A metal-free electrocatalyst for carbon dioxide reduction to multi-carbon hydrocarbons and oxygenates

Abstract

Electroreduction of carbon dioxide into higher-energy liquid fuels and chemicals is a promising but challenging renewable energy conversion technology. Among the electrocatalysts screened so far for carbon dioxide reduction, which includes metals, alloys, organometallics, layered materials and carbon nanostructures, only copper exhibits selectivity towards formation of hydrocarbons and multi-carbon oxygenates at fairly high efficiencies, whereas most others favour production of carbon monoxide or formate. Here we report that nanometre-size N-doped graphene quantum dots (NGQDs) catalyse the electrochemical reduction of carbon dioxide into multi-carbon hydrocarbons and oxygenates at high Faradaic efficiencies, high current densities and low overpotentials. The NGQDs show a high total Faradaic efficiency of carbon dioxide reduction of up to 90%, with selectivity for ethylene and ethanol conversions reaching 45%. The C2 and C3 product distribution and production rate for NGQD-catalysed carbon dioxide reduction is comparable to those obtained with copper nanoparticle-based electrocatalysts.

ARTICLE

Received 9 Jun 2016 | Accepted 8 Nov 2016 | Published 13 Dec 2016

Jingjie Wu1,*, Sichao Ma2,3,*, Jing Sun4,*, Jake I. Gold2, ChandraSekhar Tiwary1, Byoungsu Kim2,3, Lingyang Zhu2, Nitin Chopra5, Ihab N. Odeh5, Robert Vajtai1, Aaron Z. Yu2, Raymond Luo2, Jun Lou1, Guqiao Ding4,Paul J.A. Kenis2,3 & Pulickel M. Ajayan1

Electroreduction of carbon dioxide into higher-energy liquid fuels and chemicals is a promising but challenging renewable energy conversion technology. Among the electrocatalysts screened so far for carbon dioxide reduction, which includes metals, alloys, organometallics, layered materials and carbon nanostructures, only copper exhibits selectivity towards formation of hydrocarbons and multi-carbon oxygenates at fairly high efciencies, whereas most others favour production of carbon monoxide or formate. Here we report that nanometre-size N-doped graphene quantum dots (NGQDs) catalyse the electrochemical reduction of carbon dioxide into multi-carbon hydrocarbons and oxygenates at high Faradaic efciencies, high current densities and low overpotentials. The NGQDs show a high total Faradaic efciency of carbon dioxide reduction of up to 90%, with selectivity for ethylene and ethanol conversions reaching 45%. The C2 and C3 product distribution and production rate for NGQD-catalysed carbon dioxide reduction is comparable to those obtained with copper nanoparticle-based electrocatalysts.

1 Department of Materials Science and NanoEngineering, Rice University, 6100 Main Street, Houston, Texas 77005, USA. 2 School of Chemical Sciences, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801, USA. 3 International Institute for Carbon-Neutral Energy Research (WPI-I2CNER), Kyushu University, 744 Moto-oka, Nishi-ku, Fukuoka 819-0395, Japan. 4 State Key Laboratory of Functional Materials for Informatics, Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, Shanghai 200050, China. 5 Saudi Basic Industries Corporation (SABIC),Sugar Land, Texas 77478, USA. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to G.D. (email: mailto:[email protected]

Web End [email protected] ) or to P.J.A.K. (email: mailto:[email protected]

Web End [email protected] ) or to P.M.A. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:13869 | DOI: 10.1038/ncomms13869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

DOI: 10.1038/ncomms13869 OPEN

A metal-free electrocatalyst for carbon dioxide reduction to multi-carbon hydrocarbons and oxygenates

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13869

Electroreduction of carbon dioxide (CO2) with protons from water into valued commodity chemicals offers a potential route towards a carbon-neutral society1. However, the

development of this technology is hampered by the lack of active catalysts converting CO2 into more-energy embedded fuels or useful chemicals. Most metallic catalysts prefer the competitive off-pathway reduction of water to hydrogen, whereas others such as Au, Ag, Pb and Sn primarily catalyse production of carbon monoxide (CO) or formate (HCOO ) via a two-electron transfer pathway2,3. On the other hand, organometallic, two-dimensional materials and nitrogen (N)-doped nanostructured carbon materials in the form of few layer graphene, carbon nanotubes (CNTs) and carbon bres have been found only capable of catalysing CO2 reduction into CO48. To date, among the huge amount of screened materials, only Cu or copper oxide-derived catalyst are known to drive the electrolysis of CO2 into low-carbon hydrocarbons and oxygenates at fairly high Faradaic efciencies (FEs) through a multiple-step electron transfer pathway911. However, the overpotentials for hydrocarbons and oxygenates production are usually high ( 0.7 V) on most Cu

catalysts, whereas the overall current densities are typically below 50 mA cm 2 (refs 2,1113). Therefore, more efforts have to be devoted, to develop more powerful catalysts for efcient CO2 conversion to multi-carbon hydrocarbons and oxygenates. Reducing the structural features to sub-nanometre dimensions often signicantly enhances the hydrogenation activity for metal catalysts and it even has been demonstrated to transform a non-catalytic active bulk counterpart into a highly active catalyst towards CO2 reduction1417. Separately, the introduction of additional defects in carbon nanostructures by heteroatom doping gives rise to activity toward CO2 activation68,18,19.

Here we combine a high inherent defect density, achieved by tuning the dimensions and morphology of carbon nanostructure into nanometre scale, with foreign N-doping to obtain an advanced metal-free catalyst for electroreduction of CO2 to value-added chemicals. The resulting catalyst, specically N-doped graphene quantum dots (NGQDs), has substantially enriched density of N-doping defects at edge sites. These NGQDs exhibit high activity towards the electrochemical reduction of CO2 as evidenced by high reduction current densities at low overpotentials and, more importantly, they preferentially produce multi-carbon hydrocarbons and oxygenates, especially the C2 products ethylene (C2H4)

and ethanol (C2H5OH) at FEs comparable to those obtained using Cu nanoparticle-based catalysts.

ResultsNGQD synthesis. To expose the edge sites and increase their density in carbon nanostructures while simultaneously introducing non-metal heteroatom dopants at the edge locations, we intended to synthesize NGQDs through exfoliating and cutting graphene oxide (GO) precursor, and in-situ N doping in the dimethylformamide (DMF) solvent at elevated temperature and pressure (for synthesis details, see Supplementary Methods)20. The obtained NGQDs have a thickness between 0.7 and 1.8 nm, corresponding to one to three atomic layers (Supplementary Fig. 1), and a narrow lateral size distribution of 13 nm (Fig. 1a,b and Supplementary Fig. 1). The high-resolution transmission electron microscopy (TEM) images show the hexagonal-like morphology of NGQDs and their honeycomb framework (inset in Fig. 1b showing a representative zigzag edge), indicating the preservation of the graphene crystal structure. High-resolution TEM images of NGQDs with distinguishable atoms are difcult to take, as the NGQDs are even thinner than the carbon support lm in the TEM grid. We do not have a large quantity sample base to analyse the percentage of NGQDs with zigzag edges over

armchair edges, but the exigence of NGQDs with armchair edges could not be excluded in our sample. Pristine GQDs, without N-doping, were also synthesized from the same GO precursor using the identical hydrothermal analogous treatment in a solvent mixture of IPA/H2O (1:1 by volume) instead of DMF21. The pristine GQDs exhibit a similar morphology (thickness and lateral size) as the NGQDs (Supplementary Fig. 2). The Raman spectrum of NGQDs shows the characteristic D band at 1,350 cm 1 and G band at 1,589 cm 1, whereas the second order band (2D) has been quenched compared with pristine

GQDs (Fig. 1c). The ratio of ID/IG (D band intensity to G band intensity) for NGQDs is much higher than that for the pristine GQDs, due largely to the introduction of additional defects on N doping. The survey scan of X-ray photoelectron spectroscopy (XPS) clearly shows the presence of N in the NGQDs, whereas the N 1s peak was not observed for pristine GQDs (Supplementary Fig. 3). Analysis of the ne scan of the N 1s region provides information on the specic conguration of the N dopants in the carbon lattice network (Fig. 1d). Deconvolution of the N 1s peak reveals three peaks that can be assigned to pyridinic (398.5 eV), pyrrolic (400.0 eV) and graphitic N (401.2 eV)22. The total N content in NGQDs based on N/(C N) is around 6.0 at.%, in

which pyridinic N is predominant with B3.9 at.% (Fig. 1d and Supplementary Fig. 3).

Electrocatalytic activity and selectivity towards CO2 reduction. The electrocatalytic activity and selectivity of NGQDs towards CO2 reduction were evaluated in a ow cell in which NGQDs were deposited on a gas diffusion electrode (Supplementary Fig. 4)9. The electrolysis was performed in an electrolyte of 1 M KOH in the potentiostatic mode with cathodic potentials in the range of 0.20 to 1.10 V (versus reversible hydrogen

electrode (RHE)). The total FE of CO2 reduction increases as a more negative potential is applied, from 18% at 0.26 V to a

maximum of 90% at 0.75 V and, subsequently, it declines to

64% at 1.03 V (Fig. 2a). Only CO and HCOO are observed at

a low cathode potential of 0.26 V. However, when the applied

potential sweeps more negatively, various hydrocarbons of methane (CH4) and C2H4, and multi-carbon oxygenates including C2H5OH, acetate (CH3COO ) and n-propanol (n-

C3H7OH) are produced in addition to CO and HCOO (Fig. 2a). The fact that NGQDs favour catalysing CC bond formation beyond the potential of 0.61 V, leading to predominant

production of C2 and C3 products, is striking (Fig. 2a and Supplementary Fig. 5). C2H4 is the major hydrocarbon product with a maximum FE of 31% at 0.75 V, whereas CH4 has a

maximum FE of only 15% at 0.86 V. In addition to C2

hydrocarbons, another prominent feature of the NGQDs lies in its outstanding selectivity to formation of multi-carbon oxygenate liquid fuels. At 0.78 V, the maximum FE for multi-carbon

oxygenates reaches 26%, with the major component C2H5OH

accounting for 16%. In comparison, Cu nanoparticle-based catalysts yield trace CH4 (o1%), but comparable C2H4 (31.2%)

and C2H5OH (11.8%) at the similar potential of 0.74 V

(Supplementary Fig. 6)9. Overall, the NGQD catalyst exhibits electrocatalytic performance towards CO2 reduction that is similar to what has been observed for Cu nanoparticle catalysts in terms of the FE for C2 and C3 products.

To verify that the CO2 reduction reaction was catalysed by NGQDs, multiple control experiments were conducted. In one control experiment, argon gas was supplied to the cathode compartment instead of CO2, which resulted in the exclusive production of hydrogen over the cathodic potential window of

0.2 B 1.0 V. In a second control experiment using a bare

gas diffusion electrode as the cathode (no NGQD catalyst),

2 NATURE COMMUNICATIONS | 7:13869 | DOI: 10.1038/ncomms13869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13869 ARTICLE

a b

c

d

H

G

NGQDs

GQDs Graphitic N(0.9%)

Pyrrolic N

(1.2%)

D

2D

N

Intensity (a.u.)

Intensity (a.u.)

N

N

Pyridinc N (3.9%)

NGQDs

GQDs

1,000 1,500 2,000 2,500 3,000

404 402 400 398 396

Binding energy (eV)

Raman shift (cm1)

Figure 1 | Nanostructure and specic nitrogen dopant congurations of NGQDs. (a) Low-magnication TEM image of NGQDs. Scale bar, 50 nm.(b) High-magnication TEM image of NGQDs. Scale bar, 2 nm. Inset shows a single NGQD containing zigzag edges as circled. The yellow line outlines the zigzag edge. Scale bar in inset, 1 nm. (c) Raman spectrum of NGQDs as compared with that of pristine GQDs. (d) High-resolution N 1s spectrum for NGQDs, deconvoluted into three sub-peaks representing pyridinic, pyrrolic and graphitic N. The value in parentheses is the corresponding N atomic concentration calculated based on N/(N C). The inset is a schematic demonstrating zigzag edges and the N-bonding conguration with respect to

pyridinic (black), pyrrolic (blue) and graphitic (pink) N. In comparison, the N 1s peak does not appear in GQDs sample.

hydrogen formation prevailed over the same potential region. Furthermore, in an experiment using an isotopically labelled

13CO2 gas feed, mass spectrometry showed signals for 13CO (m/z 29), 13CH4 (m/z 17) and 13C2H4 (m/z 30), whereas a

spin doublet appeared in 1H NMR due to proton coupling to 13C in the liquid products (Supplementary Fig. 7), all conrming that these products are derived from 13CO2 electroreduction.

In addition, inductively coupled plasma optical emission spectroscopy revealed that the NGQDs samples only contained1.02 p.p.m. Cu, corresponding to a Cu loading of around 5 10 3 mg cm 2 on the cathode, which is too low to catalyse

CO2 reduction to hydrocarbons and oxygenates at the levels observed here, as previously reported23. These results conrm the activity of the NGQDs studied here towards efcient reduction of CO2 to multi-carbon hydrocarbons and oxygenates.

Origin of electrocatalytic activity and selectivity. To unveil the origin of the activity and selectivity of NGQDs, pristine GQDs were also tested for CO2 reduction under identical experimental conditions. The hydrogen evolution reaction dominates over the CO2 reduction reaction for the GQDs electrode. GQDs not only initialize CO2 reduction at a more negative potential, but also primarily produce CO and HCOO at even lower FEs than NGQDs (Fig. 2b). Compared with NGQDs, the FEs of hydrocarbons such as CH4 and C2H4 observed with the GQDs are low not more than 56% each. Moreover, only a trace amount of CH3COO and no C2H5OH or n-C3H7OH were observed.

In addition, the Tafel plots derived from the partial current

density of CO2 reduction (jCO2 reduction) versus cathode potential show that the GQDs electrode has a larger apparent Tafel slope than the NGQDs electrode (371 mV dec 1 for GQDs versus 198 mV dec 1 for NGQDs), indicating that the GQDs electrode exhibits much slower kinetics than the NGQDs electrode (Fig. 2d). This NGQDs versus GQDs performance comparison underscores the signicance of the N-doping defects in the NGQD catalyst in determining its activity. Previously, the incorporation of N defects to sp2 carbon nanostructures to induce charge density was reported to be crucial for actively catalysing electrochemical reactions such as oxygen reduction reaction (ORR)2426. Among the most common N congurations, pyridinic N is experimentally proven to create the most active site for the ORR2729. Similar to ORR, our prior work on N-doped CNTs and graphene suggests that pyridinic N is the leading active site for CO2 reduction6,7. The acidic CO2 molecule prefers adsorbing onto the Lewis basic pyridinic N group in carbon nanostructures27.

The NGQDs samples were analysed by ex-situ post-reaction XPS. After testing for CO2 reduction, the relative concentration of pyridinic N (398.5 eV) decreases from 65 to 38% and another N component with a peak of 400.0 eV (pyrrolic or pyridonic N) increases from 20 to 50%, whereas the percentage of graphitic N (401.2 eV) remains relatively unchanged (from 15 to 12%; Supplementary Fig. 8). The change of N component fraction before and after CO2 reduction measurement possibly results from the adsorption of CO2 onto the active pyridinic N that causes the upshift of binding energy of pyridinic N to a similar value of pyrrolic N. By applying potential, the adsorbed CO2 is

NATURE COMMUNICATIONS | 7:13869 | DOI: 10.1038/ncomms13869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13869

a b

NGQDs

NRGOs

GQDs

Faradaic efficiency (%)

100 80 60 40 20

n-C3H7OH

CH3COO C2H5OH

C2H4

CH4

CO

HCOO

Faradaic efficiency (%)

j CO 2 reduction(mA cm

30

20

10

0

0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0

E (V versus RHE) E (V versus RHE)

E (V versus RHE)

c d

2 )

Faradaic efficiency (%)

100 80 60 40 20

100

10

1

0.1

0.2 0.4 0.6 0.8 1.0 1.2

NRGOs489 mV dec1

GQDs371 mV dec1

NGQDs198 mV dec1

0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

E (V versus RHE)

Figure 2 | Electrocatalytic activity of carbon nanostructures towards CO2 reduction. (a) FEs of carbon monoxide (CO), methane (CH4), ethylene (C2H4), formate (HCOO ), ethanol (EtOH), acetate (AcO ) and n-propanol (n-PrOH) at various applied cathodic potential for NGQDs. (b) FE of CO2 reduction products for pristine GQDs. (c) Selectivity to CO2 reduction products for NRGOs. (d) Tafel plots of partial current density of CO2 reduction versus applied cathodic potential for three nanostructured carbon catalysts. The error bar represents the s.d. of three separate measurements for an electrode.

C2H4

a

b

60 NGQDs

0.4 0.6 0.8 1.0 1.2

15 GQDs

CO

j(mA cm2 )

HCOO

CH4

j(mA cm2 )

C2H4

CH4

C2H5OH

CH3COO

CO

HCOO

n-C3H7OH

CH3COO

0 0.2

0 0.4 0.6 0.8 1.0 1.2

E (V versus RHE)

E (V versus RHE)

Figure 3 | Partial current densities as a function of cathode potential for various products from electrochemical CO2 reduction. (a) When using NGQDs as cathode catalyst. (b) When using GQDs as cathode catalyst.

further reduced to products that are released from the pyridinic N site, then the pyridinic N regenerates for next round of CO2 adsorption/reduction.

To further explore the active site, the N-doped reduced GOs (NRGOs) were prepared by doping GOs with NH3 at 800 C. The

NRGOs contain a concentration of each specic N conguration that is similar to the distribution in NGQDs, but the NRGOs have larger micrometre-scale lateral size (Supplementary Figs 9 and10). The onset potential for CO2 reduction on the NRGOs electrode is identical to that for the NGQDs electrode (Fig. 2a,c). However, the NRGOs electrode primarily catalyses CO2 reduction to form CO (Fig. 2c), similar to previously reported observations when using N-doped graphene and N-doped CNTs6,7. The formation of noticeable amounts of hydrocarbons CH4 (B6%), C2H4 (B5%) and oxygenates C2H5OH (B4%)

starts at 0.90 V on NRGOs, a signicantly more negative

potential than that ( 0.61 V) on NGQDs. Compared with

NRGOs, pristine RGOs exclusively catalyse the hydrogen

evolution reaction, underlining the ability of N doping to transform carbon nanostructures from being inert to highly catalytically active. The kinetics for CO2 reduction on NRGOs is also slower than that on NGQDs, for example, a Tafel slope of 489 mV dec 1 for NRGOs versus 198 mV dec 1 for NGQDs (Fig. 2d). This comparison of NGQDs and NRGOs clearly suggests that the morphology also plays a key role in determining the activity of carbon nanostructures towards CO2 reduction in addition to doping-induced defects. When 13 mm diameter NRGOs sheets are pulverized into 13 nm quantum dots, the density of exposed edge sites increases by three orders of magnitude: 0.05% for 1 mm NRGO sheet versus 44% for 1 nm

NGQD (Supplementary Fig. 11). Owing to the lower formation energy of N doping at the edge sites than the basal planes, especially for pyridinic N, most of the N dopants would locate at the edge sites in NGQDs30,31. In contrast, in the case of NRGOs, N dopants mostly locate in the basal planes because insufcient edge sites are present to accommodate the dopants. Although

4 NATURE COMMUNICATIONS | 7:13869 | DOI: 10.1038/ncomms13869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13869 ARTICLE

they are similar in pyridinic N content, NGQDs tend to have higher density of pyridinic N at edge sites than NRGOs. The pyridinic N at the edge site is believed to be more active to induce CC bond formation than those at the basal plane, which leads to a higher yield of C2 and C3 products on NGQDs electrode than on NRGOs electrode32.

Production rate of CO2 reduction products. In addition to the promising selectivity with respect to the formation of multi-carbon hydrocarbons and oxygenates, the NGQDs electrode also exhibits a high CO2 reduction current density of 100 mA cm 2 at relative low potentials (Supplementary Fig. 12). The partial current densities for production of CO, C2H4 and C2H5OH reach 23, 46 and 21 mA cm 2 at 0.86 V, respectively (Fig. 3a),

which are on the same order of magnitude compared with corresponding partial current densities observed for commercial Cu nanoparticles (around 2040 nm) under identical testing conditions9. Accordingly, CO, C2H4 and C2H5OH could be produced at rates of 4.2, 1.4 and 0.7 mol h 1 m 2 at 0.86 V,

respectively (Supplementary Fig. 13). In contrast, the GQDs exhibit a partial current density or production rate for C2H4 that

is at least one order of magnitude lower at comparable potentials (Fig. 3b and Supplementary Fig. 13).

DiscussionA metal-free catalyst comprising NGQDs is discovered to exhibit extraordinary activity towards CO2 reduction. Interestingly,

NGQDs show predominant selectivity to the production of multi-carbon hydrocarbons and oxygenates, whereas GQDs and NRGOs primarily yield CO and HCOO . Detailed characterization reveals that the unique nanostructure in combination with utmost exposure of edge sites and heteroatom N doping grants NGQDs the unprecedented activity and selectivity. This nding expands the horizon for the design of high-performance electrocatalysts for CO2 reduction. However, the reaction mechanism of electroreduction of CO2 over NGQDs remains elusive and the unravelling requires further work. The combination of high-sensitive operando spectroscopy and rst-principle calculation would be an efcient way to provide insight into the reaction intermediates to reveal the reaction pathways. In addition, the activity difference between the zig-zag and armchair edge necessitates exploration as well in the future.

Methods

GQD synthesis. NGQDs were synthesized by heat treating the GO dispersion in DMF in a PTFE-lined autoclave at 200 C for 10 h. The pristine GQDs were prepared in a similar way, except using a mixture of isopropanol and water (1:1 by volume) as the solvent. The NRGO was prepared by doping GO in a tube furnace at 800 C, while owing ammonia for 1 h. More details about the synthesis and characterization are in the Supplementary Information.

Materials characterization. The morphology and crystallinity of N-doped and pristine GQDs was characterized by high-resolution eld-emission gun TEM (JEOL 2,100 FEG TEM). The TEM samples were prepared by dropping the QDs solution onto the ultrathin carbon lm TEM grid followed by vacuum drying at 100 C. The thickness of QDs was measured by atomic force microscopy with tapping mode (Bruker Multimode 8). The atomic force microscopy samples were prepared by dropping the QDs solution onto the Mica substrate. The morphology of NRGOs was analysed by scanning electron microscope (FEI Quanta 400 FEG ESEM). The Raman spectra were taken with a Renishaw inVia Raman microscope with 514 nm laser excitation. XPS measurements were performed to analyse the element component and oxide state of QDs and NRGOs at ambient temperature using PHI Quantera with Al-Ka X-ray source. The Cu content in the NGQDs was determined by using inductively coupled plasma optical emission spectroscopy (PerkinElmer-Optima 2000DV). One millilitre of original liquid sample that is dissolved in DMF was diluted with 5 ml water, then one drop (B0.05 ml) of concentrated HNO3 was used to adjust the pH to be o7. Two different emission lines: l 224.7000 nm and l 213.5970 nm were used to detect the Cu element.

The detection limit for Cu is 10 p.p.b.

Electrochemical measurement of CO2 reduction. The CO2 reduction reaction was conducted in an electrochemical ow cell composed of targeted quantum dots or NRGO-based gas diffusion electrodes (see Supplementary Fig. 4).

The electrochemical measurement was carried out at ambient pressure and temperature. The electrolysis was performed under potentiostatic mode with a full cell voltage ranging from 1.6 to 3.5 V controlled by a potentiostat (Autolab

PGSTAT-30, EcoChemie). Both the catholyte and anolyte were 1 M KOH(pH 13.48, as calibrated by a pH meter (Thermo Orion, 9106BNWP). Individual electrode potentials were recorded using multimeters (AMPROBE 15XP-B) connected to each electrode and a reference electrode (Ag/AgCl; RE-5B, BASi) placed in the electrolyte exit stream. The measured potentials after iR compensation were rescaled to the RHE by E (versus RHE) E (versus Ag/AgCl) 0.209 V 0.0591

V/pH pH. The current reported here was obtained by averaging the span of time

(at least 180 s) for each applied voltage.

For each applied voltage, after the cell reached steady state, 1 ml of the efuent gas stream was periodically sampled and diverted into a gas chromatograph (Thermo Finnegan Trace GC) equipped with both the thermal conductivity and ame ionization detector, and a Carboxen 1,000 column (Supelco). Meanwhile, the exit catholyte was collected at each applied voltage followed by identifying and quantifying using 1H NMR (nuclear magnetic resonance, UI500NB, Varian).

Data availability. The data that support the ndings of this study are available from the corresponding author upon request.

References

1. Whipple, D. T. & Kenis, P. J. A. Prospects of CO2 utilization via direct heterogeneous electrochemical reduction. J. Phys. Chem. Lett. 1, 34513458 (2010).

2. Hori, Y., Wakebe, H., Tsukamoto, T. & Koga, O. Electrocatalytic process of CO selectivity in electrochemical reduction of CO2 at metal electrodes in aqueous media. Electrochim. Acta 39, 18331839 (1994).

3. Qiao, J., Liu, Y., Hong, F. & Zhang, J. A review of catalysts for the electroreduction of carbon dioxide to produce low-carbon fuels. Chem. Soc. Rev. 43, 631675 (2014).

4. Lin, S. et al. Covalent organic frameworks comprising cobalt porphyrins for catalytic CO2 reduction in water. Science 349, 12081213 (2015).

5. Asadi, M. et al. Robust carbon dioxide reduction on molybdenum disulphide edges. Nat. Commun. 5, 4470 (2014).

6. Wu, J. et al. Achieving highly efcient, selective and stable CO2 reduction on nitrogen doped carbon nanotubes. ACS Nano 9, 53645371 (2015).

7. Wu, J. et al. Incorporation of nitrogen defects for efcient reduction of CO2 via two-electron pathway on three-dimensional graphene foam. Nano Lett. 16, 466470 (2016).

8. Kumar, B. et al. Renewable and metal-free carbon nanobre catalysts for carbon dioxide reduction. Nat. Commun. 4, 2819 (2013).

9. Ma, S. et al. One-step electrosynthesis of ethylene and ethanol from CO2 in an alkaline electrolyzer. J. Power Sources 301, 219228 (2016).

10. Kuhl, K. P., Cave, E. R., Abram, D. N. & Jaramillo, T. F. New insights into the electrochemical reduction of carbon dioxide on metallic copper surfaces. Energy Environ. Sci. 5, 70507059 (2012).

11. Li, C. W. & Kanan, M. W. CO2 reduction at low overpotential on Cu electrodes resulting from the reduction of thick Cu2O lms. J. Am. Chem. Soc. 134, 72317234 (2012).

12. Reske, R., Mistry, H., Behafarid, F., Roldan Cuenya, B. & Strasser, P. Particle size effects in the catalytic electroreduction of CO2 on Cu nanoparticles. J. Am.

Chem. Soc. 136, 69786986 (2014).13. Hori, Y., Murata, A. & Takahashi, R. Formation of hydrocarbons in the electrochemical reduction of carbon dioxide at a copper electrode in aqueous solution. J. Chem. Soc. Faraday Trans. 1 85, 23092326 (1989).

14. Liu, C. et al. Carbon dioxide conversion to methanol over size-selected Cu4 clusters at low pressures. J. Am. Chem. Soc. 137, 86768679 (2015).

15. Liu, C., He, H., Zapol, P. & Curtiss, L. A. Computational studies of electrochemical CO2 reduction on subnanometer transition metal clusters.

Phys. Chem. Chem. Phys. 16, 2658426599 (2014).16. Gao, S. et al. Partially oxidized atomic cobalt layers for carbon dioxide electroreduction to liquid fuel. Nature 529, 6871 (2016).

17. Kauffman, D. R., Alfonso, D., Matranga, C., Qian, H. & Jin, R. Experimental and computational investigation of Au25 clusters and CO2: a unique interaction and enhanced electrocatalytic activity. J. Am. Chem. Soc. 134, 1023710243 (2012).

18. Zhang, S. et al. Polyethylenimine-enhanced electrocatalytic reduction of CO2 to

formate at nitrogen-doped carbon nanomaterials. J. Am. Chem. Soc. 136, 78457848 (2014).19. Sreekanth, N., Nazrulla, M. A., Vineesh, T. V., Sailaja, K. & Phani, K. L. Metal-free boron-doped graphene for selective electroreduction of carbon dioxide to formic acid/formate. Chem. Commun. 51, 1606116064 (2015).

NATURE COMMUNICATIONS | 7:13869 | DOI: 10.1038/ncomms13869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13869

20. Sun, J. et al. Ultra-high quantum yield of graphene quantum dots: aromatic-nitrogen doping and photoluminescence mechanism. Part. Part. Syst. Charact. 32, 434440 (2015).

21. Shen, J. et al. Liquid phase exfoliation of two-dimensional materials by directly probing and matching surface tension components. Nano Lett. 15, 54495454 (2015).

22. Shari, T., Hu, G., Jia, X. & Wgberg, T. Formation of active sites for oxygen reduction reactions by transformation of nitrogen functionalities in nitrogen-doped carbon nanotubes. ACS Nano 6, 89048912 (2012).

23. Lum, Y. et al. Trace levels of copper in carbon materials show signicant electrochemical CO2 reduction activity. ACS Catal. 6, 202209 (2016).

24. Gong, K., Du, F., Xia, Z., Durstock, M. & Dai, L. Nitrogen-doped carbon nanotube arrays with high electrocatalytic activity for oxygen reduction. Science 323, 760764 (2009).

25. Qu, L., Liu, Y., Baek, J.-B. & Dai, L. Nitrogen-doped graphene as efcient metal-free electrocatalyst for oxygen reduction in fuel cells. ACS Nano 4, 13211326 (2010).

26. Zhang, L. & Xia, Z. Mechanisms of oxygen reduction reaction on nitrogen-doped graphene for fuel cells. J. Phys. Chem. C 115, 1117011176 (2011).

27. Guo, D. et al. Active sites of nitrogen-doped carbon materials foroxygen reduction reaction claried using model catalysts. Science 351, 361365 (2016).

28. Wu, J. et al. Nitrogen-doped graphene with pyridinic dominance as a highly active and stable electrocatalyst for oxygen reduction. ACS Appl. Mater. Interfaces 7, 1476314769 (2015).

29. Xing, T. et al. Observation of active sites for oxygen reduction reaction on nitrogen-doped multilayer graphene. ACS Nano 8, 68566862 (2014).

30. Yu, S. S., Zheng, W. T., Wen, Q. B. & Jiang, Q. First principle calculations of the electronic properties of nitrogen-doped carbon nanoribbons with zigzag edges. Carbon 46, 537543 (2008).

31. Li, Y., Zhou, Z., Shen, P. & Chen, Z. Spin gapless semiconductor metal half-

metal properties in nitrogen-doped zigzag graphene nanoribbons. ACS Nano 3, 19521958 (2009).32. Shen, A. et al. Oxygen reduction reaction in a droplet on graphite: direct evidence that the edge is more active than the basal plane. Angew. Chem. Int. Ed. 53, 1080410808 (2014).

Acknowledgements

This research is nancially supported by SABIC. We acknowledge Dr Mike Wallar at

University of South Carolina for the mass spectroscopy measurement.

Author contributions

J.W. and P.M.A. conceived the idea of catalysis of CO2 reduction. S.M., P.J.A.K. and J.W.

designed the catalysis measurement experiments. S.M., J.I.G., A.Z.Y. and R.L. performed

the electrochemical measurement. B.K. executed the isotopic labeling test. L.Z. carried

out the coupling NMR test. J.W. and J.S. synthesized and characterized the catalysts.

C.S.T. assisted the TEM characterization. P.M.A., P.J.A.K. and G.D. supervised the

experiment work. J.W. wrote the manuscript with the inputs from N.C., I.N.O., J.L. and

R.V. All authors discussed the results and commented on the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/

http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/

http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Wu, J. et al. A metal-free electrocatalyst for carbon dioxide

reduction to multi-carbon hydrocarbons and oxygenates. Nat. Commun. 7, 13869

doi: 10.1038/ncomms13869 (2016).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in

published maps and institutional afliations.

This work is licensed under a Creative Commons Attribution 4.0

International License. The images or other third party material in this

article are included in the articles Creative Commons license, unless indicated otherwise

in the credit line; if the material is not included under the Creative Commons license,

users will need to obtain permission from the license holder to reproduce the material.

To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2016

6 NATURE COMMUNICATIONS | 7:13869 | DOI: 10.1038/ncomms13869 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications