1
Formation and stabilization of elemental sulfur through
2
organomineralization
3
Julie Cosmidisa*1, Christine W. Nimsa1, David Diercksb, Alexis S. Templetona
4
a
Department of Geological Sciences, University of Colorado, Boulder, CO 80309, USA
5
b
Department of Metallurgical and Materials Engineering, Colorado School of Mines, Golden, CO
6
80401, USA
7
*Corresponding author. Email address:
[email protected]
8
Abstract
9
Elemental sulfur (S(0)) is an important intermediate in the biogeochemical cycle of sulfur that is
10
formed by chemical or biological oxidation of more reduced sulfur species. Given the restricted
11
geochemical conditions under which S(0) should persist, the mechanisms whereby S(0) can be
12
stabilized in the environment are not fully understood. Here we identify a process called “S(0)
13
organomineralization”, by which S(0) minerals are produced and stabilized following the
14
oxidation of hydrogen sulfide in the presence of numerous types of dissolved organics, including
15
simple sugars and amino acids. The S(0) particles formed through this mechanism are closely
16
associated with organics, which often form an envelope around the mineral. The organic envelopes
17
are formed by self-assembly of the dissolved organic molecules in the presence of hydrogen sulfide
18
and oxygen, and play in a role in the stabilization of S(0). Organic compound sulfurization
19
probably plays an important role in the self-assembly mechanism, by causing the polymerization
1
Present address: Department of Geosciences, Pennsylvania State University, University Park, PA 16802, USA
1
20
of relatively small dissolved organic molecules into solid, macromolecular, polymeric organics.
21
The organomineralized S(0) particles present unique and complex morphologies, which are
22
controlled by the type of dissolved organic compound present in the experimental media.
23
Depending on the organics present, organomineralized S(0) can exist as different combinations of
24
several crystal structures, including the non-thermodynamically stable β- and γ-S8 allotropes,
25
which are most likely stabilized by their close association with the organic phase. We propose that
26
complex particle morphology combined with the presence of metastable S(0) allotropes could be
27
used as a signature of S(0) organomineralization in natural settings. S(0) organomineralization was
28
obtained in the laboratory under a wide range of experimental conditions that span across
29
geochemical conditions which can be encountered in many sulfidic environments. It is possible
30
that the reaction between reduced sulfur species and organics may significantly affect the
31
production and preservation of S(0) in numerous natural systems.
32
2
1. Introduction
33 34
1.1.
Elemental sulfur formation in the environment
35
Elemental sulfur (S(0)) is an important intermediate in the biogeochemical cycle of sulfur that is
36
formed by the oxidation of more reduced species such as sulfide (Zopfi et al., 2004). S(0) is found
37
in a variety of Earth surface environments such as marine sediments (Jørgensen and Nelson, 2004;
38
Zopfi et al., 2004) and water columns (Jørgensen et al., 1991; Luther et al., 1991; Findlay et al.,
39
2014), euxinic lakes (Zerkle et al., 2010; Kamyshny et al., 2011), caves (Galdenzi et al., 2008;
40
Hamilton et al., 2015), aquifers (Einsiedl et al., 2015), hydrothermal vents (Taylor et al., 1999;
41
Embley et al., 2007; Findlay et al., 2014; Gilhooly et al., 2014), as well as sub-glacial or hot springs
42
(Boyd et al., 2007; Gleeson et al., 2011; Gleeson et al., 2012; Kamyshny et al., 2014). S(0) is
43
consumed by a wide diversity of microorganisms that can gain energy from its oxidation (Schmidt
44
et al., 1987; Suzuki et al., 1993; Franz et al., 2007; Marnocha et al., 2016), reduction (Boyd and
45
Druschel, 2013) or disproportionation (Thamdrup et al., 1993; Canfield and Thamdrup, 1994;
46
Finster et al., 1998).
47
The chemical oxidation of sulfide to S(0) by molecular oxygen occurs at rates that are several
48
orders of magnitude lower than the rate of microbial sulfide oxidation (Luther et al., 2011), and so
49
it is assumed that in low-temperature environments, most S(0) formation results from microbial
50
oxidation. Many different S-oxidizing bacteria are indeed known to biomineralize S(0) in the form
51
of intracellular or extracellular S(0) globules (Kleinjan et al., 2003), or as extracellular S(0)
52
filaments (Wirsen et al., 2002; Sievert et al., 2007). However, the microbial and enzymatic
53
pathways for S(0) formation are still not completely understood. S(0) is considered a rather
54
unstable and dynamic constituent of the sulfur pool of sediments (Troelsen and Jørgensen, 1982;
3
55
Zopfi et al., 2004), and it is only thermodynamically stable under a very restricted range of Eh and
56
pH conditions (Gleeson et al., 2010). S(0) can however persist and accumulate in some
57
environments where it forms conspicuous deposits, although it is unclear what are the mechanisms
58
responsible for its stabilization.
59
1.2.
S(0) organomineralization: definition
60
Recently, it was discovered that S(0) could be formed abiogenically through the interaction of
61
hydrogen sulfide with dissolved organics in the presence of an oxygen gradient (Cosmidis and
62
Templeton, 2016). The S(0) particles produced through this process presented filamentous and
63
spherical morphologies and were encapsulated within an organic envelope. This reaction, i.e. the
64
formation of S(0) where organics appear to play a necessary role, can be qualified as an
65
organomineralization process.
66
Organomineralization is a relatively recent concept in Earth Sciences, since its first occurrence in
67
the literature of this field only dates back from about two decades ago (Défarge and Trichet, 1995;
68
Trichet and Défarge, 1995). Contrary to microbial biomineralization, which has been studied since
69
at least the late 19th century (Robinson, 1889), the importance of organomineralization in mineral
70
formation in the environment has been widely under investigated. There has been some vigorous
71
debate over the definition of this term (Perry et al., 2007; Altermann et al., 2009; Défarge et al.,
72
2009; Perry and Sephton, 2009; Défarge et al., 2010), but a consensus seems to have been reached
73
in
74
“Organomineralization is the process of mineral formation mediated by organic matter,
75
independent of the living organisms which the organic matter derives from”. The organic
76
compounds may be by-products of biotic activity, relics of dead and decaying organisms, or non-
77
biological
the definition provided by the
organic
compounds.
Encyclopedia
Importantly,
the
of
terms
Geobiology
(Défarge, 2011):
“organomineralization”
and 4
78
“organominerals” should be used only when organic compounds have played an active role in the
79
mineral precipitation, and not in the case when they have been passively entombed or complexed
80
during crystal growth (Défarge et al., 2009).
81
Some organomineralization processes have been thoroughly researched by chemists and material
82
scientists, such as the synthesis of mesoporous organic-silica materials (Hoffmann et al., 2006) or
83
organic–inorganic zeolites (Meng and Xiao, 2014), for applications such as heterogeneous
84
catalysis or sensor technologies. Most studies published by geoscientists so far describing
85
organomineralization processes have focused on the organically-mediated formation of calcium
86
carbonates in modern and ancient microbialites (e.g,, Riding, 2000; Perry et al., 2007; Dupraz et
87
al., 2009; Pacton et al., 2015; Bindschedler et al., 2016; Diaz et al., 2017), However,
88
organomineralization reactions could also play a role in the formation of sulfides (Maclean et al.,
89
2008), and sulfate minerals (Cámara et al., 2016) in a diversity of natural environments.
90
The control exerted by organics on mineral nucleation and/or crystallization and growth has been
91
investigated in many experimental studies with calcium carbonates (Meldrum and Hyde, 2001;
92
Reitner, 2004; Braissant et al., 2007; Gallagher et al., 2013), calcium phosphates (Silverman and
93
Boskey, 2004; Kniep and Simon, 2007; Gallagher et al., 2013; Crosby and Bailey, 2017), silica
94
(Behrens et al., 2007; Delclos et al., 2008; Ahmed et al., 2010), iron oxides (Mirabello et al., 2016),
95
iron sulfides (Grimes et al., 2001), sulfate minerals (Heywood and Mann, 1994; Borah et al., 2006),
96
and metallic nanoparticles (Rashid et al., 2007; Alexandridis and Tsianou, 2011; Shah et al., 2015).
97
Organomineralization often results in the formation of minerals with specific properties such as
98
shape (Cody and Cody, 1991; Mann and Ozin, 1996; Meldrum and Hyde, 2001; Meldrum and
99
Cölfen, 2008; Wu et al., 2010; Biacchi and Schaak, 2015), size (Ananikov et al., 2007; Kuwahara
100
et al., 2014; Nakaya et al., 2014), crystal structure (Tong et al., 2004; Sand et al., 2012; Tobler et 5
101
al., 2014; Lu et al., 2015; Moliner, 2015), chemical composition (Tobler et al., 2015) or even
102
isotopic composition (Harouaka et al., 2017) that differ from minerals precipitated under purely
103
inorganic conditions. Such special properties might be used as signatures of organomineralization
104
processes in the rock record or in extraterrestrial samples (Reitner, 2004).
105
The mechanism of S(0) organomineralization, as well as the potential contribution of this process
106
to environmental S(0) formation, remains elusive. Most of what is known about the geochemical
107
interactions of sulfur with organics has been derived from studies of the sulfurization of organic
108
matter in sediments (e.g., Werne et al., 2004) or water columns (e.g., Raven et al., 2016). Some
109
previous studies have shown that redox-active functional groups in organic matter such as
110
quinones could oxidize sulfide to S(0) and thiosulfate (Heitmann and Blodau, 2006; Yu et al.,
111
2015). However, the mechanism by which hydrogen sulfide reacts with non redox-active dissolved
112
organics to produce S(0) needs to be elucidated. A better understanding of the mechanism of S(0)
113
organomineralization, and of its potential role in the formation and/or preservation of S(0) in the
114
environment, has the potential to reshape our knowledge of the sulfur biogeochemical cycle and
115
its interactions with the carbon cycle. This requires better constraints on the geochemical
116
conditions that are conducive to S(0) organomineralization. It also requires identifying the specific
117
signatures of organomineralized S(0) that might be used to discriminate it from microbially- and
118
inorganically-formed S(0) in the environment.
119
In this study, we performed new experiments where several different types of dissolved organic
120
compounds were reacted with hydrogen sulfide and oxygen, forming organomineralized S(0). In
121
the original experiments by Cosmidis and Templeton (2016), S(0) organomineralization was
122
performed using yeast extract and/or peptone – which are both complex mixtures of organic
123
compounds – as the source of organics. Here we used well-defined organic compounds, such as 6
124
simple sugars and amino acids, or more complex organics representative of natural organic matter
125
(humic acids). The experiments were also performed under a wide range of geochemical
126
conditions, including conditions representative of natural sulfidic environments. The combined
127
results enable us to determine the geochemical parameters and specific types of organics conducive
128
to S(0) organomineralization. We furthermore performed new characterizations of
129
morphology, chemical speciation, and crystal structure of the organomineralization products, that
130
allowed us to identify the specific signatures of organomineralized S(0). These experiments
131
strongly demonstrate that dissolved organic compounds can directly influence the formation and/or
132
stabilization of S(0) under our experimental conditions, and pave the way for future studies that
133
will investigate the prevalence of S(0) organomineralization in the environment.
2. Methods: experimental
134 135
the
2.1.
Gradient tubes experiments
136
S(0) organomineralization experiments were performed in sulfide gradient tubes (Fig. EA-1A)
137
prepared as described in Cosmidis and Templeton (2016). This setup was originally derived from
138
a microbial culture experiment (Gleeson et al., 2011). In short, the gradient tubes consisted of a
139
bottom layer containing an artificial mineral medium (modified EM medium), agar (1% w/v) and
140
Na2S (5 mM), and a top layer containing the same mineral medium supplemented with vitamins
141
and trace elements (see section 2.2 below) as well as different types of dissolved organic
142
compounds at a 5 g.L-1 concentration: D(+)-glucose (Sigma-Aldrich), D(+)-cellobiose (Acros
143
Organics), glycine (Acros Organics), casamino acids (Fisher Scientific), yeast extract (Fisher
144
Scientific) or humic acids (humic acid sodium salt, Aldrich). In the case of humic acids, the top
145
layer solution was filtered through 0.2 μm filters, to remove particulate material. The top layer
7
146
solutions were de-oxygenated by sparging with N2:CO2 (80:20) before being added to the tubes.
147
The use of a N2:CO2 gas mix instead of pure N2 was inherited from the original gradient tube
148
protocol, which was designed for the growth of autotrophic bacteria (Gleeson et al., 2011). The
149
top layer of each tube was oxygen-free when the experiment was started, but oxygen from the
150
atmosphere was allowed to diffuse from the top, creating an oxygen gradient opposing the sulfide
151
gradient. All the solutions, as well as the glass tubes themselves, were sterilized by autoclaving
152
prior to the preparation of the gradient tubes, which was performed under aseptic conditions. The
153
gradient tubes were kept in the dark (to exclude the possibility of photochemical reactions) and at
154
room temperature during the duration of the experiments.
155
2.2.
Batch experiments
156
S(0) organomineralization was also performed in simplified “batch” experiments (Fig. EA-1B) in
157
glass bottles (100 mL to 1 L) containing modified EM medium supplemented with dissolved
158
organics at different concentrations ranging from 0.1 mg.L-1 to 15 g.L-1. Phosphate (K2HPO4) was
159
omitted in order to avoid phosphate mineral precipitation. The organics tested were glucose,
160
glycine, humic acids, and yeast extract. The bottles were autoclaved, allowed to cool down, and
161
sparged with N2:CO2 (80:20) under aseptic conditions, before filter-sterilized solutions of vitamins
162
(Wolfe’s vitamin solution; Atlas, 2010) and trace elements (0.52 g EDTA, 0.15 g FeCl2·4H2O, 7
163
mg ZnCl, 10 mg MnCl·4H2O, 6.3 mg H3BO3, 19 mg CoCl2·6H2O, 1.7 mg CuCl2·2H2O, 24 mg
164
NiCl2·6H2Oand 36 mg Na2MoO4·2H2O per 100mL deionized water) were added, both at 1 mL.L-
165
1
166
The solutions containing humic acids needed to be filtered through 0.2 μm filters (Millipore
167
Steritop bottle top filter unit) to remove particulate material before sulfide was added in. The
. Filter-sterilized Na2S was finally added at final concentrations ranging from 50 μM to 5 mM.
8
168
bottles were loosely capped, allowing for oxygen from the atmosphere to diffuse in. The
169
experiments were stored in the dark at room temperature.
170
Alterations to this protocol were made in order to test the effects of different geochemical
171
parameters on S(0) organomineralization. In a set of experiments, the solutions were not
172
autoclaved but were instead sterilized by vacuum filtration through 0.2 μm filters (Millipore
173
Steritop bottle top filter units). An experiment was performed in which the bottles were stirred
174
with sterile magnetic stirrers, in order to test the effect of agitation (i.e. preventing the formation
175
of sulfide and oxygen gradients) on S(0) organomineralization. In another set of experiments, the
176
N2:CO2 gas mix used to sparge the bottles was replaced with pure N2. Finally, experiments were
177
performed in which the chemical composition of the medium was altered: trace elements and/or
178
vitamins were omitted, or the EM medium was replaced with distilled water, or distilled water
179
with 2.75 g.L-1 NaCl. These different experimental conditions are summarized in Table EA-1.
3. Methods: analytical
180 181
3.1.
Light Microscopy
182
Light microscopy was performed with a Zeiss Axio Imager Z1 on wet samples collected from the
183
gradient tube and batch experiments. For each sample, a ~20 μL drop of liquid was mounted
184
between a glass slide and a cover slip and imaged before it dried..
185
3.2.
Scanning Electron Microscopy
186
Samples from the gradient tubes and batch experiments were collected at different time intervals
187
throughout the course of the experiments, deposited on polycarbonate filters (GTTP Isopore
188
membrane filters, Merck Millipore, pore size 0.2 μm), and rinsed three times with distilled water.
189
The filters were allowed to dry at ambient temperature and were coated with carbon or gold prior 9
190
to analysis. The analyses were conducted on a JSM-7401F field emission Scanning Electron
191
Microscope (SEM) at the Nanoscale Fabrication Laboratory at the University of Colorado at
192
Boulder. Images were acquired in the secondary electron mode with the microscope operating at
193
3 kV and a working distance (WD) of ~3 mm, and in the backscattered electron mode at 10 kV
194
and WD ~8 mm. Energy-Dispersive X-ray Spectrometry (XEDS) analyses were performed at 20
195
kV and WD ~8 mm.
196
3.3.
Transmission Electron Microscopy
197
Samples were collected by centrifugation from batch experiments, and rinsed three times with
198
deionized water. They were then deposited on 200 mesh Cu grids covered with a lacy
199
formvar/carbon film (Ted Pella), and allowed to air-dry at ambient temperature.
200
A Philips CM200 transmission electron microscope operated at 200keV was used for imaging and
201
selected area electron diffraction (SAED) of the sulfur particles. The particles proved to be fairly
202
sensitive to electron beam exposure, with S(0) sublimating within a few minutes in the high-
203
vacuum TEM chamber while being analyzed. The beam intensity was thus decreased through
204
adjusting the spot size to allow more time for imaging and diffraction. A series of particle images
205
and diffraction patterns were collected. The d-spacings, d-spacing ratios, and angles between
206
planes observed in the diffraction patterns were compared to the known crystal structures of α-, β-
207
, and γ-cyclooctasulfur from the American Mineralogist Crystal Structure Database. From this,
208
diffraction patterns of several particles were indexed using the JEMS software package
209
(http://www.jems-saas.ch/).
210
3.4.
Scanning Transmission X-ray Microscopy
10
211
Samples from the gradient tubes experiments were rinsed three times in deionized water and a
212
small drop (~2 μL) was deposited on a Formvar-coated 200 mesh Cu TEM grid (Ted Pella) and
213
allowed to air-dry at ambient temperature. Scanning Transmission X-ray Microscopy (STXM)
214
analyses were performed on beamline 10ID-1 (SM) of the Canadian Light Source (Saskatoon,
215
Canada) (Kaznatcheev et al., 2007) and beamline 11.0.2 of the Advanced Light Source (Berkeley,
216
USA) (Bluhm et al., 2006). The X-ray beam was focused on the samples using a Fresnel zone plate
217
objective and an order-sorting aperture yielding a focused X-ray beam spot of ~30 nm. After
218
sample insertion in STXM, the chamber was evacuated to 100 mTorr and back-filled with He at
219
~1 atm pressure. Energy calibration was achieved using the well-resolved 3p Rydberg peak of
220
gaseous CO2 at 294.96 eV. Images, maps and image stacks were acquired in the 260-340 eV (C
221
K-edge)
222
(unicorn.mcmaster.ca/aXis2000.html) was used for data processing. A linear background
223
correction was applied to the spectra at the C K and S L-edges, in the 260-280 eV region and 155-
224
160 eV region, respectively, to eliminate the contribution of lower energy absorption edges. All
225
spectra were normalized to the same maximum.
226
Maps of organic C were obtained by subtracting an image obtained at 280 eV (i.e. below the C K-
227
edge) and converted into optical density (OD) from an OD-converted image at 288.2 eV (1s →π*
228
electronic transitions in peptides). Maps of S were obtained by subtracting an OD-converted image
229
obtained at 160 eV (i.e. below the S L-edge) from an OD-converted image at 163.5 eV (energy of
230
the S L3-edge). X-ray absorption near edge structure (XANES) spectra were extracted from image
231
stacks as explained in Cosmidis and Benzerara (2014). The S L-edge spectra of a reference
232
elemental sulfur compound (precipitated sulfur, Fisher Scientific) as well as of several
233
organosulfur compounds (L-cystine, L-methionine, and L-cysteine) were also acquired using the
and
155-190
eV
(S
L-edge)
energy
ranges.
The
aXis2000
software
11
234
same methods. Similarly, the C K-edge spectra of reference organic compound powders (glucose,
235
cellobiose, glycine, casamino acids, and humic acids), were also obtained.
236
3.5.
Fourier-Transform Infrared Spectroscopy
237
Samples from the batch experiments were harvested by centrifugation, rinsed three times with
238
deionized water, and air-dried. The powdered samples were then pressed into pellets in
239
approximately 5% KBr. Fourier-Transform Infrared (FTIR) spectra were recorded between 4000
240
and 400 cm−1 on a Thermo Nicolet NEXUS 670 FTIR spectrometer. Each spectrum was an
241
integration of 150 spectral scans, with a wavenumber resolution of 1 cm−1. Background corrections
242
were applied to the data following each measurement to compensate for instrumental noise and
243
contributions from atmospheric CO2 and H2O by dividing the absorbance of the sample spectrum
244
by the background spectrum at each data point. The spectra of reference organic compounds
245
powders (glucose, glycine, yeast extract and humic acids) were acquired using the same method.
246
For comparative purposes, all the spectra were normalized to a maximum absorbance of 1.
247
3.6.
X-ray Diffraction
248
Samples from the batch experiments were rinsed three times and concentrated in deionized water.
249
The slurry was pipetted on a Si wafer to form a thin layer on the wafer surface. The wafer was
250
gently heated on a slide warmer to dry the sample and create a thin film of sample on the wafer
251
surface. In order the check to stability of the sulfur phases in the atmosphere, some of the samples
252
were “aged” by letting them air dry on the Si wafer surface for ten days before being analyzed.
253
Samples were analyzed using a Siemens D500 X-ray diffractometer from 5 to 65 degrees two theta
254
using Cu Kα X-ray radiation, with a step size of 0.02 degrees and a dwell time of 2 seconds per
12
255
step. Different S0 crystal structures were identified using Jade software (MDI, version 9) and the
256
International Centre for Diffraction Data (ICDD) 2003 database.
257
3.7.
Geochemical profiles in gradient tube experiments
258
The concentrations of dissolved sulfide and oxygen, as well as the pH, were measured at different
259
time intervals in gradient tube experiments performed either without added organics, or with added
260
glucose, glycine, humic acids or yeast extract.
261
The dissolved hydrogen sulfide (H2S) and oxygen profiles were measured using Unisense (Aarhus,
262
Denmark) H2S-100 and OX-100 microsensors, respectively. The oxygen microsensor was
263
calibrated using freshly prepared de-oxygenated water (0% oxygen saturation) as well as a well-
264
aerated water (100% oxygen saturation). The H2S microsensor was calibrated using freshly
265
prepared de-oxygenated solutions of Na2S in an acetate buffer (pH