MNRAS 521, 1556–1569 (2023) h ps://doi.o g/10.1093/mn as/s ad621
Ad ance Access publica ion 2023 Ma ch 02
Assessing model-based ca bon and oxygen abundance de i a ion om
ul a iole emission lines in AGNs
E. P
´
e ez-Mon e o,
1 ‹R. Amo
´
ın ,
2 , 3 B. P
´
e ez-D
´
ıaz ,
1 J. M. V
´
ılchez
1 and R. Ga c
´
ıa-Beni o
1
1
Ins i u o de As o
´
ısica de Andaluc
´
ıa, CSIC, Apa ado de co eos 3004, 18080 G anada, Spain
2
Ins i u o de In es igaci
´
on Mul idisciplina en Ciencia y Tecnolog
´
ıa, Uni e sidad de La Se ena, Raul Bi
´
an 1305, La Se ena, Chile
3
Depa amen o de As onom
´
ıa, Uni e sidad de La Se ena, A . Juan Cis e nas 1200 No e, La Se ena, Chile
Accep ed 2023 Feb ua y 23. Recei ed 2023 Feb ua y 23; in o iginal o m 2022 Oc obe 24
A B S T R A C T
We p esen an adap ed e sion o he code HII-CHI-MISTRY-UV o de i e chemical abundances om emission lines in he
ul a iole , o use in na ow line egions (NLR) o ac i e galac ic nuclei (AGN). We e alua e di e en ul a iole emission line
a ios and how di e en assump ions abou he models, including he p esence o dus g ains, he shape o he inciden spec al
ene gy dis ibu ion, o he hickness o he gas en elope a ound he cen al sou ce, may a ec he inal es ima es as a unc ion
o he se o emission lines used. We compa e ou esul s wi h o he published ecipes o de i ing abundances using he same
emission lines and show ha de i ing he ca bon- o-oxygen abundance a io using C III ] λ1909 Å and O III ] λ1665 Å emission
lines is a obus indica o o he me al con en in AGN ha is nea ly independen o he model assump ions, simila o he case
o s a - o ming e gions. Mo eo e , we show ha a p io de e mina ion o C/O allows o a much mo e p ecise de e mina ion
o he o al oxygen abundance using ca bon UV lines, as opposed o assuming an a bi a y ela ionship be ween O/H and C/O,
which can lead o non-negligible disc epancies.
K ey wo ds: galaxies: ac i e –galaxies: abundances –galaxies: e olu ion –galaxies: ISM – galaxies: nuclei – galaxies: Sey e .
1 INTRODUCTION
Ac i e Galac ic Nuclei (AGN) hos one o he mos powe ul ene gy
sou ces in he Uni e se. The in ense adia ion ield emana ing om
he ho acc e ion discs a ound supe massi e black holes in galac ic
cen es p oduces e y b igh emission lines om he su ounding
gas. F om his, a ious physical p ope ies and chemical abundances
in hese egions can la e be in e ed up o a e y high edshi , which
can se e as indica o s o he e olu ion o he uni e se a di e en
cosmic epochs.
The UV egion is o ou s anding impo ance in his con ex . Se e al
b igh emission lines p oduced by AGNs, such as N V , He II , C IV ,
O III ], and C III ], which a e sensi i e o he ioniza ion condi ions
and physical p ope ies (i.e. elec on densi y and empe a u e) o
he wa m ISM (e.g. K e wley, Nicholls & Su he land 2019 ), can be
iden i ied in he λ∼1000–2000 Å egion. While his ange a z ࣠
2 is ypically explo ed using he Hubble Space Telescope ( HST ; e.g.
Rigby e al. 2018 ; Be g e al. 2022 ), hese lines a e edshi ed o es -
op ical wa eleng hs a z ∼2–4. The e o e, deep op ical spec oscopic
su e ys wi h g ound-based elescopes o 8–10 m class (e.g. S eidel
e al. 2003 ; Shapley e al. 2003 ; Lilly e al. 2007a ; Ku k e al. 2013 ;
Le F
`
e e e al. 2015 ; McLu e e al. 2018 ) a e ypically e icien a
de ec ing he la ges samples o UV line emi e s (e.g. S eidel e al.
2014 ; Maseda e al. 2017 ; Amo
´
ın e al. 2017 ; Nakajima e al. 2018a ;
Le F
`
e e e al. 2019 ; Fel e e al. 2020 ; Saxena e al. 2020 , 2022 ;
Schmid e al. 2021 ; Lle ena e al. 2022 ).

E-mail: [email p o ec ed]
Classical me hods o de e mining chemical abundance a e in-
s ead mos ly based on es -op ical emission lines (e.g. Maiolino &
Mannucci 2019 ). Ho we e , o galaxies a z ∼2–3, hese a e
edshi ed o he nea -in a ed (NIR) and de ec ions a e o en
limi ed o he ew b igh lines (e.g. S eidel e al. 2014 ; Shap-
ley e al. 2015 ). Join analysis o he obse ed es -UV and
es -op ical emission spec a wi h p edic ions om de ailed pho-
oioniza ion models o bo h s a - o ming galaxies (e.g. Gu kin,
Cha lo & B uzual 2016 ; P
´
e ez-Mon e o & Amo
´
ın 2017 ; Byle
e al. 2018 ) and AGNs (e.g. Fel e, Cha lo & Gu kin 2016b ;
Nakajima e al. 2018b ; Do s e al. 2019 ; Hi schmann e al. 2019 ;
Mignoli e al. 2019 ) hus eme ge as powe ul diagnos ics o
he ioniza ion and chemical abundances o galaxies (e.g. Pa
´
ıcio
e al. 2016 ; Vanzella e al. 2016 ; Amo
´
ın e al. 2017 ; Byle
e al. 2020 ) and pa e he way o simila analyses a highe
edshi s wi h he JWST (e.g. Che alla d e al. 2019 ; Rigby e al.
2021 ).
The de i a ion o he me al con en in he Na ow Line Region
(NLR) o AGN galaxies om op ical collisionally exci ed lines
(CELs) is much mo e complex han in he case o s a - o ming
egions, since he di ec me hod (i.e. based on he p io de e mina ion
o he elec on empe a u e) can signi ican ly unde es ima e he
de i ed oxygen abundances (e.g. Do s e al. 2015 ), which a e o en
conside ed as p oxies o gas me allici y in galaxies. Ins ead, he e a e
se e al s udies dealing wi h he de e mina ion o di e en elemen al
abundances in AGN based on pu e pho oioniza ion models (e.g.
S o chi-Be gmann e al. 1998 ; Cas o e al. 2017 ; Thomas e al. 2019 )
o on models assuming shocks in combina ion wi h pho oioniza ion
(e.g. Do s e al. 2021 ).
© 2023 The Au ho (s)
Published by Ox o d Uni e si y P ess on behal o Royal As onomical Socie y
Downloaded om h ps://academic.oup.com/mn as/a icle/521/1/1556/7068104 by Ins . As o isica Andalucia CSIC use on 14 May 2024
Model-based abundances in AGN om UV 1557
MNRAS 521, 1556–1569 (2023)
In he case o UV emission lines, since di ec de e mina ion o
chemical abundances in AGN is s ill no eliable, many o he a ious
app oaches pu sued a e also based on pho oioniza ion models,
including de ailed models o indi idual objec s (e.g. Da idson
1977 ; Osme 1980 ; Gaskell, Shields & Wample 1981 ; Uomo o
1984 ; Hamann & Fe land 1992 ; Fe land e al. 1996 ; Die ich &
Wilhelm-E kens 2000 ; Hamann e al. 2002 ; Shin e al. 2013 ; Fel e,
Cha lo & Gu kin 2016a ; Yang e al. 2017 ). Fo a la ge numbe o
objec s, o he au ho s p opose calib a ions o some UV emission line
a ios sensi i e o chemical abundance in hese de ailed models (e.g.
Do s e al. 2014 , 2019 ), o compa e hem in model-based diagnos ic
diag ams (e.g. Nagao, Maiolino & Ma coni 2006 ; Ma suoka e al.
2009 , 2018 ).
Ano he s a egy sui able o use in la ge su e ys is o use
a Bayesian-like compa ison be ween he adequa e emission line
a ios and he p edic ions om la ge model g ids (e.g. Mignoli
e al. 2019 ). This me hod has he ad an age o be e quan i ying
unce ain ies and easily speci ying he assump ions equi ed when
he numbe o emission lines inpu is limi ed. Fo example, he use o
pho oioniza ion models, which include a ious combina ions o C/O
and O/H has g ea signi icance o es ima ing gas-phase me allici y in
cases whe e e y ew C lines (e.g. C IV λ1549 Åand C III ] λ1909 Å)
UV emission lines can be measu ed.
In his wo k, we p esen an adap ed e sion o he code HII-CHI-
MISTRY-UV (he ea e , HCM-UV , P
´
e ez-Mon e o & Amo
´
ın 2017 ),
o iginally de eloped o de i ing O/H and C/O om UV emission
lines in s a - o ming egions, o applica ion o he NLR o AGN.
The code is based on HCM (P
´
e ez-Mon e o 2014 ), which deals wi h
op ical emission lines o de i ing O/H and N/O a ios in s a - o ming
egions, and has been ex ended o use in he NLR o AGN in P
´
e ez-
Mon e o e al. ( 2019 ).
The abo e wo k has shown ha when applied o s a - o ming
galaxies, he abundances p o ided by he HCM e sion using op ical
emission lines a e in comple e ag eemen wi h hose p o ided by
he di ec me hod, while o AGNs hey a e in ag eemen wi h he
expec ed me allici ies a he cen es o hei hos galaxies (Do s e al.
2020 ). This new e sion o he HCM-UV code is po en ially use ul
o cons aining he me allici y o he NLR o AGN up o e y high
edshi , and also p o ides a solu ion o he cons ain imposed by
using ca bon emission lines o de i e oxygen abundances, since i
also es ima es C/O.
The pape is o ganized as ollows: In Sec ion 2 , we desc ibe he
me hod o de i ing chemical oxygen and ca bon abundances in he
NLR o AGN om hei UV emission lines, based on he code HCM-
UV . In he same sec ion, we desc ibe ou g ids o pho oioniza ion
models and we discuss hei alidi y unde di e en assump ions and
o using di e en se s o emission lines. In Sec ion 3 , we apply he
me hod o a sample o compiled da a on UV lines o NLR in AGNs,
and in Sec ion 4 , we discuss he esul s and compa e hem wi h
o he published calib a ions based on he same a ailable emission
lines. Finally, in Sec ion 5 , we summa ize ou esul s and d aw ou
conclusions.
2 DESCRIPTION OF THE METHOD
The me hod discussed in his pape is he adap a ion o he code HCM-
UV , o iginally de eloped o use wi h s a - o ming objec s (P
´
e ez-
Mon e o & Amo
´
ın 2017 ), o he NLR o AGN. The ex ension o
di e en e sions o he code o use on AGN has al eady been done
o op ical lines by P
´
e ez-Mon e o e al. ( 2019 ) and o in a ed lines
by P
´
e ez-D
´
ıaz e al. ( 2022 ). In his sec ion, we discuss he model
g ids ha he code uses o es ima e O/H, C/O, and he ioniza ion
pa ame e (log U ) in hese objec s, as well as he de ined obse ables
based on he mos ypical UV emission line lux a ios ha he code
uses o de i e hese p ope ies using a Bayesian-like me hodology.
Finally, we discuss how di e en assump ions o he models and he
emission lines used a ec he esul s.
The e sion o he code desc ibed he e ( . 5.0) along wi h o he
e sions p epa ed o use in o he spec al domains, is publicly
a ailable.
1
2.1 Desc ip ion o he models
The en i e g id o models used in HCM-UV was calcula ed using
he pho oioniza ion code CLOUDY .17.01 (Fe land e al. 2017 ),
which assumes a poin -like cen al ioniza ion sou ce su ounded
by a gas dis ibu ion. The models pa ially co espond o hose o
he op ical e sion o he code desc ibed in P
´
e ez-Mon e o e al.
( 2019 ), aking in o accoun a spec al ene gy dis ibu ion (SED)
wi h wo componen s: one o he Big Blue bump peaking a 1 Ryd,
and he o he ep esen ed by a powe law wi h spec al index αx
=
−1 o he non- he mal X- ays. Fo he con inuum be ween 2 keV
and 2500 Å, we conside ed a powe law wi h wo possible alues
o he spec al index αox
= −0.8 and −1.2. The i s alue i s
be e he emission line luxes in ailo ed models by Do s e al.
( 2017 ), while he second alue i s be e he a e age alue ound
by Mille e al. ( 2011 ) in a sample o Sey e -2 galaxies. Fo he
gas, we assumed a illing ac o o 0.1 and a cons an densi y o
500 cm
−3
, as is ypical in he NLRs a ound ype 2 AGNs (Do s e al.
2014 ). As discussed in P
´
e ez-Mon e o e al. ( 2019 ) o he op ical
e sion o he code, assuming a highe alue by 2 ×10
3
cm
−3
does
no lead o la ge de ia ions in he ob ained esul s. Di e en g ids
we e calcula ed conside ing he c i e ion o s opping he models and
measu ing he esul ing spec um as a unc ion o he ac ion o
ee elec ons (
e
) in he las zone (i.e. he ou e mos wi h espec
o he ionizing sou ce). In some g ids, we conside ed a ac ion
o 98 pe cen and in o he s a ac ion o 2 pe cen . All chemical
abundances we e scaled o oxygen acco ding o he sola p opo ions
gi en by Asplund e al. ( 2009 ), excep o ni ogen and ca bon,
which a e in sola p opo ion, bu whose ela ion o oxygen was
le in he models as an addi ional ee inpu pa ame e . Finally, we
conside ed models wi h a dus - o-gas a io ha assumes he de aul
alue o he Milky Way, and also o he models ha do no conside
he exis ence o dus mixed wi h he gas. Al hough a la ge a ie y in
he assump ions o dus composi ion and p opo ions would be mo e
ealis ic, since hese canno be cons ained om he inpu emission-
line luxes, hese ha e no been mo e deeply explo ed in ou g id
o models, a oiding an unnecessa ily high numbe o models in ou
g ids.
O e all, he models in each g id co e he ange o 12 + log (O/H)
om 6.9 o 9.1 in bins o 0.1 dex, and alues o log (C/O) om −1.4
o 0.6 in bins o 0.125 dex. In addi ion, all models conside alues
o log U om −4.0 o −0.5 in bins o 0.25 dex. This gi es a o al
numbe o 5865 models o each o he esul ing g ids. Conside ing
addi ionally he wo possible alues o αOX
( −0.8 and −1.2), he wo
possible alues o
e
in he ou e mos zone (2 o 98 pe cen ) and he
exis ence o absence o dus g ains mixed wi h he gas, we ob ain a
numbe o 8 di e en calcula ed g ids.
1
In he webpage h p:// www.iaa.csic.es/ ∼epm/ HII- CHI- mis y.h ml
Downloaded om h ps://academic.oup.com/mn as/a icle/521/1/1556/7068104 by Ins . As o isica Andalucia CSIC use on 14 May 2024
1558 E. P
´
e ez-Mon e o e al.
MNRAS 521, 1556–1569 (2023)
Figu e 1. Rela ionship be ween emission line a io C3O3 and log (C/O) o model sequences wi h di e en alues o inpu pa ame e s (le o ixed 12 +
log (O/H) = 8.7, and igh o ixed log U = −2.0). The solid lines ep esen models o αOX
= −1.2,
e
in he ou e mos zone o 98 pe cen and conside ing
dus g ains. The o he lines change only one pa ame e wi h espec o his sequence. The dashed line ep esen s models wi h αOX
= −0.8, he do –dashed line
ep esen s models wi h
e
o 2 pe cen , and he hin line ep esen s models wi hou g ains. In he le -hand panel, he black dashed line ep esen s he linea i
o models o s a - o ming galaxies ha a e in P
´
e ez-Mon e o & Amo
´
ın ( 2017 ).
2.2 The HCM-UV code adap ed o AGNs
The e sion o he code we desc ibe he e o i s use o NLRs in AGNs
ollows a p ocedu e simila o ha used in o he e sions o o he
spec al egions, such as he op ical (P
´
e ez-Mon e o 2014 ) o he IR
(Fe n
´
andez-On i e os e al. 2021 ). In all cases, he code pe o ms a
Bayesian-like calcula ion ha compa es ce ain emission line luxes
and hei e o s wi h he esul s o he models in each g id. He e, we
discuss he esul s ob ained wi h he model g ids desc ibed abo e.
Ho we e , his ne w e sion o he code allo ws he use o use al e -
na i e g ids as inpu models ha ha e he same o ma as he de aul
models.
The abo e compa ison pe o med by HCM-UV o calcula e he
co esponding inal mean alues uses as weigh s o each model
he χ2 alues o some de ined speci ic emission line a ios ha
a e sensi i e o he p ope ies we wan o de i e, as desc ibed
in P
´
e ez-Mon e o ( 2014 ). The unce ain ies o he de i ed abun-
dances and U a e calcula ed as he quad a ic addi ion o he
weigh ed s anda d de ia ion and he dispe sion o he esul s a e
a Mon e Ca lo simula ion, using he inpu e o s as andom pe -
u ba ions a ound he nominal in oduced alues o each emission
lux.
The lis o UV emission lines allowed as inpu by HCM-UV , o
bo h SF and AGN, includes: Ly αa λ1216 Å, C IV λ1549 Å, O III ]
λ1665 Å, and C III ] λ1909 Å. In addi ion, he code p o ides he
abili y o inpu op ical emission lines H βand [O III ] λ5007 Å o
ob ain es ima es o abundances using emission line a ios ha a e
sensi i e o elec on empe a u e, as is he case wi h 5007/1665
(P
´
e ez-Mon e o & Amo
´
ın 2017 ).
Unlike he p e ious e sion o s a - o ming galaxies, his e sion
also includes he emission lines [N V ] a λ1239 Åand He II a λ
1640 Å, since he use o hese lines in combina ion wi h he o he
ca bon emission lines in he UV can be used o p o ide es ima es
o he o al oxygen abundance in AGNs, as desc ibed in Do s e al.
( 2019 ). In any case, hese wo lines can also be used as inpu o
he calcula ion o abundances in s a - o ming galaxies, since he
p esence o hese lines can be conside ed a s ong disc imina ing
ac o in he exci a ion o he gas when only massi e s a s a e
conside ed.
2.2.1 C/O de i a ion
Following he same p ocedu e as de ined o SF objec s, he code
compu es C/O in a i s i e a ion, aking ad an age o he ac ha he
emission line a io C3O3 depends e y li le on U . This obse able
can be de ined as al eady used by P
´
e ez-Mon e o & Amo
´
ın ( 2017 )
o s a - o ming galaxies o de i e C/O:
C3O3 = log
I(C III ]1909)
I(O III ]1665)
(1)
In Fig. 1 , we show he ela ion be ween his pa ame e and C/O
o model sequences wi h αOX
= −1.2 and assuming
e
= 98 pe cen
and g ains a a ious log ( U ) a a ixed 12 + log (O/H) = 8.7 (le -
hand panel) and o a ious C/O alues a a ixed log ( U ) = −2.0
( igh -hand panel). We also show addi ional sequences o models
wi h αOX
= −0.8, wi h
e
= 2 pe cen , and wi hou dus g ains o
assess how a ying hese pa ame e s a ec s his ela ion.
Fig. 1 shows ha despi e a sligh dependence on O/H, U , o
αOX
, he e is a well de ined linea ela ionship be ween he C3O3
pa ame e and C/O. Only models wi hou g ains seem o p edic a
sligh ly lowe slope compa ed o all o he sequences. Compa ed o
he linea ela ion de i ed in P
´
e ez-Mon e o & Amo
´
ın ( 2017 ) o
models o s a - o ming galaxies, he sequences o he models o
AGN a e e y close, albei wi h a sligh ly lowe slope, simila o ha
ob ained o AGN wi hou g ains.
We conclude ha C3O3 appea s o be a obus indica o o C/O
o AGN galaxies using UV emission lines, and ha he code can
subsequen ly be used o es ima ion.
A e es ima ing C/O and i s e o , he code cons ains he model
g id o seek a solu ion o O/H and U in a second i e a ion. This
p ocedu e ensu es ha C lines can be used wi hou any p io a bi a y
assump ion abou he ela ionship be ween O/H and C/O i he
la e can be in e ed. I C/O canno be es ima ed because some
o he equi ed lines a e no a ailable, he code assumes an expec ed
ela ionship be ween O/H and C/O, as is he case o s a - o ming
egions. By de aul , he code assumes a sola C/N a io and adop s he
empi ical ela ion be ween O/H and N/O de i ed in P
´
e ez-Mon e o
( 2014 ) o s a - o ming objec s. Ho we e , in he cu en upda ed
e sion o he code, o he ela ionships can be conside ed.
Downloaded om h ps://academic.oup.com/mn as/a icle/521/1/1556/7068104 by Ins . As o isica Andalucia CSIC use on 14 May 2024
Model-based abundances in AGN om UV 1559
MNRAS 521, 1556–1569 (2023)
Figu e 2. Rela ionship be ween emission line a io C34 using he Ly αline and o al oxygen abundance o di e en model se ies a ixed log (C/O) = −0.4
(le -hand panel) and log U = −2.0 ( igh -hand panel). The solid lines ep esen models o αOX
= −1.2,
e
in he ou e mos zone o 98 pe cen and conside ing
dus g ains. The o he lines change only one pa ame e in ela ion o his sequence.
2.2.2 O/H and U de i a ion
Among he a ious emission line a ios used in he second i e a ion
o calcula e he χ2
weigh s, he code uses pa ame e C34, de ined as
ollows:
C34 = log
I(C IV 1549) + I(C III ]1909)
I(H
i
) . (2)
This pa ame e was also used o s a - o ming objec s in P
´
e ez-
Mon e o & Amo
´
ın ( 2017 ). In his case, I( Hi ) e e s o he in ensi y o
he hyd ogen ecombina ion emission lines. The closes and b igh es
objec in his spec al ange is Ly αa λ1216 Å, bu he code also
allows he use o H βa λ4861 Å, al hough in his las case, he
ex inc ion co ec ion becomes much mo e c i ical.
In Fig. 2 , we show he ela ion be ween his pa ame e , aking he
Ly αin ensi y o he pa ame e , and O/H o some sequences o
models om he g id a ixed log (C/O) = −0.4 (le -hand panel) and
log U = −2.0. ( igh -hand panel), wi h αOX
= −1.2,
e
= 98 pe cen
in he las zone and wi h dus g ains.
I is no ed ha he ela ionship be ween his pa ame e and he
o al oxygen abundance in ce ain sequences can be bi alued, since
i inc eases o low alues o O/H, emains almos cons an and inally
s a s o dec ease when he me allici y is e y high. This beha iou is
in con as o ha obse ed in models o s a - o ming galaxies, whe e
a mono onic inc ease in he pa ame e wi h me allici y is obse ed
h oughou he ange s udied, as discussed in P
´
e ez-Mon e o &
Amo
´
ın ( 2017 ). Ho we e , he me allici y ange whe e he u no e
o he cu e is obse ed depends on he assumed C/O and U , and
can be nea ly linea o log U = −1.0. In ac , hese w o pa ame e s
ha e a e y la ge in luence on he ela ion o he pa ame e o he
me allici y.
Despi e he la ge dependence o his pa ame e on U o C/O, he
models p edic a negligible de ia ion i we conside ano he ha de
SED wi h αOX
= −0.8. Ho we e , he assump ion o absence o dus
g ains seems c i ical, since C34 shows a much mo e p onounced
bi alued beha iou when no g ains a e conside ed, wi h e y lowe
alues a high me allici ies, since Ly αis much mo e a ec ed by dus
ex inc ion.
In addi ion, he inclusion o e y low exci a ion zones also a ec s
he beha iou o his pa ame e , since Ly αis much mo e s ongly
abso bed, so ha C34 is much highe o he same O/H. This is
consis en wi h he ac ha Ly α1216 λÅis o en abso bed by he
neu al gas a ound he ionized gas nebulae and is he e o e di icul
o de ec in many objec s.
An al e na i e is o de ine he same pa ame e as a unc ion o he
line He II a λ1640 Å, which is also used in Do s e al. ( 2019 ). This
akes ad an age o he ac ha his line is o en well de ec ed wi h
high exci a ion in AGN and has less abso p ion by he su ounding
neu al gas. The co esponding obse able based on he same lines
can be de ined as ollows,
C34He2 = log
I(C IV ]1549) + I(C III ]1909)
I(He II 1640) , (3)
which is also used o he NLR o AGNs galaxies by Do s e al.
( 2014 ) named C43. Ho we e , in his wo k, we e e o i as C34He2
o dis inguish i om ou C34 pa ame e de ined abo e. The ela ion
be ween C34He2 and O/H depending on he model is shown in Fig. 3
o he same model se ies a ixed C/O = −0.4 (le -hand panel) and
ixed log U = −2.0 ( igh -hand panel) and o a alue o αOX
= −1.2,
e
= 98 pe cen and wi h dus g ains. Fu he mo e, as in he p e ious
plo s, we show addi ional sequences wi h αOX
= −0.8, wi h
e
=
2 pe cen in he las zone and also wi hou dus g ains.
As shown, he pa ame e has a simila end o ha obse ed
o C34: i inc eases wi h inc easing me allici y, al hough i en e s
sa u a ion and begins o dec ease a e y high O/H alues. The
me allici y a which his con e sion is expec ed o occu is so
dependen on U ha i is less p onounced o highe U alues. This
addi ional dependence o O/H on U s eng hens he basis o he
p ocedu e in he code by which hese wo pa ame e s a e calcula ed
simul aneously in his second i e a ion.
The ela ion be ween C34He2 and O/H does no appea o change
signi ican ly when o he ha de alues o αOX a e conside ed and,
as expec ed, when a la ge hickness o he nebula is assumed in
he models, since He II , unlike Ly α, is no abso bed by neu al gas.
Ho we e , he models wi hou g ains p edic much mo e di e en
alues o he pa ame e o he same O/H alue, since he in e nal
ex inc ion o he gas a ec s he opaci y o lines wi h high exci a ion,
such as C IV ] o He II , leading o highe alues o he pa ame e , in
con as o C34.
Ano he obse able ha he code can use in his second i e a ion
is also de ined and used by Do s e al. ( 2019 ) o de i e O/H based
on he highly exci ed emission lines ha can be obse ed in he UV
spec a o AGNs. I is he N5He2 pa ame e , which can be de ined
Downloaded om h ps://academic.oup.com/mn as/a icle/521/1/1556/7068104 by Ins . As o isica Andalucia CSIC use on 14 May 2024
1560 E. P
´
e ez-Mon e o e al.
MNRAS 521, 1556–1569 (2023)
Figu e 3. Rela ionship be ween C34/He2 emission line a io and o al oxygen abundance o di e en model sequences a ixed log (C/O) = −0.4 (le -hand
panel) and ixed log U = −2.0 ( igh -hand panel). The solid lines ep esen models o αOX
= −1.2,
e
in he ou e mos zone o 98 pe cen and conside ing dus
g ains. The o he lines change only one pa ame e wi h espec o his sequence.
Figu e 4. Rela ionship be ween he emission line a io N5/He2 and 12 +
log (O/H) o model sequences wi h di e en alues o he inpu pa ame e s
a ixed log (C/O) = −0.4. The solid lines ep esen models o αOX
= −1.2,
e
in he ou e mos zone o 98 pe cen and conside ing dus g ains. The o he
lines change only one pa ame e wi h espec o his sequence. The dashed line
ep esen s models wi h αOX
= −0.8, he do –dashed line ep esen s models
wi h
e
o 2 pe cen , and he hin line ep esen s models wi hou g ains.
as ollows,
N5He2 = log
I ([ NV ]1239)
I ( He II 1640)
. (4)
In Fig. 4 , we show he ela ion be ween his pa ame e and O/H o
model sequences a ixed C/O = −0.4 and a ying log U , assuming
an AGN SED wi h αOX
= −1.2, a inal zone wi h
e
= 98 pe cen ,
and g ains. Addi ional sequences also conside o he alues o hese
inpu pa ame e s.
Simila o C34 and C34He2, he N5He2 pa ame e also shows a
s ong dependence on U and a bi alued ela ionship wi h oxygen
abundance. The pa ame e has a lowe alue o lowe U , bu on he
con a y, i does no appea o change he me allici y o he u no e
poin .
Mo eo e , since i depends on he lines wi h high exci a ion, i
is mo e sensi i e o he shape o SED, so i has lowe alues when
we a y αOX
= −0.8 and does no change signi ican ly o a lowe
alue o
e in he las zone o he models. Finally, he absence o
dus g ains is no negligible, bu he di e ence is no as la ge as in
he case o C34, due o he same cause. Mo eo e , since N is also a
seconda y elemen , N5He2 has a s ong dependence on N/O, which
can be pa ially educed by assuming ha C/N emains unchanged
in he models, al hough his is no ully jus i ied since he s ella
mass ange o he nucleosyn he ic p oduc ion o hese wo elemen s
is no exac ly he same (Hen y, Edmunds & K
¨
oppen 2000 ), bu can
be simila ly a ec ed by p ocesses o hyd odynamic gas exchange
(Edmunds 1990 ). Changing N/O independen ly o C/O, in he g ids
o models, would allow us in p inciple o explo e in mo e de ail he
impac o N/O on he inal me allici y de i a ion, bu he absence o
any UV emission-line a io s ic ly dependen on N/O, does no allow
us o inco po a e i o he code and would unnecessa ily enla ge he
numbe o models in each g id.
As commen ed, he dependence on C/O o he emission line
a ios de ined abo e o de i e O/H can be educed by he p io
de e mina ion o C/O in he i s i e a ion using he C3O3 pa ame e .
As o he exci a ion, he dependence o he a ious obse ables on
i can also be educed by he C3C4 emission line a io de ined as
ollows,
C3C4 = log
I(C III ]1909)
I(C IV ]1549)
. (5)
This a io is al eady used by he code o s a - o ming objec s
and also by Do s e al. ( 2019 ) o he case o NLRs in AGNs. This
a io was al eady p oposed by Da idson ( 1972 ) as an indica o o
exci a ion in gas nebulae, al hough Do s e al. ( 2014 ) poin ou ha
i also depends on me allici y o low alues o log U .
The ela ion o his emission line a io wi h he ioniza ion pa ame-
e is shown in Fig. 5 o di e en a bi a y model sequences a ixed
log (C/O) = −0.4 (le -hand panel) and ixed 12 + log (O/H) = 8.7
( igh -hand panel) and assuming αOX
= −1.2,
e
= 98 pe cen and
wi h g ains. As in he p e ious cases, models wi h ha de SED a e
also shown wi h αOX
= −0.8, wi h
e
= 2 pe cen and also wi hou
g ains in he igu es.
As can be seen, highe alues o C3C4 indica e lo we alues o
U o mos sequences, excep o e y high alues o O/H, which
sho w a bi alued beha iou . This is a weake e ec compa ed o he
ela ionship ound o he O32 pa ame e o AGNs using he same
models in he op ical (P
´
e ez-Mon e o e al. 2019 ), which exhibi s
bi alued beha iou a all alues o O/H wi h a maximum alue o
log U = −2.5. This mono onically dec easing ela ionship be ween
Downloaded om h ps://academic.oup.com/mn as/a icle/521/1/1556/7068104 by Ins . As o isica Andalucia CSIC use on 14 May 2024

Model-based abundances in AGN om UV 1561
MNRAS 521, 1556–1569 (2023)
Figu e 5. Rela ionship be ween emission line a io C3C4 and ioniza ion pa ame e o di e en model sequences a ixed log (C/O) = −0.4 (le -hand panel)
and ixed 12 + log (O/H) = 8.7 ( igh -hand panel). The solid lines ep esen models o αOX
= −1.2, he a io o ee elec ons in he ou e mos zone o 98 pe cen
and conside ing dus g ains. The o he lines change only one pa ame e wi h espec o his sequence; he black dashed line ep esen s he quad a ic i o Do s
e al. ( 2019 ). The e ical dashed magen a line ma ks he dis ance assumed o he C3C4 pa ame e o accoun o dus in he models o subsequen analysis.
C3C4 and U makes i easie o he code o use he g id o models
o all alues o U . This end does no change no iceably when we
conside a alue o αOX
= 0.8 o a lowe
e o he e mina ion
c i e ion.
On he o he hand, he C3C4 pa ame e is e y sensi i e o he
p esence o dus g ains mixed wi h he gas. Neglec ing he p esence
o dus in he models esul s in much lo we alues o C3C4. Indeed,
he ela ionship ha eme ges om hese models wi hou dus is
simila o he quad a ic i o Do s e al. ( 2019 ), whe e he au ho s
do no conside dus g ains in hei models. Fo his eason, some
au ho s, such as. Nagao e al. ( 2006 ) o Do s e al. ( 2019 ), do no
include dus in hei models because many o he analysed objec s
ha e e y lo w alues o C3C4. Ho we e , acco ding o he model
sequences shown in Fig. 5 , models wi h dus canno be excluded o
alues o he C3C4 pa ame e > −0.3, a h eshold ma ked wi h a ed
line in he le -hand panel o he igu e. In any case, o he possible
explana ions o his e y low alues o C3C4 canno be excluded and
dese e u u e in es iga ion, such as he possible exis ence o pho on
leaking, al eady obse ed in some ex eme s a - o ming galaxies
(e.g. Schae e e al. 2022 ).
2.3 Tes ing he me hod wi h model emission lines
To e i y he esul s o ou model-based me hod, we used as inpu
o he code he same emission line in ensi ies p edic ed by he
models o check ha we ob ain he same chemical abundances
and ioniza ion pa ame e s used o hem. Fo his pu pose, we also
included a 10 pe cen unce ain y in he p edic ed model-based
luxes o simula e he e ec o an addi ional sou ce o e o on he
es ima es. As explained abo e, he code uses he e o o he lux as
he s anda d de ia ion o a no mal dis ibu ion o cases a ound he
nominal alue o pe o m a se ies o i e a ions ha also e eal he
associa ed unce ain y in he esul s.
In he panels o he i s ow in Fig. 6 , we show he inpu
abundances and U assumed in he models o he g id o αOX
=
−1.2, a e mina ion c i e ion when
e
= 98 pe cen is eached, and
wi h dus g ains. These alues a e compa ed wi h he esul s o he
HCM-UV code when he same condi ions a e assumed and all lines
accep ed by he code a e used. As can be seen, he ag eemen is
excellen o mos o he s udied anges (i.e. bo h he mean o se and
he s anda d de ia ion o he esiduals a e less han 0.01 dex), so ha
bo h he abundances and he log ( U ) alues can be de e mined using
only he emission line in ensi ies. Only o 12 + log (O/H) < 7.5
does he code end o ind solu ions ha a e abou 0.1 dex abo e he
co ec inpu alue o each model, bu s ill wi hin he e o limi s (i.e.
0.17 de x). F o e y high alues o log ( U ) (i.e. > −0.8), he code also
ends o ind esul s ha a e below he co ec alues, wi h de ia ions
g ea e han he e o s ob ained (wi h a mean o 0.2 dex), bu o he
es o he examined ange, he e is ag eemen ha is be e han he
associa ed e o s.
We also in es iga e he po en ial impac in he esul s o changing
he assumed condi ions in he lib a y used by HCM-UV . Thus, we
an he code assuming he same condi ions as abo e, bu using as
inpu he emission lines p edic ed by he models when we assume a
alue αOX
= −0.8, and lea ing he o he pa ame e s equal o hose
assumed by he code. The co esponding compa isons a e shown in
he second line o Fig. 6 . In he case o O/H, he mean de ia ion
is la ge han he median unce ain y (0.18 dex) only a e y low
alues, al hough he e is also a sys ema ic de ia ion o O/H > 8.0,
in he sense ha he code de e mines O/H alues ha a e 0.1–0.2 dex
below he co ec alues. In con as o O/H, he mean de ia ion
o C/O is al w ays lowe han he ob ained unce ain y (0.11 dex) in
all anges. This con i ms ha C/O can be de e mined mo e eliably
han O/H when a di e en ionizing SED is assumed. Ho we e , he U
es ima e is mo e a ec ed by he shape o he SED as i is signi ican ly
o e es ima ed, especially in he ange −3 < log U < −1.
In he hi d line o Fig. 6 , we show he compa isons when we
change he unca ion c i e ion in he models used as inpu o a
ac ion o ee elec ons = 2 pe cen and lea e he o he pa ame e s
he same. Opposi e o he inpu , he code s ill conside s he same
condi ions as hose assumed in he i s case (i.e. wi h 98 pe cen
ac ion o ee elec ons in he las model). In his case, when he
code assumes a di e en highe
e alue, he main o se occu s
a high O/H alues, whe e he code inds solu ions abou 0.3 dex
below he co ec inpu alue, p o ing he impo ance o he ela i e
in ensi y o he HI lines o de i e me allici y. On he o he hand, no
la ge de ia ion is ound o he C/O de i a i e, bu in he case o U
he code sys ema ically inds much lowe alues o log U > −1.0.
Finally, in he bo om ow o Fig. 6 , we show he compa isons
be ween he inpu alues o models ha do no conside dus and
he alues de i ed by he code assuming dus is p esen . As sho wn,
in his si ua ion, he code o e es ima es O/H wi h a mean o se
Downloaded om h ps://academic.oup.com/mn as/a icle/521/1/1556/7068104 by Ins . As o isica Andalucia CSIC use on 14 May 2024
1562 E. P
´
e ez-Mon e o e al.
MNRAS 521, 1556–1569 (2023)
Figu e 6. Compa isons be ween he chemical abundances and U assumed in he model g id and he same quan i ies calcula ed by HCM-UV when he p edic ed
emission lines a e used as inpu . The panels in he le column show he compa ison o 12 + log (O/H), he middle column o log (C/O), and he igh column
o log U . In all cases, HCM-UV AGN models wi h αOX
= −1.2,
e
= 98 pe cen and conside ing g ains. In he panels o he i s o w, we sho w he esul s when
he same condi ions a e assumed o he inpu ; in he second, we change αOX
o −1.2; in he hi d ow, we conside
e
= 2 pe cen ; and inally, in he bo om
ow, wi hou dus g ains. The solid ed line ep esen s he 1:1 ela ion in all panels.
Downloaded om h ps://academic.oup.com/mn as/a icle/521/1/1556/7068104 by Ins . As o isica Andalucia CSIC use on 14 May 2024
Model-based abundances in AGN om UV 1563
MNRAS 521, 1556–1569 (2023)
Table 1. Median de ia ions and oo mean squa es (RSME) o esiduals be ween heo e ical abundances and log ( U ) alues (AGN model inpu s) and es ima es
om HCM-UV using di e en se s o emission lines and o di e en g ids.
Se o lines Models 
OH RSME
OH 
CO RSME
CO 
U RSME
U
All lines All + 0 .01 0 .21 + 0 .02 0 .14 −0 .01 0 .26
Ly α, N V ], C IV ], He II , O III ] λ1665 Å, C III ] All + 0 .08 0 .25 −0 .05 0 .14 + 0 .03 0 .11
N V ], C IV ], He II , O III ] λ1665 Å, C III ] All + 0 .08 0 .25 −0 .05 0 .14 + 0 .03 0 .11
N V ], C IV ], He II , C III ] All + 0 .36 0 .57 – – + 0 .11 0 .34
N V ], C IV ], He II , C III ] O/HC/O cons . + 0 .01 0 .15 – – + 0 .01 0 .13
C IV ], He II , C III ] All + 0 .32 0 .64 – – + 0 .04 0 .46
C IV ], He II , C III ] O/HC/O cons . + 0 .03 0 .33 – – −0 .03 0 .31
N V ], He II All + 0 .27 0 .66 – – + 0 .08 0 .48
N V ], He II O/HC/O cons . + 0 .00 0 .52 – – −0 .00 0 .42
la ge han he mean unce ain y (i.e. 0.17 dex) o O/H < 8.0, bu
unde es ima es i o O/H > 8.5, albei wi h a la ge sp ead in each
case. Fo C/O, he de ia ion be ween he solu ions and he heo e ical
alues is al w ays wi hin unce ain y (abou 0.2 de x), e xcep o e y
lo w alues o C/O. I is also no ed ha he e is a sys ema ic endency
o ind highe C/O alues o high inpu alues o C/O, bu hese
de ia ions a e al w ays wi hin e o s. Finally, in he case o U , we ind
be e ag eemen , al hough he de ia ions a e la ge o he highes
alues o U .
F om hese compa isons, i appea s ha he C/O a io is mo e
obus han he O/H a io when he assumed condi ions in he models
a y ela i e o hose assumed by he code. In he case o O/H,
his is especially he case a high me allici ies when he assump ion
abou he p esence o dus g ains is inco ec , while o U , he la ges
disc epancies a e ound a e y high alues in all he compa isons
es ed. Finally, o log U < −1.5, all disc epancies a e ound wi hin
he de i ed unce ain ies.
2.4 Resul s as a unc ion o di e en se s o inpu emission lines
Ano he impo an aspec in he e alua ion o ou code a ises om he
ac ha i can p o ide solu ions o di e en se s o emission lines. To
his end, in his subsec ion, we compa e he heo e ical and esul ing
abundances, again using as inpu s o he code di e en se s o
emission line luxes wi h an addi ional 10. The co esponding mean
o se s and s anda d de ia ion o he esiduals ob ained om he code
o O/H, C/O, and log U compa ed o he heo e ical alues assumed
by he models a e shown in Table 1 , o di e en combina ions o
emission lines ha can be gi en as inpu o he code, and o di e en
cons ain s on he model g id used.
As shown, when all lines allowed by he code a e speci ied, he
mean o se s o he esul s a e be e han he mean sca e in all
cases. This compa ison is he same as ha in he i s line o Fig. 6 .
This compa ison is sligh ly wo se when he op ical line [O III ] a λ
5007 Å, which is allowed by he code ela i e o O III λ1665 Åis
no p o ided, bu he esul s s ill ag ee be e han he associa ed
unce ain ies. A simila esul is ob ained i he emission line Ly α
a λ1216 Åis emo ed, since He II can be used ins ead. This has no
in luence on he C/O de e mina ion.
On he o he hand, i he O III ] λ1665 Åis no p o ided, HCM-UV
canno calcula e he C3O3 a io o he C/O es ima e. In his case,
e y la ge de ia ions and e o s esul o bo h O/H and log U when
he ull model g id is used o calcula ion. In his scena io, he code
mus assume a ce ain O/H–C/O a io o ge a be e solu ion o O/H
when C-lines a e used. As shown in Table 1 , bo h he mean o se s
and he s anda d de ia ion o O/H and log U a e much be e unde
his addi ional assump ion. Ho we e , i is impo an o emembe
ha his ela ion may be somehow a bi a y o alid o a pa icula
galaxy popula ion and no alid o a pa icula objec . In any case,
e en assuming such a ela ionship in he absence o he lines equi ed
o he p e ious de i a ion o C/O, much be e esul s a e ob ained i
he lines C III ], C IV ], and N V ] a e used simul aneously wi h espec
o He II a he han calib a ing hem indi idually and in combina ion.
The abo e esul s again highligh he impo ance o a p io
de e mina ion o he C/O con en in o de o ob ain an accu a e
de e mina ion o he o al me al con en o he gas, and a e in
ag eemen wi h he esul s al eady ob ained using he same app oach
o s a - o ming galaxies in P
´
e ez-Mon e o & Amo
´
ın ( 2017 ).
3 APPLICATION TO GALAXY SAMPLES
We compiled om he li e a u e s ong UV na ow emission line
luxes o a sample o AGN consis ing o Sey e s 2 (10), quasa s
(33), quie adio galaxies (2), high- z AGN (1), and high- z adio
galaxies (96). We ake ad an age o p e ious compila ion wo k,
(e.g. Nagao e al. 2006 ; Do s e al. 2019 ) o s ong UV emission
lines in AGN, namely N V λ1239 Å, C IV λ1549 Å, He II λ1640 Å,
and C III ] λ1909 Å, bu we also include in o ma ion on he Ly αand
[O III ] λ1665 Å emission lines when hei measu emen s a e gi en in
he o iginal e e ences. Table 2 shows all he in o ma ion we ound
o ou sample, as well as he o iginal e e ences.
The Sey e s 2 (S2) sample consis s o 9 objec s aken di ec ly
om Nagao e al. ( 2006 ) and one addi ional galaxy, IZw 92, om
K aeme e al. ( 1994b ). The sample o high- z adio galaxies (HZRG)
w as ak en di ec ly
2 om he compila ion p esen ed by De B euck
e al. ( 2000 ), supplemen ed by 8 sou ces obse ed by Bo nancini e al.
( 2007 ) and 9 galaxies analysed by Ma suoka e al. ( 2009 ). Ou sample
o quasa s (QSO) consis s o h ee di e en ypes o galaxies: Type
II ype quasa s (11), 10 o hem lis ed by Nagao e al. ( 2006 ) plus
J142331.71 −001809.1, ecen ly analysed by Onoue e al. ( 2021 );
ex emely ed quasa s (21) om he compila ion o Villa Ma
´
ın
e al. ( 2020 ); and one in e media e Type I–II quasa , obse ed by
Lin e al. ( 2022 ). We added o he abo e lis o objec s 2 quie adio
galaxies (QRG) obse ed by Ma suoka e al. ( 2018 ), and a high- z
A GN (zA GN) whose b oad and na ow componen s we e iden i ied
by Tang e al. ( 2022 ).
Some o he compiled obse a ions we e al eady co ec ed o
Galac ic ex inc ion in he o iginal publica ions. These co ec ions
we e no e y la ge [i.e. E ( B −V ) ∼0.01–0.1] due o hei ela i e
posi ion on he sky, and do no lead o changes in he esul ing
2
F om he o iginal lis o 167 sou ces, we omi nine sou ces obse ed by
Ma suoka e al. ( 2009 ) ha p o ide be e accu acy, and 79 sou ces ha do
no ha e enough UV spec oscopic in o ma ion o be used as inpu o HCM-UV .
Downloaded om h ps://academic.oup.com/mn as/a icle/521/1/1556/7068104 by Ins . As o isica Andalucia CSIC use on 14 May 2024
1564 E. P
´
e ez-Mon e o e al.
MNRAS 521, 1556–1569 (2023)
Table 2. Lis o UV luxes o ou sample o AGN.
Name z Type Ly αN V1239 Å; C IV1549 Å; He II1640 Å; O III ]1665 Å; C III]1908 Å; Re .
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
NGC 1068 0.004 S2 713 ±100 224 ±41 520 ±80 187 ±29 22 ±8 240 ±35 NA06,SNI86
I ZW 92 0.029 S2 121.2 ±40.3 –25.2 ±8.4 3.84 ±1.28 – – KRA94
CDFS-901 2.578 QSO 3.71e
−3
±6e
−5 6.5e
−4
±8e
−5 1.97e
−3
±1e
−4 1.87e
−4
±9.3e
−5 ∗–3.3e
−4
±9e
−5 NA06,SKO04
CXO52 3.288 QSO 1.89e
−2
±4e
−4 6e
−4
±1.2e
−4 3.5e
−3
±2e
−4 1.7e
−3
±2e
−4 9e
−5
±3e
−5 2.1e
−3
±2e
−4 NA06,STE02
S82-20
1 3.062 QSO 36.10 ±0.21 1.34 ±0.12 8.67 ±0.05 1.52 ±0.04 0.38 ±0.04 2.04 ±0.04 LIN22
TXS J2334 + 1545 2.480 HZRG 3.1e
−2
±5e
−3 7e
−4
±3e
−4 ∗1e
−3
±1e
−3 – – 8e
−3
±3e
−3 BRE00,BRE01
TN J1941-1951 2.667 HZRG 3.7e
−2
±4
−3 –9e
−3
±1e
−3 4e
−3
±1e
−3 –2e
−3
±1e
−3 BOR07
COSMOS 05162 3.524 QRG 7.68e
−2
±1.3e
−3 3.89
−3
±1.95e
−3 ∗1.19e
−2
±8e
−4 2.6e
−3
±3e
−4 – 4.05e
−2
±6.5e
−3 MAT18
COSMOS 10690 3.100 QRG 8.1e
−3
±4.1e
−3 ∗4.5e
−3
±2.3e
−3 ∗2.25e
−3
±6.8e
−4 8.2e
−4
±4.1e
−4 ∗–1.7e
−3
±5e
−4 MAT18
No e. Column (1): name o he galaxy. Column (2): edshi . Column (3): spec al ype. Columns (4)–(9): UV emission line luxes and hei e o s in 1e
−14
e g s
−1
cm
−2
.
Column (10): e e ences: AS88 (Alling on-Smi h e al. 1988 ), BER81 (Be ge on, Maccaca o & Pe ola 1981 ), BES99 (Bes , R
¨
o ge ing & Lehne 1999 ), BES00 (Bes ,
R
¨
o ge ing & Longai 2000 ), BOR07 (Bo nancini e al. 2007 ), BRE00 (De B euck e al. 2000 ), BRE01 (De B euck e al. 2001 ), CIM98 (Cima i e al. 1998 ), GK95
(Gopal-K ishna e al. 1995 ), KI93 (Kinney e al. 1993 ), KRA94 (K aeme e al. 1994a ), LAC94 (Lacy e al. 1994 ), LAC99 (Lacy e al. 1999 ), LIL88 (Lilly 1988 ),
LIL07 (Lilly e al. 2007b ), LIN22 (Lin e al. 2022 ), MAL83 (Malkan & Oke 1983 ), MAT09 (Ma suoka e al. 2009 ), MAT18 (Ma suoka e al. 2018 ), MCC90 (McCa hy
e al. 1990a ), MCC90b (McCa hy e al. 1990b ), MCC91 (McCa hy 1991 ), MCC91b (McCa hy e al. 1991 ), NA06 (Nagao e al. 2006 ), ONO21 (Onoue e al. 2021 ),
RAW96 (Rawlings e al. 1996 ), ROB88 (De Robe is, Hu chings & Pi s 1988 ), R O T97 (Roe ge ing e al. 1997 ), SIM99 (Simpson e al. 1999 ), SNI86 (Snijde s, Ne ze &
Boksenbe g 1986 ), SPI95 (Spin ad, Dey & G aham 1995 ), STE99 (S e n e al. 1999 ), STE02 (S e n e al. 2002 ), SZO (Szokoly e al. 2004 ), TAN22 (Tang e al. 2022 ),
THU84 (Thuan 1984 ), VM99 (Villa -Ma
´
ın e al. 1999 ), VM20 (Villa Ma
´
ın e al. 2020 ), WAD99 (Wadding on e al. 1999 ).
∗s ands o emission lines p o ided by hei
uppe limi .
1
Emission lines gi en as luminosi y in uni s 10
43
e g s
−1
. The comple e e sion o his able is a ailable a he CDS.
emission-line luxes la ge han he e o limi s. On he o he hand, we
assumed ha hese objec s we e no co ec ed o in e nal ex inc ion,
so we pe o med his co ec ion only o hose objec s wi h C3C4 >
−0.3, which canno be conside ed as dus ee. This c i e ion, shown
in Fig. 5 , is consis en wi h he p edic ions om models in which he
absence o in e nal ex inc ion is he only way o achie e he e y low
C3C4 alues obse ed o ce ain objec s (Nagao e al. 2006 ; Do s
e al. 2019 ). Fo highe C3C4 alues, on he o he hand, he absence o
ex inc ion does no seem o be jus i ied in ei he he emission line lux
co ec ion o he pho oioniza ion models. The e o e, o he es o he
compiled galaxies, we assumed an a e age isual ex inc ion o A
V
=
1 mag and an e x inc ion la w o Ca delli, Clay on & Ma his ( 1989 ).
Anyw ay, we check ed ha he assump ion o o he common ex inc ion
laws does no imply a a ia ion in he esul ing chemical abundances
la ge han he epo ed e o s when only UV emission lines a e used
as an inpu , as in ou sample. Fo he case o a simul aneous use o
bo h op ical and UV lines, his co ec ion would be mo e c i ical, bu
a mo e accu a e es ima e o he ex inc ion co ec ion could also be
pe o med using he Balme dec emen .
Some o he lis ed luxes o ain emission lines (e.g. N V λ1239 Å,
[O III ] λ1665 Å) a e gi en as uppe limi s (no ed in Table 2 ). Ho we e ,
hey a e ea ed as eal lines, assuming he uppe limi equi alen o
3 σ, whe e he code conside s as inpu he lux wi h a nominal alue
o 2 σwi h an e o o 1 σ(i.e. an uppe limi o 3 is in oduced as
2 ±1), o educe he p obabili y ha hey a e jus noise, al hough
wi h a es ic ion o alues > 3 σ(i.e. la ge han he uppe limi ) in
ou Mon e Ca lo simula ions).
We applied he HCM-UV code o his sample, assuming an AGN
SED wi h αOX
= −1.2 and a s opping c i e ion in he models o
e
= 98 pe cen . The choice o hese inpu pa ame e s is a bi a y
and co esponds mainly o he same condi ions used o he op ical
e sion o he code in P
´
e ez-Mon e o e al. ( 2019 ), wi h an αOX
alue
close o he median de i ed by Mille e al. ( 2011 ). Anyway, he
e ec s o changing hese pa ame e s on he esul s a e discussed in
Sec ion 2 . The only di e ence is ha we only conside ed dus in
he models o he objec s whose lines we e co ec ed o in e nal
ex inc ion, in ag eemen wi h he c i e ion based on C3C4, while
o he es , we used only he g ids o he models wi hou dus . The
esul ing O/H, C/O, and log U alues a e p o ided in elec onic o m
along wi h he co esponding de i ed e o s, as shown in Table 3 .
4 DISCUSSION
4.1 O/H and U dis ibu ions
In Fig.
7 , we show he de i ed me allici y in he compiled sample
as a unc ion o edshi o he objec s ma ked o di e en
ca ego ies. We see no clea co ela ion wi h edshi , bu a la ge
sca e in he ange o me allici y 7.77 < 12 + log (O/H) <
8.97, wi h a mean o 8.52 ( ∼2/3 Z
), which is sligh ly highe
o he objec s assumed o be dus y (i.e. 8.55) han o he ob-
jec s we assume o be dus ee (8.46). The main end obse ed
in his dis ibu ion would no be hen e y di e en assuming
di e en inpu condi ions in he models chosen o de i e he
abundances.
In Table 4 , we show he mean O/H alues ob ained o each galaxy
class. While hese alues do no allow us o d aw i m conclusions,
since hese ypes a e e y une enly popula ed and some o hem
could be e y he e ogeneous in hei p ope ies, hey can p o ide
some clues. F o e xample, a e y low z, Sey e -2 galaxies ha e
abundances a ound he mean o he whole sample, e en i hei mean
is sligh ly subsola . This is almos cong uen wi h he sample o
adio galaxies a high z, whose mean edshi is highe han ha o
he smalle Sey e 2 subsample. On he o he hand, o he o he
ypes o galaxies a high z, e y di e en alues o O/H a e ound,
as in he case o quasa s, wi h a mean O/H alue signi ican ly lowe
[i.e. 12 + log (O/H) = 8.35] han he mean o he whole sample.
Ho we e , o he o he h ee de ined classes, he numbe o compiled
objec s is so small ha no s a is ically signi ican conclusion can be
d awn.
In Fig. 8 , we show he ela ionship be ween he C3C4 pa ame e
as ob ained om he compiled emission line luxes and he alue o
log U as ob ained om HCM-UV . As desc ibed in p e ious sec ions,
we conside ed all objec s wi h C3C4 < −0.3 o be possibly dus ee,
as discussed by Nagao e al. ( 2006 ) o Do s e al. ( 2019 ). In Table 4 ,
we also gi e he ac ion o objec s in each ca ego y ha all wi hin
Downloaded om h ps://academic.oup.com/mn as/a icle/521/1/1556/7068104 by Ins . As o isica Andalucia CSIC use on 14 May 2024