreground
AGN without finding any significant difference, as can be seen in Galactic extinction using the pystarlight5 library and Schlegel et al.
Table 2. Finally, we compared the F814W magnitudes and found (1998) maps of dust IR emission. For the intrinsic extinction we
that AGN are slightly brighter with 50% of galaxies being in the used the Cardelli et al. (1989) law (e.g., Cid Fernandes et al. 2011;
range 21.17 - 21.45, in comparison to non-AGN being in the range Pović et al. 2016; Werle et al. 2019; Wild et al. 2020). We fitted
21.05 - 21.82. the wavelength range from 3092 Å to 6534 Å, that in most of cases
In summary, our FIR AGN in the M20 sample are in average present minimum and maximum wavelengths in our spectra.
at a slightly higher redshift, with slightly higher SFRs, and a bit The spectral used basis consists of 25 elements set, which is
brighter in comparison to FIR non-AGN. a sample of the Evolutionary Population Synthesis models given
by Bruzual & Charlot (2003). It covers 25 ages of 0.001, 0.00316,
0.00501, 0.00661, 0.00871, 0.01, 0.01445, 0.02512, 0.04, 0.055,
0.10152, 0.16090, 0.28612, 0.50880, 0.90479, 1.27805, 1.434, 2.5,
3 STARLIGHT SPECTRAL FITTINGS 4.25, 6.25, 7.5, 10.0, 13.0, 15.0 and 18.0 Gyr to provide enough res-
In order to obtain the stellar population of AGN and non- olution in age and solar metallicity which were computed with the
AGN green valley galaxies, we model the stellar contribution in Salpeter (1955) initial mass function. The reason why we chose
the LEGA-C spectra through the stellar population synthesis code Bruzual & Charlot (2003) is that it has a wide range of wave-
STARLIGHT4 (Cid Fernandes et al. 2005). We checked our AGN lengths and it is also widely used in the literature, which allows us
spectra and found that all sources are type-2 AGN. Therefore, since to compare our results and previous works directly. Furthermore,
we are not having type-1 AGN, and we are in addition masking all the Bruzual & Charlot (2003) is the base model for STARLIGHT,
emission lines in spectra, we do not expect any significant AGN so it is convenient to use them together.
contribution to the continuum, as suggested in previous studies Nevertheless, to test the robustness of results, we fitted a random
1.0
0.5
0.0 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5
Classification Classification Classification
3.0
Normalised count
2.5
2.0
1.5
1.0
0.5
0.00.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
Photometricredshift Photometricredshift Photometricredshift
Normalised count
1.5
1.0
0.5
0.0 0 1 2 3 0 1 2 3 0 1 2 3
logSFR[M /yr] logSFR[M /yr] logSFR[M /yr]
2.0
Normalised count
1.5
1.0
0.5
0.0 9 10 11 12 9 10 11 12 9 10 11 12
logM*[M ] logM*[M ] logM*[M ]
Normalised count
1.0
0.5
0.019 20 21 22 23 19 20 21 22 23 19 20 21 22 23
mag(F814W) mag(F814W) mag(F814W)
Figure 1. Left column: Normalised distributions of morphological types (first row), photometric redshift (second row), SFR (third row), stellar-mass (fourth
row) and apparent F814W magnitude (bottom row) of AGN in the M20 subsample (dashed green histograms) and M19 sample (solid magenta histograms).
Middle column: Same as previous, but for non-AGN. Right column: Same as previous, but comparing FIR AGN (blue-dashed lines) and non-AGN (solid red
lines) of M20 subsample studied in this paper.
log t =
P
xt,Z log t,
t,Z 4.2 Stellar ages
In this section we compare the light-weighted mean ages of
where xt,Z is a fraction of light at stellar age t in our best-fitting FIR AGN and non-AGN green valley galaxies measured from
model and metallicity Z (which is fixed in our case as Z ). The STARLIGHT output, as explained in Section 3. Figure 4 shows
derived ages are a ’light-weighted’ mean of the total light. the normalised distributions of the light-weighted mean age in the
To obtain a general view of the stellar populations of AGN M20 sample, along with best-fit Gaussian functions fitted to the
and non-AGN, we arranged their stellar populations into three two distributions, highlighting their similarity. In line with previ-
groups: old with age > 109 yr, intermediate with age between ous section, AGN and non-AGN span a similar range of ages. The
108 yr < age ≤ 109 yr, and young populations with age ≤ 108 yr, median age of AGN is log t = 8.5 [yr], with 50% of sources cov-
in line with previous studies (e.g., Cid Fernandes et al. 2005, 2013; ering the range of log t = 8.4 − 8.6 [yr], while the median age of
Pović et al. 2016; Cai et al. 2020). non-AGN is log t = 8.4 [yr], with 50% having ages in the range of
F (normalized)
2 = 1.5 2 = 0.8
1.5 adev = 2.4 1.5 adev = 2.8
1.0 1.0
0.5 0.5
0.03600 3800 4000 4200 4400 4600 4800 5000 0.03600 3800 4000 4200 4400 4600 4800 5000
Rest frame wavelength [Å] Rest frame wavelength [Å]
0.4 0.4
0.2 0.2
residual
residual
0.0 0.0
0.2 0.2
0.4 0.4
3600 3800 4000 4200 4400 4600 4800 5000 3600 3800 4000 4200 4400 4600 4800 5000
Rest frame wavelength [Å] Rest frame wavelength [Å]
Figure 2. Example of STARLIGHT spectral fitting of AGN (left plot) and non-AGN galaxy (right plot). The top panel shows the data (Oλ, black line) and
STARLIGHT fit (Mλ, red line). The bottom panel presents the Oλ − Mλ residual spectrum. Magenta lines correspond to masked windows.
70
Non-AGN
60 AGN
Non-AGN
50 1.6 AGN
Fraction (%)
40 1.4
30 1.2
Normalisedcount
20
1.0
10
0.8
0
Young Intermediate Old 0.6
Stellar population
0.4
Figure 3. Stack bar plot showing the fractional number of FIR AGN (blue
hatched) and non-AGN (red solid) green valley galaxies whose average age 0.2
falls into the three defined ranges: young (age [yr] ≤ 108 ), intermediate (108
< age [yr] ≤ 109 ), and old (age [yr] > 109 ). 0.0
7 8 9 10
logages [yrs]
log t = 8.2 − 8.9 [yr]. Similar stellar ages are also obtained when
comparing the AGN and non-AGN subsamples within the same Figure 4. Normalised distribution of light-weighted mean ages of FIR green
stellar mass and redshift ranges, as indicated in the previous sec- valley AGN (blue dashed line) and non-AGN (red solid line) galaxies with
tion. the respective Gaussian that best fits the distributions. The vertical dashed
Since in STARLIGHT fittings we assumed a fixed solar metal- lines represent the median values for each histogram, where each median
licity (see Section 3), we did an independent and additional check has the same colour as one of the corresponding histograms.
of our analysis using the “rmodel”6 code (Cardiel et al. 2003). This
code uses different SSPs (Bruzual & Charlot 2003; Thomas et al.
2003; Vazdekis et al. 2003; Lee & Worthey 2005) and determines to 0.0) taken from the public catalogue of Straatman et al. (2019).
the stellar populations’ parameters, such as age, metallicity, the It can be seen that the locus of distribution in Figure 5 as far as
slope of the initial mass function, etc., using a pair of line inten- metallicities are concerned (green numbers of the grid) aligns with
sity indices as an input. We ran the rmodel using the Thomas et al. results obtained by STARLIGHT in Figure 4, where we assumed
(2003) SSPs. Figure 5 shows the positions occupied by our sources solar metallicity.
in HδA vs. Fe4383 parameter space (the [α/Fe] parameter was fixed However, the locus of the distribution of points in Figure 5
indicates stellar ages of approximately 1 Gyr (the red figures of
the grid). This apparent discrepancy is explained by the fact that
6 https://rmodel.readthedocs.io/en/latest/index.html the absorption line indices probe mass-weighted ages, rather than
[alpha/H]=0.0 [alpha/H]=0.0
Figure 5. HδA vs. Fe4383 diagram for FIR AGN (left) and FIR non-AGN (right) green valley galaxies plotted over the Thomas et al. (2003) models. The
model grid is the same in both plots. Dashed and dotted lines of the grid correspond to the predictions for constant age (denoted by red numbers in Gyr) and
metallicity (green numbers), respectively.
9.0
10 10 9.0
8.8 8.8
logage (yr) from STARLIGHT
8.4
H
8.2
8.2
0 0 8.0
8.0 7.8
Gonçalves et al. 2012; Schawinski et al. 2014; Hahn et al. 2017; 50% of our AGN and non-AGN have time-scale values in a range
Nogueira-Cavalcante et al. 2018, 2019; Carleton et al. 2020). In of 53 Myr - 103 Myr and 36 Myr - 194 Myr, respectively. More
this section, we measure the SF time-scale (τ) using SFR(t) ∼ exp(- discussion on quenching time-scales is given in Section 5.1.
t/τ), where t is time (i.e., our light-weighted measured stellar age),
SFR is as measured in Mahoro et al. (2017, 2019), and the exponen-
tial index (τ) is a time-scale in units of Gyr−1 (e.g., Bell & de Jong
2000; Taylor et al. 2011). Using the above equation, we obtained 5 DISCUSSION
similar median time-scales for FIR AGN and non-AGN green val- In general, galaxies are composed of multiple stellar popula-
ley galaxies of ∼ 71 Myr and ∼ 69 Myr, respectively. In addition, tions of different ages, and stellar populations are some of the key
AGN Non-AGN
Class1 Class1
9.5
Class2 2.5 Class2
2.5
Class4 Class4
2.0 1.5
8.5
1.0
1.5 8.0
0.5
7.5
1.0 10.6 10.8 11.0 11.2 11.4 11.6 11.8 0.0 10.0 10.5 11.0 11.5
logM [M /yr] logM [M /yr]
Figure 7. SFR vs. stellar mass diagram for our sample of FIR green valley AGN (left) and non-AGN (right) green valley galaxies. The MS of star-forming
galaxies (Elbaz et al. 2011) and 1 σ range are given with solid and dashed lines, respectively. The symbols are colour coded by the light-weighted stellar ages
from STARLIGHT, while the symbol type depicts the galaxies’ visual morphological class from Mahoro et al. (2019), where class 1 are elliptical, S0 or
S0/S0a galaxies, class 2 are spirals, class 4 are peculiar galaxies, and class 5 are unclassified galaxies. We do not have in our M20 sample galaxies classified
as class 3 (irregular) in Mahoro et al. (2019).
parameters for understanding the properties of galaxies and their in line with the one obtained in Alonso-Herrero et al. (2010) for
evolution. This paper has aimed to understand the stellar popula- local luminous infrared galaxies at z < 0.26.
tion properties of AGN and non-AGN galaxies with FIR emission
in the green valley, and to compare the obtained results with those The SFR vs. M∗ diagram (SFR-M∗ ) has been broadly explored
obtained previously in Mahoro et al. (2017, 2019) where enhanced in the literature (e.g., Belfiore et al. 2018; Sánchez et al. 2018,
SFRs have been found in AGN host galaxies, in contradiction with 2019; Lacerda et al. 2020) and has served as an important diag-
previous results carried out in optical. Section 4.1 and Section 4.2 nostic tool for studying galaxy evolution and transformation. As
presented comparisons of stellar populations and stellar ages of expected from their intermediate colours and physical properties,
AGN and non-AGN green valley galaxies at redshifts of 0.5 < z green valley galaxies in general lie mostly below the MS of star-
< 1.0), showing that their stellar populations are dominated by in- forming galaxies (e.g., Noeske et al. 2007; Salim et al. 2007; Ren-
termediate stellar ages. Our results are consistent with Pan et al. zini & Peng 2015; Smethurst et al. 2015). In our sample, however,
(2013), who examined green valley galaxies using COSMOS field most of the sources are located on the MS of SF, as can be seen
data at 0.2 < z < 1.0, and found intermediate stellar populations in Figure 7 and Table 3 and in Mahoro et al. (2017, 2019). For
and intermediate stellar ages. Our results are also in agreement non-AGN we can observe a trend where the galaxies with older
with a lower redshift sample of Phillipps et al. (2019), who used stellar ages have a tendency to be below the MS, as compared to
the Galaxy and Mass Assembly (GAMA) data at 0.1 < z < 0.2 those with intermediate and young stellar ages. These results are
finding that green valley galaxies show intermediate stellar popu- consistent with e.g., Ciesla et al. (2017), who used simulations and
lation ages. Moreover, Angthopo et al. (2020) used Sloan Digital star-forming galaxies in the GOODS-South field at 1.5 < z < 2.5,
Sky Survey (SDSS) data with green valley being defined through finding younger stellar ages on the MS and above, while older
the 4000 Å break strength, finding again that green valley galaxies galaxies are located below MS. This trend, however, is not so clear
are characterised by intermediate stellar ages, in line with Trussler in the case of our AGN sample, located mainly on and above the
et al. (2020) results when using SDSS DR7 galaxies in the redshift MS of SF, independently on their stellar ages. We need larger AGN
range of 0.02 < z < 0.085. spectroscopic samples to confirm this, however, results obtained in
In this work, we confirm that intermediate stellar populations this paper go in line with those obtained in Mahoro et al. (2017,
and intermediate stellar ages dominate also in the case of FIR se- 2019), where enhanced SFRs (measured in FIR) have been found
lected green valley galaxies, both AGN and non-AGN. In addition, in AGN host galaxies in comparison to non-AGN, within the same
in Section 4.3 we investigated the distribution of our sample in the stellar mass range and independently on morphology. Therefore,
Dn4000 vs. HδA diagram, and found that a high fraction of both stellar population analysis carried out in optical in this paper also
AGN (62%) and non-AGN (66%) show evidence of both bursts suggests that FIR AGN in the green valley do not show clear signs
and continuous SF in their star-formation histories. Our findings of quenched star formation and negative AGN feedback, as sug-
suggest that both AGN and non-AGN green valley galaxies in our gested previously in different X-ray and optical studies (e.g., Nan-
sample are star-forming galaxies, which explains their location on dra et al. 2007; Pović et al. 2012; Leslie et al. 2016). Moreover,
the main sequence of SF. It is also important to note that our M20 recently it was also suggested that local AGN are not just a simple
sample shows FIR emission, and that our Dn4000 median value is transition type between star-forming and quiescent galaxies, and