AEROSOL CHEMISTRY
A surface-promoted redox reaction occurs
spontaneously on solvating inorganic
aerosol surfaces
Xiangrui Kong^1 , Dimitri Castarède^1 , Erik S. Thomson^1 , Anthony Boucly^2 , Luca Artiglia^2 ,
Markus Ammann^2 , Ivan Gladich^3 , Jan B. C. Pettersson^1 *
A surface-promoted sulfate-reducing ammonium oxidation reaction was discovered to spontaneously
take place on common inorganic aerosol surfaces undergoing solvation. Several key intermediate
species—including elemental sulfur (S^0 ), bisulfide (HS−), nitrous acid (HONO), and aqueous ammonia
[NH3(aq)]—were identified as reaction components associated with the solvation process. Depth profiles
of relative species abundance showed the surface propensity of key species. The species assignments
and depth profile features were supported by classical and first-principles molecular dynamics
calculations, and a detailed mechanism was proposed to describe the processes that led to unexpected
products during salt solvation. This discovery revealed chemistry that is distinctly linked to a solvating
surface and has great potential to illuminate current puzzles within heterogeneous chemistry.
G
as–particle interfaces play essential roles
in the atmosphere and directly influence
many atmospheric processes ( 1 ), includ-
ing gas uptake ( 2 ), halogen chemistry ( 3 ),
ozone depletion ( 4 ) and heterogeneous
ice nucleation ( 5 ). Because interfacial processes
take place on molecular scales, classical bulk
thermodynamic theories are often insufficient
to describe interfaces. For example, hygro-
scopic substrates are understood to solvate
at a deliquescence relative humidity (DRH),
but modern observational techniques have re-
vealed that reversible water adsorption and
ion solvation already take place at a relative
humidity (RH) lower than the DRH ( 6 , 7 ). Ex-
planations for such surface-specific behavior
can be rooted in strictly accounting for the
distinctive physicochemical characteristics
of surfaces ( 8 – 10 ).
Redox reactions are essential for many key
processes in the atmosphere and regulate
the formation of gas molecules and aerosol
particles. However, the current redox chem-
istry framework cannot explain the chemical
cycles of some key atmospheric components,
such as nitrous acid (HONO) ( 11 ) and sulfate
( 12 ). Unexpected redox reactions may take
place on surfaces and interfaces, but interfa-
cial processes are challenging to characterize
and have often been overlooked in atmospheric
chemistry. In this study, a surface-promoted
redox reaction during salt solvation was dis-
covered with the use of ambient-pressure x-ray
photoelectron spectroscopy. Classical and first-
principles molecular dynamics (FPMD) simu-
lations were performed to understand and
characterize the solvation process. Ammonium
sulfate is a common atmospheric compound
and was used as the salt in this study, as a
proxy for atmospheric inorganic aerosol ( 13 ).
Figure 1 shows the binding energies mea-
sured using x-ray photoelectron spectroscopy
(XPS) for sulfur 2p and nitrogen 1s electrons at
dry (RH = 3%), pre-deliquescence (RH = 48%),
and full deliquescence (RH = 78%, including
transient and steady state) conditions. At the
lowest RH, the only features observed were
the expected sulfate (SO 42 −) doublet with the
sulfur 2p3/2peak around 167.5 eV (Fig. 1C, spec-
truma) and the ammonium (NH 4 +) peak at
~401.0 eV ( 14 ). As the RH was increased to
48%, additional doublets of reduced sulfur
species (Sr), including elemental sulfur (S^0 with
2p3/2around 163.2 eV) ( 14 ) and bisulfide (HS−
with 2p3/2around 162.5 eV), became visible
(Fig. 1D, spectrumb). Two additional nitrogen-
containing components—nitrous acid (HONO)
at ~404 eV and dissolved ammonia (NH 3 )( 15 )
at ~399 eV ( 14 )—also appeared with increasing
RH (Fig. 1F, spectrumd). At RH = 78%, where
full deliquescence occurred ( 16 ), a full conver-
sion from SO 42 −to S^0 was observed (Fig. 1E,
spectrumg) during the deliquescing transient
period (fig. S1). As the deliquescence process
proceeded and a new steady state was reached,
the SO 42 −reappeared and only a small amount
of HS−remained on the surface. Identification
of the spectroscopic signatures of different
sulfur and nitrogen species has been supported
and/or guided by both molecular dynamics
(MD) and extensive density of states (DOS)
calculations: MD trajectories were consist-
ent with the formation and stability of the
identified experimental species, and theircalculated spectroscopic signatures showed
excellent agreement with the experimental
peak assignments [see the supplementary
materials (SM) for a detailed discussion, as
wellastablesS2andS3].Notably,aseriesof
measurements was performed to evaluate
the influence of x-ray beam damage, and the
results confirmed that beam-induced artifacts
and adventitious carbon had not contributed
to and/or influenced the observations [see
SM for details (supplementary text section
“Exclusion of Contributions from Radiation
Damage and Adventitious Carbon”), as well
as figs. S2 to S5). In addition, oxygen gas was
dosed to check its influence on the redox re-
actions, but no effects from oxygen were ob-
served (fig. S6).
During the initial stage of deliquescence
at RH = 78%, all nitrogen species completely
vanished from the spectrum (Fig. 1B). Photo-
emission wide-range spectra measured during
deliquescence (fig. S7) showed the absence of
nitrogen species and a simultaneous increase
of Srduring the salt solvation. Notably, this
phenomenon, during which the deliquesced
surface was pure water mixed with Sr, was
transient, and the SO 42 −, NH 4 +, and NH3(aq)
species reappeared in 1 hour, once the system
was stabilized (fig. S1). Our MD simulations
(Fig. 2B and fig. S9) showed that a water sub-
monolayer on top of the ammonium sulfate
crystal was sufficient to solvate SO 42 −and NH 4 +
ions from the crystal into the adsorbed liquid-
like layer. Thus, the temporary complete de-
pletion observed by means of XPS during the
transient period indicated that additional
chemical mechanisms were at play.
A sulfate-reducing ammonium oxidation
(SRAO) reaction scheme, involving NH 4 +and
SO 42 −in aqueous systems, has been hypothe-
sized previously to explain nitrogen depletion
and sulfate reduction in anaerobic environ-
ments ( 17 ) and has been widely applied in
water treatment contexts ( 18 ). At present, the
reaction scheme with Gibbs free energy (DG^0 )
is described as ( 17 )3SO^24 þ6NHþ 4 →3S^0 þ3N 2 þ
12H 2 O DG^0 ¼ 135 :9 kJ=mol
ð 1 Þwhich consists of three subreaction steps ( 19 )3SO^24 þ4NHþ 4 →3S^2 þ4NO 2 þ4H 2 Oþ
8Hþ DG^0 ¼þ 1714 :1 kJ=mol
ð 2 Þ2NO 2 þ2NHþ 4 →2N 2 þ
4H 2 O DG^0 ¼ 720 :2 kJ=mol
ð 3 Þ3S^2 þ2NO 2 þ8Hþ→3S^0 þN 2 þ
4H 2 O DG^0 ¼ 1129 :8 kJ=mol
ð 4 ÞAlthough the overall reaction (1) is thermo-
dynamically favorable, it is not spontaneous
because not all of the elementary steps areSCIENCEscience.org 5 NOVEMBER 2021•VOL 374 ISSUE 6568 747
(^1) Atmospheric Science Research Division, Department of
Chemistry and Molecular Biology, University of Gothenburg,
SE-41296 Gothenburg, Sweden.^2 Laboratory of Environmental
Chemistry, Paul Scherrer Institute, CH-5232 Villigen PSI,
Switzerland.^3 Qatar Environment and Energy Research
Institute, Hamad Bin Khalifa University, P.O. Box 31110,
Doha, Qatar.
*Corresponding author. Email: [email protected] (X.K.);
[email protected] (I.G.); [email protected] (J.B.C.P.)
RESEARCH | REPORTS