Tailoring Nitrogen-Doped Carbons as Hosts for Single-Atom Catalysts

Article abstract of DOI:10.1002/cctc.201900547

The application of nitrogen-doped carbons (NDCs) as host materials for single-atom catalysts (SACs) is increasing because of the strong binding affinity of the heteroatom with transition metals. Establishment of the relation between the properties of NDCs, their interaction with metals, and the performance of the resulting catalysts is crucial to guide the design of more effective SACs but has not been critically addressed. Here, a series of NDCs is prepared and studied as hosts for palladium atoms. The amount of nitrogen incorporated primarily depends on the choice of carbon followed by the doping temperature and nitrogen source. Pyridinic and pyrrolic species predominate in all cases, especially at lower doping temperatures. The stabilization of palladium atoms is successful above a critical nitrogen content that depends on the carbon type. Evaluation in the catalytic semi-hydrogenation of 2-methyl-3-butyn-2-ol readily distinguishes the reactivity of single atoms and nanoparticles, which correlates with the electronic properties of palladium described by the average oxidation state. Comparison with SACs based on compositionally-related carbon nitride hosts shows that similarly high stability and tunability of the metal is achievable over NDCs, despite their lower nitrogen contents and greater heterogeneity of coordination sites.

Full text of DOI:10.1002/cctc.201900547

Accepted Article
Title: Tailoring nitrogen-doped carbons as hosts for single-atom
catalysts
Authors: Simon Büchele, Zupeng Chen, Sharon Mitchell, Roland
Hauert, Frank Krumeich, and Javier Pérez-Ramírez
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: ChemCatChem 10.1002/cctc.201900547
Link to VoR: http://dx.doi.org/10.1002/cctc.201900547
A Journal of
www.chemcatchem.org

10.1002/cctc.201900547
ChemCatChem
FULL PAPER
Tailoring nitrogen-doped carbons as hosts for single-atom
catalysts
Simon Büchele,[a] Zupeng Chen,[a] Sharon Mitchell,*[a] Roland Hauert,[b] Frank Krumeich[a], and
Javier Pérez-Ramírez*[a]
Abstract: The application of nitrogen-doped carbons (NDCs) as host
materials for single-atom catalysts (SACs) is increasing because of
the strong binding affinity of the heteroatom with transition metals.
Establishment of the relation between the properties of NDCs, their
interaction with metals, and the performance of the resulting catalysts
is crucial to guide the design of more effective SACs but has not been
critically addressed. Here, a series of NDCs is prepared and studied
as hosts for palladium atoms. The amount of nitrogen incorporated
primarily depends on the choice of carbon followed by the doping
temperature and nitrogen source. Pyridinic and pyrrolic species
predominate in all cases, especially at lower doping temperatures.
The stabilization of palladium atoms is successful above a critical
nitrogen content that depends on the carbon type. Evaluation in the
catalytic semi-hydrogenation of 2-methyl-3-butyn-2-ol readily
distinguishes the reactivity of single atoms and nanoparticles, which
correlates with the electronic properties of palladium described by the
average oxidation state. Comparison with SACs based on
compositionally-related carbon nitride hosts shows that similarly high
stability and tunability of the metal is achievable over NDCs, despite
their lower nitrogen contents and greater heterogeneity of
coordination sites.
the host should be optimized since this determines its electronic
properties. In this context, hosts with well-defined surface
functional groups and tunable properties offer great potential.[3]
Graphitic carbon nitride (C3N4) has proven an effective host to
stabilize various single metal atoms due to the six-fold nitrogen
coordination sites intrinsic to its structure.[4] The coordination of
single atoms on carbon nitride can be tailored by controlling the
properties of the host, leading to enhanced performance in
targeted reactions.[5] However, the ordered heterocyclic
framework also imposes limits on the tunability of C3N4, since the
structural integrity can only be retained up to a certain extent of
modification.
Various nitrogen-doped carbon materials (NDCs), including
carbon nanotubes,[6] graphene,[7] and other porous carbons,[8]
have also been employed as hosts for SACs, providing a wider
flexibility for tuning the nitrogen amount and functionality via a
variety of direct- and post-synthetic approaches.[9] Additionally,
NDCs are often electrically conductive, which favors the surface
diffusion of reactants and enables their use in electrocatalytic
applications.[10] The synthesis of NDCs can be generally
categorized into three distinct routes (Scheme 1). The direct
pyrolysis of nitrogen-containing polymers, metal-organic
frameworks (MOFs), or other small molecules (e.g. heterocycles,
aminated sugars, amino acids, ionic liquids), is straightforward
and can yield nanostructured NDCs with high nitrogen content
(> 30 wt.%).[9,11] However, drawbacks of this route include the
relatively small number of cheap precursors available and the
poor control over the heteroatom distribution in the resulting
material. Thermal treatment of commercial carbons with gaseous
ammonia at high temperature generates surface nitrogen groups.
However, the extent of nitrogen incorporation is limited and the
process is costly.[12] In contrast, the use of cheaper solid nitrogen
sources like dicyandiamide, melamine, or urea, constitutes an
effective and broadly applicable post-synthetic doping approach
to incorporate nitrogen species of varying type (e.g. pyridinic,
pyrrolic, quaternary or pyridine-N-oxide).[13] Although pyridinic or
pyrrolic nitrogen are frequently postulated as the desired sites for
stabilizing metal atoms,[6b,10d,14] the effect of the synthetic
approach or specific properties of a host material (e.g. surface
area and functionality) on the metal coordination and associated
performance of SACs has received little attention.[15] The lack
of systematic study restricts the establishment of property-
performance relationships for the design of more efficient
SACs based on NDCs.
Introduction
Single-atom heterogeneous catalysts (SACs) that integrate
isolated metal atoms on appropriate hosts have received huge
attention due to the prospects of utilizing every metal atom
efficiently in multiple industrially relevant catalytic processes.[1]
The selection of a host material that firmly anchors metal atoms
reducing the tendency to aggregate and sinter is critical due to the
frequent thermodynamic instability of metals atomically-dispersed
on common catalyst carriers such as metal oxides, metals, and
conventional carbons.[2] Furthermore, to ensure high performance
in targeted reactions the strength of interaction of the metal with
[a]
S. Büchele, Dr. Z. Chen, Dr. S. Mitchell, Dr. F. Krumeich,
Prof. Dr. J. Pérez-Ramírez
Department of Chemistry and Applied Biosciences
ETH Zürich
Vladimir-Prelog-Weg 1, 8093 Zürich, Switzerland
E-mails: msharon@chem.ethz.ch; jpr@chem.ethz.ch
Dr. R. Hauert
[b]
Empa, Swiss Federal Laboratories for Materials Science and
Technology
Überlandstrasse 129, 8600 Dübendorf, Switzerland
To address this, here we study the controlled
incorporation of nitrogen into three distinct carbon materials
(carbon nanofibers (CNF), reduced graphene oxide (RGO)
derived from graphene oxide (GO), and mesoporous carbon
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900547
ChemCatChem
FULL PAPER
Scheme 1. Bottom-up (left to center) and top-down (right to center) strategies to prepare NDCs containing distinct nitrogen species, namely pyridinic (Pyrid), pyrrolic
(Pyrrol), quaternary (Quat), and pyridine-N-oxide (N-ox). Bottom-up approaches involve doping common carbon materials (e.g. CNF, GO, or MPC) with nitrogen-
rich molecules (e.g. D, U, M, or NH3) while top-down routes are based on the decomposition of a nitrogen-containing polymer/precursor. Color codes: grey, C; blue,
N; white, H; red, O.
(MPC)) using dicyandiamide (D), melamine (M), urea (U), or
ammonia as nitrogen sources (Scheme 1). The comparative
properties of the carbon hosts are characterized in depth
before and after the introduction of nitrogen and subsequent
deposition of palladium. Detailed analysis by multiple
techniques confirms the incorporation of different amount of
nitrogen with distinct species, which influences the
dispersion (e.g. single atoms or nanoparticles) and average
oxidation state of palladium. To elaborate and compare
synthesis-property-performance relationships the catalysts
are subsequently evaluated in the semi-hydrogenation of
discriminate single atoms due to an increased barrier for
hydrogen splitting below a minimum ensemble size of
three,[16] and compared to a palladium SAC based on
exfoliated carbon nitride (ECN).[4a]
Results and Discussion
Throughout this manuscript, the nitrogen doped carbons (NDCs)
are coded xyz, where x denotes the carbon type (CNF, RGO, or
MPC), y the doping temperature (923, 1023, or 1123 K) and z the
nitrogen source (D, M, U, or NH3). Pd-ECN refers to a single-atom
2-methyl-3-butyn-2-ol,
a
relevant reaction known to
Figure 1. a) TEM images and b) XRD patterns of the pristine carbons (dashed lines) and metal-free NDCs (solid lines) resulting from doping with dicyandiamide at
1023 K. c) The variation of the ID/IG ratio with the doping temperature for metal-free NDCs doped with D. The solid lines show the corresponding values for the
pristine materials.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900547
ChemCatChem
FULL PAPER
Figure 2. Variation in the amount of nitrogen introduced in the carbons studied as a function of a) the doping temperature using D as the nitrogen source and b) the
nitrogen source at 1023 K. AN denotes treatment in the absence of a nitrogen source. c) N 1s XPS spectra of differently treated MPC materials, showing the fitted
results (solid black lines) of the raw data (open circles), and the fitted nitrogen species (yellow, green, blue, and red shaded peaks). Variation in the amounts of
surface nitrogen species as a function of d) the doping temperature for MPC using D as a nitrogen source and e) the carbon type when doped with D at 1023 K.
ECN is shown for reference. The color codes in (c, d) also apply to (e).
catalyst based on exfoliated carbon nitride.
show higher ID/IG values, reflecting increased structural disorder
after nitrogen doping.
Host Synthesis and Properties
The higher nitrogen contents of the NDCs compared to the
pristine carbons (nitrogen content < 0.1 wt.%) confirms the
success of the doping procedure (Table S1). Independently of the
type of carbon, lower amounts of nitrogen are introduced upon
increasing the doping temperature from 923 to 1123 K (Figure 2a).
At any given condition, the resulting nitrogen content depends on
the identity of the carbon, following the order of RGO
(19.4 - 13.5 wt.%) > MPC (13.0 - 7.8 wt.%) > CNF (3.0 - 0.7 wt.%),
which is not directly linked to the surface area. This is likely due
to the distinct concentrations of reactive sites such as surface
oxygen-containing groups (Table S1) and different degrees of
crystallinity in the corresponding carbons prior to nitrogen doping
(i.e. CNF, MPC, and GO).[19] Comparing different nitrogen
sources (Figure 2b), similar nitrogen contents are achieved using
dicyandiamide and melamine, while urea is on average 28 % less
effective, agreeing well with previous studies.[13] As expected,
treatment of MPC with gaseous ammonia at 923 and 1023 K
(MPC923NH3, MPC1023NH3) leads to even lower nitrogen
contents (Table S1). These results highlight the superiority of
using larger molecules such as dicyandiamide or melamine as
nitrogen sources. This likely relates to the fact that these solid
nitrogen sources polymerize in close proximity to the carbon
matrices before being thermally decomposed, enabling an
efficient incorporation of nitrogen. In contrast, gaseous ammonia
molecules can only be incorporated when they encounter reactive
surface groups by chance.
The incorporation of nitrogen applying solid molecular sources
causes no obvious alteration of the carbon morphology
(Figure 1a). The conical tubes of CNF, thin sheets of RGO, and
mesoporous structure of MPC remain clearly visible by
transmission electron microscopy (TEM). Analysis by argon
sorption also confirms the comparable porous properties of the
pristine and nitrogen-doped materials (Figure S1). These
observations are attractive as they demonstrate the applicability
of the doping strategy to materials with pre-defined pore
structures. Slightly increased adsorption capacity is observed in
the samples treated at higher temperature with MPC exhibiting
the highest specific surface area (639-933 m2 g−1), followed by
RGO (63-78 m2 g−1) and CNF (47-53 m2 g−1). Analysis by X-ray
diffraction (XRD) identifies reflections at 26.5, 42.4, 43.8, and
54.6° 2θ characteristic of the (002), (100), (101), and (004) planes
in graphitic carbon, which are most prominent in CNF-based
samples indicative of a higher crystallinity (Figure 1b).[17] The
degree of structural disorder can be estimated by Raman
spectroscopy from the intensity ratio of the bands at 1340 cm−1
(ID) and 1880 cm-1 (IG) attributed to the stretching vibrations of
defect sites and of the graphitic carbon lattice, respectively
(Figure S2).[18] Pristine RGO (ID/IG = 1.39) and MPC (ID/IG = 1.37)
are found to be highly disordered, while consistent with the XRD
results, CNF (ID/IG = 0.06) displays the fewest defects, which is
also confirmed by another prominent (2D) band at 2680 cm−1
(Figure 1c). Comparatively, samples doped with dicyandiamide
The speciation of surface nitrogen can be assessed by XPS.
High-resolution N 1s spectra (Figure 2c and S3) evidence four
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900547
ChemCatChem
FULL PAPER
Figure 3. AC-STEM images of palladium-containing NDCs based on CNF, RGO, and MPC doped with D at different temperatures (indicated). Where visible, the
size distribution (based on more than 140 counts) of nanoparticles is shown inset.
peaks which are assigned to pyridinic (398.4 ± 0.2 eV), pyrrolic
(400 ± 0.2 eV), quaternary (401.2 ± 0.1 eV), and pyridine-N-oxide
(403.1 ± 0.2 eV) based on literature values.[9,20] Comparison of
the NDCs derived from MPC reveals high fractions of pyridinic
and pyrrolic nitrogen species, which have been proposed as the
primary anchoring sites for single atoms, both in MPC1023D and
MPC1023M (Figure S4).[6b,14,21] In view of the similar
effectiveness, further analysis focusses on dicyandiamide as the
nitrogen source. A general decrease in the contribution of
pyridinic and pyrrolic nitrogen species is visible upon increasing
the doping temperature from 923 to 1123 K in all NDCs (Figure 2d
and S4). In contrast, the fraction of quaternary and pyridine-N-
oxide species remains relatively low over the whole temperature
range, and minor variations lie within the experimental error. Both
the amount and the contributions of individual nitrogen species
observed in the NDCs clearly differ from those of graphitic carbon
nitride (Figure 2e and S5). Bearing the heptazine as the building
blocks, ECN shows the highest contribution of combined pyridinic
and pyrrolic species, followed by NDCs based on MPC, CNF, and
RGO, and thus differences in the stabilization of metal atoms may
be envisaged between these hosts. Oxygen 1s spectra evidence
a peak around 532.5 eV for all analyzed samples which can be
assigned to the presence of surface hydroxyl groups (Figure S6).
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900547
ChemCatChem
FULL PAPER
Figure 4. Pd 3d core level XPS spectra of palladium-containing NDC catalysts, based on a) CNF, b) RGO, and c) MPC doped with D at different temperatures. The
solid lines represent the fitted result of the raw data (open symbols), whereas the green, blue, and red peaks correspond to the fitted components. The dashed lines
indicate the positions formally assigned to Pd(IV), Pd(II) and Pd(0) species. Numbers next to the spectra indicate the corresponding average palladium oxidation
state.
might appear more oxidized than bulk materials due to quantum
size effects (Note S1).[22] Hence the assignment of an average
oxidation state merely serves as a tool to distinguish the degree
of oxidation for different catalysts.
Metal Stabilization
To study the effectiveness of the NDCs as single-atom hosts,
palladium was introduced onto the carriers targeting a metal
loading of 0.5 wt.%. Herein, microwave-irradiation-assisted
deposition was found to efficiently disperse the metals on MPC
and ECN. However, this route likely favors the formation of large
Property-Performance Relationships
To compare the performance of catalysts based on NDCs with
those based on carbon nitride, the synthesized materials are
evaluated in the three-phase semi-hydrogenation of
2-methyl-3-butyn-2-ol. This application is of interest since the
activity of palladium-based catalysts is known to be sensitive to
metal agglomerates over CNF and RGO and so
a wet
impregnation approach was used instead for these materials
(Figure S7). No obvious morphological changes are observed in
the metal-containing samples (Figure S8). Analysis by aberration-
corrected scanning transmission electron microscopy (AC-STEM)
in high-angle annular dark field mode confirms the atomic
dispersion of palladium on the hosts with higher nitrogen content.
d
Nanoparticle formation (average diameter,
the hosts with lower nitrogen content and surface area (i.e.
CNF1023D) ( = 3.9 nm), CNF1123D ( = 4.3 nm), and
RGO1123D ( = 3.1 nm) (Table 1, Figure 3). A broad particle
) is observed on
d
d
d
size distribution featuring the coexistence of small palladium
nanoparticles and some large aggregates is evidenced in the
case of RGO1123D. On MPC1023 and MPC1023U large
palladium particles are observed, which in the latter case coexist
with atomically-dispersed metal species (Figure S9). Similarly, up
to 50 nm palladium particles are found in MPC1023NH3, likely
caused by the low nitrogen content reached by doping with
ammonia (Figure S10).
Further insight into the chemical state of deposited metal species
and their interaction with NDCs is obtained from the high-
resolution Pd 3d XPS spectra (Figure 4). The Pd 3d5/2 peak can
be fitted with three components which are attributed to metallic
Pd(0) (335.5 ± 0.05 eV), Pd(II) (336.7 ± 0.05 eV), and Pd(IV)
(338.3 ± 0.05 eV), respectively (Table S2). Consistent with the
microscopy analysis, the Pd(0) contribution is absent in the
samples containing atomically-dispersed palladium species,
supporting the absence of nanoparticles. In general, higher
average palladium oxidation states are obtained with decreasing
atomic population yielding OSPd above three for samples
presenting primarily isolated metal centers, reflecting the strong
metal-host interaction. Note that the attribution of precise
oxidation state is debatable and atomically dispersed species
Figure 5. Contour plot displaying the reaction rate (r) in the batch
semi-hydrogenation of 2-methyl-3-butyn-2-ol as a function of the bulk nitrogen
OS
content and average palladium oxidation state ( Pd ). The contour lines are
fitted by spline interpolation of the 12 measured samples. The dashed line
indicates the reaction rate observed over Pd-ECN. Open symbols correspond
to SACs, closed symbols to nanoparticle-containing catalysts. Reaction
conditions: mcat = 15 mg, VL = 1.5 cm3 (2-methyl-3-butyn-2-ol in toluene), and
PH2 = 3 bar.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900547
ChemCatChem
FULL PAPER
Table 1. Synthesis method and characterization and performance data of the Pd/NDC catalysts.
Catalyst
M-deposition method[a]
M species[b]
SA
BET[c] / m2 g−1
654 (701)
651 (685)
593 (639)
651 (680)
750 (756)
922 (933)
52 (53)
Pd[d] / wt.%
0.54 (1.1)
0.53 (1.1)
0.60 (1.3)
0.61 (1.3)
0.57 (1.2)
0.28 (1.5)
0.63 (2.4)
0.63 (3.1)
0.69 (4.0)
0.70 (1.8)
0.55 (2.3)
0.75 (2.9)
0.40 (0.9)
r[e] / molalkene molPd−1 h−1
S[f] / %
94
MPC1123D
MPC1023D
MPC923D
MPC1023M
MPC1023U
MPC1023
CNF1123D
CNF1023D
CNF923D
RGO1123D
RGO1023D
RGO923D
Pd-ECN
MW
MW
MW
MW
MW
MW
WI
0.70
0.05
0.04
0.04
1.53
3.92
1.00
0.87
0.03
2.63
0.28
0.18
0.27
SA
95
SA
95
SA
93
NP, SA
NP
94
95
NP
86
WI
NP
49 (52)
88
WI
SA
44 (47)
92
WI
NP
75 (78)
93
WI
SA
62 (66)
91
WI
SA
61 (63)
92
MW
SA
150 (170)
97
[a]MW: microwave-irradiation-assisted deposition, WI: wet impregnation. [b]SA: single atom, NP: nanoparticle. [c]BET method (in parentheses, the surface area
of the metal-free carbon hosts). [d]ICP-OES (in parenthesis, surface metal content determined by XPS). [e]Rate and [f]selectivity of 2-methyl-3-buten-2-ol
formation at conversions of 15-47 % in batch mode/tests.
the presence of single atoms or nanoparticles since the
mechanism of hydrogen dissociation changes from homolytic to
heterolytic below a minimum ensemble size of three.[16b] All
samples show high selectivity towards the desired alkene,
reaching values above 90 % in most cases (Table 1). The
correlation between the physicochemical properties (i.e. the
significantly lower nitrogen content of these samples and more
heterogeneous speciation.
Consistently, the average palladium oxidation state is found
to be a good descriptor for the rate of formation of the alkene in
the hydrogenation reaction over NDC supported catalysts
(Figure S11), showing higher activity with more metallic palladium
species.[5b] Thus, all SACs exhibit similar relatively low rates of
product formation (0.04 − 0.70 molalkene molPd−1 h−1) independent
of the type of carbon applied, while palladium particle-containing
samples show higher rates (0.87 – 3.92 molalkene molPd−1 h−1).
Additionally, consistent with literature data,[23] within particle-
containing samples an increase in reaction rate for larger
OS
nitrogen content, and
Pd ) and the reaction rate of the prepared
catalysts can be readily visualized in a contour plot (Figure 5). A
direct trend is seen for the average oxidation state of palladium,
increasing with the amount of nitrogen incorporated. This
suggests a higher degree of electron transfer from the metal to
the coordination sites at higher nitrogen content, consistent with
the ability to stabilize higher dispersions of the metal. Interestingly, palladium particles or agglomerates is observed. When compared
similar but shifted correlations are observed for the different
carbon materials. For example, the successful obtainment of
SACs based on RGO requires approximately 4 times higher
nitrogen content than CNF. The differences between individual
carbons could relate to both distinctions in the type, amount, and
speciation of surface functional groups.
However, no clear correlation can be derived from these variables
alone pointing to the need for improved insight into the local
structure of the coordination sites. Between the SACs obtained
to the reference Pd-ECN SAC which has been demonstrated to
be a selective alternative to the commercial Lindlar catalyst for the
semi-hydrogenation
of
alkynes,[5b]
some
NDCs-based
single-atom catalysts (e.g. MPC1123D and RGO1023D) show
significantly higher activity. Analysis of the selected used
catalysts reveals that the stability of the catalysts also depends
on the palladium oxidation state (Figure 6). In the case of
MPC1023D with OSPd of 3.36 no change in the oxidation state is
observed after reaction and the atomic dispersion of palladium is
preserved. The stability of this material was further confirmed by
the constant conversion and selectivity observed in a catalytic test
for 10 h in flow. However, the single-atom catalysts with lower
OSPd of 3.20 (MPC1123D) and 3.28 (RGO1023D) are slightly
reduced under reaction conditions thus forming visible
nanoparticles on the surface. On the contrary, the reference
OSPd
the average oxidation state varies between 3.2 <
<3.5
illustrating some tunability of the metal centers, but in all cases
they remain highly cationic. Interestingly, this range is narrower
but within the same values as previously observed for palladium
species in distinct SACs based on carbon nitride,[5b] despite the
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900547
ChemCatChem
FULL PAPER
prepare highly porous host materials. By changing the doping
temperature, the nitrogen content could be tuned using different
molecular nitrogen sources, with dicyandiamide and melamine
yielding the highest amount of heteroatom incorporation at low
temperatures. Atomic dispersions of palladium could be stabilized
above a critical nitrogen content that depended on the type of
carbon. The strong interaction required to stabilize single atoms
resulted in a significant degree of oxidation of the metal species.
Despite the lower abundance of nitrogen and greater
heterogeneity of coordination sites, the structural and electronic
properties of SACs could be tuned within a similar range as
reported for palladium systems based on carbon nitride. The
semi-hydrogenation of 2-methyl-3-butyn-2-ol was used to readily
discriminate the reactivity of single atoms and nanoparticles on
NDCs, reaching comparable catalytic performance to palladium
single atoms on ECN for the same electronic state of the metal.
The presented results hence widen the scope of single-atom
hosts by providing improved understanding into the stabilization
of palladium atoms on NDCs and a general strategy to further
tailor the properties of SACs for specific applications.
Experimental Section
Materials
Graphite flakes and silicon carbide were obtained from Alfa Aesar,
tetraaminepalladium(II) nitrate (Pd(NH3)4(NO3)2) solution (containing
5 wt.% Pd) from abcr GmbH, and hydrogen peroxide solution (35 %) from
Acros Organics. All other chemicals were obtained from Sigma-Aldrich.
Host Synthesis
MPC was synthesized by pyrolysis of
a
silica-templated
resorcinol-formaldehyde resin, adapted from a reported recipe.[24] In a
typical synthesis resorcinol (10 g) was mixed with formaldehyde solution
(15.8 g, 34.5 wt.%) and an aqueous dispersion of SiO2 nanoparticles
(20.45 g, 40 wt.%, 12 nm average diameter) and heated to 358 K while
stirring. The obtained dark-red resin was kept at the same temperature
overnight. The resultant solid was ground and heated to 1123 K
(5 K min−1) in a N2 atmosphere for 3 h. Pristine MPC was obtained by
grinding the black powder again (sieve fraction < 200 μm for batch
experiments and 200-400 μm for catalytic tests in flow) and treating with
aqueous NH4HF2 (4 m, 1 dm3) at 293 K for 48 h to remove the silica
template. Graphene oxide (GO) was synthesized according to the modified
Hummer´s method.[25] Briefly, a mixture of concentrated H2SO4:H3PO4
(360:40 cm3) was added to graphite flakes (natural, 325 mesh, 99.8 %)
and KMnO4 (18 g) and heated to 323 K for 15 h. The mixture was cooled
with ice and poured onto ice (400 g) containing H2O2 (3 cm3, 35wt.%). The
resultant brown solid was separated by centrifugation and successively
washed with water (200 cm3), HCl (200 cm3, 30 wt.%), and ethanol
(2×200 cm3). The remaining material was re-dispersed in deionized water
and freeze-dried to obtain GO. ECN was synthesized as previously
described.[1d] Briefly, dicyandiamide (10 g) was heated to 823 K
(2.3 K min−1) in a N2 atmosphere for 4 h. ECN was obtained by thermal
exfoliation of the resulting material at 773 K (5 K min−1) for 5 h in static air.
Figure 6. a) AC-STEM images and b) comparative average palladium oxidation
OSPd
state (
) of selected fresh and used catalysts after the batch semi-
hydrogenation tests evaluated in Figure 5. c) Stable conversion (X) and
selectivity (S) in the semi-hydrogenation of 2-methyl-3-butyn-2-ol in flow over
MPC1023D (TOS, time on stream).
Pd−ECN retains atomic dispersion despite being slightly
reduced.[4b,5b] This demonstrates the suitability of specifically
tailored carbon-based materials as SACs and shows that the
reactivity and stability of the single atoms on carbon materials can
be tuned in a certain range by varying the metal-host interaction
strength via nitrogen doping.
Conclusions
The tunability of single-atom catalysts based on nitrogen-doped
carbons was explored to establish synthesis-property-
performance relationships to guide their improved design. A
practical doping procedure was applied to three carbon materials
of distinct morphology demonstrating its broad applicability to
Nitrogen doping
Before the doping step, pristine MPC and CNF were activated to introduce
surface oxygen groups enabling a more effective incorporation of nitrogen,
whereas GO was applied as synthesized.[11a,26] MPC was refluxed in HNO3
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900547
ChemCatChem
FULL PAPER
(4 m, 20 cm3 gcarbon−1) at 383 K for 16 h and CNF was treated in a mixture
of concentrated H2SO4:HNO3 (66:33 cm3 gcarbon−1) at 343 K for 2 h. Both
materials were washed with water after acid treatment (1 dm3 gcarbon−1) and
dried at 338 K. For nitrogen doping the desired carbon (1 g) was mixed
with nitrogen source (3 g, dicyandiamide (D) or melamine (M), or 4.3g,
urea (U)) in water (40 cm3) and sonicated for 0.5 h at 323 K. The resulting
dispersions were freeze dried and subsequently heated in a tubular oven
(Carbolite) under N2 flow to 723 K (5 K min−1) for 2 h and successively to
either 923 K, 1023 K, or 1123 K (5 K min−1) for 3 h. For MPC and GO the
same procedure was also applied but without nitrogen source (1023 K),
yielding the materials referred to as MPC1023 and pristine RGO
(RGO1023), respectively. The amount of nitrogen source was chosen to
obtain a high nitrogen content while still retaining good doping efficiency
(amount of nitrogen incorporated per gram of nitrogen source). To keep
the moles of added nitrogen atoms constant, 3 wt. equivalents with respect
to carbon are used for D and M whereas 4.3 wt. equivalents are used for
U. Two additional MPC-based samples were synthesized by treatment in
5 % NH3/He flow (20 cm3 min−1) at 923 K or 1023 K for 3 h (5 K min−1),
coded as MPC923NH3 and MPC1023NH3, respectively.
an angular step size of 0.05° and an acquisition time of 1.5 s per step was
recorded. Raman spectroscopy was performed with a confocal Raman
microscope (WITec CRM 200) using a wavelength of 532 nm and a 100×
objective lens. The machine was operated in backscattering mode at a
power of 6 mW. Elemental analysis was determined using a LECO
TruSpec Micro (C, H, N) instrument and a LECO 628 O Micro (O)
instrument. X-ray photoelectron spectroscopy (XPS) was measured using
a Physical Electronics Quantum 2000 instrument. Monochromatic Al Ka
radiation (1486.6 eV) generated from an electron beam operated at 15 kV
and 32.3 W was applied. The residual pressure during measurements was
5 × 10−8 Pa. For high-resolution spectra, pass energies of 46.95 eV were
used (full width at half maximum for Ag 4f7/2 = 1.0 eV at this conditions). A
neutralizer was operating during analysis. All spectra measured on
Pd-ECN were referenced to the C1s peak at 288.3 eV. Spectra of carbon
materials were analyzed without adjustment since no charging was
observed due to their electrical conductivity. All samples were mounted
onto the sample holder by pressing the powders onto indium foil. The
palladium average oxidation state was calculated based on the relative
content of different palladium species (Pd(0), Pd(II), and Pd(IV)) fitted to
the peak area of Pd 3d core-level XPS spectra (Note S11). Inductively
coupled plasma–optical emission spectrometry (ICP–OES) was measured
with a Horiba Ultra 2 instrument equipped with a photomultiplier tube
detector. For each measurement sample (0.02 g) was digested by heating
a mixture of HNO3 (3 cm3) and H2O2 (1 cm3) in a Teflon-lined stainless-
steel autoclave at 458 K overnight. Solutions were then filtered (pore size,
0.45 μm) and diluted to 25 cm3 prior to analysis.
Metal introduction
All catalysts were prepared with a nominal loading of 0.5 wt.% palladium.
For ECN and MPC carriers, palladium was introduced using microwave-
irradiation-assisted deposition. The carrier (0.5 g) was mixed with water
(20 cm3) and sonicated for 1 h. An aqueous Pd(NH3)4(NO3)2 solution
(5 wt.% Pd, 0.05 cm3) was added and stirred overnight before a cyclic
program (20 cycles) of 15 s irradiation at 100 W followed by 3 min cooling
was applied in a CEM Discover SP microwave reactor. The resulting
catalyst was collected by filtration, washed with deionized water (500 cm3)
and ethanol (250 cm3), and dried at 338 K overnight. For CNF and RGO,
a standard wet deposition was applied to prevent the strong interaction of
these conductive materials with microwave irradiation. The same aqueous
dispersions as previously described were heated at 338 K until the solvent
completely evaporated. The resulting powders was heated to 573 K
(5 K min−1) for 5 h in an IR tubular oven (behr Labor-Technik) under N2
flow.
Catalyst testing
The hydrogenation of 2-methyl-3-butyn-2-ol was performed in a CEM
Discover SP microwave reactor under continuous stirring. Typically, the
catalyst (0.015 g) was mixed with a solution of the substrate (0.4 m,
1.5 cm3) in toluene, pressurized with hydrogen (3 bar) and heated to 323 K
until a pressure drop to 2.7 bar (corresponding to approximately 40 %
conversion of the alkynol) was observed. The resulting solution was filtered
(pore size, 0.45 μm). Catalytic tests under flow conditions were conducted
at T = 323 K, P = 3 bar, FL(alkyne + toluene) = 1 cm3 min–1
,
and
FG(H2) = 36 cm3 min–1 in a microreactor (H-Cube Pro, ThalesNano).
Hydrogen generated in situ by electrolysis and the feed solution flow
concurrently through a cylindrical cartridge (I.D. = 3.5 mm) loaded with
catalyst (0.09 g, particle size = 200-400 μm) and filled with silicon carbide
(particle size = 200–400 μm).The products of the reaction were analyzed
Characterization
Samples for transmission electron microscopy (TEM) were prepared by
finely grinding catalyst powders and dusting them onto holey carbon
coated copper mesh grids. TEM measurements were performed on a
Talos F200X instrument operated at 200 kV. For the aberration corrected
scanning TEM (AC-STEM) investigation, the material was dispersed in
ethanol and a few drops of the suspension were deposited onto a
perforated carbon foil supported on a copper grid. AC-STEM investigations
were performed on a Hitachi HD-2700CS. The microscope was operated
at an acceleration potential of 200 kV (cold field emitter). A probe corrector
(CEOS) that is incorporated in the microscope column between the
condenser lens and the probe forming objective lens provides a resolution
below 0.1 nm. AC-STEM images are generated in high-angle annular dark
field mode with incoherently scattered electrons resulting in an intensity
strongly increasing with the atomic number (Z−contrast). The images
(1024 × 1024 pixels) were recorded with frame times between 10 and 20 s.
Due to the challenges of analyzing atomically-dispersed metal samples on
three-dimensional hosts,[4a] imaging was undertaken in multiple regions of
the specimen to ensure representativeness of the observations. Argon
sorption at 77 K was measured using a Micrometrics 3Flex instrument,
after evacuating the samples at 423 K for 7 h. The surface area was
estimated using the Brunauer-Emmett-Teller (BET) method calculated in
a range of relative pressure from 10−3 to 10−1. Powder X-ray diffraction
without further purification using
a gas chromatograph (HP-6890)
equipped with a HP-5 capillary column and a flame ionization detector.
Acknowledgements
This work was supported by the Swiss National Science
Foundation (200021-169679). Evgeniya Vorobyeva for
assistance with TEM analysis. ScopeM at ETH Zurich for use of
their facilities. The Laboratory of Nanometallurgy and the
Complex Materials group at ETH Zurich for providing access to a
Raman spectrometer and a freeze drier, respectively.
Keywords: single-atom catalysis • nitrogen-doped carbons•
synthesis-property-performance relationships • palladium •
metal-carrier interaction
[1]
a) A. Wang, J. Li, T. Zhang, Nat. Rev. Chem. 2018, 2, 65-81; b) J. Liu,
ACS Catal. 2017, 7, 34-59; c) H. Zhang, G. Liu, L. Shi, J. Ye, Adv. Energy
Mater. 2018, 8, 1701343; d) Z. Chen, E. Vorobyeva, S. Mitchell, E. Fako,
M. A. Ortuño, N. López, S. M. Collins, P. A. Midgley, S. Richard, G. Vilé,
patterns were obtained using
diffractometer operated in the Bragg−Brentano geometry applying Ni-
filtered Cu Kα (l =0.1541 nm) radiation. Data in the range 5–70° 2q with
a
PANalytical X’Pert PRO-MPD
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900547
ChemCatChem
FULL PAPER
J. Pérez-Ramírez, Nat. Nanotechnol. 2018, 13, 702-707; e) B. C. Gates,
[10] a) S. Liu, H. Yang, X. Huang, L. Liu, W. Cai, J. Gao, X. Li, T. Zhang, Y.
Huang, B. Liu, Adv. Funct. Mater. 2018, 28, 1800499; b) W. Ju, A. Bagger,
G.-P. Hao, A. S. Varela, I. Sinev, V. Bon, B. Roldan Cuenya, S. Kaskel,
J. Rossmeisl, P. Strasser, Nat. Commun. 2017, 8, 944; c) X. Li, W. Bi, M.
Chen, Y. Sun, H. Ju, W. Yan, J. Zhu, X. Wu, W. Chu, C. Wu, Y. Xie, J.
Am. Chem. Soc. 2017, 139, 14889-14892; d) Y. Chen, S. Ji, Y. Wang, J.
Dong, W. Chen, Z. Li, R. Shen, L. Zheng, Z. Zhuang, D. Wang, Y. Li,
Angew. Chem., Int. Ed. 2017, 56, 6937-6941; e) N. Cheng, S. Stambula,
D. Wang, M. N. Banis, J. Liu, A. Riese, B. Xiao, R. Li, T.-K. Sham, L.-M.
Liu, G. A. Botton, X. Sun, Nat. Commun. 2016, 7, 13638.
[11] a) W. Shen, W. Fan, J. Mater. Chem. A 2013, 1, 999-1013; b) L. Yang,
X. Zeng, W. Wang, D. Cao, Adv. Funct. Mater. 2018, 28, 1704537.
[12] a) X. Wang, G. Sun, P. Routh, D.-H. Kim, W. Huang, P. Chen, Chem.
Soc. Rev. 2014, 43, 7067-7098; b) H. Fei, J. Dong, Y. Feng, C. S. Allen,
C. Wan, B. Volosskiy, M. Li, Z. Zhao, Y. Wang, H. Sun, P. An, W. Chen,
Z. Guo, C. Lee, D. Chen, I. Shakir, M. Liu, T. Hu, Y. Li, A. I. Kirkland, X.
Duan, Y. Huang, Nat. Catal. 2018, 1, 63-72.
M. Flytzani-Stephanopoulos, D. A. Dixon, A. Katz, Catal. Sci. Technol.
2017, 7, 4259-4275; f) B. Qiao, A. Wang, X. Yang, L. F. Allard, Z. Jiang,
Y. Cui, J. Liu, J. Li, T. Zhang, Nat. Chem. 2011, 3, 634.
M. Moliner, J. E. Gabay, C. E. Kliewer, R. T. Carr, J. Guzman, G. L. Casty,
P. Serna, A. Corma, J. Am. Chem. Soc. 2016, 138, 15743-15750.
S. Mitchell, E. Vorobyeva, J. Pérez-Ramírez, Angew. Chem., Int. Ed.
2018, 57, 15316-15329.
a) Z. Chen, S. Mitchell, E. Vorobyeva, R. K. Leary, R. Hauert, T. Furnival,
Q. M. Ramasse, J. M. Thomas, P. A. Midgley, D. Dontsova, M. Antonietti,
S. Pogodin, N. López, J. Pérez-Ramírez, Adv. Funct. Mater. 2017, 27,
1605785; b) G. Vilé, D. Albani, M. Nachtegaal, Z. Chen, D. Dontsova, M.
Antonietti, N. Lopez, J. Pérez-Ramírez, Angew. Chem., Int. Ed. 2015, 54,
11265-11269.
a) E. Vorobyeva, Z. Chen, S. Mitchell, R. K. Leary, P. Midgley, J. M.
Thomas, R. Hauert, E. Fako, N. López, J. Pérez-Ramírez, J. Mater.
Chem. A 2017, 5, 16393-16403; b) Z. Chen, E. Vorobyeva, S. Mitchell,
E. Fako, N. López, S. M. Collins, R. K. Leary, P. A. Midgley, R. Hauert,
J. Pérez-Ramírez, Natl. Sci. Rev. 2018, 5, 642–652.
[2]
[3]
[4]
[5]
[6]
[7]
[13] M. Vikkisk, I. Kruusenberg, U. Joost, E. Shulga, I. Kink, K. Tammeveski,
Appl. Catal., B 2014, 147, 369-376.
a) D. A. Bulushev, M. Zacharska, A. S. Lisitsyn, O. Y. Podyacheva, F. S.
Hage, Q. M. Ramasse, U. Bangert, L. G. Bulusheva, ACS Catal. 2016,
6, 3442-3451; b) R. Arrigo, M. E. Schuster, Z. Xie, Y. Yi, G. Wowsnick,
L. L. Sun, K. E. Hermann, M. Friedrich, P. Kast, M. Hävecker, A. Knop-
Gericke, R. Schlögl, ACS Catal. 2015, 5, 2740-2753.
[14] a) L. Yang, D. Cheng, H. Xu, X. Zeng, X. Wan, J. Shui, Z. Xiang, D. Cao,
Proc. Natl. Acad. Sci. U. S. A. 2018, 115, 6626-6631; b) J. Wang, Z.
Huang, W. Liu, C. Chang, H. Tang, Z. Li, W. Chen, C. Jia, T. Yao, S. Wei,
Y. Wu, Y. Li, J. Am. Chem. Soc. 2017, 139, 17281-17284.
[15] X. Huang, H. Yan, L. Huang, X. Zhang, Y. Lin, J. Li, Y. Xia, Y. Ma, Z. Sun,
S. Wei, J. Lu, J. Phys. Chem. C 2018, DOI: 10.1021/acs.jpcc.8b07181.
[16] a) S. Vernuccio, P. R. von Rohr, J. Medlock, Ind. Eng. Chem. Res. 2015,
54, 11543-11551; b) D. Albani, M. Shahrokhi, Z. Chen, S. Mitchell, R.
Hauert, N. López, J. Pérez-Ramírez, Nat. Commun. 2018, 9, 2634.
[17] a) A. Ramos, I. Cameán, A. B. García, Carbon 2013, 59, 2-32; b) Z. Q.
Li, C. J. Lu, Z. P. Xia, Y. Zhou, Z. Luo, Carbon 2007, 45, 1686-1695.
[18] A. C. Ferrari, J. Robertson, Philos. Trans. R. Soc., A 2004, 362, 2477-
2512.
a) D. Liu, C. Wu, S. Chen, S. Ding, Y. Xie, C. Wang, T. Wang, Y. A.
Haleem, Z. ur Rehman, Y. Sang, Q. Liu, X. Zheng, Y. Wang, B. Ge, H.
Xu, L. Song, Nano Res. 2018, 11, 2217-2228; b) H. Fei, J. Dong, C. Wan,
Z. Zhao, X. Xu, Z. Lin, Y. Wang, H. Liu, K. Zang, J. Luo, S. Zhao, W. Hu,
W. Yan, I. Shakir, Y. Huang, X. Duan, Adv. Mater. 2018, 30, 1802146; c)
H. Yan, H. Cheng, H. Yi, Y. Lin, T. Yao, C. Wang, J. Li, S. Wei, J. Lu, J.
Am. Chem. Soc. 2015, 137, 10484-10487; d) C. Zhang, J. Sha, H. Fei,
M. Liu, S. Yazdi, J. Zhang, Q. Zhong, X. Zou, N. Zhao, H. Yu, Z. Jiang,
E. Ringe, B. I. Yakobson, J. Dong, D. Chen, J. M. Tour, ACS Nano 2017,
11, 6930-6941; e) H. B. Yang, S.-F. Hung, S. Liu, K. Yuan, S. Miao, L.
Zhang, X. Huang, H.-Y. Wang, W. Cai, R. Chen, J. Gao, X. Yang, W.
Chen, Y. Huang, H. M. Chen, C. M. Li, T. Zhang, B. Liu, Nat. Energy
2018, 3, 140-147.
a) D. A. Bulushev, M. Zacharska, E. V. Shlyakhova, A. L. Chuvilin, Y.
Guo, S. Beloshapkin, A. V. Okotrub, L. G. Bulusheva, ACS Catal. 2016,
6, 681-691; b) X. X. Wang, D. A. Cullen, Y.-T. Pan, S. Hwang, M. Wang,
Z. Feng, J. Wang, M. H. Engelhard, H. Zhang, Y. He, Y. Shao, D. Su, K.
L. More, J. S. Spendelow, G. Wu, Adv. Mater. 2018, 30, 1706758; c) A.
Zitolo, N. Ranjbar-Sahraie, T. Mineva, J. Li, Q. Jia, S. Stamatin, G. F.
Harrington, S. M. Lyth, P. Krtil, S. Mukerjee, E. Fonda, F. Jaouen, Nat.
Commun. 2017, 8, 957; d) L. Jiao, G. Wan, R. Zhang, H. Zhou, S.-H. Yu,
H.-L. Jiang, Angew. Chem., Int. Ed. 2018, 57, 8525-8529.
[19] C. Li, A. Zhao, W. Xia, C. Liang, M. Muhler, J. Phys. Chem. C 2012, 116,
20930-20936.
[20] R. Lin, S. K. Kaiser, R. Hauert, J. Pérez-Ramírez, ACS Catal. 2018, 8,
1114-1121.
[21] Y. Chen, S. Ji, Y. Wang, J. Dong, W. Chen, Z. Li, R. Shen, L. Zheng, Z.
Zhuang, D. Wang, Y. Li, Angew. Chem. Int. Ed. 2017, 56, 6937-6941.
[22] W. P. Zhou, A. Lewera, R. Larsen, R. I. Masel, P. S. Bagus, A.
Wieckowski, J. Phys. Chem. B 2006, 110, 13393-13398.
[23] N. Semagina, A. Renken, D. Laub, L. Kiwi-Minsker, J. Catal. 2007, 246,
308-314.
[24] S. Han, K. Sohn, T. Hyeon, Chem. Mater. 2000, 12, 3337-3341.
[25] D. C. Marcano, D. V. Kosynkin, J. M. Berlin, A. Sinitskii, Z. Sun, A.
Slesarev, L. B. Alemany, W. Lu, J. M. Tour, ACS Nano 2010, 4, 4806-
4814.
[8]
[9]
L. Borchardt, Q.-L. Zhu, M. E. Casco, R. Berger, X. Zhuang, S. Kaskel,
X. Feng, Q. Xu, Mater. Today 2017, 20, 592-610.
[26] D. Hulicova-Jurcakova, M. Seredych, G. Q. Lu, T. J. Bandosz, Adv. Funct.
Mater. 2009, 19, 438-447.
This article is protected by copyright. All rights reserved.

10.1002/cctc.201900547
ChemCatChem
FULL PAPER
Entry for the Table of Contents
FULL PAPER
All about the nitrogen: The influence
of different synthetic variables on the
amount and type of nitrogen species
introduced in nitrogen-doped carbons
and the subsequent ability to stabilize
palladium atoms is studied. The
stabilization of single atoms depends
on the carbon type and nitrogen
content, while the performance in
alkyne semi-hydrogenation correlates
with the electronic properties of the
metal (color map).
Simon Büchele, Zupeng Chen, Sharon
Mitchell*, Roland Hauert, Frank
Krumeich, and Javier Pérez-Ramírez*
Page 1 – Page 9
Tailoring nitrogen-doped carbons as
hosts for single-atom catalysts
This article is protected by copyright. All rights reserved.