Role of Heteroatoms in S,N-Codoped Nanoporous Carbon Materials in CO2 (Photo)electrochemical Reduction

Article abstract of DOI:10.1002/cssc.201801073

Thiourea-modified wood-based activated carbon materials were evaluated as catalysts for the CO₂ electrochemical reduction reaction (CO₂ERR). The materials obtained at 950 °C showed long-term stability. The results indicated that thio

Full text of DOI:10.1002/cssc.201801073

Accepted Article
Title: Role of heteroatoms in S, N-codoped nanoporous carbons in
CO2 (photo)electrochemical reduction
Authors: Teresa Bandosz and Wanlu Li
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemSusChem 10.1002/cssc.201801073
Link to VoR: http://dx.doi.org/10.1002/cssc.201801073
A Journal of
www.chemsuschem.org

10.1002/cssc.201801073
ChemSusChem
Role of heteroatoms in S, N-codoped nanoporous carbons in CO2
(photo)electrochemical reduction
Dr. Wanlu Li, and Prof. Teresa J. Bandosz*
Department of Chemistry and Biochemistry, The City College of New York, New York, NY 10031,
USA, and Ph.D. Program in Chemistry, The Graduate Center of the City University of New York,
New York, NY 10016, USA
* Whom correspondence should be addressed to: Tel; (212) 650-607; Fax: (212) 650-6107;
tbandosz@ccny.cuny.edu
Abstract
Thiourea-modified wood-based activated carbons were evaluated as catalysts for CO2
electrochemical reduction reaction (CO2ERR). The materials obtained at 950oC showed a long
stability. The results indicated that thiophenic sulfur provides catalytic activity for CO formation.
However, it was not as active for CH4 formation as was pyridinic-N. Tafel plots suggested that the
nanoporous structure enhanced the kinetics for CO2 reduction. The electric conductivity limited
the activity for CO2ERR in the materials modified at 600, 800 and 900oC. The effect of visible
light on CO2ERR was also investigated in this study. Upon irradiation, photocurrent was generated,
and a current density increased during CO2 reduction process. Combined with a band-gap
alignment, the results indicate that thiophenic-S in the carbon matrix contributed to sample’s
photoactivity in visible light. These species enhance the overall reduction process promoting both
hydrogen evolution reaction and CO2 reduction to CO.
1
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
Introduction
The rising concentration of CO2 in the atmosphere becomes an environmental issue, which leads
to the global climate change. It also results in an increased ocean acidification affecting many
marine ecosystems. Especially it is detrimental for the creatures with calcium carbonate in their
shells or skeleton. One of the way to decrease/eliminate these effects is CO2 electrochemical
reduction. It emerged as a potential strategy for achieving a carbon-neutral cycle. The key
challenge for this application is to develop selective, stable, efficient and inexpensive
electrocatalysts. Metal-free nitrogen-doped carbon nanotube,[1–5] nanofibers[6], nanowires[7],
graphene[8,9] and carbon frameworks[10] have been studied in recent years as catalysts for the
electrochemical CO2 reduction. It has been reported that CO is formed on N-doped carbon
nanofibers and N-doped carbon nanotubes (NCNTs).1-4 Formate is another desired product and
Wang and co-workers[8] reported that the Faradaic efficiency (FE) for formate formation on N-
doped graphene reached its maximum of 73% at −0.84 V vs. RHE. The catalyst indicated stability
for 12 hours when tested at -0.84 V. It showed enhanced reduction reaction kinetics, increased
selectivity and improved stability of N-graphenes during electrocatalysis in comparison to those
measured on unmodified graphene oxides. It was also found that the pyridinic-N configuration
provides catalytic activity. N-doped graphene quantum dots were also reported as highly active
catalysts for a hydrocarbon and multi-carbon oxygenates formation when used as for CO2ERR.
They showed the total 90% FE for the reduction CO2 process, which includes the 45% FE for the
CO formation and the 45% of FE for the C2 products (ethylene and ethanol) formation. It was
concluded that the pyridinic-N groups at the edge site were more active to induce C–C bond
formation than those at basal planes. This led to a higher yield of C2 and C3 products than those
on N-doped graphene oxides materials.
2
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
Recently, porous carbons have gained attention as potential catalysts for the CO2 electrochemical
reduction (CO2ERR). Song and coworkers[11] studied a series of mesoporous N-doped carbons
with highly uniform cylindrical pore structures. The Faradaic efficiency for ethanol formation was
77% at -0.56 vs. RHE with a 6h stable performance. The cylindrical pore structure facilitated the
dimerization of the key CO* intermediate resulting in the formation of ethanol. The DFT
calculation suggested the both pyridinic-N and pyrrolic-N sites were the active sites for its
formation. The C-C bond formation was more favorable at pyridinic-N sites than at pyrolic-N.
Attempts were also made to explore CO2 electroreduction on nanoporous carbon membranes to
avoid an addition of any binder. Wang and coworkers[12] studied N-doped nanoporous
carbon/carbon-nanotube composite membrane as binder-free electrodes for electrochemical
reduction of CO2. Hierarchically structured porous N-doped carbon membranes (HNCMs)
provided a large surface area with incorporated active nitrogen species. CNTs were processed into
a membrane to further improve an electron transfer efficiency. The composite showed the FE for
the formate formation of 81% at -0.8V vs RHE and maintained it for 36 h without any decay in
the performance. It is found that pyridinic-N was active and selective for the reduction process.
Wood-based N-doped nanoporous carbons as CO2 electroreduction catalysts have been studied by
our group.[13] The best sample showed the FE for CO and CH4 formation of 24% and 0.76%,
respectively. Besides, on the electrochemically reduced form of the carbon catalyst the FE for CO
and CH4 increased to 39 % and 1.2%, respectively. The results supported the previous finding
indicating that pyridinic-N was the most active site for CO2ERR among all the nitrogen-containing
functional groups present on the surface of carbons. The surface basicity and a high volume of
ultramicropores improved the CO2 reduction process. There was an indication that a nanoporous
structure is crucial for the CH4 formation.[13] It was suggested that the ultramicropores work as
3
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
Fisher-Tropsch nanoreactors where methane is formed owing to a strong adsorption of CO* and
an involvement of hydrogen from hydrogen evolution.
In another study molten salt synthesized N-doped carbons were used as CO2 electrocatalysts. They
showed a long stability (48h) with the FE for CO formation of 22%.[14] The results suggested that
a unique pyrazinic-N species on the surface might act as sacrificial agents preventing the active
site, especially pyridinic-N, from being oxidized during the reduction process. These species might
also promote a C-C bond formation since C3 products (acetone and propanol) were detected in the
electrolyte.
Sulfur containing catalysts have seldom been tested for CO2ERR. Metal-sulfur clusters have been
studied around 30 years ago. The presence of the [Fe4S4(SR)4]2- (R=PhCH2 or But) cluster shifted
0.7V to a positive direction compared to a potential without the cluster on an mercury electrode in
dimethylformamide (DMF) electrolyte.[15] The products were mainly formate and phenylacetate.
The optimum efficiency for the formate formation was 27% on the [Fe4S4(SCH2Ph)4]2- cluster.
[M2Fe6S8(SEt)9]3-(M=Mo or W) also showed the 0.5V potential shift to the positive direction for
the formate formation.[15] On rhodium cluster [(RhCp*)3(u3-S)2]2+ complex formate was detected
at -1.5V vs SCE in the Bu4NBF4 containing CH3CN electrolyte. Besides, in a LiBF4 containing
CH3CN electrolyte, an oxalate formation was reported.[16] It was proposed that both low valent
metals and electron rich sulfur ligands are the possible active sites for CO2 reduction on the metal-
sulfur clusters.
Recently, nitrogen and sulfur co-doped iron-containing highly porous carbon-gel has been studied
for the CO2 electroreduction reaction in an aqueous electrolyte.[17] CO was the main product and
trace of methane was also detected. The catalyst showed the highest FE of 85% at -1.3V vs. RHE
for CO. However, the role of sulfur in the CO2ERR was not thoroughly discussed. Metal-free
4
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
sulfur containing nanoporous carbon has been tested in our previous study.[18] The Faradaic
efficiency for the CO and CH4 formation was only 2 and 0.1%, respectively. Even though the FE
was very small, the XPS analysis suggested that the thiophenic-S was one of the activate sites for
CO2ERR. It is important to mention that sulfur-containing carbons have been reported as
photocatalysts for various applications.[19–24] The sulfur in thiophenic configurations was indicated
as the most crucial for photoactivity.[23] Therefore, it would be interesting to explore the effect of
light on the electrochemical reduction of CO2 on the sulfur-containing carbon catalysts.
The objective of the paper is to evaluate the role of sulfur species present in the microporous carbon
matrix in the process of the CO2 (photo) electroreduction. A series of microporous carbons
containing both nitrogen and sulfur was prepared. The samples’ porosity and surface chemistry
were extensively examined. The trends in the catalytic behavior were analyzed and the role of
various nitrogen and sulfur configurations, especially the latter species, was pointed out. Since
photoactivity can be helpful for CO2 reduction, the effect of visible light exposure on the samples’
performance as CO2 reduction catalysts has been also evaluated.
Results and discussion
The surface chemistry of the initial activated carbons was analyzed by X-ray photoelectron
spectroscopy (XPS). As seen from Table 1, all BAX samples contain C, O, N and S. With an
increase in the heating temperature the nitrogen and sulfur content decreased. While the content
of sulfur decreased gradually, a dramatic drop in the content of nitrogen was found at 900oC. Then
the nitrogen content remained almost constant with further carbonization at 950oC. Interestingly,
o
the highest content of oxygen is in the sample heated at 800 C, which indicates that this
temperature results in the highest oxidation level of the BAX carbons modified with thiourea.
5
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
Table 1. Atomic concentration of elements on the surface determined by XPS analysis of the
samples studied.*
Sample
(atomic %)
C
O
N
S
82.8
81.4
81.2
84.2
88.3
88.2
86.6
89.3
8.2
5.8
4.1
3.6
3.0
1.3
1.2
1.5
1.5
3.2
2.1
2.2
1.0
1.6
0.9
1.2
0.6
BAX-TU-600
BAX-TU-600-E
BAX-TU-800
BAX-TU-800-E
BAX-TU-900
BAX-TU-900-E
BAX-TU-950
BAX-TU-950-E
12.4
13.0
11.8
8.8
9.7
10.7
8.6
* The contribution from fluorine to the carbon 1s XPS spectrum has been subtracted.
The detailed deconvolution of C 1s, O 1s, N 1s, S 2p core energy level spectra (Figure 1) revealed
the differences in the specific configurations of the surface groups. Taking into account the
deconvolution of C 1s core energy level spectra (Figure S1 of the Supporting Information, Table
2) and the atomic concentration (at. %) of carbon, the contribution of sp2 carbon in BAX-600, 800,
900 and 950 are 60.9 at. %, 63.7 at. %, 67.0 at. %, 67.0 at. %. This suggests the number of carbon
atoms involved in C-C bonds gradually increased with an increase in the carbonization temperature.
For the sample heated to 950oC, the amount of C-C bond remained the same at that in the sample
heated to 900oC, indicating similar carbonization level of the carbon matrices. Based on the
deconvolution of O 1s core energy level spectra (Figure S1 of the Supporting Information, Table
2) and the surface elemental composition, and knowing that the contributions of O=C and C-O in
BAX-TU-800 are similar to those in samples heated to 900oC and 950oC, the actual atomic
concentration of the oxygen species on the BAX-TU-800 surface is higher than on those of other
6
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
samples. The marked contributions of carboxylic and carbonyl groups on this sample are expected
to bring the acidic character to its surface.
The deconvolution of N 1s core energy level spectra (Table 2, Figure 1) revealed a transformation
of nitrogen functionalities with an increase in the treatment temperature. The carbon heated at
600oC showed mainly pyridinic (N-6) and pyrrolic (N-5) nitrogen. The treatment at 800oC resulted
in a decrease in the contribution of N-5 and the appearance of quaternary-N (N-Q) species. These
results suggest that N-5 is converted to N-Q at 800oC. Moreover, the ratio of the contributions of
N-Q to N-6 increased from 0.36 at 800oC to 3.2 at 950oC indicating that N-6 is partially converted
to N-Q at a high temperature. This process is accompanied by a significant loss of oxygen during
heating, which was also reported from other studies on nitrogen containing carbons[25]. Both
atomic concentration and relative concentration of pyridinic-N decreased with an increase in the
heating temperature. Heating at 950oC resulted also in the formation of N-oxides.[26]
7
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
Figure 1. N 1s and S 2p core energy level spectra of the initial sample
8
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
Table 2. The results of deconvolution of C 1s, O 1s, N 1s, S 2p core energy level spectra for the carbons studied.
BAX-TU- BAX-TU- BAX-TU- BAX-TU-
BAX-TU-
900
BAX-TU- BAX-TU- BAX-TU-
Energy, eV Bond assignment
600
600-E
800
800-E
900-E
950
950-E
284.7-284.8 C-C (sp2, graphitic carbon)
73.5
80.8
78.5
73.3
75.9
15.5
77.2
77.4
66.7
C 1s
C-O, (in phenolic, alcoholic,
286.3-286.4
16.4
17.5
1.7
15.2
24.1
2.6
19.8
3.0
14.0
31.5
1.8
etheric), C-N
288.0-288.2 C=O (in carbonyl or quinone) 6.7
289.6-290.1 O-C=O (in carboxyl or ester) 3.4
4.3
2.0
5.4
3.2
5.6
3.0
O=C (in carbonyl/
O 1s
quinone) or O=S (in
sulfoxides/ sulfones) or
O=N
531.9-532.2
73.6
78.8
21.2
83.7
16.3
71.7
28.3
80.6
78.9
21.1
85.9
75.1
24.9
O-C (in phenol/epoxy or
533.4-533.8 O-S (in thioethers
/sulfonic) or O-N
16.4
19.4
39.1
14.1
18.0
398.7-398.8 N-6 (in pyridines)
400.4-400.6 N-5 (in pyrroles)
401.1-401.2 N-Q (in quaternary)
70.0
30.0
52.8
47.2
52.0
29.0
19.0
27.0
50.0
23.0
31.3
47.7
20.9
18.3
70.8
N 1s
60.9
57.8
24.2
403.2
N-oxides
10.8
57.1
R-S-S-, C-S-C (in
bisulfides/thiophenes)
S
2p3/2
164.0-164.3
82.1
56.7
88.7
11.3
73.9
85.9
14.1
61.8
73.4
26.6
R2-S=O (in sulfoxides) or
R-SO2 (in sulfones)
167.2-168.2
169.8
9.9
8.0
21.1
22.2
16.2
9.9
21.8
16.4
39.9
11.0
9
R-SO3H
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
The deconvolution of S 2p energy level spectra of the sample modified at 600oC (Table 2, Figure
1) indicates that the sulfur species on the surface are mainly in thophenic-S, sulfoxides (or sulfones)
and sulfonic acid configurations. Further treatment at 800oC decreased the contribution of the latter
two species. However, with the further increase in the heating temperature, the contribution of
thiophenic-S decreased which was associated with an increase in the contribution of
sulfoxides/sulfones.
The texture of the carbons has been also studied in detail. The parameters of the porous structure
are presented in Table 3. All BAX carbons are micro/mesoporous with 62%-66% of the total pore
volume existing in micropores. The impregnation with thiourea followed by heat treatment at
various temperatures resulted in an alteration in the porous structures. Generally, an increase in
the heating temperature led to a decrease in the surface area and total pore volume. Especially after
heating at 950oC, the surface area and total volume decreased considerably. When the heating
temperature reached 900oC and above, the microporous volume increased because of the partial
gasification of the carbon and removal of the carbon atoms from the pore walls together with
nitrogen, sulfur and oxygen atoms.
Table 3 also includes the conductivity of the carbon chips. The electric conductivity increased
exponentially with an increase in the carbonization temperature, which is attributed to the
transformation of disordered carbons to their higher order of aromaticity and condensation. Based
on the contributions of C-C bonds from the XPS analysis, the sample heated at 600 and 800oC
showed less carbon in the sp2 configuration than those of 900 and 950oC, which is consistent with
the lower conductivity of the former samples. The lower conductivity of the sample heated to
900oC in comparison to that heated at 950oC could be due to the higher sulfur content in the carbon
matrix of the former sample. It is possible that the slight expansion of the d spacing in the carbon
10
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
matrix by sulfur decreases conductivity.[26] Both, the carbon in sp2 configurations and surface
chemistry affect the conductivity of the carbons.
Table 3. The parameters of the porous structure for the materials studied.
SBET
Vt
Vmeso
V<0.7 nm V<1 nm Vmic
Conductivity
Sample
(m2/g) (cm3/g) (cm3/g) (cm3/g) (cm3/g) (cm3/g) Vmic/Vt (S/m)
BAX-TU-600
1370
0.849 0.326 0.163 0.286 0.523 0.62
24±3
BAX-TU-800 1354
BAX-TU-900 1356
0.840 0.317 0.188 0.294 0.523 0.62
77±4
0.823 0.299 0.189 0.314 0.533 0.65
277±8
BAX-TU-950
1146
0.690 0.234 0.173 0.261 0.456 0.66
400±10
Figure 2. a) N2 adsorption isotherms on the samples prepared at various temperatures; b) Pore
size distributions; c) The relationship between the conductivity of the BAX-TU samples and
the heat treatment.
The SEM and HR-TEM images revealed that our samples typical amorphous carbons from the
view point of their microstructure. The examples of the images for BAX-TU-800 are included in
Figure 3.
11
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
b)
a)
Figure 3. SEM (A) and HR-TEM (B) images of the BAX-TU-800 sample.
To evaluate the acid-base properties of the carbons’ surfaces, potentiometric titration experiments were
carried out. Table 4 summarizes surface pH and pKa values for the species present on the carbons’ surfaces.
All four carbons show similar types of acidic groups but differ in their amounts. On BAX-TU-800 the
largest number of acidic groups is detected, especially in the category of strongly acidic groups with the
pKa<7. As the XPS analysis indicated, the high amount of strong acidic groups on BAX-TU-800 are
assigned to carboxylic acids. These acidic groups were decomposed at high temperature. The low surface
pH of BAX-TU-950 are due to the formation of strongly acidic groups, which corresponds to the presence
of the N-oxides groups (pKa=0.79).[27] as indicated from the surface XPS results. This species is outside the
detection range of the titration technique applied.
12
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
Figure 4. a) Proton binding curves; b) pKa distributions of the studied samples.
Table 4. Surface pH values, peak positions and numbers of groups (in parentheses, [mmol/g])
for the material studied.
Strong
Surface
pH
All
Acidic
Groups
(pH<7)
Sample
pKa 4-5 pKa 5-6 pKa 6-7 pKa 7-8 pKa 8-9 pKa 9-10 pKa 10-11
6.01
Groups
4.69
(0.044) 7.68
9.26
BAX-TU-600 4.86
BAX-TU-800 4.60
0.240
0.288
0.582
0.619
(0.080)
6.83
(0.080)
(0.262)
(0.116)
4.20
5.44
6.83
8.12
9.27
10.15
(0.090) (0.065) (0.133)
(0.088) (0.141)
(0.103)
6.01
4.75
(0.036)
6.96
8.29
9.57
BAX-TU-900 4.62
BAX-TU-950 3.75
0.252
0.256
0.534
0.560
(0.071)
(0.086) (0.196)
(0.145)
4.56
5.73
6.92
7.86
9.37
10.17
(0.059) (0.030) (0.167) (0.070)
(0.188)
(0.045)
The CO2 electroreduction was monitored under controlled potentials and the FE for CO2 reduction
were calculated based on the moles of products detected. Figures 5a and 5c show the trends in the
FE for CO and CH4 formation at the 5h reduction process. All potentials reported here are
referenced to a reversible hydrogen electrode (RHE). CO was the main reaction product of
CO2EER. The potential at which the specific product was first detected was considered as an onset
13
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
potential. As seen in Figure 5a, CO was first detected at -0.37V on BAX-TU-950, which
corresponds to the onset overpotential of 0.27V for the CO formation given the equilibrium
potential of -0.11V (vs. RHE). This onset potential is lower than those reported for metal
electrodes, metal nanostructured, and metal-free electrocatalysts.[1,28,29] For example, the onset
potential for copper was reported to be -1V.[28] The onset potential of nanostructured Ag and
nitrogen-doped carbon nanotube catalysts were reported as 0.6V and 0.5 V, respectively.[1,29] The
low onset overpotential on BAX-TU-950 could be due to its porous structure that favors CO2
reduction via employing the adsorption forces.[18] The FE for CO formation increased with the
potential becoming more negative. It reached the maximum of 29% at -0.67 V at the 5 h reduction
process. In comparison to the previous studies,[13,18] BAX-TU-950 showed the highest FE for CO
among all nanoporous carbons used as-received, without an additional step of the surface
electroreduction.[13,30] In our previous study, one of the samples tested was obtained at a similar
treatment procedure as was BAX-TU-950 but it was impregnated with melamine instead of
thiourea.[13] The actual atomic concentrations of pyridinic-N, pyrrolic-N, quaternary-N and N-
oxides on that carbon were 1.3 at. %, 1.3 at. %, 0.87 at. % and 0.43 at. %, respectively (Figure. 6).
Its best FE for the CO formation was 24% at -0.66V at 6h CO2 reduction. Thus, the sample treated
with thiourea overperforms that one treated with melamine. The XPS data of thiourea treated
BAX-TU-950 indicates actual atomic concentrations of pyridinic-N, quaternary-N and N-oxides
as 0.27 at. %, 0.87 at. % and 0.36 at. %, respectively. The quaternary-N, which was identified
previously as catalytically active, is present on the surface in a similar quantity as that on the
melamine treated samples. Interestingly, pyridinic-N that is the most active for the CO2 reduction
is present in lesser quantity than that on the surface of the melamine treated sample. This suggests
that besides pyridinic-N and quaternary-N there are other active species catalyzing the CO
14
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
formation on the surface of BAX-TU-950. This might be related to the activity of the sulfur-
containing species. It has been reported previously that thiophenic-S group brought the activity for
CO2 reduction.[18] The thiophenic-S induces slightly positively charge to the neighboring
-
carbon,[31,32] which can stabilize the COOH*/CO2 ·intermediate. Sulfur in such configuration can
be present in small pores, similar in size to CO2 molecules and can increase the strength of the
interactions of the intermediates with the carbon surface in these pores.
Figure 5. a) Faradaic efficiency for CO formation b) partial current density of CO c) Faradaic
efficiency for CH4 formation at 5h in CO2-saturated 0.1 M KHCO3
Figure 6. The atomic concentration/distribution of nitrogen species on the surface BAX treated
with melamine and thiourea. The data for BAX-M-950 is taken from Ref. [9].
15
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
At more negative potentials, the FE of CO formation decreased, which is attributed to higher
extent of hydrogen evolution reaction (HER) than that of CO2ERR.[29,33] The partial CO current
densities (jco) as a function of the applied potential were extracted from the data of the controlled
potential electrolyses (Figure. 5b). As seen, the kinetics of the reduction process for the CO
formation increases linearly as the applied potential becomes more negative than -0.5V. A similar
linear relationship for heteroatom doping into carbon materials has been reported in the reduction
of CO[1,2,6] and formate[8]. The FEs for CO decreased dramatically for the samples treated at low
temperatures. They are 15.0%, 6.3% and 4.8%, BAX-TU-900, 800 and 600, respectively, at 5h
reduction at the potential of -0.67V, which is the potential of the maximum FE for CO on BAX-
TU-950.
Even though the lower temperature of the heat treatment preserved more heteroatoms (both
nitrogen and sulfur) in the carbon matrix, the conductivity of BAX-TU-600 and 800 samples is
low, which likely hinders the reduction process involving the electron transfer to the electrode
surface. Thus, in spite of the favorable surface chemistry of these samples, their performance might
be greatly limited by the low electric conductivity. The results suggest that the most developed
porous structure of BAX-TU-600 cannot compensate the limitation of conductivity. These results
also suggest that CO2ERR is not directly related to the porosity.
CH4 was formed on BAX-TU-950 with the maximum FE of 0.23% at 0.54V, which corresponds
to the overpotential of 0.37V. The change in the efficiency with the potential is very similar to that
for the CO formation. The FE of BAX-TU-950 for CH4 formation is the highest among all tested
samples. Besides, its overpotential is lower than that of BAX-TU-900. In comparison to our
previous results,[13] the samples addressed here are less favorable for the CH4 formation than were
the samples of BAX modified with melamine. The highest FE of CH4 for the melamine-modified
16
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
sample was reported as 0.76% at 6h at -0.66V. Since both melamine and thiourea treated samples
showed similar amount of quaternary-N and N-oxides, the lower FE for the methane formation
could due to the smaller amount of pyridinic-N on BAX-TU-950. Thus, CO is less likely to be
stabilized and further reduced to CH4 without pyridinic-N. It also infers that thiophenic-S is not as
active as is pyridinic-N in the stabilizing CO* intermediate and thus less CH4 is formed. Since the
FE for CH4 is smaller on BAX-TU-950 than that on BAX-M-950, in spite of the higher volume of
micropores in the former sample, more pyridinic-N located in micropores might compensate the
enhancing pore size/volume effect on methane formation in the case of BAX-M-950.
Another important feature of an ideal CO2 reduction electrocatalyst is its stability, given the fact
that many polycrystalline metals[34] deactivate in the minute timescale in this catalytic process. We
have tested the long-term durability of our carbon catalyst at a constant applied electrode potential
of −0.67 V (Fig .5). The best stability was found for BAX-TU-950 with a current density remained
steady at -1.5mAcm-2 throughout the 40h period (Figure S3). However, the FEs of CO formation
first increased and then decreased with the progress of the reduction time. As seen in Figure 8a,
the FE of CO for BAX-TU-950 increased from 23% (1h) to 29% (5h) within the first 5h, and then
gradually decreased during the reduction process. The sample was detached from the current
collector after 40h reduction, which ceased its catalytic performance. On the other hand, the FE of
CH4 for BAX-TU-950 reached its maximum after 24h and then gradually decreased. The stability
dramatically decreased for BAX-TU-600, 800 and 900. They lasted for about 18h, 23.8h and 21h,
respectively. Figure 8d shows the LSV curves of our catalysts in N2-saturated KHCO3 electrolyte,
at the potential of -0.67V. The higher current density of BAX-TU-600, 800 and 900 indicates the
higher HER activity than that of BAX-TU-950. The results indicate that when the catalyst is more
favorable to HER, its stability for CO2 reduction decreases. The detachment of the catalyst film
17
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
from the current collector was observed, which could be due to the mechanical strain and stress
created from the rapid release of H2 product from HER.
After CO2 reduction, a marked decrease in the content of sulfur is found for all samples, likely as
a result of surface oxidation during CO2 reduction (Table 1). As indicated previously, sulfur
species incorporated to the carbon matrix are much less stable than are nitrogen ones [13, 18, 35, 36]
.
Using the catalyst in CO2ERR caused a slight reduction of the surface of all samples but BAX-
TU-600, for which the content of oxygen increased. The small change in the content of nitrogen
indicate the stability of this element in the carbon matrix.
The deconvolution of C 1s core energy level spectra of the used catalysts confirms the general
trend in surface reduction. A detailed look at N 1s core energy level spectra shows that while the
contribution of pyridines decreased for BAX-TU-600 and BAX-TU-800, their amounts are very
stable for BAX-TU-900 and BAX-TU-950. Since these species were indicated before as most
catalytically active for CO2 reduction[13,14,18] these XPS results, combined with the good
performance of these samples, support the mechanism proposed previously on the important role
-
of positively charged centers in stabilization of COOH*/ CO2 · intermediates in CO2
electroreduction.[13, 18] On the other hand, sulfur in the best performing showed the highest level of
[13]
oxidation and the reduced sulfur species, besides providing centers for CO2 reduction
also protect nitrogen catalytic sites by undergoing oxidation reactions.
might
Since we hypothesized that the adsorptive interactions are important for CO2 electroreduction, CO2
adsorption isotherms at 20 oC were measured up to atmospheric pressure and they are presented
in Figure 7. Even though BAX-TU-600, which has the smallest volume of micropores, revealed
the smallest quantity of CO2 adsorbed, the lack of a direct relationship between the amount
adsorbed and the volume of small pores indicates the importance of N- and S- rich surface for
18
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
CO2 specific interaction/reactions with carbon.[35,36] The volume of small pores in all samples but
that modified at 600 oC are similar and the slightly higher amount of CO2 adsorbed on BAX-TU -
800 might be related to its high content of sulfoxide and sulfones which might attract CO2 in a
specific way.[37]
Figure 7. CO2 adsorption isotherms at 20 oC.
The surface pH is also expected to have an effect of CO2 electroreduction with acidic surface
increasing Faradaic efficiency for H2 and CH4 formation. It is seen in Figure 6c, surface acidity
might enhance formation of CH4 on BAX-950 compared to that on BAX-900. Their porosities are
quite similar, but the pH of the former samples is much lower that of the latter one.
To exclude that the decomposition of the carbon catalysts themselves is responsible for the
formation of CO or CH4, the reduction experiments were carried out in N2 saturated electrolyte at
-0.67 V vs. RHE. No CO and CH4 were detected for all the BAX-TU materials through the whole
reduction process.
19
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
Figure 8. a) Faradaic efficiency for CO formation; b) Faradaic efficiency for CH4 formation during
CO2 reduction at the potential -0.67 V vs. RHE; c) The FE for detected gas phase of BAX-TU-950
for 40h; d) LSV for the materials studied in N2 saturated 0.1M KHCO3
The total FE was accounted for 80%-90% throughout the reduction process on the BAX-TU-950
catalyst (Figure 8c). The capacitive charging behavior of carbons, the adsorption of the produces
in pores and the presence of dissolved products in the electrolyte can contribute to missing 10-
20 %. In addition, the leakage of H2 gas after 24h reduction could take place due to the imperfect
sealing of the electrochemical cell. To verify whether or not liquid products were formed, NMR
20
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
analysis of the electrolytes after CO2 reduction was carried out. The 1H NMR spectra are presented
in Figure S4 of the Supporting Information. Trace amounts of C1 products (methanol, formic acid)
and C3 products (acetone) were detected in the electrolytes of all the samples. Their quantification
is beyond the scope of this paper.
To test the effect of light on the CO2 electroreduction, BAX-TU-950 and BAX-TU-600 were
chosen since they showed the lowest and highest content of sulfur among four samples studied.
Their photosenistivities were first tested. The repeatable transient photocurrents upon on/off
irradiation are shown in Figure 9. As seen, the photocurrent increased upon light irradiation on
both materials. The photocurrent generated (based on the initial current) was 3.3% and 46.8% for
BAX-TU-950 and BAX-TU-600, respectively. Even though the initial current of BAX-TU-600
(0.44mA) is 2 order magnitude less than that of BAX-TU-950 (43mA), the percentage of the
generated photocurrent showed that BAX-TU-600 was much more photoactive than was BAX-
TU-950.
Figure 9. Photocurrent generation upon on/off visible light radiation at voltage 1 V on a) BAX-
TU-600 and b) BAX-TU-950.
21
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
To find out how light affects the CO2 reduction, the reduction process was first performed in the
dark conditions for 4h at the potential of -0.67V. Then the electrode was exposed to solar light
(150 W Xe lamp). Upon irradiation, the absolute value of the current density for both samples
increased and remained stable at the higher current density (Figure 10a) than those under dark.
The FE for the CO formation at 5h was 27.0% and 4.4% for BAX-TU-950 and BAX-TU-600,
respectively. When the CO2 reduction was conducted in the dark for 5h, the FE for CO on BAX-
TU-950 and BAX-TU-600 was 29.0% and 4.8%, respectively. Thus, the FE of CO slightly
decreased under irradiation for both materials. While the partial CO current density of BAX-TU-
950 increased 21%, for BAX-TU-600 that increase was 28%, in comparison to those measured in
the dark (Figure 10b). The formation rate of CO on BAX-TU-950 and BAX-TU-600 under
irradiation were 15.8 mol/h and 3.9 mol/h, respectively. The formation rate under irradiation
are higher than those under dark (Table S1). These results are very promising in comparison with
those reported in the literature. Thus on plasmonic Au/p-GaN the rate of CO formation was 5
nmol/h for CO formation.[38] On CoII(Ch)-modified cathode the highest rate measured was 0.58
mol/h at pH of 6.7[39] and on Ag-loaded boron-doped diamond (BDD) electrodes the reported
rate with 5.97mol/h.[40]
Assuming that the rest of the current is used for HER, the partial H2 current density of BAX-TU-
950 and BAX-TU-600 increased 28% and 41%, respectively, in comparison to those in the dark.
The results suggest that the generated photoelectrons contributed to both CO2 and HER reduction.
The photoinduced electrons were accepted by both CO2 and water and more of them was used for
HER than that for CO2 reduction. Based on XPS analysis, more thiophenic-S was present on the
BAX-TU-600 surface, and this configuration seems to enhance the generation of photocurrent
along with an increase in the current linked to the CO formation.
22
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
Figure 10. a) The current density and FE of CO upon light for BAX-TU-600 and BAX-TU-950;
b) Comparison of the total current density and CO partial current density measured in the dark and
at light for both BAX-TU-950 and BAX-TU-600.
Tafel analyses (Figure 11) have been performed to elucidate the kinetics associated with the
reduction of CO2 to CO. The obtained Tafel slope is 153 mV dec−1 and 146mV dec−1 for the BAX-
TU-900 and BAX-TU-950, respectively. Those values are close to 118 mV dec−1, which is
expected for a rate-limiting single-electron transfer at an electrode.[41] For nitrogen-doped carbon
nanotube(NCNT) the Tafel slope was reported to be 203 mV/dec.[1] Another study reported the
lowest value of 160 mV dec−1 among the NCNTs.[2] This suggests the nanoporous carbons present
fast kinetics for CO formation and faster than those on CNT. This could be due to the porous
structure of the carbon. Even though the developed pore structure is not a necessary factor
governing the catalytic activity for the CO formation, it seems beneficial for the overall reduction
process. The previous studies indicated that the positively charged sites, especially the positively
charged carbon atoms introduced from pyridinic-N and quanternary-N are the active sites for CO
formation.[13,18] The thiophenic-S also introduces the positive charge to the carbon atoms although
it is less electronegative than nitrogen is, which results in a lower the activity for the CO formation
23
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
than that brought by pyridinic-N. The introduction of sulfur, on the other hand, also lowers the
conductivity of the carbon. Since the electric conductivity of plays a crucial role for CO2 reduction,
increasing simultaneously the positively charged sites as well as conductivity would be the future
goal to achieve a high efficiency of CO2 reduction.
Figure 11. Tafel plots for BAX-TU-900 and BAX-TU-950.
To check if exposure to light affect the electron transfer on the samples tested, the impedance
spectra of BAX-TU-950 and BAX-TU-600 were measured in the KHCO3 electrolyte saturated
with CO2 (Figure 5S of the Supplementary Information). While for BAX-YU-950 exposure to
light resulted in a visible decrease the high frequency arch indicating an increase in charge
transfer[42], the results for BAX-TU-600 do show any effect of light on electron transfer. This
might be caused by limitation of the conductivity of the carbon and also in its chemical instability,
as shown in the XPS data.
24
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
To date, this is the first report on the solar light influence on the CO2 electroreduction on
carbonaceous catalysts. The results suggest that upon light, thiophenic-S increases carbon
photsensitivity and promotes the CO formation. To further understand the nature of the process,
the band gaps and band alignments were estimated from the Mott-Schottky plots (Figure S6 of the
Supporting Information) from electrochemical impedance spectroscopy (EIS) measurements
(Figure 12a). The photoactive catalyst should have a band gap of less than 3.18eV in order to
adsorb visible light. BAX-TU-600 and BAX-TU-950 shows band gap of 1.91eV and 3.02eV,
respectively. The smaller bandgap of BAX-TU-600 in comparison to that of BAX-TU-950 might
explain its higher photoactivity. Moreover, the position of the conduction band (CB) of BAX-TU-
600 is more favorable than that of BAX-TU-950 for both the CO2 reduction and H2 evolution
reaction, which is consistent with the higher extent of photocurrent generation at light on BAX-
TU-600 as compared to that on BAX-TU-950.
Figure 12. a) Band-gap diagram of BAX-TU-600 and BAX-TU-950 (The CO2 redution reactions
must involve protons from water); b)XPS valence spectra of BAX-TU-950 and BAX-TU-600.
25
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
To further confirm the difference in the bandgap, the XPS valence spectra for BAX-TU-600 and
BAX-TU-950 were collected[43]. The results are presented in Figure 12b. Since in the case of
carbon the valence and conduction bands are rather the apparent values related to local electronic
structures[44] [49], we consider these results as a rather semiquanitative than quantitative estimation
of the valence band position. Nevertheless, the results indicated that the valance band of BAX-
TU-950 is at higher energy than that of BAX-TU-600 and this trend is in agreement with our
estimation illustrated in Figure12a.
Conclusions
From all three N and S-co-doped carbon addressed in this research, the sample modified at 950 oC
showed the highest activity at a low overpotential for CO formation in the process of CO2ERR.
This sample also showed the longest stability. Besides pyridinic-N, the catalytic activity of
thiophenic-S was found to contribute to its high FE for the CO formation. The results were better
than those obtained on the previously studied nitrogen-doped carbon. However, the activity of that
carbon for the CH4 formation was smaller than that of pyridinic-N. The increase in the performance
with the increase in the heat treatment temperature indicates the importance of an electric
conductivity for the CO2ERR process.
In this first study of sulfur modified carbons as CO2 (photo)electrocatalysts for CO2ERR.
Thiophenic-S was found to play a crucial role in generating photocurrent upon light, which is
beneficial for the overall reduction process (both hydrogen evolution reaction and CO2 reduction).
Thus, the obtained results advance the development of state-of-art metal-free catalyst for the CO2
reduction through engaging the catalytic activity and photoactivity provided by the incorporation
26
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
of sulfur to the carbon matrix. The findings have significant implications for the understanding of
the CO2 reduction process and functional sulfur-carbon moieties.
Experimental Section
Materials
A wood-based commercial activated carbon BAX-1500 (Mead Westvaco), which is referred to as
BAX was chosen for this study. The BAX was soaked in a thiourea suspension (mass ratio=1:1)
in water for 24h under stirring at a room temperature. The mixture was then heated to evaporate
water and dried at 120 oC. As the next step, the samples were heated to 600, 800, 900 and 950 oC
with rate of 20 oC/min (under N2). The final temperature was kept for 30 min. After carbonization,
the samples were washed with deionized (DI) water at 80 oC for 4h in a 2-liter beaker to remove
the excess of the thiourea and water-soluble compounds. Then they were filtered and washed with
o
DI water until the pH of leachate was 7. The samples were dried at 120 C overnight. Following
the treatment, the samples are referred to as BAX-TU-600, BAX-TU-800, BAX-TU-900, BAX-
TU-950, where last three digits refer to the heating temperature.
X-ray Photoelectron Spectroscopy (XPS)
The XPS spectra were collected using the Physical Electronics PHI 5000 VersaProbe II
spectrometer with a monochromatic Al Kα (1486. 6 eV) radiation operating at 15 kV and 50 W in
the analyzer chamber. Prior to analysis, all samples were outgassed until 10-8 Torr at room
temperature. High-resolution spectra of samples were collected with the constant pass energy
values of 29.35 eV and a 200 μm diameter analysis area with a take-off angle of 45o. 117.4 eV
pass energy was used for the survey spectra and the calculation of surface atomic concentration.
The spectrometer energy scale was calibrated using Cu 2p3/2, Ag 3d5/2, and Au 4f7/2 photoelectron
27
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
lines at 932.7, 368.3, and 84.0 eV, respectively. The SmartSoft-VP2.6.3.4 software package was
used for acquisition and data analysis and Multipack software was used to fit deconvoluted
photoelectron spectra. A Shirley-type background was subtracted from the signals and recorded
spectra were fitted using Gauss–Lorentz method in order to determine the binding energy of the
different element core levels accurately. XPS valence spectra was collected with 11.75 eV pass
energy at which the instrumental resolution is about 0.05 eV.
Sorption of nitrogen/porosity evaluation
Sorption of nitrogen was measured on ASAP 2020 Surface Area and Porosity Analyzer
o
(Micromeritics) at -196 C. Before analysis, the samples were outgassed at 120°C to constant
vacuum (10-4 Torr). At these conditions only physically adsorbed species were removed without
changing the surface chemistry. The surface areas (NLDFT method), total pore volumes (Vt), the
volumes of mesopore (Vmeso), micropore (Vmic), ultramicropore (V<0.7nm) and the volume of pores
smaller than 1 nm (V<1 nm), were calculated from the isotherms using the Non-Linear Density
Functional Theory (NLDFT) www.nldft.com.[45,46] The pore size distributions were also calculated
using this method.
Adsorption of CO2
o
Adsorption of CO2 was measured at 20 C until 760 torr on ASAP 2020. Before analysis, the
samples were outgassed at 120°C to constant vacuum (10-4 Torr). The temperature was controlled
using a circulating water bath.
SEM and TEM Electron microscopy
SEM images were obtained with a Zeiss Supra 55 VP at a magnification 100 KX with an
accelerating voltage of 5.00 kV. Scanning was performed in situ on a sample of powder without
28
This article is protected by copyright. All rights reserved.

10.1002/cssc.201801073
ChemSusChem
coating. HR-TEM was performed using a JEOL 2100 LaB6 instrument operating at 200 kV.
Analyses were performed after the carbon samples were resuspended in isopropyl alcohol.
DC Conductivity measurements
A film of a sample dispersed N-methyl-2-pyrrolidone (NMP) was coated on a 0.5 cm × 0.45 cm
gold interdigitated electrode. The samples were dried at 120 °C for 24 h to evaporate the NMP
solvent. The sheet resistance was measured with a 4-probe method using Keithley 2400 Multimeter.
Then the DC conductivity was calculated in the following formula:
1
퐿
휎 =
∙
푅 푊 ∙ 푡
where L is the length of the electrode, W is the wideness of the electrode, t is the thickness of the
electrode, σ is the conductivity of the material and R is the measured sheet resistance. The sample
film’s thickness was measured by a digital micrometer (Pittsburgh®). Each material was tested
three times and the average values were reported.
Potentiometric titration
Potentiometric titration analysis was carried out on the DMS Titrando 888 automatic titrator
(Metrohm). The samples (~ 0.050 g) were added to 25 mL of NaNO3 (0.01 M), and equilibrated
overnight at a room temperature. The suspension was saturated with N2 throughout the
measurements under a continuous stirring to eliminate the influence of atmospheric CO2. Each
sample was titrated with volumetric standard NaOH (0.1 M) after adding 0.1M HCl to the sample
suspension. The initial pH of the suspension was used as the surface pH value. The experimental
data were transformed into proton binding curves, Q, and then to pKa distributions. The pKa
distributions were calculated using SAIEUS numerical procedure [45,47]
.
CO2 electrochemical reduction measurement
29
This article is protected by copyright. All rights reserved.