A&A, 693, A250 (2025)
https://doi.org/10.1051/0004-6361/202450894
c© The Authors 2025
Astronomy
&Astrophysics
Euclid preparation
LVII. Observational expectations for redshift z < 7 active galactic nuclei in the
Euclid Wide and Deep surveys
Euclid Collaboration: M. Selwood1,? , S. Fotopoulou1 , M. N. Bremer1, L. Bisigello2,3 , H. Landt4 ,
E. Bañados5 , G. Zamorani6 , F. Shankar7 , D. Stern8 , E. Lusso9,10 , L. Spinoglio11 , V. Allevato12 ,
F. Ricci13,14 , A. Feltre9 , F. Mannucci9 , M. Salvato15 , R. A. A. Bowler16 , M. Mignoli6 , D. Vergani6 ,
F. La Franca13 , A. Amara17, S. Andreon18 , N. Auricchio6 , M. Baldi19,6,20 , S. Bardelli6 , R. Bender15,21 ,
C. Bodendorf15, D. Bonino22 , E. Branchini23,24,18 , M. Brescia25,12,26 , J. Brinchmann27 , S. Camera28,29,22 ,
V. Capobianco22 , C. Carbone30 , J. Carretero31,32 , S. Casas33 , M. Castellano14 , S. Cavuoti12,26 ,
A. Cimatti34, G. Congedo35 , C. J. Conselice16 , L. Conversi36,37 , Y. Copin38 , F. Courbin39 ,
H. M. Courtois40 , M. Cropper41 , A. Da Silva42,43 , H. Degaudenzi44 , A. M. Di Giorgio11 , J. Dinis42,43,
F. Dubath44 , X. Dupac37, S. Dusini45 , M. Farina11 , S. Farrens46 , S. Ferriol38, M. Frailis47 , E. Franceschi6 ,
S. Galeotta47 , B. Gillis35 , C. Giocoli6,48 , A. Grazian49 , F. Grupp15,21, L. Guzzo50,18 , S. V. H. Haugan51 ,
H. Hoekstra52 , M. S. Holliman53, W. Holmes8, I. Hook54 , F. Hormuth55, A. Hornstrup56,57 , P. Hudelot58,
K. Jahnke5 , E. Keihänen59 , S. Kermiche60 , A. Kiessling8 , B. Kubik38 , M. Kümmel21 , M. Kunz61 ,
H. Kurki-Suonio62,63 , R. Laureijs64, S. Ligori22 , P. B. Lilje51 , V. Lindholm62,63 , I. Lloro65, D. Maino50,30,66,
E. Maiorano6 , O. Mansutti47 , O. Marggraf67 , K. Markovic8 , N. Martinet68 , F. Marulli69,6,20 , R. Massey4 ,
E. Medinaceli6 , S. Mei70 , M. Melchior71, Y. Mellier72,58, M. Meneghetti6,20 , E. Merlin14 , G. Meylan39,
M. Moresco69,6 , L. Moscardini69,6,20 , E. Munari47,73 , S.-M. Niemi64, J. W. Nightingale74,75 , C. Padilla76 ,
S. Paltani44 , F. Pasian47 , K. Pedersen77, W. J. Percival78,79,80 , V. Pettorino64, G. Polenta81 , M. Poncet82,
L. A. Popa83, L. Pozzetti6 , F. Raison15 , R. Rebolo84,85, A. Renzi3,45 , J. Rhodes8, G. Riccio12, H.-W. Rix5 ,
E. Romelli47 , M. Roncarelli6 , E. Rossetti19 , R. Saglia21,15 , D. Sapone86 , B. Sartoris21,47 ,
R. Scaramella14,87 , M. Schirmer5 , P. Schneider67 , T. Schrabback88 , M. Scialpi9,10,147 , A. Secroun60 ,
G. Seidel5 , S. Serrano89,90,91 , C. Sirignano3,45 , G. Sirri20 , L. Stanco45 , C. Surace68 ,
P. Tallada-Crespí31,32 , D. Tavagnacco47 , A. N. Taylor35, H. I. Teplitz92 , I. Tereno42,93, R. Toledo-Moreo94 ,
F. Torradeflot32,31 , I. Tutusaus95 , L. Valenziano6,96 , T. Vassallo21,47 , A. Veropalumbo18,24 , Y. Wang92 ,
J. Weller21,15 , E. Zucca6 , A. Biviano47,73 , M. Bolzonella6 , E. Bozzo44 , C. Burigana2,96 ,
C. Colodro-Conde84, G. De Lucia47 , D. Di Ferdinando20, J. A. Escartin Vigo15, R. Farinelli6, K. George21 ,
J. Gracia-Carpio15, M. Martinelli14,87 , N. Mauri34,20 , C. Neissner76,32 , Z. Sakr97,95,98 , V. Scottez72,99,
M. Tenti20 , M. Viel73,47,100,101,102 , M. Wiesmann51 , Y. Akrami103,104 , S. Anselmi45,3,105 ,
C. Baccigalupi100,47,101,73 , M. Ballardini106,6,107 , M. Bethermin108,68 , A. Blanchard95 , L. Blot109,105 ,
S. Borgani110,73,47,101 , S. Bruton111 , R. Cabanac95 , A. Calabro14 , G. Canas-Herrera64,112 , A. Cappi6,113,
C. S. Carvalho93, G. Castignani6 , T. Castro47,101,73,102 , K. C. Chambers114 , S. Contarini15,69 , T. Contini95 ,
A. R. Cooray115 , O. Cucciati6 , S. Davini24 , B. De Caro45,3, G. Desprez116, A. Díaz-Sánchez117 ,
S. Di Domizio23,24 , H. Dole118 , S. Escoffier60 , A. G. Ferrari34,20 , I. Ferrero51 , F. Finelli6,96 , A. Fontana14 ,
F. Fornari96 , L. Gabarra119 , K. Ganga70 , J. García-Bellido103 , V. Gautard120, E. Gaztanaga90,89,121 ,
F. Giacomini20 , G. Gozaliasl122,62 , A. Hall35 , H. Hildebrandt123 , J. Hjorth124 , J. J. E. Kajava125,126 ,
V. Kansal127,128 , D. Karagiannis129,130 , C. C. Kirkpatrick59, L. Legrand131 , G. Libet82, A. Loureiro132,133 ,
J. Macias-Perez134 , G. Maggio47 , M. Magliocchetti11 , R. Maoli135,14 , C. J. A. P. Martins136,27 , S. Matthew35,
L. Maurin118 , R. B. Metcalf69,6 , P. Monaco110,47,101,73 , C. Moretti100,102,47,73,101 , G. Morgante6,
S. Nadathur121 , L. Nicastro6 , N. A. Walton137 , L. Patrizii20, A. Pezzotta15 , M. Pöntinen62 , V. Popa83,
C. Porciani67 , D. Potter138 , I. Risso139 , P.-F. Rocci118, M. Sahlén140 , A. G. Sánchez15 , A. Schneider138 ,
E. Sefusatti47,73,101 , M. Sereno6,20 , P. Simon67, A. Spurio Mancini141,41 , J. Steinwagner15, G. Testera24,
R. Teyssier142 , S. Toft57,143 , S. Tosi23,24,18 , A. Troja3,45 , M. Tucci44, C. Valieri20, J. Valiviita62,63 ,
G. Verza144,145 , J. R. Weaver146 , and I. A. Zinchenko21
(Affiliations can be found after the references)
Received 28 May 2024 / Accepted 4 November 2024
? Corresponding author; matthew.selwood@bristol.ac.uk
Open Access article, published by EDP Sciences, under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This article is published in open access under the Subscribe to Open model. Subscribe to A&A to support open access publication.
A250, page 1 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
ABSTRACT
We forecast the expected population of active galactic nuclei (AGN) observable in the Euclid Wide Survey (EWS) and Euclid Deep Survey (EDS).
Starting from an X-ray luminosity function (XLF), we generated volume-limited samples of the AGN expected in the Euclid survey footprints.
Each AGN was assigned a spectral energy distribution (SED) appropriate for its X-ray luminosity and redshift, with perturbations sampled from
empirical distributions. The photometric detectability of each AGN was assessed via mock observations of the assigned SED. We estimate 40
million AGN will be detectable in at least one Euclid band in the EWS and 0.24 million in the EDS, corresponding to surface densities of
2.8× 103 deg−2 and 4.7× 103 deg−2. The relative uncertainty on our expectation for Euclid detectable AGN is 6.7% for the EWS and 12.5% for the
EDS, driven by the uncertainty of the XLF. Employing Euclid-only colour selection criteria on our simulated data we select a sample of 4.8× 106
(331 deg−2) AGN in the EWS and 1.7× 104 (346 deg−2) in the EDS, amounting to 10% and 8% of the AGN detectable in the EWS and EDS.
Including ancillary Rubin/LSST bands improves the completeness and purity of AGN selection. These data roughly double the total number of
selected AGN to comprise 21% and 15% of the Euclid detectable AGN in the EWS and EDS. The total expected sample of colour-selected AGN
contains 6.0× 106 (74%) unobscured AGN and 2.1× 106 (26%) obscured AGN, covering 0.02 ≤ z . 5.2 and 43 ≤ log10(Lbol/erg s−1) ≤ 47. With
these simple colour cuts expected surface densities are already comparable to the yield of modern X-ray and mid-infrared surveys of similar area.
The EWS sample is most comparable to the WISE C75 AGN selection and the EDS sample is most similar to the yield of the collated Spitzer
cryogenic surveys when considering Euclid bands alone, or the XXL-3XLSS survey AGN sample when also considering selection with ancillary
optical bands. We project that 15% (7.6%) of the total Euclid detectable population in the EWS (EDS) will exhibit X-ray fluxes that could be
detected in the XMM-COSMOS survey, showing that the vast majority of Euclid-detected AGN would not be detectable in modern medium-depth
X-ray surveys.
Key words. surveys – galaxies: active – quasars: general
1. Introduction
Active galactic nuclei (AGN) mark a phase of luminous accre-
tion of matter onto a central supermassive black hole (SMBH).
AGN therefore indicate periods of growth for the massive com-
pact objects, thought to ubiquitously occupy the centers of mas-
sive galaxies (Magorrian et al. 1998). A co-evolution between
AGN and their host-galaxies has long been invoked through a
series of observational relations linking their respective physical
properties (e.g. Ferrarese & Merritt 2000; Gültekin et al. 2009).
The interaction of AGN and their host-galaxies, or AGN feed-
back (Fabian 2012, and references therein), is a crucial mech-
anism to reproduce observed galaxy properties (Granato et al.
2004; Shankar et al. 2006; Marulli et al. 2008) and quench star
formation in massive galaxies (Gabor et al. 2010; Dubois et al.
2013; Piotrowska et al. 2022). In cosmological simulations AGN
feedback has been identified as a necessary ingredient to repli-
cate present day distributions of galaxies (e.g. Bower et al.
2006; Croton et al. 2006; Sijacki et al. 2007; Schaye et al. 2015;
Rosito et al. 2021; Ward et al. 2022).
The statistical evolution of galaxy and AGN populations over
cosmic time can be described by a luminosity function (LF).
AGN LFs have been of central importance since quasars (i.e.
high-luminosity unobscured AGN) were first discovered as dis-
tinct cosmological objects, as early as 1968 (Schmidt 1968).
The six decade-spanning field of work has seen AGN LFs con-
structed in the radio (e.g. Dunlop & Peacock 1990), submillime-
ter (e.g. Vaccari et al. 2010), infrared (IR; e.g. Gruppioni et al.
2013; Lacy et al. 2015), ultraviolet (UV)/optical (e.g. Boyle et al.
1987; Pei 1995; Kulkarni et al. 2019; Adams et al. 2023), X-
ray (e.g. Miyaji et al. 2000; Ueda et al. 2003; La Franca et al.
2005; Ueda et al. 2014; Aird et al. 2015; Buchner et al. 2015;
Fotopoulou et al. 2016a), and bolometric (e.g. Hopkins et al.
2007; Shen et al. 2020) domains, spanning a wide range of red-
shifts.
By deriving and comparing LFs from differently selected
samples of galaxies or AGN, their formation histories can be
compared and inferred (e.g. Dai et al. 2009; Gruppioni et al.
2013; Lacy et al. 2015). For example, the evolution of nuclear
obscuration in AGN can be probed by measuring and comparing
the LFs of obscured and unobscured AGN. Such studies found
that high-luminosity obscured AGN peak in space density at a
higher redshift than their unobscured counterparts, hinting at an
evolutionary scenario (e.g. Lacy et al. 2005; Hopkins et al. 2008;
Glikman et al. 2018).
Almost all studies have shown that the AGN LF has a strong
evolution with redshift, which is an ubiquitous feature across
different wavebands (e.g. Boyle et al. 2000; Ueda et al. 2003;
Richards et al. 2006a; Hasinger et al. 2005; Hopkins et al. 2007;
Gruppioni et al. 2013; Buchner et al. 2015; Fotopoulou et al.
2016a; Shen et al. 2020). This evolution is not a simple change
of normalisation but also affects the slope of the parameter-
ising function. For example, X-ray and some optical, radio,
and IR studies report a flattening of the faint-end slope of the
AGN LF with increasing redshift. This implies the peak space-
density of low-luminosity AGN is at a lower redshift than that of
bright quasars, indicating a “cosmic downsizing” of AGN (e.g.
Cowie et al. 1996; Hasinger et al. 2005; Aird et al. 2015). AGN
feedback is often invoked as the mechanism behind this phe-
nomenon, shutting down the supply of in-falling cold gas to the
central SMBH via galactic-scale dust ejection with powerful out-
flows or by the gradual heating of the host-galaxy’s dark matter
halo (e.g. Fabian 2012, and references therein).
The European Space Agency (ESA) Euclid space telescope
(Laureijs et al. 2011; Euclid Collaboration: Mellier et al.
2025), successfully launched in July 2023, is the pre-
mier dark energy mission of ESA. Euclid will observe
∼14 500 deg2 of the extra-Galactic sky over a six-year
period, providing high spatial resolution imaging sampled at
0′′.1 pixel−1 in the optical (Euclid Collaboration: Cropper et al.
2025) and ∼0′′.3 pixel−1 in the near-infrared (NIR;
Euclid Collaboration: Jahnke et al. 2025) for billions of
astrophysical sources (Euclid Collaboration: Scaramella et al.
2022). Despite having a primary mission focussed on cosmol-
ogy, the rich data set generated by the Euclid surveys will drive
significant progress in many areas of astronomy. To effectively
exploit Euclid data for AGN legacy science we must first
understand the available sample size and characteristics of the
AGN detectable with Euclid photometry.
This work aims to forecast the expected number of z < 7
unobscured and obscured AGN observable with Euclid pho-
tometry in the Euclid Wide Survey (EWS) and the Euclid
Deep Survey (EDS). We first determine the sample size and
properties of AGN detectable with Euclid photometry, before
A250, page 2 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
focussing on the sample of AGN we can select with Euclid
and Vera C. Rubin Observatory/Large Synoptic Survey Tele-
scope (Rubin/LSST; Ivezic´ et al. 2019) photometric criteria. For
a comprehensive analysis of 7 ≤ z ≤ 9 AGN in Euclid we
refer the reader to Euclid Collaboration: Barnet et al. (2019). We
compared the output of our simulations at z ∼ 7 with those of
Euclid Collaboration: Barnet et al. (2019) in the Appendix.
In Sect. 2 we introduce the X-ray LF (XLF) utilised in this
work. Section 3 outlines our AGN sample generation method
from input data to resulting photometry. Our main results for the
number of detected AGN in the Euclid surveys are presented
in Sect. 4. The drivers and impact of uncertainty within our
framework, the expected sample sizes of AGN selected using
Euclid photometric colour criteria, and comparisons to the yield
of AGN from surveys in different wavebands are discussed in
Sect. 5.
Throughout this work we refer to AGN as unobscured
or obscured based on their optical properties, that is unob-
scured AGN denote Type 1 AGN with broad permitted emis-
sion lines (full-width at half-maximum; FWHM &2000 km s−1)
and obscured AGN signifies Type 2 AGN with no observable
broad line components. We assume a ΛCDM cosmology with
H0 = 70 km s−1 Mpc−1, Ωm = 0.3 and ΩΛ = 0.7. All magnitudes
are in the AB system (Oke & Gunn 1983) unless stated other-
wise.
2. X-ray AGN luminosity function
The differential LF, dφ(L, z)/dL, is defined as the number of
AGN per unit comoving volume per unit luminosity interval
dφ(L, z)
dL
=
d2N
dVc dL
(L, z), (1)
where N is the number of AGN with luminosity L in the comov-
ing volume Vc at redshift z and φ is the comoving number density
(dN/dVc). The expected number of AGN, 〈N〉, in luminosity and
comoving volume interval ∆L∆Vc(∆z) can be extracted from a
differential LF as defined in Eq. (1) via the calculation
〈N〉 =
∫ Lmax
Lmin
∫ zmax(L)
zmin
dφ(L, z)
dL
dVc
dz dΩ
(z) Ω(L, z) dz dL, (2)
where Lmin and Lmax are the minimum and maximum of the lumi-
nosity interval ∆L respectively, zmin is the minimum of the red-
shift interval ∆z, zmax(L) is either the maximum redshift at which
an object of luminosity L can still be detected or the maximum
of the redshift interval, dVc/dz dΩ is the differential comoving
volume element of the Universe at redshift z, and Ω(L, z) is the
sky coverage available for an object of luminosity L and redshift
z. The sky coverage available for an object is calculated from the
flux-area curve of the survey(s) for which the expectation value
of the number of objects is being derived.
This work concerns the determination of the total popula-
tion of z < 7 AGN detectable with Euclid. We therefore require
an analysis that represents both unobscured and obscured AGN
as they are observed in the Universe. Given these considera-
tions, we adopted the observed 5–10 keV XLF constructed in
Fotopoulou et al. (2016a). The 5–10 keV band avoids absorption
of the X-ray spectrum up to obscuring hydrogen column densi-
ties of NH ∼ 1023 cm−2. The observed 5–10 keV fluxes of AGN
with NH = 1023 cm−2 are >90% of the intrinsic flux for z > 1 and
>80% at lower redshifts. This means the 5–10 keV band selects
an unbiased population of Compton-thin (NH . 1024 cm−2)
AGN (Della Ceca et al. 2008). Even though the 5–10 keV XLF
is constrained by observations up to z ∼ 4, we show in Fig. 1
that our extrapolations are consistent with the latest XLF con-
straints at much higher redshifts (z ∼ 6; e.g. Wolf et al. 2021;
Barlow-Hall et al. 2023). AGN LFs constructed in IR bands,
which also incorporate both unobscured and obscured AGN,
are not well constrained beyond z ∼ 3 (e.g. Gruppioni et al.
2013; Lacy et al. 2015) and hence require greater extrapolation
than the XLF with no high-redshift constraints to compare to.
Whilst UV/optical AGN LFs are constrained by observations up
to z ∼ 7 (e.g. Wang et al. 2019; Matsuoka et al. 2023), this wave-
band probes only the unobscured (quasar) population. Thus,
strong assumptions are required to incorporate obscured AGN.
By using an extrapolated XLF the total AGN space density,
which is consistent with the latest observational constraints, is
preserved and treated self-consistently.
Furthermore, the 5–10 keV band is straightforwardly rec-
onciled with the 2–10 keV band assuming a power-law spec-
trum with a representative photon index. This is advanta-
geous because there are a wide range of established models
and empirical analyses connecting 2–10 keV emission of AGN
with bolometric luminosity (Shen et al. 2020; Duras et al. 2020),
UV/optical emission (Lusso et al. 2010), obscuration properties
(Merloni et al. 2014), and broad-band spectral energy distribu-
tion (SED) shapes (Salvato et al. 2009; Fotopoulou et al. 2016b;
Shen et al. 2020). We leverage these models in our work to pro-
duce robust multiwavelength photometry that represent the best
of our current knowledge in connecting X-ray luminosity and
redshift of AGN with observed multiwavelength emission from
empirical data.
Whilst the 5–10 keV band is highly complete for Compton-
thin AGN, the effect of absorption is more substantial for AGN
with NH = 1024 cm−2, especially at low redshifts. At z > 2 more
than 80% of the intrinsic flux can still be observed. Our choice
of the 5–10 keV XLF is therefore incomplete for Compton-thick
AGN (NH & 1024 cm−2) in the local Universe. The fraction
of missed Compton-thick AGN in hard X-ray observations is
uncertain, with estimates ranging from 20–40% depending on
redshift (Civano et al. 2015; Laloux et al. 2023; Pouliasis et al.
2024) and up to 80% depending on the selection (Fiore et al.
2008). Modelling of the cosmic X-ray background averaged
across redshifts and luminosities suggests Compton-thick AGN
should be at least as abundant as Compton-thin AGN probed by
2–10 keV XLFs (Gilli et al. 2007). The nature of Compton-thick
objects can in principle be confirmed by subsequent observa-
tions at higher X-ray energies (e.g. 14–195 keV) and/or at IR
wavelengths, which correct the usual 2–10 keV X-ray emission
(e.g. Spinoglio et al. 2022). The number of confirmed hard X-
ray identified Compton-thick AGN in the z ≤ 1.5 Universe is
only in the tens of sources (e.g. Ajello et al. 2012; Civano et al.
2015; Marchesi et al. 2018), which if missed by our analysis
would make a negligible difference to our results. Comparing
IR and hard X-ray selected samples of AGN at 0.2 < z < 1.2,
Mendez et al. (2013) derive an upper limit suggesting that ∼10%
of IR-selected AGN are Compton-thick. Compton-thick objects,
at least those not optically obscured, will be detected using
Euclid photometry and spectroscopy. We therefore assess that
the numbers of detectable AGN presented in this work may be a
conservative estimate due to the selection effects biased against
heavily obscured and Compton-thick AGN in the X-ray regime.
It is plausible that Compton-thick AGN will add up to an addi-
tional ∼10% to our Euclid detected AGN estimates at z < 2.
In Fotopoulou et al. (2016a) XLF parameters were estimated
using a sample of 1 115 X-ray selected AGN with 0.01 < z < 4.0
A250, page 3 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
and 41 < log10(LX/erg s
−1) < 46. The sample was compiled
from a mixture of wide area, medium area, and pencil-beam X-
ray fields: The Monitor of All-sky X-ray Image extra-Galactic
survey (MAXI; Ueda et al. 2011), The XMM-Newton Hard
Bright Serendipitous Survey (HBSS; Della Ceca et al. 2004),
XMM-COSMOS (Cappelluti et al. 2009), XMM-Lockman Hole
(Brunner et al. 2008), XMM-Chandra Deep Field South (XMM-
CDFS; Ranalli et al. 2013), Chandra-COSMOS (Elvis et al.
2009), AEGIS-X Deep (AEGIS-XD; Nandra et al. 2015), and
Chandra-Chandra Deep Field South (Xue et al. 2011). Using
Bayesian model selection a luminosity-dependent density evo-
lution (LDDE) model was shown to best describe the data. In
LDDE models the number density of AGN changes over cosmic
time with low-luminosity and high-luminosity AGN evolving on
different timescales. This evolution is implemented through the
luminosity dependence of the critical redshift, zc.
The LDDE model presented in Fotopoulou et al. (2016a)
uses the formalism introduced by Ueda et al. (2003) and is
described by the following
dφ(L, z)
d log10 L
=
dφ(L, z = 0)
d log10 L
(L, z), (3)
where dφ(L, z)/d log10 L represents the differential LF, dφ(L, z =
0)/d log10 L describes the local (z ∼ 0) LF and (L, z) denotes
the LF evolution factor. The local XLF is well described by a
broken power-law distribution
dφ(L, z = 0)
d log10 L
=
A(
L
L0
)γ1
+
(
L
L0
)γ2 , (4)
where A is the LF normalisation, L0 is the luminosity at which
the break occurs and γ1, γ2 are the slopes of the power-law above
and below L0. The evolution factor has the form
(L, z) =
(1 + zc)p1 + (1 + zc)p2(
1+z
1+zc
)−p1
+
(
1+z
1+zc
)−p2 , (5)
where p1, p2 are slopes of the evolution factor broken power law
and zc is the luminosity-dependent critical redshift, expressed by
zc(L) =
{
z∗c for L ≥ Lα
z∗c
(
L
Lα
)α
for L < Lα
, (6)
where z∗c is the high-luminosity critical redshift. The α exponent
and Lα luminosity are parameters calculated in the fit of the XLF.
For the main simulation, we used the mode of the parameter pos-
teriors presented in Fotopoulou et al. (2016a) given in Table 1.
The impact of the parameter uncertainties as well as extrapola-
tion of the XLF is examined in Sect. 5.1.
Throughout this manuscript we primarily consider X-ray
luminosity and fluxes in the 2–10 keV domain in erg s−1 and
erg s−1 cm−2, respectively. When used as input to the XLF we
convert to the 5–10 keV domain by assuming an AGN X-ray
spectrum following a power-law distribution, F(E) ∝ E−Γ where
Γ is the photon index. We take Γ = 1.9, the midpoint of the range
of Γ = 1.8−2.0 found for samples of radio-quiet AGN (e.g.
Nandra & Pounds 1994; Reeves & Turner 2000; Piconcelli et al.
2005; Page et al. 2005; Young et al. 2009).
We compare the XLF invoked in this work with a selection of
AGN LFs in Fig. 1. Over the full redshift range considered in this
work we compared to the well established log10(NH/cm
−2) ≤ 24
hard XLF of Ueda et al. (2014) and the bolometric quasar LF
of Shen et al. (2020). At z > 4 we regard the Harikane et al.
Table 1. Parameter values employed for the XLF of Fotopoulou et al.
(2016a).
Parameter Value
log10(L0/erg s
−1) 43.77
γ1 0.87
γ2 2.40
p1 5.89
p2 −2.30
z∗c 2.12
log10(Lα/erg s
−1) 44.51
α 0.24
log10(A/Mpc
−3) −5.97
Notes. These values denote the mode of the posterior draws sampled in
the Bayesian analysis of the LDDE XLF.
(2023) faint unobscured AGN LF derived using The James Webb
Space Telescope (JWST; Gardner et al. 2006) Near InfraRed
Spectrograph (NIRSpec) deep spectroscopy. In the high-redshift
(z & 6) regime we consider the z > 6 Sloan Digital
Sky Survey (SDSS; York et al. 2000) quasar LF of Jiang et al.
(2016), which was used to derive z > 7 AGN expecta-
tions for Euclid in Euclid Collaboration: Barnet et al. (2019),
the recent Schindler et al. (2023) z ∼ 6 quasar LF derived
from Pan-STARRS1 (PS1; Bañados et al. 2016, 2023) and Sub-
aru High-z Exploration of Low-Luminosity Quasars (SHELLQs;
Matsuoka et al. 2018) survey observations, as well as constraints
placed on the XLF at z > 6 by Wolf et al. (2021) using the
extended ROentgen Survey with an Imaging Telescope Array
(eROSITA; Merloni et al. 2012).
We converted the Shen et al. (2020) bolometric LF to the
2–10 keV domain using the X-ray bolometric correction of
Duras et al. (2020), which is adopted throughout this work due to
its universal applicability to obscured and unobscured AGN over
7 dex in luminosity. The bolometric correction KX = Lbol/LX is
parameterised as
KX(Lbol) = a
[
1 +
(
log10(Lbol/L)
b
)c]
, (7)
where (a, b, c) = (10.96, 11.93, 17.79).
To convert the high-redshift quasar UV LFs of Jiang et al.
(2016) and Schindler et al. (2019) to the 2–10 keV domain
we followed the inverse of the method outlined in Ricci et al.
(2017). Briefly, we calculated the UV monochromatic luminos-
ity at 1450 Å, L1450, from the absolute magnitude at 1450 Å,
M1450, using
L1450 = 4pid210−0.4M1450 f0, (8)
where d = 10 pc = 3.0857 × 1019 cm and f0 = 3.65 ×
10−20 erg s−1 cm−2 Hz−1 is the zero-point. Next, a UV power-law
SED Lν ∝ ναν (e.g. Giallongo et al. 2015) with αν = −0.44 for
1200 Å < λ < 5000 Å (Natali et al. 1998; Vanden Berk et al.
2001) and αν = −1.57 for 228 Å < λ < 1200 Å (Telfer et al.
2002) was adopted to obtain the monochromatic luminosity at
2500 Å, L2500. This was converted into a monochromatic lumi-
nosity at 2 keV, L2 keV, through the equation
log10 L2 keV = (0.952 ± 0.033)log10 L2500 − (2.138 ± 0.975), (9)
derived in Lusso et al. (2010). Both monochromatic luminosities
are in erg s−1 cm−2 Hz−1. Finally, an assumed X-ray power law
A250, page 4 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
10
8
6
4
2
0.01 z < 0.2
Fotopoulou+16
Ueda+14
Shen+20
0.2 z < 0.5 0.5 z < 1.5
10
8
6
4
2
1.5 z < 2.0 2.0 z < 3.0 3.0 z < 4.0
42 44 46
10
8
6
4
2
4.0 z < 5.0
JWST - Harikane+23
42 44 46
5.0 z < 6.0
42 44 46
6.0 z < 7.0
SDSS - Jiang+16
eROSITA - Wolf+21
PS1 - Schindler+23
0.0 0.2 0.4 0.6 0.8 1.0
log10 (LX / ergs 1)
0.0
0.2
0.4
0.6
0.8
1.0
lo
g 1
0
(d
/d
lo
g 1
0
L X
/M
pc
3 )
17 21 25 29 17 21 25 29
M1450
17 21 25 29
Fig. 1. Comparison of different AGN LFs homogenised to the 2–10 keV X-ray band. Corresponding absolute UV magnitudes at 1450 Å, M1450,
are displayed on the upper axes. In each panel the LFs are realised for the central redshift value. The hard X-ray LF of Fotopoulou et al. (2016a)
employed in this work is shown in black. The grey shaded regions depict the 1σ uncertainty. For reference, we plot the Fotopoulou et al. (2016a)
XLF evaluated at z = 0.1 as orange dotted lines in each panel. The hard XLF of Ueda et al. (2014) is shown in green, with the green shaded regions
corresponding to the 1σ uncertainty generated with sampling from the published parameter uncertainties. The magenta lines portray the bolometric
quasar LF of Shen et al. (2020), converted to the X-ray domain. The light blue lines portray the Harikane et al. (2023) z > 4 faint unobscured AGN
LF derived with JWST/NIRSpec deep spectroscopy. In the final panel the Jiang et al. (2016) z > 6 SDSS quasar LF is represented by the blue
curve. The solid orange curve gives the Schindler et al. (2023) z ∼ 6 quasar LF derived from Pan-STARRS1 and SHELLQs observations. The red
uncertainty interval represents eROSITA high-redshift constraints on the XLF (Wolf et al. 2021). In all cases dashed curves and hatched uncertainty
intervals indicate extrapolation.
with photon index Γ = 1.9 was used to acquire the integrated
2–10 keV X-ray luminosity.
Across the redshift range probed here, the Fotopoulou et al.
(2016a) and Ueda et al. (2014) XLFs are consistent within
1σ uncertainties. Both of these XLFs are also consistent
within 1σ with the recent z < 4 XLF determination of
Peca et al. (2023). At z ≥ 4, where constraining observa-
tional data are sparse and we must extrapolate the XLFs, there
is some tension on the normalisation. The Ueda et al. (2014)
and Shen et al. (2020) parametrisations show a steeper decline
in space density at log10(L2−10 keV/erg s−1) > 44 compared
to that of Fotopoulou et al. (2016a). The UV/optically derived
quasar LFs of Jiang et al. (2016) and Schindler et al. (2023)
agree with this decline. The recent X-ray derived result of
Wolf et al. (2021) however, advocates for a higher space den-
sity of log10(L2−10 keV/erg s−1) ∼ 46 AGN at z > 6, demonstrat-
ing consistency with our extrapolation of the Fotopoulou et al.
(2016a) XLF. Through similar means, Barlow-Hall et al. (2023)
also derived constraints on the high-redshift XLF that are
consistent with Wolf et al. (2021). Our extrapolation of the
Fotopoulou et al. (2016a) XLF is therefore in agreement with the
most recent XLF constraints across the entire redshift range con-
sidered in this work.
We verified that the space density of our simulated Euclid-
detectable unobscured AGN are consistent with empirical
UV/optical quasar samples at 1 ≤ z ≤ 6. In the 6 ≤ z ≤ 7
regime there is a space-density excess of up to an order of mag-
nitude in our simulated sample at −26 . M1450 . −23. Recent
results from observations with JWST point towards a steeper
than expected AGN LF at low luminosities (M1450 > −22)
A250, page 5 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
.
XLF
Fotopoulou et al. (2016a)
Euclid Area (EWS / EDS)
EC: Scaramella et al. (2022)
Integrate XLF
43 ≤ log10(Lbol/erg s−1) ≤ 50, δ log10 Lbol = 0.1
0.01 ≤ z ≤ 7, δz = 0.01
AGN volume-limited Sample
Optically Obscured?
Merloni et al. (2014)
Mean Quasar SED
Shen et al. (2020)
AGN SED Class (LX, z)
Fotopoulou et al. (2016b)
XXL E(B − V) Distributions
Fotopoulou et al. (2016b)
IGM Extinction
Madau (1995)
Mock Observations
(Euclid, Rubin/LSST, DECam, WISE,
GALEX, Spitzer, 2MASS, VISTA)
Final AGN Photometric Catalogue
NAGN,EWS = 1.2 × 108
NAGN,EDS = 4.0 × 105
Assess Euclid Detectable AGN (5σ)
NAGN,EWS = 4.0× 107
NAGN,EDS = 2.4× 105
Assess Euclid Selected AGN Sample
EC: Bisigello et al. (2024)
NAGN,EWS = 8.1 × 106
NAGN,EDS = 3.5 × 104
Unobscured Obscured
Fig. 2. Sketch outlining the method adopted in this work to attain obser-
vational expectations for AGN in the Euclid surveys.
in the z & 3 regime (e.g. Harikane et al. 2023; Kocevski et al.
2023; Maiolino et al. 2024), with some results suggesting that
the space density of AGN is up to an order of magnitude greater
than extrapolations of quasar UV LFs (Matthee et al. 2024).
Additionally, there are hints that the UV/optical AGN LF gives
an underestimated space density at z ∼ 4 due to the incomplete-
ness of canonically used colour selections (e.g. Boutsia et al.
2018). We also observe an excess of Euclid-detectable unob-
scured AGN in our simulation compared to empirical UV/optical
quasars across all probed luminosities at z < 1. It is known that
XLFs give a higher space density of AGN at low redshifts com-
pared to UV/optical quasar LFs (compare to e.g. Kulkarni et al.
2019). As explored in Ricci et al. (2017), much of this excess is
due to the obscured AGN incorporated in the XLF. Some of these
AGN, particularly those with high-luminosities, were assigned
as unobscured by the Merloni et al. (2014) probabilistic model
in our SED assignment prescription (Sect. 3.3), driving this com-
parative excess.
3. Method
We now describe the main methodology, going from input XLF
to output AGN photometry. We explain the adopted Euclid sur-
vey parameters and modelling in Sect. 3.1. In Sect. 3.2 we
explain the calculation to generate the volume-limited samples
of AGN. The SED assignment model is presented in Sect. 3.3.
Finally, Sect. 3.4 describes how we derive the final photomet-
ric measurements and assess Euclid detectability of each AGN
in our sample. We present a flowchart outlining the adopted
methodology in Fig. 2.
3.1. Euclid surveys
The Euclid space telescope possesses two photometric instru-
ments: the Visible Imager (VIS, Euclid Collaboration: Cropper
et al. 2025) and the Near Infrared Spectrograph and Photome-
ter (NISP, Euclid Collaboration: Jahnke et al. 2025). The VIS
instrument carries a single broadband optical filter, IE (5300–
9200 Å) which covers the wavelength range of the traditional
riz bands. NISP possesses three near-infrared (NIR) photometric
filters (Euclid Collaboration: Schirmer et al. 2022): YE (9500–
12 120 Å), JE (11 680–15 670 Å), and HE (15 220–20 210 Å).
Over its six year nominal lifetime, Euclid will perform two
core surveys. The EWS is a program observing ∼14 500 deg2
of the extra-Galactic sky with Euclid’s four photometric filters
(Euclid Collaboration: Scaramella et al. 2022). The EDS will
provide observations two magnitudes deeper than the EWS for
several distinct fields totalling 50 deg2. The EDS not only helps
with calibrations of the EWS data but also extends the scientific
scope of Euclid to fainter galaxies and AGN. Throughout this
work we adopt the expected EWS photometric depths presented
in Euclid Collaboration: Scaramella et al. (2022) for a 5σ point
source detection, summarised in Table 2.
The high Galactic latitude of the extragalactic observations
made with Euclid means that there will be minimal Galactic
extinction, with only 7–8% of the Euclid sky exceeding a Galac-
tic E(B − V) of 0.1 (Galametz et al. 2017). Due to this and the
assumption that observed magnitudes will be corrected accord-
ingly, there is no need to account for the effects of Galactic
extinction on observed magnitudes throughout this work.
The EWS and EDS areas1 of 14 500 deg2 (4.42 sr) and
50 deg2 (0.015 sr, Euclid Collaboration: Scaramella et al. 2022)
were adopted as the available area applied to every source
within our 2–10 keV flux limits for our sample generation
1 In X-ray studies a full flux-area curve governed by the detector char-
acteristics would be considered. Here, we assume homogeneous sensi-
tivity across the full Euclid survey coverage for simplicity.
A250, page 6 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Table 2. Euclid photometric filters.
Band λeff (Å) EWS Limiting Mag EDS Limiting Mag
IE 7200 26.2 28.2
YE 10 810 24.3 26.3
JE 13 670 24.5 26.5
HE 17 710 24.4 26.4
Notes. Corresponding effective wavelengths (λeff) are given in
angstrom. Observational depths (5σ point source) for the EWS and EDS
are based on values reported in Euclid Collaboration: Scaramella et al.
(2022).
calculation. We imposed an upper flux limit of F2−10 keV, upper =
10−11 erg s−1 cm−2 to prevent sources with un-physically large
incident fluxes being recovered by the XLF integration for our
volume-limited sample. The upper flux limit is the observed
flux of the brightest source in the ROtogen SATellite (ROSAT;
Truemper 1982) All Sky Catalog (RASS; Boller et al. 2016).
We also applied a lower flux limit of F2−10 keV, lower =
10−19 erg s−1 cm−2. This value was empirically determined
to probe beyond Euclid’s optical magnitude limits with
an appropriate margin. The lower flux limit was deter-
mined as the incident flux when the Shen et al. (2020) mean
quasar SED was scaled to the minimum X-ray luminosity
(log10[L2−10 keV/erg s−1] = 41.8) and maximum redshift (z = 7)
probed in this work. Determining an area minimum flux limit
ensures the completeness of our simulated AGN samples at the
lowest fluxes, close to the Euclid detectability threshold, whilst
saving on computation by not simulating a large amount of
sources that cannot be detected with Euclid.
3.2. AGN sample generation
We integrated the XLF to create a volume-limited sample of
statistically expected X-ray AGN present in the EWS and EDS
footprints. We performed this calculation following Eq. (2) with
the EWS and EDS areas described in Sect. 3.1. In each instance
we integrated the XLF over 0.01 ≤ z ≤ 7 with a constant redshift
interval of δz = 0.01.
We defined the X-ray luminosity integration range to
correspond to a bolometric luminosity interval of 43 ≤
log10(Lbol/erg s
−1) ≤ 50, corresponding to the approximate
range for which observed AGN are present in the bolomet-
ric LF data of Shen et al. (2020). We calculated the analogous
2–10 keV luminosity range by adopting the 2–10 keV AGN
bolometric correction of Duras et al. (2020). This hard X-
ray bolometric correction is a universal bolometric correction
equally applicable to X-ray selected obscured and unobscured
AGN population over the range 41 ≤ log10(Lbol/erg s−1) ≤ 48,
the greatest range in luminosity available for such a relation. We
therefore integrate our XLF in the 2–10 keV luminosity range
41.8 ≤ log10(L2−10 keV/erg s−1) ≤ 46.3, with a constant interval
of δ log10 L2−10 keV = 0.1. Adopting the conversion introduced
in Sect. 2 using Eq. (8) and (9), this L2−10 keV integration range
corresponds to the M1450 range −29.0 ≤ M1450 ≤ −17.2.
The result of our integration is a set of distinct
(δ log10 L2−10 keV, δz) bins with the expected number of AGN,〈N〉, within the corresponding X-ray luminosity and redshift
range. We neglected all integration bins where 〈N〉 < 1 as it does
not make sense to treat parameter ranges where we statistically
expect fewer than one AGN. These cases arise due to the statisti-
cal nature of the operation. For our final volume-limited sample,
we re-sampled our integration bins with 〈N〉 > 1. Accordingly,
bin i corresponding to an X-ray luminosity, redshift interval of
(δ log10 L2−10 keV, i, δzi) and expectation value 〈N〉 = Ni first has
Ni rounded to an integer and is then re-sampled to become Ni
AGN in our data set. Each separate AGN is assigned an X-ray
luminosity and redshift value sampled uniformly from the parent
integration bin parameter range (δ log10 L2−10 keV, i, δzi).
3.3. SED allocation model
AGN have variations in spectral shape which can correlate with
the luminosity and redshift. The shape of the SED is a result of
the interplay between the black hole accretion rate (BHAR) of
the AGN and the star-formation rate (SFR) of the host galaxy.
In cases with high obscuration of the AGN, low BHAR or high
SFR, observed AGN SEDs can have significant contributions
from its host galaxy which results in composite sources.
Aiming to encapsulate the diversity of observed AGN SEDs
and their variations, a number of characteristic SEDs are adopted
in this work. We incorporate probabilistic spectral variations
(e.g. dust extinction, optical to X-ray slope) sampled from empir-
ical distributions to recreate the heterogeneity of observed AGN
fluxes and colours. Our model ensures that AGN in our sam-
ple with similar luminosities and redshifts have realistic multi-
wavelength variation in their SEDs, creating a range of photo-
metric detectability when normalised at the same wavelength.
3.3.1. AGN optical class assignment
As our XLF incorporates unobscured and obscured AGN in an
unbiased fashion up to NH ∼ 1023 cm−2, we introduced a proba-
bilistic model to assign a particular AGN as either unobscured or
obscured, optically. For this purpose, we leveraged the optically
obscured AGN fraction evolution of Merloni et al. (2014).
In Merloni et al. (2014), the optically obscured AGN frac-
tion ( fobsc) as a function of redshift and L2−10 keV was derived
from a sample of 1310 X-ray selected AGN from the XMM-
COSMOS field in the redshift range 0.3 ≤ z ≤ 3.5. Within their
sample AGN were classified as optically unobscured or obscured
based on their optical/NIR properties. AGN were considered
unobscured if there were broad emission lines (FWHM >
2000 km s−1) in their optical spectra and obscured otherwise. If
no optical spectra were available for a given AGN they were
instead classified via the best-fit template class obtained by
SED fitting. The optically obscured fraction evolution is param-
eterised as
fobsc = B(1 + z)δ, (10)
where B and δ are luminosity-dependent parameters. The 2–
10 keV luminosities and redshift values in our work exceed
the ranges considered in Merloni et al. (2014), we therefore
extrapolated their function to higher and lower luminosities
and to higher redshift. This results in low X-ray luminosi-
ties giving obscured AGN fractions greater than unity at z &
3. We therefore truncate the maximum obscured fraction at
100%. This modification implies that objects with z > 3 and
log10(L2−10 keV/erg s−1) ≤ 43.5 are assured to be assigned as
obscured AGN. The resulting implementation of the model
obeys
fobsc = min[B(1 + z)δ, 1]. (11)
The luminosity dependent values of the parameters B and δ are
presented in Table 3. The redshift evolution of the optically
A250, page 7 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Table 3. Luminosity-dependent parameter values used in this work for
the Merloni et al. (2014) optically obscured AGN fraction (Eq. (10)).
log10(L2−10 keV/erg s−1) B δ
≤43.5 0.71 0.26
43.5–44.3 0.46 0.17
≥44.3 0.05 1.27
0 1 2 3 4 5 6 7
z
0.0
0.2
0.4
0.6
0.8
1.0
O
pt
ic
al
ly
 o
bs
cu
re
d 
AG
N
 fr
ac
ti
on
  f
ob
sc
log10 (L2 10keV / ergs 1) = 42.0
log10 (L2 10keV / ergs 1) = 44.0
log10 (L2 10keV / ergs 1) = 46.0
Fig. 3. Evolution of the Merloni et al. (2014) optically obscured AGN
fraction as a function of redshift for different L2−10 keV values. The
data points with error bars show the empirical data reproduced from
Merloni et al. (2014), which was used to derive their model. The dashed
lines depict extrapolation of the model.
obscured fraction of AGN for three L2−10keV values, spanning
the range considered in this work are depicted in Fig. 3.
In practise for each simulated AGN we calculated fobsc cor-
responding to its (L2−10 keV, z) co-ordinate and used this as the
probability that the AGN is optically obscured. We drew a ran-
dom number from a uniform distribution between 0 and 1. If the
number was below the corresponding obscured AGN fraction the
AGN was assigned as obscured and vice versa for unobscured.
This is a binary choice between AGN optical types which in real-
ity is a continuum of different obscuration fractions. The binary
optical AGN classification is for simplicity. Empirically sampled
E(B−V) for intrinsic reddening of our AGN SEDs is introduced
in Sect. 3.3.4.
3.3.2. Unobscured AGN
To model unobscured AGN we adopted the broad-band mean
quasar SED assembled in Shen et al. (2020) that spans from
ultra-hard X-rays to far-infrared (FIR). This SED represents
the average continuum emission of unobscured AGN neglecting
emission line contributions. The SED was utilised in Shen et al.
(2020) for their bolometric AGN LF derivation, calculating bolo-
metric corrections and reconciling unobscured AGN emission
10 1 100 101 102 103 104 105 106
Wavelength / Å
10 1
100
101
F
/A
rb
Richards et al. (2006)
Krawczyk et al. (2013)
Lusso et al. (2015)
Hopkins et al. (2007)
Fig. 4. Broad-band mean quasar SED introduced in Shen et al. (2020)
which we assign to unobscured AGN in this work, normalised at 1 µm.
This SED is a combination of multiple templates from the literature: IR
template of Richards et al. (2006b, red), optical/UV SED template of
Krawczyk et al. (2013, blue), EUV power-law model based on the spec-
tral index of Lusso et al. (2015, orange), which is directly connected
(magenta) to the X-ray template of Hopkins et al. (2007, green). The
template in this figure is a rest-frame realization of the SED with no
IGM absorption, generated with the mean αox of Lusso et al. (2010),
〈αox〉 = −1.37. The blue shaded region indicates the range in possible
realisations of the optical portion of the SED considering αox values
with the reported dispersion of 0.18.
between different wavebands. The full SED model, across all
wavelengths is shown in Fig. 4.
In the optical/UV, the SED template of Krawczyk et al.
(2013) was adopted. This mean template was derived from
96 716 luminous broad-lined quasars that do not show signs of
dust reddening, covering the wavelength range 912 Å–30 µm.
The incorporated quasars are at 0.064 < z < 5.46. In the IR
regime, the Krawczyk et al. (2013) SED template is extended
to 100 µm using the Richards et al. (2006b) SED. This IR SED
includes dust emission, removing the need for an additional
dust emission model. On the short wavelength side, the opti-
cal/UV SED is extended into the extreme UV (EUV; λ < 912 Å)
using a power-law model, fν ∝ ναν , with spectral index αν =
−1.70 (Lusso et al. 2015). This model extends from 600 Å to
912 Å, where the flux is directly connected to an X-ray SED
template. The X-ray SED adopted in this work is that used
in Hopkins et al. (2007). Extending shortwards of 0.5 keV, this
template follows a power law with intrinsic photon index Γ = 1.8
and an exponential cut-off at 500 keV. Following Ueda et al.
(2003), a reflection component is included with a reflection solid
angle of 2pi, inclination cos i = 0.5 and Solar abundances.
To scale the X-ray and optical portions of the SED, an
optical-to-X-ray luminosity relation must be utilised. The lit-
erature reports a non-linear correlation between the X-ray and
optical luminosities for unobscured AGN (Steffen et al. 2006;
Just et al. 2007; Lusso et al. 2010). In particular this relates the
luminosities at 2 keV and 2500 Å with the expression
log10 Lν(2 keV) = β log10 Lν(2500 Å) + C, (12)
where Lν is the luminosity density in units of erg s−1 Hz−1. This
relation can be parameterised by the optical-to-X-ray spectral
index, αox, defined as
αox = 0.384 log10
(
Lν(2 keV)
Lν(2500 Å)
)
. (13)
A250, page 8 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Lusso et al. (2010) reported a mean αox value 〈αox〉 ≈
−1.37 ± 0.01 with a dispersion of 0.18, for a sample of 545
X-ray selected unobscured AGN from the XMM-COSMOS
survey observed at 0.04 < z < 4.25 with 40.6 ≤
log10(L2−10 keV/erg s−1) ≤ 45.3. The sample used in Lusso et al.
(2010) has similar properties to the AGN expected to be probed
in this work. Therefore, for each simulated unobscured AGN we
drew a value of αox from a normal distribution centered on 1.37
with a standard deviation of 0.18. A single αox value sampled
with empirical scatter can be adopted in this case because no
significant correlation is observed between αox and redshift or
L2−10 keV for X-ray selected samples (Lusso et al. 2010). Draw-
ing αox values from an empirically derived distribution perturbs
the normalisation of the optical-IR portion of the SED with
respect to the X-ray luminosity, allowing us to introduce nat-
ural variations in AGN flux within this single SED model for
unobscured AGN.
3.3.3. Obscured AGN
X-ray selected obscured AGN display a considerable range of
spectral shapes. Hard X-rays allow AGN to be selected even with
a substantial portion of their optical AGN emission obscured by
dust. Spectral shapes therefore range from Seyfert 2-like SEDs
with high AGN contribution to that of quiescent and star-forming
galaxies (SFG) when host-galaxy contamination is high.
To assign appropriate obscured AGN SEDs depending on
both X-ray luminosity and redshift we leveraged the SED fit-
ting results of Fotopoulou et al. (2016b). SED fitting was per-
formed with UV–mid-infrared (MIR) photometry of the 1000
brightest X-ray sources in the XXL survey (Pierre et al. 2016),
which includes 972 unobscured and obscured AGN in the red-
shift range 0.01 < z < 4. This XXL survey sample covers
an area of 50 deg2 with a medium X-ray flux limit of 4.8 ×
10−14 erg s−1 cm−2 in the 2–10 keV band, providing a middle
ground between deep and all-sky surveys. Properties reported
for AGN in this field should thus provide a realistic represen-
tation of those expected to be encountered with Euclid. The
extensive multiwavelength followup programme in the XXL
fields has provided a means to connect X-ray AGN emission
with empirical broad-band SEDs in a self-consistent manner.
The SED templates used in the fitting are drawn from those
used in the Ilbert et al. (2009) SED fitting analysis of Cosmic
Evolution Survey (COSMOS; Scoville et al. 2007) sources and
Salvato et al. (2009) SED fitting of XMM-COSMOS sources,
therefore ensuring compatibility in colour distributions and AGN
populations with the AGN used to derive the Merloni et al.
(2014) optically obscured AGN fraction (Sect. 3.3.1), which
incorporates data from the XMM-COSMOS survey. The tem-
plates in Fotopoulou et al. (2016b) are categorised into QSO,
AGN, Starburst (SB), SFG, and Passive categories. The sources
were assigned in each category prior to the SED fitting, using a
Random Forest classifier as described in Fotopoulou & Paltani
(2018). The SED templates used in this work have therefore
been demonstrated to well approximate the empirical colour-
distributions of X-ray selected AGN. In all cases, rather than the
assignment of a particular SED meaning that the AGN has solely
the properties of its assigned class, it means the observed pho-
tometry of the AGN is best described by the allocated template.
For the purposes of assigning obscured AGN spectra we
dropped all sources which have a best-fit template in the QSO
category, leaving a sample of 652 AGN. Within the SED cat-
egory denominated ‘AGN’ in Fotopoulou et al. (2016b) there
are three characteristic SED shapes; Seyfert 2-like, QSO2-like,
103 104 105 106 107
Wavelength / Å
lo
g 1
0(
F
/A
rb
)
H[O ]
PASS
SFG
SB
QSO2
SEY2
SB-AGN
Fig. 5. Representative SEDs utilised for each obscured AGN SED class
(‘PASS’: pink, ‘SFG’: brown, ‘SB’: purple, ‘QSO2’: orange, ‘SEY2’:
green, ‘SB-AGN’: red) in this work. Based on the SED shapes in
Fotopoulou et al. (2016b). Grey dotted lines highlight the rest-frame
wavelength of the [O iii]λ5007 Å and Hα emission lines. Templates are
assigned based on the relative probability of each template class in high-
and low-luminosity groups at a given redshift.
and SB-AGN composite-like. We therefore split the AGN cate-
gory into three further classes based on the characteristic shapes,
resulting in six SED classes considered for obscured AGN in
this work: (1) PASS, (2) SFG, (3) SB, (4) QSO2, (5) SEY2, and
(6) SB-AGN. These six obscured AGN classes provide a range
of different AGN-galaxy contributions and stellar populations as
observed for X-ray AGN. We selected a singular SED from the
set of SEDs belonging to each defined class which was used to
represent its class throughout this work. The selected represen-
tative SED for each class is presented in Fig. 5.
All the obscured AGN SEDs adopted here are empirical and
therefore include nebular emission lines. We note the presence of
particularly prominent emission lines in the SED templates for
our AGN-dominated classes (QSO2, SEY2, and SB-AGN). All
three of these representative SEDs include the Hα emission line.
The QSO2 template additionally features the Hβ line. The SEY2
template includes [O iii]λ5007 Åand the template for SB-AGN
incorporates a range of prominent emission lines from Lyα at
1216 Å to [S iii]λ9533 Å.
To include an X-ray luminosity dependence in our obscured
AGN SED assignment we split the XXL obscured sources
in each of our six SED classes into high and low X-
ray luminosity groups, creating 12 SED groups in total.
We defined high X-ray luminosity sources as those with
log10(L2−10 keV/erg s−1) ≥ 44 and low X-ray luminosity sources
as those with log10(L2−10 keV/erg s−1) < 44. We fit log-normal
probability distributions to the redshift-space distribution of
our 12 SED groups. The resulting normalised probability dis-
tributions for each class were then extrapolated to z = 7.
The normalised and extrapolated redshift-space probability dis-
tributions are presented in Appendix A. For a given source
with a (log10 L2−10 keV, z) co-ordinate we used our log-normal
A250, page 9 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
distribution fits in the appropriate X-ray luminosity group to find
the probability of the source occurring in each SED class at its
given redshift. These six probabilities were scaled to provide a
summed total probability of unity. We used these scaled proba-
bilities to draw an SED class for each obscured AGN.
The adopted methodology provides a template appropriate
for the X-ray luminosity and redshift of the obscured AGN.
SED templates are more probable if they were assigned more
frequently in the XXL SED fitting analysis at a given redshift,
therefore reflecting the available empirical data. In both the high-
and low-luminosity groups the AGN-dominated SED classes
(QSO2, SEY2, and SB-AGN) are most probable at z > 1.5. The
SB SED class has a similar probability as the AGN-dominated
classes in the high-luminosity group in this redshift regime. For
both luminosity groups at z < 1.5 there are no dominantly prob-
able SED classes but a balanced chance of any of the classes
being assigned to an AGN.
None of the SED templates assigned to obscured AGN in
this work extend to X-ray wavelengths. We therefore adopted
bolometric corrections to appropriately scale the SEDs con-
sidering their respective L2−10 keV values. We first transformed
from 2–10 keV luminosity to the bolometric domain with the
bolometric correction of Duras et al. (2020). Following this,
we leveraged the zero-intercept 2 µm bolometric correction of
Runnoe et al. (2012) to transform from a bolometric luminosity
to the λLλ(2 µm) luminosity, which we used to scale our SEDs.
We opted to scale our SEDs at 2 µm in the rest-frame as the cor-
responding portion of the SED is free from major effects of dust
reddening, non-continuum emission features and polycyclic aro-
matic hydrocarbon (PAH) emission. Furthermore, because our
obscured AGN SEDs are derived from X-ray observations, three
of our classes (PASS, SFG, SB) do not contain significant AGN
contributions to their optical–MIR AGN emission (Fig. 5). Due
to this, scaling these SEDs at wavelengths longer than ∼2 µm
results in erroneous photometry.
The Runnoe et al. (2012) 2 µm bolometric correction was
originally derived from a sample of UV-bright unobscured AGN,
which is not the context in which we are applying it here. We
argue however that at rest-frame NIR wavelengths the average
SEDs of obscured and unobscured AGN are remarkably similar
(e.g. Alonso-Herrero et al. 2006; Polletta et al. 2007, see Fig. 3
in Hickox et al. 2017). In this regime there is a minimum in AGN
emission where the accretion disc power-law continuum emission
in unobscured AGN falls off and the black body emission of the
torus begins, which is ubiquitous to all AGN. NIR bolometric cor-
rections for unobscured or obscured AGN are seldom investigated
in the literature, largely due to complications in deconvolving the
AGN and host-galaxy contribution to spectra at these wavelengths
(Elvis et al. 1994; Richards et al. 2006b; Runnoe et al. 2012). Due
to the compatibility of obscured and unobscured AGN spectra
at 2 µm and the sparsity of alternative options we elected to use
the Runnoe et al. (2012) 2 µm bolometric correction in this work
whilst accepting that it is a limitation of our method.
We apply both the X-ray and 2 µm bolometric corrections
for the bulk of this work employing the nominal parameter val-
ues reported in their respective works (i.e. neglecting parameter
uncertainty). We adopted this procedure so that we can fully con-
strain the impact of the bolometric correction dispersion on our
resulting AGN number estimates (see Sect. 5.1).
3.3.4. Intrinsic and IGM extinction
Extinction is the combination of the absorption and scattering of
photons by intervening dust and gas along the line of sight. We
treat two different sources of extinction in AGN SEDs; intrinsic
extinction and intergalactic medium (IGM) extinction. Intrinsic
extinction is caused by dust and gas originating at the redshift
of the source, often in the host-galaxy. This form of extinction,
often referred to as reddening, is most severe in the UV/optical
but acts on wavelengths up to the IR regime, causing variations
in observed SEDs and therefore observed colours. IGM extinc-
tion considers the absorption and scattering of photons by inter-
vening gas and dust in the line of sight to astrophysical sources at
cosmological distances. Specifically, photons with wavelengths
shorter than the rest-frame Lyα transition (1216 Å) are attenu-
ated by intergalactic H i.
To apply intrinsic reddening with realistic E(B − V) val-
ues for AGN SEDs, we again leveraged the results of the XXL
SED fitting analysis of Fotopoulou et al. (2016b). Their analy-
sis was conducted on 972 X-ray selected AGN in the XXL sur-
vey. Intrinsic X-ray properties of the AGN were derived directly
from their X-ray spectra, while optical and longer wavelength
properties related to the AGN host-galaxies were ascertained via
broad-band SED fitting of the multiwavelength photometry of
each source. Relevant to this section, the SED fitting allowed the
best-fit SED template for each AGN to be determined along with
estimations of the intrinsic E(B − V).
We divided the SED fitting results into two classes; (1) XXL-
QSO and (2) XXL-AGN, defined as all sources with a best-fitting
SED template belonging to the QSO class or not, respectively.
We constructed E(B − V) probability distributions for the XXL-
QSO and XXL-AGN classes based on source frequency at each
discrete E(B − V) value used in the analysis of Fotopoulou et al.
(2016b). The discrete colour excess values lie in the range 0.0 ≤
E(B − V) ≤ 0.5 in steps of 0.05. The results of 1000 draws of
each XXL AGN E(B − V) distribution created for this work is
depicted in Fig. 6. For reference, we also plot 1000 draws of
the distribution of all XXL sources analysed in Fotopoulou et al.
(2016b) and 1000 draws of the E(B − V) distribution derived
from SDSS quasars (Hopkins et al. 2004).
Our distributions show the XXL-QSO class have generally
lower E(B − V) values than the XXL-AGN class. XXL-AGN
sources account for the higher colour excess values observed in
the all-sources distribution. This is expected as QSOs are by def-
inition less obscured than AGN due to the requirement of a clear
viewing angle towards the central engine to observe a QSO-like
spectrum. Comparing the SDSS quasar E(B − V) distribution to
that of the XXL-QSO class, it is clear that X-ray selected AGN
probe higher E(B − V) values than their optically selected coun-
terparts. This is mainly due to optical selection effects. It is there-
fore beneficial to adopt an XXL-based E(B − V) distribution for
our analysis as it provides a less biased distribution of E(B − V)
values.
We found no significant trend between E(B − V) values
and L2−10 keV, redshift or SED fitting template class in the XXL
data. Therefore, in our analysis we used our XXL-QSO distri-
bution to sample E(B − V) values for dust extinction in unob-
scured AGN SEDs and the XXL-AGN distribution for obscured
AGN SEDs. Intrinsic extinction was applied to our SED mod-
els assuming a Small Magellanic Cloud (SMC) extinction law
(Prevot et al. 1984). The SMC extinction law has been demon-
strated to be appropriate for use with AGN (e.g. Hopkins et al.
2004; Salvato et al. 2009) for all but the most extremely red-
dened objects (Zafar et al. 2015).
We applied IGM extinction to all our SEDs following
the redshift-dependent prescription presented in Madau (1995).
Although updated versions IGM extinction models are avail-
able (e.g. Meiksin 2006; Inoue et al. 2014; Thomas et al. 2017)
A250, page 10 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
0.0 0.1 0.2 0.3 0.4 0.5
E(B V)
0
100
200
300
400
500
600
700
800
N
dr
aw
s
All XXL
XXL-QSO
XXL-AGN
SDSS Quasars
Fig. 6. Comparison of 1000 draws from the unobscured and obscured
AGN E(B − V) distributions used in this work. The distributions
are based on the XXL AGN E(B − V) distribution derived in
Fotopoulou et al. (2016b, grey shaded). The XXL-QSO distribution
(blue) is the distribution found for XXL AGN assigned the QSO class
template and is used for unobscured AGN in this work. The XXL-AGN
distribution (red) is constructed for all XXL sources not assigned the
QSO class and is used for obscured AGN in this work. For compari-
son, we also include the E(B − V) distribution found for SDSS quasars
(green, Hopkins et al. 2004).
we assert that the adopted method is sufficient for the analy-
ses described in this study. Only the Euclid IE band is affected
by IGM extinction at the highest redshifts probed in this work
(z & 6.5), with the largest effect being for ancillary bands con-
sidered from Rubin/LSST. We are not assessing photometric
dropouts and therefore it is beyond the scope of this work to
include a more detailed IGM extinction model.
3.4. Assessing AGN detectability
We assigned every source in our EWS and EDS AGN samples an
appropriate broad-band SED according to the models described
in Sect. 3.3. We scaled and transformed each SED to retrieve
an incident flux SED in line with the appointed luminosity and
redshift co-ordinate of the AGN. The detectability of each AGN
was assessed by performing a synthetic photometric observation
of its assigned SED using Euclid’s filter transmission curves. To
perform this mock observation we used the sedpy Python pack-
age (Johnson 2019). We additionally observed each SED with a
selection of ancillary bands from other facilities, which we out-
line in Appendix B.
When considering Euclid photometric bands, we per-
turbed the resulting fluxes based on the associated uncer-
tainty for each Euclid filter following the prescription of
Euclid Collaboration: Bisigello et al. (2024). In this formulation
fluxes are perturbed considering a normal distribution where the
standard deviation is equal to the expected photometric uncer-
tainty for each AGN. The photometric uncertainty in each Euclid
filter is described by the sum in quadrature of the photon noise
(σnoise) and the background flux density error (σbkg). The photon
noise is defined as
σnoise =
f5σ
5
r
rref
, (14)
where f5σ is the 5σ survey depth in each filter (Table 2), r is the
projected radius of each source, and rref is the median effective
radius of a source with flux densities equal to f5σ. As we do not
assign our AGN a physical size or physical properties (e.g. stel-
lar mass) that can be used to derive a physical size we assumed
that r = rref = 0′′.23. This corresponds to the radius of the pho-
tometric aperture used to measure the flux of a point source,
1.25 times the FWHM of the Euclid point spread function (PSF;
Euclid Collaboration: Scaramella et al. 2022). This assumption
neglects the dependence of σnoise on source radius. Our mod-
elling therefore results in an underestimation of the photometric
uncertainty for low-redshift extended objects and an overestima-
tion for high-redshift truly point-like objects. The background
flux error component is derived as
σbkg = σnoise
√
f
fsky pi r2
, (15)
where f is the flux of the object and fsky represents the ref-
erence sky surface background corresponding to 22.33, 22.10,
22.11, and 22.28 mag arcsec−2 for IE, YE, JE, and HE, respec-
tively. Resultant observed magnitudes generated with perturbed
fluxes were used to determine if each AGN will be detectable
with Euclid. The photometry prescription used here assumes that
all flux is captured by photometric measurements. Our derived
AGN colours are validated in Appendix C.
We stress that an AGN deemed to be detectable with Euclid
photometry in this work does not necessarily mean it will be
identified as an AGN. Rather, it is expected to be detected above
5σ in the photometric filters of the corresponding Euclid survey.
We consider the selection of Euclid detectable AGN via photom-
etry in Sect. 5.2.
4. Results
Our main results for the number and surface densities of AGN
with ≥5σ Euclid photometric detections in the EWS are pre-
sented in Table 4 and for the EDS in Table 5. We report our
forecast for each of Euclid’s four photometric filters individu-
ally, for AGN detectable in at least one filter, and simultaneously
in all filters. We present more details in Appendix D, giving full
breakdowns of the fields presented here for the total AGN sam-
ple as well as splits into unobscured and obscured AGN.
Of the AGN detectable in at least one Euclid filter in the
EWS, 3.9× 107 (98%) lie in the redshift range 0.01 ≤ z ≤
4, where our XLF is well constrained by observations, and
1.4× 106 (2%) at 4 < z ≤ 7. In the EDS, 2.3× 105 detectable
AGN (96%) are in the redshift range 0.01 ≤ z ≤ 4, while
8.7× 103 (4%) are in the range 4 < z ≤ 7. In the case of
both Euclid surveys the vast majority of the detectable AGN are
derived from the well constrained region of the input XLF. Over
the full redshift range 0.01 ≤ z ≤ 7, we find Euclid-detectable
AGN in the EWS comprise 31% unobscured sources and 69%
obscured. In the EDS we find 21% of Euclid-detectable AGN
will be unobscured with 79% obscured.
We present the distribution of our predicted AGN numbers
in redshift bins for the EWS and EDS in Figs. 7a and 7b,
respectively. Both distributions show a space-density peak of
detectable AGN at z ≈ 1. This is consistent with the redshift
A250, page 11 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Table 4. Expected number of AGN detectable with Euclid photometry in the EWS.
Band Detectable AGN Surface Density (deg−2) Unobscured AGN Obscured AGN 0.01 ≤ z ≤ 4 4 < z ≤ 7
IE 3.1× 107 2.2× 103 1.2 × 107 1.9 × 107 3.0 × 107 1.1 × 106
YE 2.2× 107 1.5× 103 8.7 × 106 1.3 × 107 2.1 × 107 8.4 × 105
JE 3.0× 107 2.1× 103 9.7 × 106 2.0 × 107 2.9 × 107 1.0 × 106
HE 3.5× 107 2.4× 103 9.9 × 106 2.5 × 107 3.4 × 107 1.1 × 106
(IE |YE | JE |HE) 4.0× 107 2.8× 103 1.2 × 107 2.8 × 107 3.9 × 107 1.4 × 106
(IE ∧YE ∧ JE ∧HE) 2.1× 107 1.4× 103 8.4 × 106 1.2 × 107 2.0 × 107 7.2 × 105
Notes. Numbers are reported on a per-filter basis as well as the number of AGN detectable in at least one Euclid filter, (IE |YE | JE |HE), and
the number of AGN detectable in all Euclid filters, (IE ∧YE ∧ JE ∧HE). Corresponding surface densities of Euclid detectable AGN are presented
as well as splits by obscured/unobscured classification and redshift range. Detailed breakdowns of this information by optical classification is
provided in Appendix D.
Table 5. Expected number of AGN detectable with Euclid photometry in the EDS.
Band Detectable AGN Surface Density (deg−2) Unobscured AGN Obscured AGN 0.01 ≤ z ≤ 4 4 < z ≤ 7
IE 1.9× 105 3.8× 103 4.9 × 104 1.4 × 105 1.8 × 105 6.8 × 103
YE 1.6× 105 3.2× 103 4.4 × 104 1.2 × 105 1.5 × 105 5.3 × 103
JE 2.0× 105 4.0× 103 4.5 × 104 1.5 × 105 1.9 × 105 6.3 × 103
HE 2.3× 105 4.5× 103 4.5 × 104 1.8 × 105 2.1 × 105 7.6 × 103
(IE |YE | JE |HE) 2.4× 105 4.7× 103 4.9 × 104 1.9 × 105 2.3 × 105 8.7 × 103
(IE ∧YE ∧ JE ∧HE) 1.6× 105 3.1× 103 4.3 × 104 1.1 × 105 1.5 × 105 4.8 × 103
Notes. Numbers are reported on a per-filter basis as well as the number of AGN detectable in at least one Euclid filter, (IE |YE | JE |HE), and
the number of AGN detectable in all Euclid filters, (IE ∧YE ∧ JE ∧HE). Corresponding surface densities of Euclid detectable AGN are presented
as well as splits by obscured/unobscured classification and redshift range. Detailed breakdowns of this information by optical classification is
provided in Appendix D.
at which the number density of moderate luminosity X-ray AGN
peaks, as described by our XLF (Fotopoulou et al. 2016a). From
Fig. 7a we see that for z . 5.5 filters IE (blue), JE (green) and
HE (black) yield similar numbers of AGN in the EWS. For red-
shifts greater than this the yield of AGN detectable in the short-
est wavelength IE filter drops dramatically. This is caused by the
portion of the AGN SED affected by IGM extinction shifting
into the IE pass-band. From z & 6.4 AGN detections in the IE
band are limited to only the most luminous sources. This effect
is also observed in the EDS redshift distributions (Fig. 7b), how-
ever with a marginally shallower fall-off at z ≈ 6 due to the
deeper limiting magnitude of EDS allowing more AGN in this
regime to be detected with the IE filter.
The predicted density distributions of Euclid detectable
AGN observed magnitudes as a function of redshift are depicted
in Figs. 8 and 9 for EWS and EDS, respectively. Comparing
between the two surveys, we see that in each filter the deeper
limiting magnitude of the EDS allows a high density of AGN
close to the detection limit to be observed at z = 2. Access to
this parameter space in the EDS will therefore allow a greater
surface density of AGN to be detected. The current deepest avail-
able NIR survey, ultra-VISTA (McCracken et al. 2012), reaches
a similar depth to the EDS over 1.5 deg2. The EDS will match the
depth of ultra-VISTA and surpass the area coverage by a factor
of 30.
For both the EWS and EDS the YE band yields the small-
est number of detectable AGN and HE is capable of detecting
the highest. There are a few effects driving this result. The YE
filter (along with JE) is the narrowest Euclid band. This means
less photons will be captured by the bandpass. The HE followed
by IE bands are the widest Euclid bandpasses, generating the
reciprocal of this effect. Alongside this consideration, YE has the
shallowest limiting magnitude of the Euclid bands which will
inhibit the detection of sources close to the detection limit com-
pared to other, marginally deeper filters. The respective wave-
length coverage plays an important role in the AGN yield of each
band. Each un-reddened SED assigned in this work normalised
at 1 µm is presented in Fig. 10, with the effective wavelength
position of each Euclid filter plotted for the redshift range cov-
ered by our samples. It is clear that the longer wavelength band-
passes (JE, HE) will capture more flux from each SED compared
to the shorter wavelength bands due to the shape of the SEDs
where the flux diminishes at shorter wavelengths. This effect
is apparent particularly at higher redshifts (z & 4). With the
addition of dust extinction to our SEDs the wavelength cover-
age driven performance difference is further accentuated for the
longer wavelength Euclid bands. We find specifically that more
obscured AGN assigned the ‘QSO2’ and ‘SEY2’ SED classes
are detected as the bandpass wavelength increases. It appears
that the increased ability to detect obscured AGN to higher red-
shifts is the main driver for the differences in the number of AGN
detected with each Euclid filter.
The order of relative performance of each Euclid band dif-
fers slightly between the EWS and EDS samples. For the EWS
we see that in descending order the most AGN are detectable
with HE > IE > JE > YE. In contrast, the results from the
EDS go as HE > JE > IE > YE. The main difference between
the two samples is that the EWS probes the bright-end slope of
the XLF, detecting the brightest sources from across the extra-
Galactic sky, whereas the EDS with its deeper limiting magni-
tudes and smaller area yields detectable AGN which lie on the
faint-end slope of the XLF. Due to our optical type assignment
model (see Sect. 3.3.1) less luminous AGN are more likely to
be assigned as obscured, with this probability increasing with
A250, page 12 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
0 1 2 3 4 5 6 7
z
103
104
105
106
dN
/d
z
IE
YE
JE
HE
(a)
0 1 2 3 4 5 6 7
z
101
102
103
104
dN
/d
z
IE
YE
JE
HE
(b)
Fig. 7. Predicted redshift distributions of Euclid detectable AGN in the
redshift range 0.01 ≤ z ≤ 7 in the EWS (top) and EDS (bottom). The
data are binned in steps of δz = 0.25. Separate distributions are pre-
sented for AGN detectable in each of Euclid’s photometric filters; IE
(blue), YE (red), JE (green), and HE (black).
redshift. We therefore find that more AGN are assigned as
obscured in our EDS sample which probes the fainter AGN pop-
ulation. Our obscured AGN SED assignment model based on
XXL SED fitting data (Sect. 3.3.3) allocates AGN2-like SEDs
(classes ‘QSO2’, ‘SEY2’, ‘SB-AGN’) with the highest proba-
bility to the low-luminosity group at higher redshifts. Pairing
these factors together results in the EDS sample having a com-
parably higher number of AGN assigned AGN2-like SEDs com-
pared to the EWS which contains more bright unobscured AGN.
Referring to Fig. 10 and as discussed above, the longer wave-
length bandpasses (JE, HE) perform better at detecting AGN
with AGN2-like SEDs and therefore in the EDS the JE fil-
ter detects more AGN than IE due to the increased fraction of
obscured AGN in the sample.
5. Discussion
5.1. Uncertainties
Throughout this work we make a number of key assumptions
regarding models and empirical values adopted. Each of these
choices introduces an element of uncertainty into our overall
estimates of AGN numbers. In this section we work through
our methodology and address four key assumptions we have
made. We assess the impact each of these has on our predictions
and attempt to quantify and understand the major drivers of the
uncertainties within our framework. Many of the assumptions
and choices we have made are due to inherent ambiguity in our
current knowledge and understanding of astrophysical relations,
particularly at higher redshifts. This work will therefore serve
to point towards relations and models that can be extended and
improved upon with data from future facilities, some of which
may be addressed with Euclid itself.
5.1.1. Extrapolation of the XLF
Through necessity we extrapolated the XLF used in this work
in redshift space. The data used to constrain the XLF lies in the
range 0.01 < z < 4.0. The lack of constraining observations
at the low-luminosity end of the XLF, particularly at high red-
shift, leads to substantial uncertainty in the XLF in this regime.
Our results for the EDS probe this area of parameter space and
therefore will be most heavily affected by the uncertainty in this
extrapolation.
To understand the impact of extrapolating the XLF to higher
redshifts we made use of parameter posterior distributions gen-
erated during the construction of the XLF in Fotopoulou et al.
(2016a). For the nine XLF parameters (given in Table 1) we
obtained the final three hundred samples before final conver-
gence, which we found effectively samples the 68% credible
interval (analogous to a ±1σ interval in frequentist statistics) of
each parameter. From these parameter sets we generated three
hundred corresponding realisations of the XLF. We followed
through our method (Sect. 3) for both the EWS and EDS using
each XLF realisation to assess the uncertainty on our numbers
of Euclid detectable AGN. We carried out our EWS analysis in
this section with reduced resolution to save on computation2.
Figure 11 presents the fractional uncertainty on the num-
ber of AGN detectable in at least one Euclid band for both
the EWS and EDS, binned in redshift (∆z = 0.25) and bolo-
metric luminosity (∆ log10[Lbol/erg s
−1] = 0.25). For refer-
ence, we plot curves corresponding to the 50% completeness
flux limits of the 7 Ms exposure of the Chandra Deep Field-
South (CDFS, F2−7 keV = 2.5 × 10−16 erg s−1 cm−2; Luo et al.
2017), representing the deepest currently available X-ray survey
and the medium-depth X-ray survey XMM-COSMOS (black,
F2−10 keV = 5.6 × 10−15 erg s−1 cm−2; Cappelluti et al. 2009).
We additionally show the sample limiting flux of the XXL
1000 brightest AGN sample (XXL-1000-AGN) as used in
Fotopoulou et al. (2016b) for SED fitting (blue; F2−10 keV =
4.8 × 10−14 erg s−1 cm−2; Fotopoulou et al. 2016b), which we
leveraged for our SED assignment procudure (Sect. 3.3). We
homogenised each flux limit to the 2–10 keV band, where appli-
cable, assuming an X-ray power law with photon index Γ = 1.9
2 Specifically, for XLF integration bins with 〈N〉 ≥ 100, we re-sampled
the data to treat one AGN in our data set to represent every 100 AGN.
Any remainder AGN were treated as individual AGN with 〈N〉 = 1, the
same as for integration bins where 〈N〉 < 100. We found that compared
to our full resolution treatment of the EWS there is a difference of only
1% in the resulting number of Euclid-detectable AGN.
A250, page 13 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
0 1 2 3 4 5 6 7
z
14
16
18
20
22
24
26
I E
101
10210
3
103
104
105
IE limiting magnitude
100
101
102
103
104
105
N
AG
N
/(
0.
2
z
0.
3
m
ag
)
0 1 2 3 4 5 6 7
z
14
16
18
20
22
24
26
Y E
101
102
103
104
105
YE limiting magnitude
100
101
102
103
104
105
N
AG
N
/(
0.
2
z
0.
3
m
ag
)
0 1 2 3 4 5 6 7
z
14
16
18
20
22
24
26
J E
101
102103
103
104
105
JE limiting magnitude
100
101
102
103
104
105
N
AG
N
/(
0.
2
z
0.
3
m
ag
)
0 1 2 3 4 5 6 7
z
14
16
18
20
22
24
26
H
E
101
102103
103
104
105
HE limiting magnitude
100
101
102
103
104
105
N
AG
N
/(
0.
2
z
0.
3
m
ag
)
Fig. 8. Observed magnitude vs redshift density distributions for observable AGN in Euclid’s four photometric filters in the EWS. Two-dimensional
histograms representing the density of observed AGN are plotted in grey-scale. Subplots correspond to IE (top left), YE (top right), JE (bot-
tom left), and HE (bottom right). The data are binned with 40 bins in the x and y domains, giving two-dimensional AGN density in units of
NAGN/(0.2 z, 0.3 mag). Contours of constant log10 NAGN are plotted for log10 NAGN = (1, 2, 3, 4, 5). Lines depicting the limiting magnitude in the
EWS for each filter are plotted in dashed magenta lines.
and converted to the bolometric domain assuming the X-ray
bolometric correction of Duras et al. (2020). In general, for both
the EWS and EDS the uncertainties are lowest in the regions
where the XLF is well constrained by data at z < 3 and
log10(Lbol/erg s
−1) ∼ 44−46. This also corresponds to the region
where we expect the majority of AGN selected from the Euclid
data via colour cuts lie (Sect. 5.2). There is a clear increase in
the fractional uncertainty with increasing redshift at all luminosi-
ties, which is expected as a manifestation of the decreasing avail-
ability of constraining observations as redshift increases. At low
bolometric luminosities (log10[Lbol/erg s
−1] < 44.5), the uncer-
tainty increases at a greater rate with increasing redshift than at
higher luminosities owing to the poorer sampling of the XLF
by data at low luminosities. We find that for both the EWS and
EDS the fractional uncertainty increases with increasing lumi-
nosity. This is because higher luminosity AGN can be probed by
Euclid to larger redshifts, which in turn have a higher associated
uncertainty. We therefore find that the uncertainties on the num-
ber of Euclid detectable AGN from the XLF are dominated by
correlations with redshift.
In our EDS analysis, we find the median number of AGN
observable in at least one Euclid band over the entire red-
shift range 0.01 ≤ z ≤ 7 is (2.4± 0.3)× 105. The associated
uncertainty corresponds to 12.5% of the total number of Euclid
observable AGN. Considering different redshift ranges, the frac-
tional uncertainty is 7.2% in the z ≤ 1 regime, 12.6% over the
range 1 < z ≤ 4, and 37.5% at 4 < z ≤ 7. As expected, we
find that the EDS probes the low-luminosity end of the XLF to
higher redshifts than the EWS. Despite some of the largest rela-
tive uncertainties (&40%) in this region, we predict that the EDS
is capable of probing AGN up to and beyond the flux limit of the
7 Ms CDF-S, across a much larger area. Therefore, even if only
a small number of AGN are indeed present in this regime and
observed by Euclid, constraints on the AGN LF can be vastly
improved.
In our EWS analysis, the median number of AGN observable
in at least one Euclid filter is (4.1± 0.2)× 107. The associated
uncertainty corresponds to 6.7% of the total number of Euclid
observable AGN across the entire probed redshift range. The
fractional uncertainties for the EWS in different redshift ranges
are 6.1% at z ≤ 1, 7.1% in the range 1 < z ≤ 4, and 28.8% in the
4 < z ≤ 7 regime. This marks an approximately halved relative
uncertainty derived from the XLF between the EDS and EWS.
In turn, this is a consequence of the EWS observing AGN that
occupy the well constrained bright end of the XLF, whilst the
EDS will probe AGN which occupy the faint end of the XLF,
where the greater uncertainties are present.
For comparison, we ran our EDS XLF uncertainty anal-
ysis utilising the hard XLF of Ueda et al. (2014). To obtain
an uncertainty using the Ueda et al. (2014) XLF we generated
three hundred realisations using three hundred sets of Gaussian
sampled parameters considering their published 1σ parameter
A250, page 14 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
0 1 2 3 4 5 6 7
z
18
20
22
24
26
28
I E
101
102
102.5
IE limiting magnitude
100
101
102
103
N
AG
N
/(
0.
2
z
0.
3
m
ag
)
0 1 2 3 4 5 6 7
z
18
20
22
24
26
28
Y E
101
101
102
102.5
YE limiting magnitude
100
101
102
103
N
AG
N
/(
0.
2
z
0.
3
m
ag
)
0 1 2 3 4 5 6 7
z
18
20
22
24
26
28
J E
101101
102
102.5
JE limiting magnitude
100
101
102
103
N
AG
N
/(
0.
2
z
0.
3
m
ag
)
0 1 2 3 4 5 6 7
z
18
20
22
24
26
28
H
E
101
101
101
102
102
102.5
HE limiting magnitude
100
101
102
103
N
AG
N
/(
0.
2
z
0.
3
m
ag
)
Fig. 9. Same as Fig. 8 for the EDS. Contours of constant log10 NAGN are plotted for log10 NAGN = (1, 2, 2.5).
uncertainties. Following our method as above, we found a total
of (2.6± 0.8)× 105 AGN detectable in at least one Euclid band.
This is consistent with our results adopting the Fotopoulou et al.
(2016a) XLF within our derived XLF uncertainty and makes an
almost indistinguishable difference to our final result. Analysing
the redshift-dependent uncertainties, we find that within the
well-constrained z ≤ 4 regime the number of Euclid detectable
AGN derived with the Fotopoulou et al. (2016a) and Ueda et al.
(2014) XLFs agree within 1σ in all redshift bins. In the extrapo-
lated z > 4 region, the Ueda et al. (2014) XLF predicts a steeply
decreasing yield of detectable AGN with increasing redshift
compared to the Fotopoulou et al. (2016a) XLF. This is consis-
tent with the space-density decline observed in Fig. 1. The dis-
parity between results with the two XLFs reaches 1 dex in the
5.5 ≤ z ≤ 6.0 bin. The relative uncertainty of the Ueda et al.
(2014) XLF predictions increases significantly with increasing
redshift, inflated by the lower numbers of detectable AGN, up
to 100% in the 5.5 ≤ z ≤ 6.0 bin. A larger overall relative
uncertainty of 30% is found with the Ueda et al. (2014) XLF,
where the relative uncertainty is larger than obtained with the
Fotopoulou et al. (2016a) XLF at all redshifts. The larger uncer-
tainty in the low-redshift regime has a greater impact when prop-
agated through our method to Euclid detectable AGN as there are
more observable AGN in this domain. In Sect. 5.2 we demon-
strate that Euclid is expected to identify ∼4 × 104 AGN in the
EWS at z > 4 using colour cuts alone. These data will serve to
point towards which extrapolation of the LF is correct.
103 104 105
Wavelength / Å
10 4
10 3
10 2
10 1
100
101
F
/
F
,1
m
HE
JE
YE
IE
z=0247
z=0247
z=0247
z=0247
Unobsc.
PASS
SFG
SB
QSO2
SEY2
SB-AGN
Fig. 10. All SEDs (unobscured and obscured) assigned to AGN in
this work, normalised at 1 µm. The redshift evolution of the effective
wavelength for each Euclid filter over the redshift range probed in this
work are depicted below as black lines. The seven SED classes in this
figure are: Unobscured AGN (‘Unobsc.’; blue), Passive (‘PASS’; pink),
Star-forming (‘SFG’; brown), Starburst (‘SB’; purple), High-luminosity
obscured AGN (‘QSO2’; orange), Seyfert 2 (‘SEY2’; green), and
Starburst-AGN composite (‘SB-AGN’; red).
A250, page 15 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
0 1 2 3 4 5 6 7
z
43
44
45
46
47
48
lo
g 1
0
(L
bo
l/
er
g
s
1 )
EWS
0 1 2 3 4 5 6 7
z
EDS
CDFS 7Ms
XMM-COSMOS
XXL-1000-AGN
5
10
20
50
100
N
AG
N
/N
AG
N
(%
)
Fig. 11. Fractional uncertainty (δNAGN/NAGN) on the number of AGN detectable in at least one Euclid band in bins of redshift (∆z = 0.25) and
bolometric luminosity (∆ log10[Lbol/erg s
−1] = 0.25) for the EWS (left) and EDS (right). The fractional uncertainties were derived using three
hundred realisations of the adopted XLF probing the ±1σ interval of each parameter given in Table 1. For reference, we plot curves corresponding
to the 50% completeness flux limits of the 7 Ms exposure of the Chandra Deep Field-South (red, F2−7 keV = 2.5 × 10−16 erg s−1 cm−2; Luo et al.
2017) and medium-depth X-ray survey XMM-COSMOS (black, F2−10 keV = 5.6×10−15 erg s−1 cm−2; Cappelluti et al. 2009). We additionally show
the sample limiting flux of the XXL 1000 brightest AGN sample (XXL-1000-AGN) as used in Fotopoulou et al. (2016b) for SED fitting (blue;
F2−10 keV = 4.8 × 10−14 erg s−1 cm−2; Fotopoulou et al. 2016b). The flux limits were homogenised to the 2–10 keV band assuming an X-ray power
law with photon index Γ = 1.9, where applicable, and converted to the bolometric domain assuming the X-ray bolometric correction of Duras et al.
(2020).
5.1.2. Extrapolation of the optically obscured AGN fraction
The extrapolation of the Merloni et al. (2014) optically obscured
AGN fraction evolution led to some modification at lower lumi-
nosities where the fraction would exceed 1.0 beyond z = 3. We
also found that when extrapolated beyond z ≈ 6.5 the obscured
fraction of the highest luminosity AGN becomes greater than
that of the medium luminosity sources (Fig. 3). This is touched
upon in Merloni et al. (2014), where it is described that the
incidence of obscuration shows significant redshift evolution
only for the most luminous AGN, which appear to be more
commonly obscured at higher redshifts. Treister & Urry (2006)
also reported the relative optically obscured fraction of X-ray
selected AGN increases with redshift in the range 0 ≤ z ≤ 4,
when corrected for optical counterpart selection effects.
In regard to X-ray obscuration, a range of works present
evidence that the absorbed but Compton-thin fraction of
X-ray AGN increases with redshift at fixed luminosity (e.g.
La Franca et al. 2005; Hasinger 2008; Treister et al. 2009;
Aird et al. 2015). Aird et al. (2015) report with low signif-
icance that the absorbed fraction of the highest luminosity
AGN may increase more substantially with redshift than lower
luminosity AGN, remarkably similarly to our extrapolation of
the Merloni et al. (2014) model, albeit over a smaller redshift
range (see Aird et al. 2015, Fig. 15). The heavily obscured
(log10[NH/cm
−2] > 23), high-redshift (3 ≤ z ≤ 6) frac-
tion of AGN across a range of luminosities is suggested
to be ∼0.6−0.8 (e.g. Vito et al. 2018; Signorini et al. 2023;
Pouliasis et al. 2024). No dependence on X-ray luminosity or
redshift is found for these values, however the derived fractions
are higher than the local Universe obscured fraction for the same
column density (53%; Burlon et al. 2011).
In the high X-ray luminosity regime, Vijarnwannaluk et al.
(2022) report that the obscured (log10[NH/cm
−2] > 22) frac-
tion of X-ray selected AGN with log10(LX/erg s
−1) > 44.5
lies around 76% at z = 2. This again marks an increase
compared to obscured fractions derived in the local Universe.
Buchner et al. (2015) similarly report that averaged over cos-
mic time, obscured AGN with log10(NH/cm
−2) > 22 account
for 77% of the number and luminosity density of the AGN
population with log10(LX/erg s
−1) > 43. They also describe
evidence of a redshift evolution, with the obscured fraction of
Compton-thin AGN increasing towards z = 3 where it is 25%
higher than the local Universe value. Malizia et al. (2012) per-
formed a study on the X-ray and optical obscuration properties
of a sample of International Gamma-Ray Astrophysics Labo-
ratory/IBIS (INTEGRAL/IBIS; Winkler et al. 2003) AGN span-
ning the redshift range 0.0014 ≤ z ≤ 3.7. They present evi-
dence that the X-ray absorbed (log10[NH/cm
−2] > 22) frac-
tion of AGN is a function of both X-ray luminosity and red-
shift, with the absorbed fraction increasing with increasing
redshift but decreasing with increasing luminosity. They also
determine that when corrected for bias, X-ray absorbed AGN
account for up to 80% of the population, in agreement with the
results discussed above. Furthermore, they determine that the
optically obscured and X-ray obscured classifications agree in
88% of AGN. Assuming the agreement of X-ray and optical
obscuration in the majority of objects (e.g. Malizia et al. 2012;
Merloni et al. 2014; Burtscher et al. 2016; Fotopoulou et al.
2016b), we expect the fraction of optically obscured AGN to
correlate with these general trends.
Obscuration in AGN is driven by attenuating gas and dust from
both the AGN torus and ISM of the host galaxy in the line-of-sight
to the observer. The contribution from the AGN torus is suggested
to decrease with increasing AGN luminosity, a trend described
by the receding torus model (e.g. Lawrence 1991; Simpson 2005;
Assef et al. 2013). A similar idea is proposed in the AGN evo-
lution scheme of Hopkins et al. (2008), in which obscured AGN
precede an unobscured AGN phase after blowing out gas and dust
accumulated from galaxy mergers. The latter model provides a
reasoning for the earlier peak in space density of obscured AGN
compared to unobscured (e.g. Lacy et al. 2015). Gilli et al. (2022)
A250, page 16 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
suggests that host galaxy ISM plays a larger role with increasing
redshift, with z & 6 AGN expected to be primarily obscured by
the ISM of their hosts.
Considering the results above, for low X-ray luminos-
ity AGN extrapolated to higher redshifts our model may
over-predict the obscured fraction of AGN by 20–40%. In
this regime, the overall trend of the optically obscured
fraction of AGN increasing from local values is pre-
served and in agreement with the literature. For medium
and high X-ray luminosities our extrapolated model agrees
with observational trends of optically obscured fractions
of AGN increasing with redshift and the highest lumi-
nosity AGN having a more significant evolution compared
to medium luminosity counterparts. Overall, the extrapolated
model used in our method is consistent with the available lit-
erature, however at high redshifts it is not feasible to discern the
true evolution of the optically obscured fraction due to lack of
observable objects with the current technological limitations.
5.1.3. AGN bolometric correction dispersion
Intrinsic scatter in bolometric corrections is to be expected as
AGN spectra are diverse and exhibit different scaling from object
to object with the population following a correlated trend rather
than an exact underlying rule. We decided to conduct our anal-
ysis using a nominal bolometric correction conversion as a pop-
ulation average. To test the impact of this assumption on our
overall estimates of Euclid observable AGN we re-ran our anal-
ysis for the EDS, separately enabling scatter for each bolometric
correction, for three hundred runs each.
We explored the dispersion for the employed bolometric cor-
rections in Duras et al. (2020) and Runnoe et al. (2012) through
Gaussian sampling of each parameterising variable considering
the 1σ uncertainties reported in their respective publications. For
the Duras et al. (2020) X-ray bolometric correction we sampled
an additional factor derived from the reported intrinsic spread of
0.27 dex. New parameter and intrinsic spread values were drawn
for each obscured AGN we consider.
Introducing scatter in the X-ray bolometric correction we
find a median number AGN detectable in at least one Euclid
band of (2.150± 0.001)× 105. The 1σ uncertainty of our anal-
ysis corresponds to 0.05% of the total number of Euclid observ-
able AGN. This difference is marginal to the overall results of
our analysis.
Completing the analysis with scatter in the 2 µm bolometric
correction we report a median number of AGN detectable in at
least one Euclid band of (2.156± 0.001)× 105. Identical to our
X-ray bolometric correction uncertainty analysis, the uncertainty
associated with the 2 µm bolometric correction corresponds to
0.05% of the total number of Euclid observable AGN.
5.1.4. Unobscured AGN emission lines
The Shen et al. (2020) mean quasar SED used to model unob-
scured AGN in this work lacks AGN emission lines. Strong AGN
emission lines, such as the Lyα and Hα hydrogen recombination
lines, are demonstrated to cause significant deviations to pho-
tometric colours. These changes are a strong function of red-
shift as the emission lines are shifted in and out of different
photometric filters (Temple et al. 2021). We focussed on Lyα
and Hα in the present analysis because these emission lines
are the most prominent emission line features in many AGN.
The findings of this section therefore provide a lower limit to
the overall effect the inclusion of emission lines has on our
results.
The Hα+[N ii] emission line complex is composed of the
broad and narrow component Hα emission line and narrow [N
ii]λ6549 and [N ii]λ6583 lines. Hα has a rest-frame wavelength
of 6563 Å and is hence redshifted beyond the extent of the HE
band at z = 2.05. The analysis of Hα in this section is there-
fore limited to the redshift range 0.01 ≤ z ≤ 2.05. Lyα is also
comprised of a broad and narrow component and has rest-frame
wavelength 1216 Å. Lyα therefore enters the IE band at z = 3.4.
Accordingly, the analysis of the impact of including Lyα in our
unobscured AGN template is constrained to 3.4 ≤ z ≤ 7.0.
To attain realistic Lyα and Hα+[N ii] flux excess values we
exploited measurements of stacked quasar templates produced in
Euclid Collaboration: Lusso et al. (2024). We refer the reader to
the aforementioned work for full details of the spectral construc-
tion and analysis, however we give a brief overview here. Nine
empirical SDSS quasar stacked spectra were generated from
a parent sample of 91 579 SDSS-DR7 quasars following the
method of Lusso et al. (2015). Each empirical stack was binned
in Hβ equivalent width (EW) and FWHM ranges. The nine tem-
plates were modified using a grid of redshifts, E(B − V), and
bolometric luminosities. The ranges of these parameters were
constructed to probe the expected observed parameter space of
unobscured AGN in Euclid. The spectra were scaled to the given
luminosity, optically reddened with the assigned E(B− V) value
and the Prevot et al. (1984) SMC law, and finally redshifted with
IGM extinction applied via the curve of Prochaska et al. (2014).
Across the parameter grid a random sub-sample of 1248 of these
spectra that satisfied the Euclid depths of VIS and NISP (5σ)
were selected.
The emission line fluxes for each of the empirical unob-
scured AGN spectra included in the final sample were
measured using the Quasar Spectral Fitting library (QSFIT;
Calderone et al. 2017; Selwood et al. 2023) AGN spectral fit-
ting package. We leveraged the QSFIT spectral fitting results to
obtain the Hα+[N ii] complex integrated luminosity excess for
each of the nine unobscured AGN templates. For each empirical
template the median luminosity and its median absolute devia-
tion (σMAD) was determined. We then assigned each unobscured
AGN in our EWS and EDS samples a random unobscured AGN
template class and sampled a corresponding Hα+[N ii] luminos-
ity based on the derived median and σMAD values determined for
the corresponding template class. Transforming to a Hα+[N ii]
complex flux excess, we added the flux perturbation to the mea-
sured incident flux in the Euclid photometric band associated
with the Hα observed-frame centroid. We then re-calculated the
apparent magnitude considering the emission line perturbation.
Adopting this procedure we assessed the number of AGN that
would become observable in any Euclid band with the addition
of the Hα+[N ii] emission line complex in our unobscured AGN
SED. We followed the same procedure for Lyα measurements,
however due to there only being ∼300 spectra with z > 3.4 in the
Euclid Collaboration: Lusso et al. (2024) sample we do not sep-
arate by template class. Instead, we take the median Lyα lumi-
nosity and σMAD of the full z > 3.4 sub-sample.
In our EWS sample, there are 1.0× 107 candidate unob-
scured AGN in the Hα+[N ii] range 0.01 ≤ z ≤ 2.05. We found
4644 (0.05%) of these AGN become newly observable with the
inclusion of Hα+[N ii]. There are 1.9× 106 unobscured AGN
at 3.4 ≤ z ≤ 7.0 of which 119 (0.006%) become observable
with the inclusion of Lyα. In total a lower limit of 4763 out of
1.2× 107 (0.04%) unobscured AGN could have been observed
A250, page 17 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
with the inclusion of broad emission lines in our unobscured
AGN template.
For the EDS sample the inclusion of Hα+[N ii] results in
an additional 60 detectable unobscured AGN out of 3.3× 104 at
0.01 ≤ z ≤ 2.05, a gain of 0.2%. There are 5.9× 103 unobscured
AGN at z > 3.4 of which only two (0.03%) become observable
by Euclid with Lyα considered. In total 62 of 3.9× 104 (0.16%)
unobscured AGN could be observed with the inclusion of broad
emission lines in our unobscured AGN template.
The enhanced proportion of newly detected AGN from
Hα+[N ii] in the EDS is due to the deeper apparent magnitudes
(i.e. fluxes) probed in the survey. As the 5σ flux threshold is
deeper, a small change in the source flux provided by emission
lines is more likely to allow an AGN to breach this threshold
and become detectable with Euclid. In both surveys the inclu-
sion of Lyα made a smaller difference to detectability compared
to Hα+[N ii]. This is due to the higher redshift range where Lyα
occupies the Euclid bands. The additive Lyα flux excess in this
regime is comparatively smaller and so has a lower impact on
detectability than with Hα+[N ii]. We observe from Figs. 8 and
9 that whilst there is a high density of AGN near the detection
threshold at 0.01 ≤ z ≤ 2.05, the parameter space is more sparse
at z > 3.4. This means there are less AGN that could become
observable due to a flux perturbation from emission lines at
higher redshifts.
Despite these numbers amounting to a lower limit, the
derived number of newly observed AGN are a negligible per-
centage of the total observable sources and fall well within the
Poisson error (σ =
√
N) of our data sets. We therefore conclude
that the omission of AGN emission lines contributes a minimal
effect and uncertainty on our total AGN number estimates in the
Euclid photometric surveys.
5.1.5. Uncertainty comparison
Here we compare the uncertainties determined for different mod-
els and assumptions in this work. In Fig. 12 we present a colla-
tion of the relative uncertainties quantified in this section on the
number of AGN with a detection in at least one Euclid band. The
blue shaded region signifies the expected relative Poisson noise
(σ =
√
N) for our EDS detectable AGN.
The dominant source of uncertainty in our analysis is the
uncertainty in the XLF. The XLF relative uncertainty is several
orders of magnitude greater than any other source of uncertainty
quantified here and is the only source that generates an uncer-
tainty greater than the derived Poisson noise for our EDS sam-
ple. Amounting to 12.5% for the EDS and 6.7% for the EWS,
we may consider this the overall uncertainty of our detectable
AGN estimates. Well within the EDS Poisson noise, the second
largest uncertainty is driven by the IR and X-ray bolometric cor-
rection dispersion. These relative uncertainties are already neg-
ligible compared to that of the XLF.
We were unable to quantify the impact that extrapolating
the Merloni et al. (2014) optically obscured AGN fraction model
has on our results. Although we have established our extrap-
olation agrees with the general trends of literature results, the
ground truth of the redshift evolution and X-ray luminosity
dependence of the optically obscured AGN fraction remains
unresolved. Despite the complexity in constraining such depen-
dencies we hope that the extension of such models in the redshift
and luminosity domain can be an area of investigation with next-
generation facilities.
In a scenario where our predictions in this analysis are
proven to be in tension with Euclid observations, we can back-
XLF IRBC XBC EL
10 4
10 3
10 2
10 1
N
AG
N
N
AG
N
EDS
EWS
Fig. 12. Relative 1σ uncertainty of the number of AGN detectable in
at least one Euclid band (NAGN) for each of the quantified sources of
uncertainty introduced in this work. Displayed in this plot are uncer-
tainties introduced by the X-ray luminosity function (XLF), infrared
bolometric correction dispersion (IRBC), X-ray bolometric correction
dispersion (XBC) and unobscured AGN emission lines (EL) for which
a lower limit was quantified using the Lyα emission line and Hα+[N ii]
complex. The expected relative Poisson noise for the EDS, calculated
as σ =
√
NAGN, is indicated by the shaded blue region.
propagate through our assumed models and framework to iden-
tify where disparities lie between our current understanding and
the ground truth. This will in turn serve to inform the community
of areas where there are gaps and inconsistencies in our current
modelling of AGN properties and demographics.
5.2. Photometric AGN selection in Euclid
Optical to MIR colour-colour cuts are often invoked to select
samples of AGN from photometric data sets (e.g. Stern et al.
2005; Donley et al. 2012; Assef et al. 2018). The colours used
in AGN selection exploit spectral features unique to AGN, such
as the MIR excess, to separate AGN from inactive galaxies and
stars in the colour-colour parameter space. Because they can be
quickly and cheaply applied to large data sets, simple colour
selection criteria are used in many cases as an initial selection
for candidates of a population before applying more complex
identification methods. This will likely be the use-case for colour
selection of AGN in Euclid.
Only a fraction of the AGN detected by Euclid (i.e. present
in at least one image) will actually be selected as AGN based on
Euclid photometry alone. Indeed, half of the obscured AGN SED
classes we assign in Sect. 3.3.3 (PASS, SFG, SB) are entirely
free from AGN spectral features in the Euclid wavelength range.
This means that without external data these AGN will appear
indistinguishable from non-active galaxies. Therefore, it will not
be possible to identify all of our Euclid detectable AGN using
photometric criteria, with Euclid data alone or otherwise.
Colour selection criteria for AGN using
Euclid photometry has been explored in depth in
Euclid Collaboration: Bisigello et al. (2024). Optimal selec-
tion criteria were derived for the EWS and EDS based on
Euclid photometry exclusively and with the inclusion of ancil-
lary photometric observations, such as with Spitzer/IRAC
A250, page 18 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Table 6. Performance of Euclid photometry AGN selection criteria (Euclid Collaboration: Bisigello et al. 2024) discussed in Sect. 5.2 applied to
our samples of EWS and EDS AGN.
Survey Target Class Photometry Total Selected AGN Surface Density (deg−2) Unobscured AGN Obscured AGN C P
EWS Unobsc. AGN Euclid 4.8× 106 331 4.4× 106 3.7× 105 0.23 0.17
EWS Unobsc. AGN Euclid, Rubin/LSST 5.7× 106 393 5.5× 106 1.8× 105 0.45 0.92
EWS All AGN Euclid, Rubin/LSST 6.0× 106 413 4.3× 106 1.7× 106 0.51 0.19
EWS ∪ Euclid, Rubin/LSST 8.1× 106 556 6.0× 106 2.1× 106 0.37 -
EDS Unobsc. AGN Euclid 1.7× 104 346 1.6× 104 400 0.11 0.23
EDS Unobsc. AGN Euclid, Rubin/LSST 2.0× 104 392 1.9× 104 390 0.45 0.92
EDS All AGN Euclid, Rubin/LSST 2.9× 104 579 2.0× 104 9.5× 103 0.32 0.58
EDS ∪ Euclid, Rubin/LSST 3.5× 104 692 2.5× 104 1.0× 104 0.22 -
Notes. We consider only AGN that are detected above the 5σ limiting magnitude for the four constituent filters in each colour criterion, which
are given in Appendix E. For each criterion we report its target class, referring to the population of AGN the criterion aims to select, the total
number of selected AGN, the surface density of selected AGN in deg−2, the completeness (C) at the 5σ level in the relevant filters, and the expected
purity (P) for the criterion, as estimated in Euclid Collaboration: Bisigello et al. (2024). The final row for each survey denotes the union (∪) of all
considered selection criteria.
(Fazio et al. 2004) and Rubin/LSST. Each selection was defined
to target either unobscured AGN or all AGN including obscured
and composite sources. Optimal criteria were derived to maximise
the F1-score, that is the harmonic mean of sample completeness
(C; fraction of selected AGN with respect to the total AGN
population) and purity (P; fraction of selected sample that are
AGN, not contaminants). Each criterion exhibits a selection
function with a unique redshift dependence. In the following, we
have applied each of the optimal colour selection criteria to our
samples of simulated AGN to assess how many Euclid detectable
AGN we can expect to select using Euclid colours. We verified
that the AGN colours derived with our models are consistent with
those derived in Euclid Collaboration: Bisigello et al. (2024)
over the relevant ranges for AGN selection.
We considered only AGN in our samples with 5σ detec-
tions in the constituent bands for each colour criterion. For
AGN selection schemes incorporating Rubin/LSST bands we
considered the 5σ point-source final co-added depths reported
in Ivezic´ et al. (2019); (u, g, r, i, z, Y) = (25.6, 26.9, 26.9, 26.4,
25.6, 24.8). Detailed breakdowns of the photometric selection
criteria and the results of applying each to our data are provided
in Appendix E. We additionally present density plots depicting
our AGN sample with each colour selection criterion considered
in this section (Figs. E.1, E.2).
Table 6 provides a summary of the performance of each pho-
tometric selection criterion applied to our AGN sample. Abso-
lute expected numbers of selected AGN and surface densities are
presented as well as an assessment of the sample completeness.
In all circumstances the quoted completeness refers to the com-
pleteness of selected AGN relative to 5σ detected AGN available
in the same bands. Due to our sample, by construction, contain-
ing only the expected colours of AGN and not other astrophysi-
cal sources we are unable to estimate the purity of our selected
samples, however we note the expected purity of each selection
criterion from Euclid Collaboration: Bisigello et al. (2024).
Adopting the Euclid-only unobscured AGN criterion, we
expect a completeness C = 0.23 in the EWS and C = 0.11 in the
EDS. When considering completeness with respect to available
unobscured AGN, the target class of the selection, the respec-
tive EWS and EDS completeness values rise to C = 0.52 and
C = 0.40. AGN selected using colour cuts with Euclid pho-
tometry alone represent 40% and 35% of the unobscured AGN
detectable in at least one Euclid band in the EWS and EDS,
respectively. The same samples correspond to 12% and 8% of
the total AGN population detectable in at least one Euclid band.
The surface densities resulting from these selections have a dif-
ference of only 15 deg−2 between the EWS and EDS, despite the
EDS probing two magnitudes deeper. The disparity stems from
the low overall completeness for the EDS selection. Although
there is around twice the surface density of detected AGN in
the EDS, the Euclid-only unobscured AGN selection has around
half the completeness. The inefficiency of the EDS selection is
noted in Euclid Collaboration: Bisigello et al. (2024), emerging
from the fact that most EDS AGN are expected to be faint, lying
on the detection boundary of each band.
The addition of Rubin/LSST bands, when available, are
expected to supplement the purity, completeness, and size of
AGN samples selected with Euclid data. Crucially, the inclu-
sion of ancillary data will allow the selection of obscured and
composite AGN as well as unobscured. To determine the total
sample sizes of selectable AGN with Euclid, we took the union
of all selected samples resulting from the criteria discussed in
this section, leveraging both Euclid and Rubin/LSST photome-
try. The total selected samples in the EWS and EDS have overall
completeness C = 0.37 and C = 0.22. Considering selected and
available unobscured AGN, the total samples have completeness
C = 0.66 and C = 0.58 in the EWS and EDS. The complete-
ness is C = 0.15 and C = 0.09 with respect to obscured AGN.
These total samples equate to 21% and 15% of the AGN found
to obtain a 5σ detection in at least one Euclid photometric band
in the EWS and EDS, respectively.
Figure 13 presents redshift distributions, redshift-dependent
completeness, and Lbol-z planes of the Euclid-only selection and
union of all selections for the EWS and EDS. The middle pan-
els of Fig. 13 show the completeness is highest for all selected
samples around cosmic noon (z ∼ 2), matching the expecta-
tions of Euclid Collaboration: Bisigello et al. (2024). This marks
the redshift at which the 4000 Å break enters the YE band,
providing one of the cruxes the photometric selection criteria
exploits to separate AGN from galaxies. An elevated complete-
ness is observed around this redshift for the total selected sam-
ple compared to the Euclid-only selected sample in both sur-
veys and is particularly accentuated for the EDS selections.
The shorter wavelength optical bands help to further distinguish
AGN colours from the generally redder galaxy colours.
The derived 5σ level completeness for each of the applied
selection criteria may appear as a point of concern for our
samples. Many analyses however do not have the benefit of
first assessing the underlying population of available detected
sources. Those that do perform such an exploration report similar
A250, page 19 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
or diminished levels of completeness. For example, Assef et al.
(2018) estimate a completeness of 0.17 and 0.28 for their colour-
colour selections of AGN from the AllWISE survey with a
reliability of 90% and 75%, respectively (see also Stern et al.
2012; Assef et al. 2013).
The redshift distributions (top panels of Fig. 13) show the
highest number of AGN are selected around around cosmic noon
in both surveys. This is congruent with the peak in the number of
detected AGN (Fig. 7) as well as the highest selection complete-
ness. AGN selected at z > 3 are dominated by those identified
by the Euclid-only unobscured AGN criterion. The addition of
ancillary photometry will help to distinguish AGN at low and
intermediate redshifts, whilst Euclid colours alone can identify
the luminous unobscured AGN at z > 3. We project ∼4 × 104
AGN at z > 4 will be selected in the EWS, with ∼2000 selected
AGN at z > 4.5.
Unobscured AGN are selected efficiently in all cases, with
around half of the available AGN of this class identified using
Euclid photometry alone and up to two thirds when employ-
ing ancillary optical colours. Obscured AGN are less efficiently
identified with only a tenth of those available in the EDS selected
and a fifth in the EWS. Indeed, across all selections for both sur-
veys only ∼15% of selected AGN have E(B − V) ≥ 0.05. In
comparison, ∼50% of the detected AGN available in the selec-
tion bands have E(B − V) ≥ 0.05. The majority of the colour
selections used here were fine-tuned to identify the unobscured
population, hence it is no surprise the reddened population is
missed. It is clear however that there is an abundance of AGN
with higher E(B − V) available to be extracted from the data
through alternative means.
The obscured AGN which are identified in either survey
have the SB-AGN, SEY2, or SB SED classes. The SB-AGN
and SB classes have optical–NIR colours similar to unobscured
AGN (see Fig. C.1). As such these SED classes are often cap-
tured in the same selection cuts identifying unobscured AGN. No
obscured AGN with the PASS, SFG or QSO2 SED classes were
selected. These are all galaxy-dominated SEDs (see Fig. 10) with
optical–NIR colours that are hard to distinguish from inactive
galaxies. As such, we would not expect these classes to be effi-
ciently selected via Euclid photometry.
AGN will be selected over the bolometric luminosity range
43 ≤ log10(Lbol / erg s−1) ≤ 47, a significant portion of the range
used as input for our simulation. The Lbol-z planes in the bot-
tom panel of Fig. 13 show that ancillary photometry allows us
to select lower-luminosity AGN in both the EWS and EDS. This
is apparent compared to the Euclid-only unobscured AGN selec-
tion at 0.5 ≤ z ≤ 3 for both surveys. The range of the 1σ verti-
cal bars suggest AGN selected around 0.5 ≤ z ≤ 3 in the EDS
are relatively lower-luminosity than those in the EWS. This is
due to the two magnitude deeper observations in the EDS reach-
ing a fainter population of AGN. The highest redshift selected
AGN from our simulation is at z = 5.16 and selected from the
EWS. Due to the much larger area probed by the EWS it is more
likely that exceptionally luminous high-redshift AGN will be
observed and selected. We discuss predictions of high-redshift
(z ∼ 7) AGN detections in more detail compared with the results
of Euclid Collaboration: Barnet et al. (2019) in Appendix F.
The large disparity between the projected number of AGN
we will detect with Euclid versus those we can identify as AGN
should not be neglected. Our findings suggest work should be
undertaken to improve upon and devise new AGN selection
methods in order to maximize the AGN yield with Euclid as
well as other facilities. This is particularly crucial for interme-
diate and high-redshift AGN selection. We show in this work
102
104
106
N
AG
N
Total selected
Euclid-only
0.0
0.2
0.4
0.6
0.8
1.0
C
Total selected
Euclid-only
0 1 2 3 4 5 6 7
z
43
44
45
46
lo
g 1
0
(L
bo
l/
er
g
s
1 )
0 1 2 3 4 5 6 7
z
Total selected
Euclid-only
Fig. 13. Redshift distributions (top), redshift-dependent completeness
(middle), and Lbol-z planes (bottom) for the selected AGN in the EWS
(left) and EDS (right). AGN selected with Euclid-only photometric cri-
teria (blue) and the total selected sample defined as the union (∪) of all
Euclid and ancillary ugrz photometric criteria discussed in Sect. 5.2 are
plotted for each panel. In all plots redshift is binned with width δz = 0.5.
In the Lbol-z plane points represent the median, vertical lines represent
1σ standard deviation and horizontal lines represent the width of the
redshift bin.
we expect to detect an abundance of AGN in this regime with
Euclid, however we lack a means to efficiently identify such
AGN from the data. Of course, this is partly due to Euclid prob-
ing wavelengths where it is difficult to distinguish AGN emission
from other populations of galaxies. A multiwavelength approach
where ancillary observations allow, or explore the viability of
machine learning driven techniques (Bisigello et al., in prep.;
Signor et al., in prep.) is necessary.
Before Rubin/LSST bands become available, it is possi-
ble to exploit readily accessible ancillary optical data to per-
form AGN selection in conjunction with Euclid bands. The
Dark Energy Survey (DES; Abbott et al. 2018), covering a foot-
print of ∼5000 deg2 in the southern hemisphere, provides opti-
cal to NIR photometry in the g, r, i, z, Y bands to 5σ pho-
tometric depths of 25.0, 24.5, 23.7, 22.6 and 21.3, respec-
tively. Unfortunately, a one-to-one mapping is not possible
with DES for the photometric selection criteria defined in
Euclid Collaboration: Bisigello et al. (2024) as DES lacks obser-
vations in a u band. The Ultraviolet Near Infrared Optical North-
ern Survey (UNIONS) however will observe ∼5000 deg2 of the
northern hemisphere in the gri bands and ∼10 000 deg2 in the
crucial u band (Ibata et al. 2017). We assessed the number of
AGN in the EWS with a 5σ detection in each DES band to be
1.2× 107 in g, 1.3× 107 in r, 1.1× 107 in i, 6.1× 106 in z and
2.1× 106 in Y . For AGN in the EDS we predict 4.2× 105 in g,
4.3× 105 in r, 3.4× 105 in i, 1.9× 105 in z and 5.9× 104 in Y .
5.3. Comparison to other surveys
We compared the surface density of AGN in a mixture of wide
field and medium area surveys described in Table 7 to those pre-
dicted for Euclid in this work in Fig. 14. In terms of detectable
AGN, Euclid will detect far more AGN in the EWS and EDS
than are identified in other surveys that cover similar areas. Of
A250, page 20 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Table 7. Characteristics and AGN identification statistics for a selection of medium area and wide field surveys across different wavebands, which
we use for comparison with our Euclid yields.
Survey Band Area (deg2) Number of AGN AGN Surface Density (deg−2) Reference(s)
EWS (detected) optical–NIR 14 500 4.0× 107 2800 This work
EWS (selected) optical–NIR 14 500 8.1× 106 556 ”
EWS (selected, Euclid-only) optical–NIR 14 500 4.8× 106 331 ”
EDS (detected) optical–NIR 50 2.4× 105 4700 ”
EDS (selected) optical–NIR 50 3.5× 104 692 ”
EDS (selected, Euclid-only) optical–NIR 50 1.7× 104 346 ”
XMM-SERVS 0.5–10 keV 13.1 10 300 786 Chen et al. (2018);
Ni et al. (2021)
XXL-3XLSS 0.5–10 keV 50 26 056 521 Chiappetti et al. (2018)
Spitzer cryogenic (a) NIR–MIR 55 ∼20 000 364 Lacy & Sajina (2020)
eFEDS 0.2–2.3 keV 142 22 079 155 Liu et al. (2022);
Salvato et al. (2022)
DES optical–NIR 4913 945 860 193 Yang & Shen (2023)
SDSS optical 9376 750 414 80 (b) Lyke et al. (2020)
eRASS1 (c) 0.5–2 keV ∼15 000 645 000 43 Salvato et al. (in prep)
AllWISE (R90) MIR 30 093 4.54× 106 151 Assef et al. (2018)
AllWISE (C75) MIR 30 093 2.09× 107 695 Assef et al. (2018)
Notes. The quoted area relates to the area considered in the survey AGN selection procedure as described in the corresponding reference(s). Euclid
survey detected and selected yields as plotted in Fig. 14 are provided. Detected samples refer to the number of AGN with a 5σ detection in at least
one Euclid filter. “Selected” samples correspond to the union of all selection criteria considered in Sect. 5.2 and “selected, Euclid-only” samples
refer to AGN selected with selection criteria that incorporate Euclid bands exclusively. (a)Collation of AGN identified in Spitzer Space Telescope
cryogenic surveys: The Spitzer Wide-Area Infrared Extragalactic Survey (SWIRE; Lonsdale et al. 2003), The AGN and Galaxy Evolution Survey
(AGES; Kochanek et al. 2012), The Spitzer First Look survey (Lacy et al. 2005; Fadda et al. 2006), and the Spitzer COSMOS survey (S-COSMOS;
Sanders et al. 2007). (b)Lower limit as the sample includes only spectroscopically confirmed unobscured quasars. (c)Extragalactic counterparts are
identified for eRASS1 soft band (0.5–2 keV) sources in the area that overlaps with the Legacy Survey DR10 (Dey et al. 2019; Zenteno et al., in
prep.). A median AGN surface density is provided, given that the depth of eRASS1 increases getting closer to the South Ecliptic Pole (Liu et al.,
in prep.).
course in this case we are comparing detectable AGN in the
Euclid photometric surveys to AGN that have been selected
through various means in different surveys.
Considering AGN that are selected with Euclid-only photo-
metric criteria (see Sect. 5.2), we expect AGN surface densities
that are on par with surveys of similar area. In the EDS AGN
selected with Euclid-only unobscured AGN criteria yields an
AGN surface density of 346 deg−2, marginally lower than that
of the collated cryogenic Spitzer surveys and the XXL-3XLSS
survey. The XXL-3XLSS is the deepest X-ray survey we have
compared with in our analysis and is likely to recover AGN with
colours that cannot be selected using our photometric criteria as
some may appear indistinguishable from non-active galaxies in
the optical–NIR. The AGN surface density of 331 deg−2 derived
for the EWS using Euclid-only unobscured AGN colour selec-
tion is roughly double the AGN surface density found for DES
and AllWISE, using their 90% reliability (R90) criterion. Work-
ing with Euclid photometry alone then, we predict that the sur-
face density of selected AGN in the EWS will be greater than
ground-based optical and space-based MIR peers.
Considering the total selected AGN sample (union of all cri-
teria discussed in Sect. 5.2), we forecast the EDS will yield a
greater AGN surface density than the XMM-3XLSS survey. The
total sample surface density is marginally below that of XMM-
SERVS, which marks the highest AGN surface density from a
survey considered in this analysis. The total selected AGN sur-
face density for the EWS will be greater than optical surveys
and comparable to the reported AGN surface density for the All-
WISE 75% completeness (C75) selection. Again, this empha-
sises that we expect Euclid to perform to a similar degree as a
space-based MIR counterpart in regard to AGN selection.
Towards the end of Euclid survey operations both the EWS
and EDS are expected to yield surface densities of selected AGN
that are similar or in excess of their wide and medium-field coun-
terparts across a range of wavebands. If AGN selection in Euclid
is improved upon by combining different techniques and fine
tuning the criteria considered here, Euclid may well surpass the
AGN selection performance of any prior surveys. This represents
a promising outcome considering that Euclid was not designed
for the facilitation of AGN identification.
5.4. Expected X-ray counterparts
We examined the range of 2–10 keV X-ray fluxes probed by
Euclid detectable AGN in our samples. Figure 15 shows the two-
dimensional density distribution of our EWS AGN sample on a
2–10 keV X-ray flux vs Euclid HE apparent magnitude plane.
HE was selected for this visualisation as we found the greatest
yield of Euclid detectable AGN in this band. The 2–10 keV flux
limits of the X-ray surveys considered in this section are plotted
for reference: XMM-COSMOS (90% completeness, F2−10 keV =
9.3 × 10−15 erg s−1 cm−2; Cappelluti et al. 2009), XXL-3XLSS
(90% completeness, F2−10 keV = 3.8 × 10−14 erg s−1 cm−2;
Chiappetti et al. 2018), XMM-SERVS W-CDF-S field (90%
completeness, F2−10 keV = 2.9 × 10−14 erg s−1 cm−2; Ni et al.
2021), and the XXL 1000 brightest AGN sample (XXL-
1000-AGN) as used in Fotopoulou et al. (2016b) for SED fit-
ting (sample flux limit, F2−10 keV = 4.8 × 10−14 erg s−1 cm−2;
Fotopoulou et al. 2016b). The over-density of sources on an
approximately linear trend in the figure is populated by our AGN
SEDs and is a consequence of using the nominal X-ray and 2 µm
bolometric corrections without considering their dispersion.
A250, page 21 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
101 102 103 104 105
Survey area / deg2
102
103
104
AG
N
 s
ur
fa
ce
 d
en
si
ty
 / 
de
g
2
EWS detected AGN
EWS selected AGN
EWS selected AGN
    (Euclid-only)
EDS detected AGN
EDS selected AGN
EDS selected AGN
    (Euclid-only)
XMM-SERVS 0.5 10keV
XXL-3XLSS 0.5 10keV
eFEDS 0.2 2.3keV
eRASS1 0.5 2keV
DES DR2 QSO
SDSS DR16 QSO
Spitzer cryogenic surveys
AllWISE R90
AllWISE C75
Fig. 14. AGN surface density versus survey area comparison for a number of wide field and medium area surveys in different wavebands. Included
AGN surface densities are: EWS detectable AGN (magenta diamond, unfilled), EWS selected AGN (magenta diamond, solid fill), EWS selected
AGN using Euclid photometry only (magenta diamond, translucent fill), EDS detectable AGN (blue diamond, unfilled), EDS selected AGN (blue
diamond, solid fill), EDS selected AGN using Euclid photometry only (blue diamond, translucent fill), XMM-SERVS (black downward-facing
triangle, unfilled), XXL-3XLSS (black cross, unfilled), eFEDS (black square, unfilled), eRASS1 (black pentagon, unfilled), DES (green circle,
unfilled), SDSS spectroscopic lower limit (green lower limit), Spitzer Space Telescope combined cryogenic surveys (red plus, unfilled), AllWISE
R90 selection (red left-facing triangle, unfilled), and AllWISE C75 selection (red right-facing triangle, unfilled). The considered Euclid detectable
AGN have ≥5σ detection in at least one Euclid band. The plotted values and references for each survey are given in Table 7.
AGN detectable in the HE band can exhibit 2–10 keV X-
ray fluxes ranging from 8.8× 10−12 to 2.5× 10−17 erg s−1 cm−2.
Therefore, the AGN population detectable with Euclid
photometry will, in part, reside beyond the X-ray flux lim-
its of all the X-ray surveys considered here. Equally, there is
also a portion of parameter space where a population of AGN
detectable in modern X-ray surveys remain undetectable in the
Euclid HE band. We therefore expect Euclid to probe a dif-
ferent population of AGN to those selected purely from X-ray
surveys, similarly to what is observed with MIR AGN selec-
tion (e.g. Eckart et al. 2010). It is likely that AGN detected
by Euclid but not in X-ray surveys are either low-luminosity
(log10[L2−10 keV/erg s−1] ∼ 42−43) AGN at intermediate (z ∼ 1)
to high redshifts (z > 3), or AGN that are absorbed in the X-
rays. The smaller converse population observable in X-ray sur-
veys but not with Euclid are likely to be largely made up of
AGN that are under-luminous compared to their host galaxies
(e.g. Mendez et al. 2013).
To the flux limit of the deepest X-ray survey considered here
(XMM-COSMOS; F2−10 keV = 9.3 × 10−15 erg s−1 cm−2, 90%
completeness Cappelluti et al. 2009), we assessed the number
of possible X-ray counterparts for our Euclid observable AGN.
We forecast 5.8× 106 (1.8× 104) AGN, corresponding to 15%
(7.6%) of the total detectable population in the EWS (EDS)
will exhibit X-ray fluxes that could be detected in the XMM-
COSMOS survey. Of these AGN, 2.6× 106 (7.9× 103) are unob-
scured and 3.2× 106 (1.0× 104) are obscured in the EWS (EDS).
This highlights the difference in AGN population that will be
probed by Euclid compared to those selected in the X-ray regime
as up to 85% (92.4%) of EWS (EDS) Euclid-detected AGN
would not be detectable in the medium-depth XMM-COSMOS
survey. The lower X-ray detection rate of the EDS sample stems
from the fainter NIR magnitudes probed corresponding to lower
luminosity AGN at the faintest X-ray fluxes.
6. Summary
In this work we made forecasts of the observational expec-
tations for z < 7 AGN in the EWS and EDS. Starting
from the Fotopoulou et al. (2016a) 5–10 keV XLF we gener-
ated volume-limited samples of the statistically expected AGN
in the Euclid survey footprints (Sect. 3.2). Our samples cover
the redshift range 0.01 ≤ z ≤ 7 and bolometric luminosity
range 43 ≤ log10(Lbol/erg s−1) ≤ 47, corresponding to 41.8 ≤
log10(L2−10 keV/erg s−1) ≤ 46.3, or −29.0 ≤ M1450 ≤ −17.2.
Each AGN in our sample was assigned an SED based on
its X-ray luminosity and redshift (Sect. 3.3). As the observed
5–10 keV XLF considers obscured and unobscured AGN in an
unbiased fashion up to NH ∼ 1023 cm−2, we employed the opti-
cally obscured AGN fraction evolution model of Merloni et al.
(2014) to assign each AGN as optically obscured or unob-
scured. Unobscured AGN were assigned the mean quasar SED
collated in Shen et al. (2020) with αox values sampled from
the empirical distribution determined in Lusso et al. (2010). For
obscured AGN we leveraged XXL AGN SED fitting results
(Fotopoulou et al. 2016b) to probabilistically allocate an empir-
ical SED template class based on X-ray luminosity and redshift
(Sect. 3.3.3). Finally, we applied dust extinction to each AGN
SED, sampling from obscured and unobscured AGN E(B − V)
distributions derived from XXL survey AGN (Fotopoulou et al.
2016b), as well as IGM extinction (Madau 1995, Sect. 3.3.4).
Through the probabilistic assignment of optical obscuration
class, E(B − V) values, αox in unobscured AGN, and SED tem-
plate class in obscured AGN, we ensured empirically driven
A250, page 22 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
18 16 14 12 10
log10 (F2 10keV / ergs 1 cm 2)
12.5
15.0
17.5
20.0
22.5
25.0
27.5
30.0
H
E
log10 (FX/Fopt) = 1
log10 (FX/Fopt) = + 1
XMM-COSMOS
XMM-SERVS (W-CDF-S)
XXL-3XLSS
XXL-1000-AGN
HE limiting magnitude
100
101
102
103
104
105
N
AG
N
Fig. 15. Density of EWS AGN in the 2–10 keV X-ray flux vs HE
observed magnitude parameter space. The Euclid detectable AGN are
shown in purple, whilst undetectable AGN are represented in grey. The
AGN population detectable with Euclid photometry will, in part, reside
beyond the X-ray flux limits of current X-ray surveys. There is also
a portion of parameter space where a population of AGN detectable
in modern X-ray surveys remain undetectable in the Euclid HE band.
X-ray flux limits for the XMM-COSMOS (black; 90% completeness,
F2−10 keV = 9.3 × 10−15 erg s−1 cm−2; Cappelluti et al. 2009), XXL-
3XLSS (orange; 90% completeness, F2−10 keV = 3.8×10−14 erg s−1 cm−2;
Chiappetti et al. 2018), the XMM-SERVS W-CDF-S field (blue; 90%
of total area, F2−10 keV = 2.9 × 10−14 erg s−1 cm−2; Ni et al. 2021) sur-
veys, and the XXL 1000 brightest AGN sample (XXL-1000-AGN) as
used in Fotopoulou et al. (2016b) for SED fitting (red; sample flux limit,
F2−10 keV = 4.8 × 10−14 erg s−1 cm−2; Fotopoulou et al. 2016b) are plot-
ted for reference. The dotted black lines represent log10(FX/Fopt) = ±1.
The majority of AGN should lie between these lines.
diversity in the resultant photometry of our AGN, even at similar
luminosities and redshifts.
Once assigned and scaled, we performed mock observa-
tions of each AGN SED in our sample, convolving with the
Euclid bands and an assortment of ancillary photometric bands
(Sect. 3.4). We utilised the resulting photometric catalogue to
investigate the observable population of z < 7 AGN in the Euclid
surveys.
Our main findings are summarised as follows.
1. We estimate 4.0× 107 AGN will have a ≥5σ detection in at
least one Euclid band in the EWS. Of these AGN 31% are
unobscured and 69% are obscured. Our predicted yield cor-
responds to a detectable AGN surface density in the EWS
of 2.8× 103 deg−2. In the EDS we expect a ≥5σ detection
in at least one Euclid band for 2.4× 105 AGN, of which
21% are unobscured AGN and 79% are obscured AGN. A
detectable AGN surface density of 4.7× 103 deg−2 is found
for the EDS. Full four-band ≥5σ Euclid photometry cov-
erage will be available for 2.1× 107 AGN in the EWS and
1.6× 105 AGN in the EDS. This is equivalent to AGN sur-
face densities of 1.4× 103 deg−2 and 3.1× 103 deg−2 for the
EWS and EDS, respectively.
2. The dominant source of uncertainty in our analysis is from
uncertainties in the XLF. We quantify the relative uncertainty
on our numbers of Euclid-detectable AGN to correspond to
6.7% for the EWS and 12.5% for the EDS. In the redshift
regimes z ≤ 1, 1 < z ≤ 4, and z > 4 the relative uncertain-
ties due to the XLF correspond respectively to 6.1%, 7.1%,
and 28.8% for the EWS and 7.2%, 12.6%, and 37.5% for the
EDS. The disparity in relative uncertainties for the two sur-
veys is due to the EWS probing the more tightly constrained
bright end of the XLF, while the EDS is able to probe the
lesser constrained faint end of the XLF.
3. Using the Euclid colour selection criteria derived in
Euclid Collaboration: Bisigello et al. (2024), we obtain
expectations on the number of AGN we will select with
Euclid data. Employing Euclid bands only we select
4.8× 106 (331 deg−2) AGN in the EWS, comprising 92%
unobscured AGN and 8% obscured AGN. In the EDS
we select a sample of 1.7× 104 (346 deg−2) AGN, which
consists of 97% unobscured AGN and 3% obscured AGN.
4. Ancillary ugrz photometric bands from Rubin/LSST
improve the completeness, purity, and size of colour-selected
AGN samples. Including these selection criteria we select
a total of 8.1 × 106 (556 deg−2) and 3.5 × 104 (692 deg−2)
AGN in the EWS and EDS, respectively. The EWS total
selected AGN sample consists of 75% unobscured AGN and
25% obscured AGN, while the EDS total selected AGN sam-
ple consists of 71% unobscured AGN and 29% obscured
AGN. These samples represent a yield of 20% and 15% of
the EWS and EDS samples of AGN with a ≥ 5σ detec-
tion in at least one Euclid band. The total expected sam-
ple of colour-selected AGN across both Euclid surveys con-
tains 6.0× 106 (74%) unobscured AGN and 2.1× 106 (26%)
obscured AGN, covering 0.02 ≤ z . 5.2 and 43 ≤
log10(Lbol/erg s
−1) ≤ 47.
5. The predicted surface densities of Euclid selected AGN are
comparable to those derived from other modern wide-field
and medium-area surveys, across a range of wavebands. Our
EWS yield is most comparable to the WISE C75 AGN selec-
tion when considering samples selected with Euclid photom-
etry alone and including ancillary photometric bands, yield-
ing a slightly lower surface density. Our EDS selected sur-
face density considering Euclid bands alone is comparable
to the yield of the combined Spitzer cryogenic surveys. Con-
sidering Euclid and ancillary optical bands the EDS selected
AGN surface density is marginally greater than that of the
XXL-3XLSS survey, which is of similar area.
6. We project that 5.8× 106 (1.8× 104) AGN, corresponding to
15% (7.6%) of the total Euclid detectable population in the
EWS (EDS) will exhibit X-ray fluxes that could be detected
in the XMM-COSMOS survey. Therefore, we assess that up
to 85% (92.4%) of EWS (EDS) Euclid-detected AGN would
not be detectable in modern medium-depth X-ray surveys.
We expect Euclid to yield a sizeable statistical sample of
AGN, in the order of tens of millions detected AGN and millions
of selected AGN. The deep optical to NIR magnitudes, high spa-
tial resolution, and large areas interrogated by the Euclid surveys
will enable a diverse range of scientific studies with AGN to be
completed in the coming years.
Acknowledgements. We warmly thank the anonymous referee for their detailed
comments and excellent suggestions on the manuscript, improving the qual-
ity of the work. We also extend our thanks to J. T. Schindler and R. Gilli for
their helpful comments and discussion on the manuscript. The Euclid Consor-
tium acknowledges the European Space Agency and a number of agencies and
institutes that have supported the development of Euclid, in particular the Agen-
zia Spaziale Italiana, the Austrian Forschungsförderungsgesellschaft funded
through BMK, the Belgian Science Policy, the Canadian Euclid Consortium,
the Deutsches Zentrum für Luft- und Raumfahrt, the DTU Space and the Niels
Bohr Institute in Denmark, the French Centre National d’Etudes Spatiales, the
A250, page 23 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Fundação para a Ciência e a Tecnologia, the Hungarian Academy of Sciences,
the Ministerio de Ciencia, Innovación y Universidades, the National Aeronau-
tics and Space Administration, the National Astronomical Observatory of Japan,
the Netherlandse Onderzoekschool Voor Astronomie, the Norwegian Space
Agency, the Research Council of Finland, the Romanian Space Agency, the State
Secretariat for Education, Research, and Innovation (SERI) at the Swiss Space
Office (SSO), and the United Kingdom Space Agency. A complete and detailed
list is available on the Euclid web site (http://www.euclid-ec.org). This
work is supported by the UKRI AIMLAC CDT, funded by grant EP/S023992/1.
This work has benefited from the support of Royal Society Research Grant
RGS\R1\231450. V. A., L. B., A. B., G. C., E. L., F. L. F., M. M., F. R. acknowl-
edge the support from the INAF Large Grant “AGN & Euclid: a close entan-
glement” Ob. Fu. 01.05.23.01.14 VA acknowledges support from INAF-PRIN
1.05.01.85.08 and INAF Large Grant 2023 “AGN and Euclid: a close entan-
glement”, Ob. Fu. 1.05.23.01.14. A. F. acknowledges the support from project
“VLT-MOONS” CRAM 1.05.03.07, INAF Large Grant 2022 “The metal circle:
a new sharp view of the baryon cycle up to Cosmic Dawn with the latest genera-
tion IFU facilities” and INAF Large Grant 2022 “Dual and binary SMBH in the
multi-messenger era”. F. R. and F. L. F. acknowledge support from PRIN 2017
“Black hole winds and the baryon life cycle of galaxies: the stone-guest at the
galaxy evolution supper” contract #2017PH3WAT.
References
Abbott, T. M. C., Abdalla, F. B., Allam, S., et al. 2018, ApJS, 239, 18
Adams, N. J., Bowler, R. A. A., Jarvis, M. J., Varadaraj, R. G., & Häußler, B.
2023, MNRAS, 523, 327
Aird, J., Coil, A. L., Georgakakis, A., et al. 2015, MNRAS, 451, 1892
Ajello, M., Alexander, D. M., Greiner, J., et al. 2012, ApJ, 749, 21
Alonso-Herrero, A., Pérez-González, P. G., Alexander, D. M., et al. 2006, ApJ,
640, 167
Assef, R. J., Stern, D., Kochanek, C. S., et al. 2013, ApJ, 772, 26
Assef, R. J., Stern, D., Noirot, G., et al. 2018, ApJS, 234, 23
Bañados, E., Venemans, B. P., Decarli, R., et al. 2016, ApJS, 227, 11
Bañados, E., Schindler, J.-T., Venemans, B. P., et al. 2023, ApJS, 265, 29
Barlow-Hall, C. L., Delaney, J., Aird, J., et al. 2023, MNRAS, 519, 6055
Boller, T., Freyberg, M. J., Trümper, J., et al. 2016, A&A, 588, A103
Boutsia, K., Grazian, A., Giallongo, E., Fiore, F., & Civano, F. 2018, ApJ, 869,
20
Bower, R. G., Benson, A. J., Malbon, R., et al. 2006, MNRAS, 370, 645
Boyle, B. J., Fong, R., Shanks, T., & Peterson, B. A. 1987, MNRAS, 227, 717
Boyle, B. J., Shanks, T., Croom, S. M., et al. 2000, MNRAS, 317, 1014
Brunner, H., Cappelluti, N., Hasinger, G., et al. 2008, A&A, 479, 283
Buchner, J., Georgakakis, A., Nandra, K., et al. 2015, ApJ, 802, 89
Burlon, D., Ajello, M., Greiner, J., et al. 2011, ApJ, 728, 58
Burtscher, L., Davies, R. I., Graciá-Carpio, J., et al. 2016, A&A, 586, A28
Calderone, G., Nicastro, L., Ghisellini, G., et al. 2017, MNRAS, 472, 4051
Cappelluti, N., Brusa, M., Hasinger, G., et al. 2009, A&A, 497, 635
Chen, C. T. J., Brandt, W. N., Luo, B., et al. 2018, MNRAS, 478, 2132
Chiappetti, L., Fotopoulou, S., Lidman, C., et al. 2018, A&A, 620, A12
Civano, F., Hickox, R. C., Puccetti, S., et al. 2015, ApJ, 808, 185
Cowie, L. L., Songaila, A., Hu, E. M., & Cohen, J. G. 1996, AJ, 112, 839
Croton, D. J., Springel, V., White, S. D. M., et al. 2006, MNRAS, 365, 11
Dai, X., Assef, R. J., Kochanek, C. S., et al. 2009, ApJ, 697, 506
Della Ceca, R., Maccacaro, T., Caccianiga, A., et al. 2004, A&A, 428, 383
Della Ceca, R., Caccianiga, A., Severgnini, P., et al. 2008, A&A, 487, 119
Dey, A., Schlegel, D. J., Lang, D., et al. 2019, AJ, 157, 168
Donley, J. L., Koekemoer, A. M., Brusa, M., et al. 2012, ApJ, 748, 142
Dubois, Y., Gavazzi, R., Peirani, S., & Silk, J. 2013, MNRAS, 433, 3297
Dunlop, J. S., & Peacock, J. A. 1990, MNRAS, 247, 19
Duras, F., Bongiorno, A., Ricci, F., et al. 2020, A&A, 636, A73
Eckart, M. E., McGreer, I. D., Stern, D., Harrison, F. A., & Helfand, D. J. 2010,
ApJ, 708, 584
Elvis, M., Wilkes, B. J., McDowell, J. C., et al. 1994, ApJS, 95, 1
Elvis, M., Civano, F., Vignali, C., et al. 2009, ApJS, 184, 158
Euclid Collaboration (Barnett, R., et al.) 2019, A&A, 631, A85
Euclid Collaboration (Scaramella, R., et al.) 2022, A&A, 662, A112
Euclid Collaboration (Schirmer, M., et al.) 2022, A&A, 662, A92
Euclid Collaboration (Bisigello, L., et al.) 2024, A&A, 691, A1
Euclid Collaboration (Lusso, E., et al.) 2024, A&A, 685, A108
Euclid Collaboration (Cropper, M. S., et al.) 2025, A&A, in press, https://
doi.org/10.1051/0004-6361/202450996
Euclid Collaboration (Jahnke, K., et al.) 2025, A&A, in press, https://doi.
org/10.1051/0004-6361/202450786
Euclid Collaboration (Mellier, Y., et al.) 2025, A&A, in press, https://doi.
org/10.1051/0004-6361/202450810
Fabian, A. C. 2012, ARA&A, 50, 455
Fadda, D., Marleau, F. R., Storrie-Lombardi, L. J., et al. 2006, AJ, 131, 2859
Fan, X. 1999, AJ, 117, 2528
Fan, X., Bañados, E., & Simcoe, R. A. 2023, ARA&A, 61, 373
Fazio, G. G., Hora, J. L., Allen, L. E., et al. 2004, ApJS, 154, 10
Ferrarese, L., & Merritt, D. 2000, ApJ, 539, L9
Fiore, F., Grazian, A., Santini, P., et al. 2008, ApJ, 672, 94
Fotopoulou, S., & Paltani, S. 2018, A&A, 619, A14
Fotopoulou, S., Buchner, J., Georgantopoulos, I., et al. 2016a, A&A, 587, A142
Fotopoulou, S., Pacaud, F., Paltani, S., et al. 2016b, A&A, 592, A5
Gabor, J. M., Davé, R., Finlator, K., & Oppenheimer, B. D. 2010, MNRAS, 407,
749
Galametz, A., Saglia, R., Paltani, S., Apostolakos, N., & Dubath, P. 2017, A&A,
598, A20
Gardner, J. P., Mather, J. C., Clampin, M., et al. 2006, Space Sci. Rev., 123, 485
Giallongo, E., Grazian, A., Fiore, F., et al. 2015, A&A, 578, A83
Gilli, R., Comastri, A., & Hasinger, G. 2007, A&A, 463, 79
Gilli, R., Norman, C., Calura, F., et al. 2022, A&A, 666, A17
Glikman, E., Lacy, M., LaMassa, S., et al. 2018, ApJ, 861, 37
Granato, G. L., De Zotti, G., Silva, L., Bressan, A., & Danese, L. 2004, ApJ,
600, 580
Gruppioni, C., Pozzi, F., Rodighiero, G., et al. 2013, MNRAS, 432, 23
Gültekin, K., Richstone, D. O., Gebhardt, K., et al. 2009, ApJ, 698, 198
Harikane, Y., Zhang, Y., Nakajima, K., et al. 2023, ApJ, 959, 39
Hasinger, G. 2008, A&A, 490, 905
Hasinger, G., Miyaji, T., & Schmidt, M. 2005, A&A, 441, 417
Hewett, P. C., Warren, S. J., Leggett, S. K., & Hodgkin, S. T. 2006, MNRAS,
367, 454
Hickox, R. C., Myers, A. D., Greene, J. E., et al. 2017, ApJ, 849, 53
Hopkins, P. F., Strauss, M. A., Hall, P. B., et al. 2004, AJ, 128, 1112
Hopkins, P. F., Richards, G. T., & Hernquist, L. 2007, ApJ, 654, 731
Hopkins, P. F., Hernquist, L., Cox, T. J., & Kereš, D. 2008, ApJS, 175, 356
Ibata, R. A., McConnachie, A., Cuillandre, J.-C., et al. 2017, ApJ, 848, 128
Ilbert, O., Capak, P., Salvato, M., et al. 2009, ApJ, 690, 1236
Inoue, A. K., Shimizu, I., Iwata, I., & Tanaka, M. 2014, MNRAS, 442, 1805
Ivezic´, Ž., Kahn, S. M., Tyson, J. A., et al. 2019, ApJ, 873, 111
Jiang, L., McGreer, I. D., Fan, X., et al. 2016, ApJ, 833, 222
Johnson, B. D. 2019, Astrophysics Source Code Library [record ascl:1905.026]
Just, D. W., Brandt, W. N., Shemmer, O., et al. 2007, ApJ, 665, 1004
Kocevski, D. D., Onoue, M., Inayoshi, K., et al. 2023, ApJ, 954, L4
Kochanek, C. S., Eisenstein, D. J., Cool, R. J., et al. 2012, ApJS, 200, 8
Krawczyk, C. M., Richards, G. T., Mehta, S. S., et al. 2013, ApJS, 206, 4
Kulkarni, G., Worseck, G., & Hennawi, J. F. 2019, MNRAS, 488, 1035
Lacy, M., & Sajina, A. 2020, Nat. Astron., 4, 352
Lacy, M., Wilson, G., Masci, F., et al. 2005, ApJS, 161, 41
Lacy, M., Ridgway, S. E., Sajina, A., et al. 2015, ApJ, 802, 102
La Franca, F., Fiore, F., Comastri, A., et al. 2005, ApJ, 635, 864
Laloux, B., Georgakakis, A., Andonie, C., et al. 2023, MNRAS, 518, 2546
Laureijs, R., Amiaux, J., Arduini, S., et al. 2011, arXiv e-prints
[arXiv:1110.3193]
Lawrence, A. 1991, MNRAS, 252, 586
Liu, T., Buchner, J., Nandra, K., et al. 2022, A&A, 661, A5
Logan, C. H. A., & Fotopoulou, S. 2020, A&A, 633, A154
Lonsdale, C. J., Smith, H. E., Rowan-Robinson, M., et al. 2003, PASP, 115, 897
Luo, B., Brandt, W. N., Xue, Y. Q., et al. 2017, ApJS, 228, 2
Lusso, E., Comastri, A., Vignali, C., et al. 2010, A&A, 512, A34
Lusso, E., Worseck, G., Hennawi, J. F., et al. 2015, MNRAS, 449, 4204
Lyke, B. W., Higley, A. N., McLane, J. N., et al. 2020, ApJS, 250, 8
Madau, P. 1995, ApJ, 441, 18
Magorrian, J., Tremaine, S., Richstone, D., et al. 1998, AJ, 115, 2285
Maiolino, R., Scholtz, J., Curtis-Lake, E., et al. 2024, A&A, 691, A145
Malizia, A., Bassani, L., Bazzano, A., et al. 2012, MNRAS, 426, 1750
Marchesi, S., Ajello, M., Marcotulli, L., et al. 2018, ApJ, 854, 49
Marulli, F., Bonoli, S., Branchini, E., Moscardini, L., & Springel, V. 2008,
MNRAS, 385, 1846
Mateos, S., Alonso-Herrero, A., Carrera, F. J., et al. 2012, MNRAS, 426, 3271
Matsuoka, Y., Strauss, M. A., Kashikawa, N., et al. 2018, ApJ, 869, 150
Matsuoka, Y., Onoue, M., Iwasawa, K., et al. 2023, ApJ, 949, L42
Matthee, J., Naidu, R. P., Brammer, G., et al. 2024, ApJ, 963, 129
McCracken, H. J., Milvang-Jensen, B., Dunlop, J., et al. 2012, A&A, 544, A156
Meiksin, A. 2006, MNRAS, 365, 807
Mendez, A. J., Coil, A. L., Aird, J., et al. 2013, ApJ, 770, 40
Merloni, A., Predehl, P., Becker, W., et al. 2012, arXiv e-prints
[arXiv:1209.3114]
Merloni, A., Bongiorno, A., Brusa, M., et al. 2014, MNRAS, 437, 3550
Miyaji, T., Hasinger, G., & Schmidt, M. 2000, A&A, 353, 25
Morganson, E., Gruendl, R. A., Menanteau, F., et al. 2018, PASP, 130, 074501
Morrissey, P., Conrow, T., Barlow, T. A., et al. 2007, ApJS, 173, 682
Nandra, K., & Pounds, K. A. 1994, MNRAS, 268, 405
A250, page 24 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Nandra, K., Laird, E. S., Aird, J. A., et al. 2015, ApJS, 220, 10
Natali, F., Giallongo, E., Cristiani, S., & La Franca, F. 1998, AJ, 115, 397
Ni, Q., Brandt, W. N., Chen, C.-T., et al. 2021, ApJS, 256, 21
Oke, J. B., & Gunn, J. E. 1983, ApJ, 266, 713
Page, K. L., Reeves, J. N., O’Brien, P. T., & Turner, M. J. L. 2005, MNRAS, 364,
195
Peca, A., Cappelluti, N., Urry, C. M., et al. 2023, ApJ, 943, 162
Pei, Y. C. 1995, ApJ, 438, 623
Piconcelli, E., Jimenez-Bailón, E., Guainazzi, M., et al. 2005, A&A, 432, 15
Pierre, M., Pacaud, F., Adami, C., et al. 2016, A&A, 592, A1
Piotrowska, J. M., Bluck, A. F. L., Maiolino, R., & Peng, Y. 2022, MNRAS, 512,
1052
Polletta, M., Tajer, M., Maraschi, L., et al. 2007, ApJ, 663, 81
Pouliasis, E., Ruiz, A., Georgantopoulos, I., et al. 2024, A&A, 685, A97
Prevot, M. L., Lequeux, J., Maurice, E., Prevot, L., & Rocca-Volmerange, B.
1984, A&A, 132, 389
Prochaska, J. X., Madau, P., O’Meara, J. M., & Fumagalli, M. 2014, MNRAS,
438, 476
Ranalli, P., Comastri, A., Vignali, C., et al. 2013, A&A, 555, A42
Reeves, J. N., & Turner, M. J. L. 2000, MNRAS, 316, 234
Ricci, F., Marchesi, S., Shankar, F., La Franca, F., & Civano, F. 2017, MNRAS,
465, 1915
Richards, G. T., Strauss, M. A., Fan, X., et al. 2006a, AJ, 131, 2766
Richards, G. T., Lacy, M., Storrie-Lombardi, L. J., et al. 2006b, ApJS, 166, 470
Rosito, M. S., Pedrosa, S. E., Tissera, P. B., et al. 2021, A&A, 652, A44
Runnoe, J. C., Brotherton, M. S., & Shang, Z. 2012, MNRAS, 426, 2677
Salvato, M., Hasinger, G., Ilbert, O., et al. 2009, ApJ, 690, 1250
Salvato, M., Wolf, J., Dwelly, T., et al. 2022, A&A, 661, A3
Sanders, D. B., Salvato, M., Aussel, H., et al. 2007, ApJS, 172, 86
Schaye, J., Crain, R. A., Bower, R. G., et al. 2015, MNRAS, 446, 521
Schindler, J.-T., Fan, X., McGreer, I. D., et al. 2019, ApJ, 871, 258
Schindler, J.-T., Bañados, E., Connor, T., et al. 2023, ApJ, 943, 67
Schmidt, M. 1968, ApJ, 151, 393
Scoville, N., Aussel, H., Brusa, M., et al. 2007, ApJS, 172, 1
Selwood, M., Calderone, G., Fotopoulou, S., & Bremer, M. N. 2023, MNRAS,
518, 130
Shankar, F., Lapi, A., Salucci, P., De Zotti, G., & Danese, L. 2006, ApJ, 643, 14
Shen, X., Hopkins, P. F., Faucher-Giguère, C.-A., et al. 2020, MNRAS, 495,
3252
Signorini, M., Marchesi, S., Gilli, R., et al. 2023, A&A, 676, A49
Sijacki, D., Springel, V., Di Matteo, T., & Hernquist, L. 2007, MNRAS, 380, 877
Simpson, C. 2005, MNRAS, 360, 565
Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131, 1163
Spinoglio, L., Fernández-Ontiveros, J. A., & Malkan, M. A. 2022, ApJ, 941, 46
Steffen, A. T., Strateva, I., Brandt, W. N., et al. 2006, AJ, 131, 2826
Stern, D., Eisenhardt, P., Gorjian, V., et al. 2005, ApJ, 631, 163
Stern, D., Assef, R. J., Benford, D. J., et al. 2012, ApJ, 753, 30
Strauss, M. A., Fan, X., Gunn, J. E., et al. 1999, ApJ, 522, L61
Sutherland, W., Emerson, J., Dalton, G., et al. 2015, A&A, 575, A25
Tee, W. L., Fan, X., Wang, F., et al. 2023, ApJ, 956, 52
Telfer, R. C., Zheng, W., Kriss, G. A., & Davidsen, A. F. 2002, ApJ, 565, 773
Temple, M. J., Hewett, P. C., & Banerji, M. 2021, MNRAS, 508, 737
Thomas, R., Le Fèvre, O., Le Brun, V., et al. 2017, A&A, 597, A88
Treister, E., & Urry, C. M. 2006, ApJ, 652, L79
Treister, E., Urry, C. M., & Virani, S. 2009, ApJ, 696, 110
Truemper, J. 1982, Adv. Space Res., 2, 241
Ueda, Y., Akiyama, M., Ohta, K., & Miyaji, T. 2003, ApJ, 598, 886
Ueda, Y., Hiroi, K., Isobe, N., et al. 2011, PASJ, 63, S937
Ueda, Y., Akiyama, M., Hasinger, G., Miyaji, T., & Watson, M. G. 2014, ApJ,
786, 104
Vaccari, M., Marchetti, L., Franceschini, A., et al. 2010, A&A, 518, L20
Vanden Berk, D. E., Richards, G. T., Bauer, A., et al. 2001, AJ, 122, 549
Vijarnwannaluk, B., Akiyama, M., Schramm, M., et al. 2022, ApJ, 941, 97
Vito, F., Brandt, W. N., Yang, G., et al. 2018, MNRAS, 473, 2378
Wang, F., Yang, J., Fan, X., et al. 2019, ApJ, 884, 30
Ward, S. R., Harrison, C. M., Costa, T., & Mainieri, V. 2022, MNRAS, 514,
2936
Warren, S. J., Hambly, N. C., Dye, S., et al. 2007, MNRAS, 375, 213
Winkler, C., Courvoisier, T. J. L., Di Cocco, G., et al. 2003, A&A, 411, L1
Wolf, J., Nandra, K., Salvato, M., et al. 2021, A&A, 647, A5
Wright, E. L., Eisenhardt, P. R. M., Mainzer, A. K., et al. 2010, AJ, 140,
1868
Xue, Y. Q., Luo, B., Brandt, W. N., et al. 2011, ApJS, 195, 10
Yang, Q., & Shen, Y. 2023, ApJS, 264, 9
York, D. G., Adelman, J., Anderson, J. E., Jr., et al. 2000, AJ, 120, 1579
Young, M., Elvis, M., & Risaliti, G. 2009, ApJS, 183, 17
Zafar, T., Møller, P., Watson, D., et al. 2015, A&A, 584, A100
1 School of Physics, HH Wills Physics Laboratory, University of
Bristol, Tyndall Avenue, Bristol BS8 1TL, UK
2 INAF, Istituto di Radioastronomia, Via Piero Gobetti 101, 40129
Bologna, Italy
3 Dipartimento di Fisica e Astronomia “G. Galilei”, Università di
Padova, Via Marzolo 8, 35131 Padova, Italy
4 Department of Physics, Centre for Extragalactic Astronomy,
Durham University, South Road, Durham DH1 3LE, UK
5 Max-Planck-Institut für Astronomie, Königstuhl 17, 69117 Heidel-
berg, Germany
6 INAF-Osservatorio di Astrofisica e Scienza dello Spazio di
Bologna, Via Piero Gobetti 93/3, 40129 Bologna, Italy
7 Department of Physics and Astronomy, University of Southampton,
Southampton SO17 1BJ, UK
8 Jet Propulsion Laboratory, California Institute of Technology, 4800
Oak Grove Drive, Pasadena, CA 91109, USA
9 INAF-Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, 50125
Firenze, Italy
10 Dipartimento di Fisica e Astronomia, Università di Firenze, via G.
Sansone 1, 50019 Sesto Fiorentino Firenze, Italy
11 INAF-Istituto di Astrofisica e Planetologia Spaziali, via del Fosso
del Cavaliere, 100, 00100 Roma, Italy
12 INAF-Osservatorio Astronomico di Capodimonte, Via Moiariello
16, 80131 Napoli, Italy
13 Department of Mathematics and Physics, Roma Tre University, Via
della Vasca Navale 84, 00146 Rome, Italy
14 INAF-Osservatorio Astronomico di Roma, Via Frascati 33, 00078
Monteporzio Catone, Italy
15 Max Planck Institute for Extraterrestrial Physics, Giessenbachstr.
1, 85748 Garching, Germany
16 Jodrell Bank Centre for Astrophysics, Department of Physics and
Astronomy, University of Manchester, Oxford Road, Manchester
M13 9PL, UK
17 School of Mathematics and Physics, University of Surrey, Guild-
ford, Surrey GU2 7XH, UK
18 INAF-Osservatorio Astronomico di Brera, Via Brera 28, 20122
Milano, Italy
19 Dipartimento di Fisica e Astronomia, Università di Bologna, Via
Gobetti 93/2, 40129 Bologna, Italy
20 INFN-Sezione di Bologna, Viale Berti Pichat 6/2, 40127 Bologna,
Italy
21 Universitäts-Sternwarte München, Fakultät für Physik, Ludwig-
Maximilians-Universität München, Scheinerstrasse 1, 81679
München, Germany
22 INAF-Osservatorio Astrofisico di Torino, Via Osservatorio 20,
10025 Pino Torinese (TO), Italy
23 Dipartimento di Fisica, Università di Genova, Via Dodecaneso 33,
16146 Genova, Italy
24 INFN-Sezione di Genova, Via Dodecaneso 33, 16146 Genova, Italy
25 Department of Physics “E. Pancini”, University Federico II, Via
Cinthia 6, 80126 Napoli, Italy
26 INFN section of Naples, Via Cinthia 6, 80126 Napoli, Italy
27 Instituto de Astrofísica e Ciências do Espaço, Universidade do
Porto, CAUP, Rua das Estrelas, PT4150-762 Porto, Portugal
28 Dipartimento di Fisica, Università degli Studi di Torino, Via P.
Giuria 1, 10125 Torino, Italy
29 INFN-Sezione di Torino, Via P. Giuria 1, 10125 Torino, Italy
30 INAF-IASF Milano, Via Alfonso Corti 12, 20133 Milano, Italy
31 Centro de Investigaciones Energéticas, Medioambientales y Tec-
nológicas (CIEMAT), Avenida Complutense 40, 28040 Madrid,
Spain
32 Port d’Informació Científica, Campus UAB, C. Albareda s/n,
08193 Bellaterra (Barcelona), Spain
33 Institute for Theoretical Particle Physics and Cosmology (TTK),
RWTH Aachen University, 52056 Aachen, Germany
34 Dipartimento di Fisica e Astronomia “Augusto Righi” - Alma
Mater Studiorum Università di Bologna, Viale Berti Pichat 6/2,
40127 Bologna, Italy
A250, page 25 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
35 Institute for Astronomy, University of Edinburgh, Royal Observa-
tory, Blackford Hill, Edinburgh EH9 3HJ, UK
36 European Space Agency/ESRIN, Largo Galileo Galilei 1, 00044
Frascati, Roma, Italy
37 ESAC/ESA, Camino Bajo del Castillo, s/n., Urb. Villafranca del
Castillo, 28692 Villanueva de la Cañada, Madrid, Spain
38 Université Claude Bernard Lyon 1, CNRS/IN2P3, IP2I Lyon, UMR
5822, Villeurbanne F-69100, France
39 Institute of Physics, Laboratory of Astrophysics, Ecole Polytech-
nique Fédérale de Lausanne (EPFL), Observatoire de Sauverny,
1290 Versoix, Switzerland
40 UCB Lyon 1, CNRS/IN2P3, IUF, IP2I Lyon, 4 rue Enrico Fermi,
69622 Villeurbanne, France
41 Mullard Space Science Laboratory, University College London,
Holmbury St Mary, Dorking, Surrey RH5 6NT, UK
42 Departamento de Física, Faculdade de Ciências, Universidade de
Lisboa, Edifício C8, Campo Grande, PT1749-016 Lisboa, Portugal
43 Instituto de Astrofísica e Ciências do Espaço, Faculdade de Ciên-
cias, Universidade de Lisboa, Campo Grande, 1749-016 Lisboa,
Portugal
44 Department of Astronomy, University of Geneva, ch. d’Ecogia 16,
1290 Versoix, Switzerland
45 INFN-Padova, Via Marzolo 8, 35131 Padova, Italy
46 Université Paris-Saclay, Université Paris Cité, CEA, CNRS, AIM,
91191 Gif-sur-Yvette, France
47 INAF-Osservatorio Astronomico di Trieste, Via G. B. Tiepolo 11,
34143 Trieste, Italy
48 Istituto Nazionale di Fisica Nucleare, Sezione di Bologna, Via Irne-
rio 46, 40126 Bologna, Italy
49 INAF-Osservatorio Astronomico di Padova, Via dell’Osservatorio
5, 35122 Padova, Italy
50 Dipartimento di Fisica “Aldo Pontremoli”, Università degli Studi
di Milano, Via Celoria 16, 20133 Milano, Italy
51 Institute of Theoretical Astrophysics, University of Oslo, P.O. Box
1029, Blindern, 0315 Oslo, Norway
52 Leiden Observatory, Leiden University, Einsteinweg 55, 2333 CC
Leiden, The Netherlands
53 Higgs Centre for Theoretical Physics, School of Physics and
Astronomy, The University of Edinburgh, Edinburgh EH9 3FD,
UK
54 Department of Physics, Lancaster University, Lancaster LA1 4YB,
UK
55 Felix Hormuth Engineering, Goethestr. 17, 69181 Leimen, Ger-
many
56 Technical University of Denmark, Elektrovej 327, 2800 Kgs. Lyn-
gby, Denmark
57 Cosmic Dawn Center (DAWN), Copenhagen, Denmark
58 Institut d’Astrophysique de Paris, UMR 7095, CNRS, and Sor-
bonne Université, 98 bis boulevard Arago, 75014 Paris, France
59 Department of Physics and Helsinki Institute of Physics, Gustaf
Hällströmin katu 2, University of Helsinki, Helsinki 00014, Finland
60 Aix-Marseille Université, CNRS/IN2P3, CPPM, Marseille, France
61 Université de Genève, Département de Physique Théorique and
Centre for Astroparticle Physics, 24 quai Ernest-Ansermet, CH-
1211 Genève 4, Switzerland
62 Department of Physics, P.O. Box 64, Helsinki 00014 University of
Helsinki, Finland
63 Helsinki Institute of Physics, Gustaf Hällströmin katu 2, University
of Helsinki, Helsinki, Finland
64 European Space Agency/ESTEC, Keplerlaan 1, 2201 AZ Noord-
wijk, The Netherlands
65 NOVA optical infrared instrumentation group at ASTRON, Oude
Hoogeveensedijk 4, 7991 PD Dwingeloo, The Netherlands
66 INFN-Sezione di Milano, Via Celoria 16, 20133 Milano, Italy
67 Universität Bonn, Argelander-Institut für Astronomie, Auf dem
Hügel 71, 53121 Bonn, Germany
68 Aix-Marseille Université, CNRS, CNES, LAM, Marseille, France
69 Dipartimento di Fisica e Astronomia “Augusto Righi” - Alma
Mater Studiorum Università di Bologna, via Piero Gobetti 93/2,
40129 Bologna, Italy
70 Université Paris Cité, CNRS, Astroparticule et Cosmologie, 75013
Paris, France
71 University of Applied Sciences and Arts of Northwestern Switzer-
land, School of Engineering, 5210 Windisch, Switzerland
72 Institut d’Astrophysique de Paris, 98bis Boulevard Arago, 75014
Paris, France
73 IFPU, Institute for Fundamental Physics of the Universe, via Beirut
2, 34151 Trieste, Italy
74 School of Mathematics, Statistics and Physics, Newcastle Univer-
sity, Herschel Building, Newcastle-upon-Tyne NE1 7RU, UK
75 Department of Physics, Institute for Computational Cosmology,
Durham University, South Road, Durham DH1 3LE, UK
76 Institut de Física d’Altes Energies (IFAE), The Barcelona Insti-
tute of Science and Technology, Campus UAB, 08193 Bellaterra
(Barcelona), Spain
77 Department of Physics and Astronomy, University of Aarhus, Ny
Munkegade 120, DK-8000 Aarhus C, Denmark
78 Waterloo Centre for Astrophysics, University of Waterloo, Water-
loo, Ontario N2L 3G1, Canada
79 Department of Physics and Astronomy, University of Waterloo,
Waterloo, Ontario N2L 3G1, Canada
80 Perimeter Institute for Theoretical Physics, Waterloo, Ontario N2L
2Y5, Canada
81 Space Science Data Center, Italian Space Agency, via del Politec-
nico snc, 00133 Roma, Italy
82 Centre National d’Etudes Spatiales – Centre spatial de Toulouse,
18 avenue Edouard Belin, 31401 Toulouse Cedex 9, France
83 Institute of Space Science, Str. Atomistilor, nr. 409 Ma˘gurele, Ilfov
077125, Romania
84 Instituto de Astrofísica de Canarias, Calle Vía Láctea s/n, 38204
San Cristóbal de La Laguna, Tenerife, Spain
85 Departamento de Astrofísica, Universidad de La Laguna, 38206 La
Laguna, Tenerife, Spain
86 Departamento de Física, FCFM, Universidad de Chile, Blanco
Encalada, 2008 Santiago, Chile
87 INFN-Sezione di Roma, Piazzale Aldo Moro, 2 - c/o Dipartimento
di Fisica, Edificio G. Marconi, 00185 Roma, Italy
88 Universität Innsbruck, Institut für Astro- und Teilchenphysik, Tech-
nikerstr. 25/8, 6020 Innsbruck, Austria
89 Institut d’Estudis Espacials de Catalunya (IEEC), Edifici RDIT,
Campus UPC, 08860 Castelldefels, Barcelona, Spain
90 Institute of Space Sciences (ICE, CSIC), Campus UAB, Carrer de
Can Magrans, s/n, 08193 Barcelona, Spain
91 Satlantis, University Science Park, Sede Bld, 48940 Leioa-Bilbao,
Spain
92 Infrared Processing and Analysis Center, California Institute of
Technology, Pasadena, CA 91125, USA
93 Instituto de Astrofísica e Ciências do Espaço, Faculdade de Ciên-
cias, Universidade de Lisboa, Tapada da Ajuda, 1349-018 Lisboa,
Portugal
94 Universidad Politécnica de Cartagena, Departamento de Elec-
trónica y Tecnología de Computadoras, Plaza del Hospital 1, 30202
Cartagena, Spain
95 Institut de Recherche en Astrophysique et Planétologie (IRAP),
Université de Toulouse, CNRS, UPS, CNES, 14 Av. Edouard Belin,
31400 Toulouse, France
96 INFN-Bologna, Via Irnerio 46, 40126 Bologna, Italy
97 Institut für Theoretische Physik, University of Heidelberg,
Philosophenweg 16, 69120 Heidelberg, Germany
98 Université St Joseph; Faculty of Sciences, Beirut, Lebanon
99 Junia, EPA department, 41 Bd Vauban, 59800 Lille, France
100 SISSA, International School for Advanced Studies, Via Bonomea
265, 34136 Trieste TS, Italy
101 INFN, Sezione di Trieste, Via Valerio 2, 34127 Trieste TS, Italy
A250, page 26 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
102 ICSC - Centro Nazionale di Ricerca in High Performance Comput-
ing, Big Data e Quantum Computing, Via Magnanelli 2, Bologna,
Italy
103 Instituto de Física Teórica UAM-CSIC, Campus de Cantoblanco,
28049 Madrid, Spain
104 CERCA/ISO, Department of Physics, Case Western Reserve Uni-
versity, 10900 Euclid Avenue, Cleveland, OH 44106, USA
105 Laboratoire Univers et Théorie, Observatoire de Paris, Université
PSL, Université Paris Cité, CNRS, 92190 Meudon, France
106 Dipartimento di Fisica e Scienze della Terra, Università degli Studi
di Ferrara, Via Giuseppe Saragat 1, 44122 Ferrara, Italy
107 Istituto Nazionale di Fisica Nucleare, Sezione di Ferrara, Via
Giuseppe Saragat 1, 44122 Ferrara, Italy
108 Université de Strasbourg, CNRS, Observatoire astronomique de
Strasbourg, UMR 7550, 67000 Strasbourg, France
109 Kavli Institute for the Physics and Mathematics of the Universe
(WPI), University of Tokyo, Kashiwa, Chiba 277-8583, Japan
110 Dipartimento di Fisica - Sezione di Astronomia, Università di Tri-
este, Via Tiepolo 11, 34131 Trieste, Italy
111 Minnesota Institute for Astrophysics, University of Minnesota, 116
Church St SE, Minneapolis, MN 55455, USA
112 Institute Lorentz, Leiden University, Niels Bohrweg 2, 2333 CA
Leiden, The Netherlands
113 Université Côte d’Azur, Observatoire de la Côte d’Azur, CNRS,
Laboratoire Lagrange, Bd de l’Observatoire, CS 34229, 06304
Nice cedex 4, France
114 Institute for Astronomy, University of Hawaii, 2680 Woodlawn
Drive, Honolulu, HI 96822, USA
115 Department of Physics & Astronomy, University of California
Irvine, Irvine, CA 92697, USA
116 Department of Astronomy & Physics and Institute for Compu-
tational Astrophysics, Saint Mary’s University, 923 Robie Street,
Halifax, Nova Scotia B3H 3C3, Canada
117 Departamento Física Aplicada, Universidad Politécnica de Carta-
gena, Campus Muralla del Mar, 30202 Cartagena, Murcia, Spain
118 Université Paris-Saclay, CNRS, Institut d’astrophysique spatiale,
91405 Orsay, France
119 Department of Physics, Oxford University, Keble Road, Oxford
OX1 3RH, UK
120 CEA Saclay, DFR/IRFU, Service d’Astrophysique, Bat. 709,
91191 Gif-sur-Yvette, France
121 Institute of Cosmology and Gravitation, University of Portsmouth,
Portsmouth PO1 3FX, UK
122 Department of Computer Science, Aalto University, PO Box
15400, Espoo FI-00 076, Finland
123 Ruhr University Bochum, Faculty of Physics and Astronomy,
Astronomical Institute (AIRUB), German Centre for Cosmological
Lensing (GCCL), 44780 Bochum, Germany
124 DARK, Niels Bohr Institute, University of Copenhagen, Jagtvej
155, 2200 Copenhagen, Denmark
125 Department of Physics and Astronomy, Vesilinnantie 5, University
of Turku, Turku 20014, Finland
126 Serco for European Space Agency (ESA), Camino bajo del
Castillo, s/n, Urbanizacion Villafranca del Castillo, Villanueva de
la Cañada, 28692 Madrid, Spain
127 ARC Centre of Excellence for Dark Matter Particle Physics, Mel-
bourne, Australia
128 Centre for Astrophysics & Supercomputing, Swinburne University
of Technology, Victoria 3122, Australia
129 School of Physics and Astronomy, Queen Mary University of Lon-
don, Mile End Road, London E1 4NS, UK
130 Department of Physics and Astronomy, University of the Western
Cape, Bellville, Cape Town 7535, South Africa
131 ICTP South American Institute for Fundamental Research, Insti-
tuto de Física Teórica, Universidade Estadual Paulista, São Paulo,
Brazil
132 Oskar Klein Centre for Cosmoparticle Physics, Department of
Physics, Stockholm University, Stockholm SE-106 91, Sweden
133 Astrophysics Group, Blackett Laboratory, Imperial College Lon-
don, London SW7 2AZ, UK
134 Univ. Grenoble Alpes, CNRS, Grenoble INP, LPSC-IN2P3, 53,
Avenue des Martyrs, 38000 Grenoble, France
135 Dipartimento di Fisica, Sapienza Università di Roma, Piazzale
Aldo Moro 2, 00185 Roma, Italy
136 Centro de Astrofísica da Universidade do Porto, Rua das Estrelas,
4150-762 Porto, Portugal
137 Institute of Astronomy, University of Cambridge, Madingley Road,
Cambridge CB3 0HA, UK
138 Department of Astrophysics, University of Zurich, Winterthur-
erstrasse 190, 8057 Zurich, Switzerland
139 Dipartimento di Fisica, Università degli studi di Genova, and
INFN-Sezione di Genova, via Dodecaneso 33, 16146 Genova, Italy
140 Theoretical astrophysics, Department of Physics and Astronomy,
Uppsala University, Box 515, 751 20 Uppsala, Sweden
141 Department of Physics, Royal Holloway, University of London,
Egham TW20 0EX, UK
142 Department of Astrophysical Sciences, Peyton Hall, Princeton Uni-
versity, Princeton, NJ 08544, USA
143 Niels Bohr Institute, University of Copenhagen, Jagtvej 128, 2200
Copenhagen, Denmark
144 Center for Cosmology and Particle Physics, Department of Physics,
New York University, New York, NY 10003, USA
145 Center for Computational Astrophysics, Flatiron Institute, 162 5th
Avenue, 10010 New York, NY, USA
146 Department of Astronomy, University of Massachusetts, Amherst,
MA 01003, USA
147 University of Trento, Via Sommarive 14, I-38123 Trento, Italy
A250, page 27 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Appendix A: Obscured AGN SED class
distributions
In Fig. A.1 we present the normalised and extrapolated
(to z = 7) redshift-space probability distribution for each
obscured AGN SED class in Fotopoulou et al. (2016b) assigned
in this work (Sect. 3.3.3). We separately present distribu-
tions for the high (log10[L2–10 keV/erg s
−1] ≥ 44) and low
(log10[L2–10 keV/erg s
−1] < 44) X-ray luminosity groups.
0.0
0.5
1.0
1.5
2.0
2.5
3.0 log10 (L2 10keV / ergs 1) < 44
0 1 2 3 4 5 6 7
0.0
0.5
1.0
1.5
2.0
2.5
log10 (L2 10keV / ergs 1) 44
PASS
SFG
SB
QSO2
SEY2
SB-AGN
0.0 0.2 0.4 0.6 0.8 1.0
z
0.0
0.2
0.4
0.6
0.8
1.0
N
or
m
al
is
ed
 p
ro
ba
bi
lit
y 
de
ns
it
y
Fig. A.1. Normalised and extrapolated (to z = 7) redshift-space proba-
bility distributions for each obscured AGN SED class assigned in this
work. Distributions are presented separately for the low (top) and high
(bottom) X-ray luminosity groups.
Appendix B: Ancillary photometry
In addition to the Euclid filter set presented in Table 2, we also
perform synthetic photometric observations with bands from
complimentary surveys covering UV–MIR. The addition of these
bands aid in our discussion and assessment of the observational
expectations for AGN with Euclid compared to existing and
upcoming surveys. The characteristics of each of the ancillary
bands utilised in our work are given in Table B.1.
Appendix C: Derived AGN colours
We validate the colours of AGN generated in this work by
showing that our derived colours are consistent with AGN
colour-colour diagrams from the literature (e.g. Stern et al. 2005;
Mateos et al. 2012; Fotopoulou & Paltani 2018). In each case we
test that our AGN occupy the expected positions on the diagrams
using our EDS data where we require a detection with S/N > 5
in at least one Euclid band.
In Fig. C.1 we plot our AGN on an adapted optical–NIR–
MIR colour-colour space used in Fotopoulou & Paltani (2018)
and Logan & Fotopoulou (2020). We substituted the Euclid YE
Table B.1. Characteristics of ancillary filters used for synthetic photo-
metric observations of AGN in this work.
Survey Filter λeff (µm) Reference
2MASS J 1.66 Skrutskie et al. (2006)
2MASS H 1.24 "
2MASS Ks 2.15 "
DES g 0.473 Morganson et al. (2018)
DES r 0.642 "
DES i 0.784 "
DES z 0.926 "
DES Y 1.01 "
GALEX FUV 0.154 Morrissey et al. (2007)
GALEX NUV 0.230 "
Rubin/LSST u 0.368 Ivezic´ et al. (2019)
Rubin/LSST g 0.478 "
Rubin/LSST r 0.622 "
Rubin/LSST i 0.753 "
Rubin/LSST z 0.869 "
Rubin/LSST Y 0.973 "
Spitzer IRAC1 3.53 Fazio et al. (2004)
Spitzer IRAC2 4.48 "
Spitzer IRAC3 5.70 "
Spitzer IRAC4 7.80 "
VISTA J 1.25 Sutherland et al. (2015)
VISTA H 1.65 "
VISTA Ks 2.15 "
WISE W1 3.35 Wright et al. (2010)
WISE W2 4.60 "
WISE W3 11.6 "
WISE W4 22.1 "
and JE bands in place of the VISTA Y and J bands and the
Rubin/LSST g band (gLSST) in place of the SDSS g band. This
colour-colour space is known to separate well population loci
of stars, galaxies and unobscured AGN. The unobscured AGN
are expected to fall in a locus on the blue side of the diagram
and galaxies (i.e. the obscured AGN classes in this work) in a
more extended locus on the red side of unobscured AGN (see
Fig. 4 in Fotopoulou & Paltani 2018). For clarity in the locations
of different SED shapes, we plot each SED class with separate
colours. We see that the different SED classes assigned to AGN
in this work fall into the expected positions in the colour space.
The unobscured AGN locus is contaminated by the SB-AGN and
SB SED classes, specifically AGN with z > 2 for the latter. This
is no surprise given the similarity of these SEDs at optical wave-
lengths (see Fig. 10). We note that a portion of the unobscured
AGN stray into the galaxy locus of the diagram. Akin to what
is reported in Logan & Fotopoulou (2020), these interlopers are
affected by extinction, either from intrinsic dust absorption with
E(B − V) & 0.2 or from IGM absorption affecting gLSST at
z & 3.4.
We plot the Spitzer/IRAC [3.6]−[4.5] and [5.8]−[8.0] Vega
system colours of our EDS AGN sample with z < 4 in Fig. C.2.
The black dashed lines in this plot denote the AGN selection cri-
terion identified by Stern et al. (2005), valid for redshifts z < 4.
We found that 58.2% of our total available AGN were identified
using this selection. By class we select 100% of the unobscured
AGN and 47.5% of the obscured AGN. This is expected as we
see that the majority of non-selected obscured AGN belong to
SED classes where the representative SED does not include a
strong AGN component at these wavelengths (i.e. PASS, SFG,
A250, page 28 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Fig. C.1. EDS AGN simulated in this work plotted on the optical-NIR-
MIR colour space used in Fotopoulou & Paltani (2018). We used the
Euclid YE and JE bands, the Rubin/LSST g band (gLSST), and the
WISE 3.4 µm (W1) band. Only AGN with S/N > 5 in at least one Euclid
band are plotted. Each of our AGN SED classes are plotted individ-
ually: Unobscured AGN (‘Unobsc.’; blue), Passive (‘PASS’; orange),
Star-forming (‘SFG’; green), Starburst (‘SB’; red), High-luminosity
obscured AGN (‘QSO2’; purple), Seyfert 2 (‘SEY2’; brown), and
Starburst-AGN composite (‘SB-AGN’; pink).
SB). Furthermore, it is reassuring that 100% of unobscured AGN
assigned with our mean quasar SED are selected using this
method. Such AGN identification methods should in theory be
well optimised towards including the ‘average’ quasar in the
resulting selected sample.
The three-band AGN colour selection criterion of
Mateos et al. (2012) utilises the WISE 3.4, 4.6 and 12 µm
bands (W1, W2, and W3, respectively). In Fig. C.3 we show
the W1-W2 and W2-W3 Vega system colours along with the
Mateos et al. (2012) three-band AGN colour selection criterion
in black dashed lines. Colours are plotted for our EDS AGN
sample with z < 2, the redshift range for which this selection is
valid. We found 37.1% of the total available AGN are identified
in this case. Again, 100% of the unobscured AGN are selected,
an outcome validating our mean quasar colours as discussed
above. With this selection criterion however, only 22.9% of
our obscured AGN are identified. The low identification rate of
obscured AGN in this case can be attributed to the MIR selection
scheme being optimised for high X-ray luminosity AGN that
have a power-law spectral shape in the MIR. The SED templates
that are well identified have the corresponding MIR spectral
shape. As we explore in Sect. 5.3, our Euclid detected AGN
probe to faint X-ray fluxes which will lead to incompleteness
in this selection. The overall positions of our different SED
classes in the colour space agree well with the redshift evolution
tracks for corresponding populations presented throughout
Mateos et al. (2012).
Fig. C.2. Spitzer/IRAC AGN selection of Stern et al. (2005) applied to
our EDS AGN. We plot only AGN with S/N > 5 in at least one Euclid
band with z < 4. Each of our AGN SED classes are plotted individually
with corresponding colours identical to Fig. C.1.
We conclude that the colours derived for AGN in this work
are consistent with observations.
Fig. C.3.WISE three-colour selection of Mateos et al. (2012) applied to
our EDS AGN. This selection uses the WISE 3.4, 4.6, and 12 µm bands
(W1, W2, and W3 respectively). We plot only AGN with S/N > 5 in at
least one Euclid band and with z < 2. Each of our AGN SED classes are
plotted individually with corresponding colours identical to Fig. C.1.
A250, page 29 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Appendix D: Detailed detectable AGN breakdown
We present detailed breakdowns of the numbers and surface den-
sities of AGN expected to be detectable with Euclid photome-
try in the EWS and EDS. This information was summarised in
Tables 4 and 5 in the main text, however here we give full break-
downs of the total numbers of AGN as well as splits by opti-
cal obscuration class (i.e. unobscured and obscured). In Table
D.1 we provide details of the numbers and surface densities of
detectable AGN in the EWS and EDS across the full redshift
range 0.01 ≤ z ≤ 7. Table D.2 presents the numbers of detectable
AGN in the EWS and EDS broken down into the redshift ranges
0.01 ≤ z ≤ 4 and 4 < z ≤ 7. The former of these ranges corre-
sponds to the region in which the XLF adopted in this work is
well constrained by data, whilst the latter marks regions where
the XLF was extrapolated.
A250, page 30 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Table D.1. Expected number and corresponding surface densities of AGN detectable with Euclid photometry in the EWS and EDS.
Survey Band Surface density (deg−2) Detectable AGN
Total Unobscured Obscured Total Unobscured Obscured
EWS IE 2.2× 103 835 1.3 × 103 3.1× 107 1.2 × 107 1.9 × 107
YE 1.5× 103 600 892 2.2× 107 8.7 × 106 1.3 × 107
JE 2.1× 103 665 1.4 × 103 3.0× 107 9.7 × 106 2.0 × 107
HE 2.4× 103 679 1.7 × 103 3.5× 107 9.9 × 106 2.5 × 107
(IE |YE | JE |HE) 2.8× 103 849 1.9 × 103 4.0× 107 1.2 × 107 2.8 × 107
(IE ∧YE ∧ JE ∧HE) 1.4× 103 581 842 2.1× 107 8.4 × 106 1.2 × 107
EDS IE 3.8× 103 973 2.8 × 103 1.9× 105 4.9 × 104 1.4 × 105
YE 3.2× 103 860 2.3 × 103 1.6× 105 4.4 × 104 1.2 × 105
JE 4.0× 103 899 3.1 × 103 2.0× 105 4.5 × 104 1.5 × 105
HE 4.5× 103 909 3.6 × 103 2.3× 105 4.5 × 104 1.8 × 105
(IE |YE | JE |HE) 4.7× 103 986 3.7 × 103 2.4× 105 4.9 × 104 1.9 × 105
(IE ∧YE ∧ JE ∧HE) 3.1× 103 849 2.3 × 103 1.6× 105 4.3 × 104 1.1 × 105
Notes. We present total numbers as well as splits by optical obscuration class. Numbers are reported on a per-filter basis as well as the number of
AGN detectable in at least one Euclid filter, (IE |YE | JE |HE), and the number of AGN detectable in all Euclid filters, (IE ∧YE ∧ JE ∧HE).
Table D.2. Expected number of AGN detectable with Euclid photometry in the EWS and EDS, broken down into the redshift ranges 0.01 ≤ z ≤ 4
and 4 < z ≤ 7.
Survey Band 0.01 ≤ z ≤ 4 4 < z ≤ 7
Total Unobscured Obscured Total Unobscured Obscured
EWS IE 3.0 × 107 1.1 × 107 1.9 × 107 1.1 × 106 7.5 × 105 3.5 × 105
YE 2.1 × 107 8.1 × 106 1.3 × 107 8.4 × 105 5.8 × 105 2.6 × 105
JE 2.9 × 107 8.9 × 106 2.0 × 107 1.0 × 106 7.3 × 105 3.1 × 105
HE 3.4 × 107 9.1 × 106 2.5 × 107 1.1 × 106 7.8 × 105 3.7 × 105
(IE |YE | JE |HE) 3.9 × 107 1.1 × 107 2.7 × 107 1.4 × 106 9.2 × 105 4.7 × 105
(IE ∧YE ∧ JE ∧HE) 2.0 × 107 7.9 × 106 1.2 × 107 7.2 × 105 5.1 × 105 2.2 × 105
EDS IE 1.8 × 105 4.5 × 104 1.4 × 105 6.8 × 103 3.3 × 103 3.4 × 103
YE 1.5 × 105 4.0 × 104 1.2 × 105 5.3 × 103 3.4 × 103 1.9 × 103
JE 1.9 × 105 4.1 × 104 1.5 × 105 6.3 × 103 3.7 × 103 2.6 × 103
HE 2.1 × 105 4.2 × 104 1.8 × 105 7.6 × 103 3.8 × 103 3.8 × 103
(IE |YE | JE |HE) 2.3 × 105 4.5 × 104 1.8 × 105 8.7 × 103 3.9 × 103 4.7 × 103
(IE ∧YE ∧ JE ∧HE) 1.5 × 105 3.9 × 104 1.1 × 105 4.8 × 103 3.1 × 103 1.7 × 103
Notes. The former of these ranges corresponds to where the XLF adopted in this work is well constrained by data, whilst the latter marks regions
where the XLF was extrapolated. We present total numbers as well as splits by optical obscuration class. Numbers are reported on a per-filter basis
as well as the number of AGN detectable in at least one Euclid filter, (IE |YE | JE |HE), and the number of AGN detectable in all Euclid filters,
(IE ∧YE ∧ JE ∧HE).
Appendix E: AGN colour selection criteria
Euclid Collaboration: Bisigello et al. (2024) derives a number
of different photometric selections for AGN in Euclid using
Euclid, Rubin/LSST and Spitzer Space Telescope/IRAC bands.
In this section we applied each criterion to our AGN sam-
ple and report the results. Throughout this section we use the
nomenclature of AND, OR corresponding to the logical AND, OR
operators.
Table E.1 gives the redshift-dependent performance
of Euclid photometry AGN selection criteria defined in
Euclid Collaboration: Bisigello et al. (2024) applied to our
samples of EWS and EDS AGN. Density plots showing the
Euclid colour AGN selection criteria discussed in Sect. 5.2
applied to our samples of EWS and EDS AGN are shown in
Figs. E.1 and E.2, respectively. In the top panels of Figs. E.1
and E.2 we additionally plot trails of the colour-space redshift
evolution of our unobscured AGN template with E(B − V) = 0
and E(B − V) = 0.1 for z ∈ [0, 7] in steps of δz = 1. These
lines illustrate how the position of our AGN on the considered
selection plots are influenced by SED and source parameters.
We observe that much of the spread of our population in the
top right portion of the diagram is occupied by high-redshift
sources on the right and more heavily optically obscured
sources on the left. Interestingly, we observe that the lowest
redshift unobscured AGN and mildly dust reddened unobscured
AGN would not be selected using these Euclid photometric
criteria.
In the EWS the optimal selection criteria for unobscured
AGN exploiting only Euclid photometry obeys
IE − YE < 0.5
AND IE − JE < 0.7
AND IE − JE < −2.1 (IE − YE) + 0.9,
(E.1)
A250, page 31 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
Table E.1. Redshift-dependent performance of Euclid photometry AGN selection criteria (Euclid Collaboration: Bisigello et al. 2024) applied to
our samples of EWS and EDS AGN.
Survey Target Class Photometry Redshift Selected AGN C log10(Lbol/erg s
−1)
EWS Unobscured AGN Euclid 0.0 ≤ z < 0.5 2.1 × 105 0.13 43.6 ± 0.6
0.5 ≤ z < 1.0 7.3 × 105 0.13 44.4 ± 0.7
1.0 ≤ z < 1.5 1.2 × 106 0.24 45.2 ± 0.6
1.5 ≤ z < 2.0 1.2 × 106 0.37 45.4 ± 0.5
2.0 ≤ z < 2.5 7.2 × 105 0.39 45.6 ± 0.5
2.5 ≤ z < 3.0 3.9 × 105 0.34 45.8 ± 0.4
3.0 ≤ z < 3.5 2.1 × 105 0.29 45.8 ± 0.4
3.5 ≤ z < 4.0 9.9 × 104 0.22 45.9 ± 0.4
4.0 ≤ z < 4.5 3.6 × 104 0.12 45.9 ± 0.3
4.5 ≤ z < 5.0 2.0 × 103 0.01 45.8 ± 0.2
5.0 ≤ z < 5.5 3 0.00 45.7 ± 0.1
EWS ∪ Euclid, Rubin/LSST 0.0 ≤ z < 0.5 4.3 × 105 0.26 43.6 ± 0.6
0.5 ≤ z < 1.0 1.7 × 106 0.28 44.1 ± 0.7
1.0 ≤ z < 1.5 2.1 × 106 0.38 44.9 ± 0.6
1.5 ≤ z < 2.0 2.0 × 106 0.56 45.3 ± 0.5
2.0 ≤ z < 2.5 1.1 × 106 0.53 45.5 ± 0.5
2.5 ≤ z < 3.0 5.2 × 105 0.44 45.7 ± 0.4
3.0 ≤ z < 3.5 2.1 × 105 0.29 45.8 ± 0.4
3.5 ≤ z < 4.0 9.9 × 104 0.22 45.9 ± 0.4
4.0 ≤ z < 4.5 3.6 × 104 0.12 45.9 ± 0.3
4.5 ≤ z < 5.0 2.0 × 103 0.01 45.8 ± 0.2
5.0 ≤ z < 5.5 3 0.00 45.7 ± 0.1
EDS Unobscured AGN Euclid 0.0 ≤ z < 0.5 3.0 × 102 0.04 43.5 ± 0.5
0.5 ≤ z < 1.0 2.4 × 103 0.06 44.0 ± 0.7
1.0 ≤ z < 1.5 5.5 × 103 0.13 44.7 ± 0.7
1.5 ≤ z < 2.0 4.5 × 103 0.17 45.2 ± 0.7
2.0 ≤ z < 2.5 2.5 × 103 0.17 45.4 ± 0.6
2.5 ≤ z < 3.0 1.2 × 103 0.14 45.6 ± 0.4
3.0 ≤ z < 3.5 6.5 × 102 0.13 45.6 ± 0.4
3.5 ≤ z < 4.0 2.2 × 102 0.08 45.6 ± 0.4
4.0 ≤ z < 4.5 78 0.04 45.9 ± 0.3
EDS ∪ Euclid, Rubin/LSST 0.0 ≤ z < 0.5 1.5 × 103 0.23 43.5 ± 0.5
0.5 ≤ z < 1.0 6.2 × 103 0.15 43.9 ± 0.7
1.0 ≤ z < 1.5 7.5 × 103 0.18 44.7 ± 0.7
1.5 ≤ z < 2.0 9.4 × 103 0.35 45.0 ± 0.6
2.0 ≤ z < 2.5 6.4 × 103 0.43 45.2 ± 0.6
2.5 ≤ z < 3.0 2.6 × 103 0.30 45.4 ± 0.5
3.0 ≤ z < 3.5 6.8 × 102 0.14 45.6 ± 0.4
3.5 ≤ z < 4.0 2.2 × 102 0.08 45.6 ± 0.4
4.0 ≤ z < 4.5 78 0.04 45.9 ± 0.3
Notes. For the Euclid-only unobscured AGN criterion and union (∪) of all Euclid and Rubin/LSST selection criteria we report the number of
selected AGN, completeness (C) at the 5σ level in the relevant filters, and the median bolometric luminosity (±1σ) in each redshift bin. These
data are visualised in Fig. 13.
and is expected to achieve F1∼ 0.2 with P = 0.2. Applied to
our Euclid detectable AGN we selected 4.8× 106 AGN, corre-
sponding to a selected AGN surface density of 331 deg−2. We
select 92% unobscured and 8% obscured AGN SEDs with this
selection. An overall completeness C = 0.23 results for our full
candidate sample. Considering only unobscured AGN, the target
class of the selection, the completeness rises to C = 0.52.
The colour selection performance of unobscured AGN in the
EWS, and indeed for all AGN colour selections in the EWS
and EDS, is improved with the addition of Rubin/LSST optical
bands. The optimal selection criteria for unobscured AGN in the
EWS including Rubin/LSST is given by
IE − HE < 1.2
AND u − z < 1.1
AND IE − HE < −1.3 (u − z) + 1.9,
(E.2)
where u and z correspond to the Rubin/LSST optical filters. The
criterion is expected to achieve F1 = 0.9 with P = 0.9. Imple-
mented on our sample this selection yields 5.7× 106 selected
AGN, which corresponds to a surface density of 393 deg−2.
With this criterion our resulting selected sample contained 97%
A250, page 32 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
unobscured SEDs and 3% obscured AGN SEDs. We derived an
overall completeness of C = 0.45, with a completeness C = 0.75
when considering unobscured AGN only.
In the EWS, the selection of all AGN targeting obscured,
unobscured, and composite systems utilising Euclid photometry
only is disregarded due to the difficulty of separating AGN pow-
ered sources from contaminants with the limited optical and NIR
filters available. This selection task is improved with the addition
of optical Rubin/LSST bands. The optimal criterion with such
external data achieves F1 = 0.3 with P = 0.2 and is formulated
as
u − r < 0.2
OR IE − YE < −0.9 (u − r) + 0.8, (E.3)
where u and r are the Rubin/LSST optical filters. We note
that this selection is for objects that fall on the outside
of the defined boundary in colour space (referred to as
“type-B” in Euclid Collaboration: Bisigello et al. 2024). Apply-
ing this colour selection to our data we selected a sam-
ple of 6.0× 106 AGN, corresponding to a surface density of
413 deg−2. Our selected sample with this criterion comprised
72% unobscured and 28% obscured AGN SEDs. As raised
in Euclid Collaboration: Bisigello et al. (2024), we note that
this colour selection has a low sample purity and therefore is
expected to select a large fraction of contaminant inactive galax-
ies. We found the overall completeness of this selected sample
is C = 0.51. For unobscured (obscured) AGN populations the
completeness achieved is C = 0.65 (0.33).
The optimal colour criteria for AGN selection presented in
Euclid Collaboration: Bisigello et al. (2024) differ between the
EWS and EDS. Different criteria are required given that the two
surveys probe different parts of the AGN LF. For unobscured
AGN in the EDS the best selection criteria using Euclid pho-
tometry alone, with F1 = 0.2 and P = 0.2, is given by
IE − YE < 0.3
AND IE − HE < 0.5
AND IE − HE < −1.6 (IE − YE) + 0.8.
(E.4)
This selection criterion yields 1.7× 104 AGN at a surface density
of 346 deg−2 when applied to our sample. Our selected sample
contained 97% unobscured SEDs and 3% obscured AGN SEDs.
The completeness derived for this selection is C = 0.11 for the
overall population and C = 0.40 considering the target class of
unobscured AGN only. We note that this selection is expected
to be particularly contaminated by dwarf irregular galaxies in
practise.
Adding Rubin/LSST bands raises the quality of unobscured
AGN selection in the EDS to an expected F1 = 0.8 with P = 0.9.
The criterion in this case is
IE − HE < 1.1
AND u − z < 1.2
AND IE − HE < −1.2 (u − z) + 1.7,
(E.5)
where u and z are the Rubin/LSST optical filters. Using
this selection on our Euclid detectable sample we selected
2.0× 104 AGN, with a corresponding surface density of
392 deg−2. Our selected sample in this case contained 98%
unobscured and 2% obscured AGN SEDs. The completeness
achieved with this selection is C = 0.45 overall and C = 0.76
when only unobscured AGN are considered.
As for the all AGN selection with only Euclid photometry
in the EWS, the EDS counterpart is also disregarded due to the
1.0 0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
IE YE
1.0
0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
I E
J E
z = 0
Selection Boundary
Quasar SED E(B V) = 0.0
Quasar SED E(B V) = 0.1
100
101
102
103
104
105
N
A
G
N
1.0 0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
u z
1.0
0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
I E
H
E
Selection Boundary
100
101
102
103
104
105
N
AG
N
1.0 0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
u r
1.0
0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
I E
Y E
Selection Boundary
100
101
102
103
104
105
N
AG
N
Fig. E.1. Optimal Euclid photometry AGN selection criteria derived in
Euclid Collaboration: Bisigello et al. (2024) for the EWS applied to our
data. We show two-dimensional density plots for AGN that are detected
above the 5σ point-source depths in all four relevant bands. The
coloured regions show the selected AGN and grey regions show non-
selected AGN. Panels show selections for unobscured AGN with Euclid
photometry only (top), unobscured AGN with Euclid and Rubin/LSST
photometry (middle) and all AGN with Euclid and Rubin/LSST pho-
tometry (bottom). In the top panel we plot the colour-space redshift
evolution of our unobscured AGN template with E(B − V) = 0 (green)
and E(B − V) = 0.1 (orange) for z ∈ [0, 7] in steps of δz = 1. White
circles show the z = 0 points and white squares denote the z = 5 points.
The E(B − V) = 0 green crosses in the selection region correspond to
z = 1 and z = 2.
A250, page 33 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
1.0 0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
IE YE
1.0
0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
I E
H
E
z = 0
Selection Boundary
Quasar SED E(B V) = 0.0
Quasar SED E(B V) = 0.1
100
101
102
N
AG
N
1.0 0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
u z
1.0
0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
I E
H
E
Selection Boundary
100
101
102
N
AG
N
1.0 0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
g r
1.0
0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
I E
Y E
Selection Boundary
100
101
102
103
N
AG
N
Fig. E.2. Same as Fig. E.1 for the EDS.
difficulty of separating a general active galaxy population from
inactive galaxies. Adding ancillary Rubin/LSST bands for the
selection of all AGN in the EDS improves the performance of
the optimal colour selection to have an expectation of F1 = 0.3
and P = 0.6. The optimal criterion in this case follows
IE − YE < 1.7
AND g − r < 0.3
AND IE − YE < −3.5 (g − r) + 0.9,
(E.6)
where g and r are the Rubin/LSST optical filters. Applied to our
sample this criterion selects 2.9× 104 of our AGN, providing a
surface density of 579 deg−2. Our resulting selected sample was
comprised of 67% unobscured SEDs and 33% obscured AGN
SEDs. We derived an overall completeness C = 0.32 for this
selection. Considering separated AGN classes we found a com-
pleteness of C = 0.51 for unobscured and C = 0.18 for obscured
AGN.
Appendix F: High-redshift predictions
The predicted yield of 7 ≤ z ≤ 9 quasars in the EWS was
explored in Euclid Collaboration: Barnet et al. (2019). Quasars
were incorporated in their work utilising the Jiang et al. (2016)
high-redshift quasar LF with two different assumed rates of
decline (modest and steep) in space density at z ≥ 6. The
decline in quasar space density is parameterised as φ ∝ 10k(z−6),
where φ is the quasar LF and k takes the values k = −0.72 or
k = −0.92 for the modest and steep rate of decline, respec-
tively. Contaminant populations such as M-type stars, L and
T-type dwarfs and compact early-type galaxies were addition-
ally modelled (Hewett et al. 2006). Quasar selection functions
were integrated over the sample of quasars and contaminants
to determine the predicted yield of quasars with 7 ≤ z ≤ 9
in the EWS. Over the full considered redshift range quasars
were successfully selected to the effective depth JE ∼ 22
when using only Euclid bands. Selection with a modest (steep)
quasar LF decline predicted 87 (51) quasars in the redshift range
7 ≤ z ≤ 7.5. This corresponds to a quasar surface density of
6.0× 10−3 deg−2 (3.5× 10−3 deg−2).
For comparison with the predicted EWS quasar yields of
Euclid Collaboration: Barnet et al. (2019), we extended our sim-
ulation to z = 7.5. To approximate the available quasar candi-
dates, we imposed conditions on our EWS sample for a detec-
tion with JE < 22, an unobscured optical classification, and
occupation of the redshift range 7.0 ≤ z ≤ 7.5. Applying these
constraints we found 1053 unobscured AGN, equal to a surface
density of 0.07 deg−2, a significantly higher yield. We consider
that we have made only a magnitude and redshift cut for our
estimates, therefore recovering the detectable quasars in the tar-
get parameter space, but not incorporating the complex selec-
tion function constructed in Euclid Collaboration: Barnet et al.
(2019). The selection function is likely to reject a number of the
unobscured AGN included in our approximated sample.
The AGN selected to define the Jiang et al. (2016) LF ful-
filled two colour cuts primarily so that contaminants were
avoided in the final quasar sample. SDSS main survey quasars
required no detection in the ugr bands and obeyed
i − z > 2.2, (F.1)
where i and z refer to the SDSS bands. These criteria
select i-band dropout objects and separate quasars (and cool
brown dwarfs) from the majority of stellar objects (Fan 1999;
Strauss et al. 1999). Final quasar candidates also satisfied the cri-
terion
z − J < 0.5(i − z) + 0.5, (F.2)
where J refers to the UKIRT Infrared Deep Sky Survey
(UKIDSS; Warren et al. 2007) band. This colour-space cut sepa-
rates quasars from the cool brown dwarf population. We applied
these criteria to our 7.0 ≤ z ≤ 7.5 unobscured AGN sample
with JE < 22, substituting the the UKIDSS J band for the
Euclid JE band. With these additional constraints we recover 668
detectable quasars, which is a factor of eight higher than forecast
in Euclid Collaboration: Barnet et al. (2019).
A250, page 34 of 35
Euclid Collaboration: A&A, 693, A250 (2025)
High-redshift quasar yield predictions are especially sensi-
tive to the assumed LF shape and redshift evolution (Tee et al.
2023). The XLF employed in this work, when extrapolated, does
not exhibit the steep and accelerating space density decline at
z & 6 seen in UV/optical and other X-ray determinations of
the AGN LF (e.g. Ueda et al. 2014; Jiang et al. 2016; Wang et al.
2019; Matsuoka et al. 2023, see Fig. 1). At z > 6 our simulation
predicts an unobscured AGN space density excess of up to 1 dex
at M1450 ∼ −24, when compared with empirical UV/optical
quasar observations. This ultimately is the driver of the appar-
ent excess of quasars we predict at 7 ≤ z ≤ 7.5 compared to the
Euclid Collaboration: Barnet et al. (2019) analysis.
Equally as impactful as the assumed LF are the integration
range and selection parameters of high-redshift quasar yield pre-
dictions. Schindler et al. (2023) make z > 7 EWS quasar predic-
tions also utilising the k = −0.7 Jiang et al. (2016) quasar LF.
Integrating to M1450 . −22.4 and using HE < 24 as the effec-
tive survey depth, they predicted 809 detectable quasars in the
EWS at 7 ≤ z ≤ 8. This sample is of comparable size to our
7 ≤ z ≤ 7.5 simulated detectable sample of 1053, which incor-
porates unobscured AGN to M1450 ∼ −22.6, albeit over a larger
redshift slice.
At present there are only 532 quasars confirmed at z ≥ 5.3
and 275 quasars known at z ≥ 6 (Fan et al. 2023, and references
therein). Our predictions suggest there are still many quasars at
these redshifts to be uncovered, with Euclid capable of detecting
many of them. We found that ∼ 600 obscured AGN will have a
JE < 22 detection in the redshift range 7.0 ≤ z ≤ 7.5. For the rea-
sons discussed above this number is uncertain, however suggests
that Euclid may uncover a population of high-redshift obscured
AGN that are half as numerous in the data as unobscured AGN.
A250, page 35 of 35