Solid frustrated-Lewis-pair catalysts constructed by regulations on surface defects of porous nanorods of CeO2

Abstract

Identification on catalytic sites of heterogeneous catalysts at atomic level is important to understand catalytic mechanism. Surface engineering on defects of metal oxides can construct new active sites and regulate catalytic activity and selectivity. Here we outline the strategy by controlling surface defects of nanoceria to create the solid frustrated Lewis pair (FLP) metal oxide for efficient hydrogenation of alkenes and alkynes. Porous nanorods of ceria (PN-CeO₂ ) with a high concentration of surface defects construct new Lewis acidic sites by two adjacent surface Ce³⁺ . The neighbouring surface lattice oxygen as Lewis base and constructed Lewis acid create solid FLP site due to the rigid lattice of ceria, which can easily dissociate H-H bond with low activation energy of 0.17 eV.

ARTICLE

Received 14 Dec 2015 | Accepted 15 Mar 2017 | Published 18 May 2017

Sai Zhang1,*, Zheng-Qing Huang2,*, Yuanyuan Ma1,3, Wei Gao1, Jing Li1, Fangxian Cao1, Lin Li4, Chun-Ran Chang2 & Yongquan Qu1,3

Identication on catalytic sites of heterogeneous catalysts at atomic level is important to understand catalytic mechanism. Surface engineering on defects of metal oxides can construct new active sites and regulate catalytic activity and selectivity. Here we outline the strategy by controlling surface defects of nanoceria to create the solid frustrated Lewis pair (FLP) metal oxide for efcient hydrogenation of alkenes and alkynes. Porous nanorods of ceria (PN-CeO2) with a high concentration of surface defects construct new Lewis acidic sites by two adjacent surface Ce3 . The neighbouring surface lattice oxygen as Lewis base and constructed Lewis acid create solid FLP site due to the rigid lattice of ceria, which can easily dissociate HH bond with low activation energy of 0.17 eV.

DOI: 10.1038/ncomms15266 OPEN

Solid frustrated-Lewis-pair catalysts constructed by regulations on surface defects of porous nanorods of CeO2

1 Center for Applied Chemical Research, Frontier Institute of Science and Technology and State Key Laboratory for Mechanical Behavior of Materials, Xian Jiaotong University, XiYi Hall, 99 Yanxiang Road, Xian, Shannxi 710049, China. 2 Institute of Industrial Catalysis, School of Chemical Engineering and Technology, Xian Jiaotong University, Xian 710049, China. 3 MOE Key Laboratory for Nonequilibrium Synthesis and Modulation of Condensed Matter, Xian Jiaotong University, Xian 710049, China. 4 State Key Laboratory of Catalysis, Dalian Institute of Chemical Physics, Chinese Academy ofSciences, Dalian 116023, China. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed toY.Q. (email: mailto:[email protected]

Web End [email protected] ) or to C.-R.C. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 8:15266 | DOI: 10.1038/ncomms15266 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15266

Catalytic hydrogenation of unsaturated substrates, a fundamental transformation process, is nding prolic industrial reactions, especially for petrochemistry13.

Hydrogen activation by either transition metal complexes or metal-based heterogeneous catalysts have been well developed and successfully applied for various reactions4. Recently, frustrated Lewis pairs (FLPs), Lewis acids and bases that are sterically prevented from interaction to form Lewis acidbase adjuncts, can efciently and cooperatively activate many small molecules (for example, H2, CO2, and NO) and even strong CH bond for important catalytic reactions including hydrogenation, hydroamination and carbon dioxide reduction525.

However, molecular-based homogeneous FLP catalytic systems raise the difculties in product purication and catalyst recovery. Therefore, the development of heterogeneous catalysts with FLP-like activity is extremely expected. Meanwhile, the combination of Au powder as Lewis acid and molecular Lewis bases (for example, imine and nitrile) successfully constructs the rst example of semi-heterogeneous FLP catalytic system26. In this catalytic system, large quantity of Lewis acids and bases were required to avoid the formation of stable Lewis acidbase complexes. The semi-heterogeneous molecular sieves/B(C6F5)3

FLPs deliver remarkable catalytic activity for hydrogenation of ketones and aldehydes27. Graphene synthesized by chemical exfoliation has been reported as efcient catalysts for hydrogen activation through an FLP-like behaviour28. When the Lewis acid and base sites are in an adequate distance, the activation of hydrogen becomes possible2931. However, they require restricted demanding reaction conditions with long reaction time and a high H2 pressure of 30 bar28. Formation of an FLP of the

Na H /hydroxyl proton O(H ) has been successfully demonstrated for Ptx/NaY zeolites by using in situ neutron diffraction and spectroscopic measurements32. Inspired by the catalytic mechanism of molecular FLP catalysts, it is also possible to develop novel heterogeneous FLP-like catalysts if independent Lewis acidic and basic species coexist in one material. Recently, Ozin and colleagues3337 has reported that the surface FLP sites of In2O3-x(OH)y created by a Lewis acidic coordinately unsaturated surface indium site proximal to an oxygen vacancy and a Lewis basic surface hydroxide site showed the catalytic activity for the CO2 reduction by H2 in both experimental results and theoretical predictions. It provides a possible strategy to construct solid FLP sites in oxides by creating oxygen vacancy.

Nanostructured ceria is characterized with a large number of surface-bound defects that are primarily oxygen vacancy or Ce3 species as active sites for heterogeneous catalysis3843. In previous investigations, the catalytic ability of ceria is driven by the ability to switch between different oxidation states (Ce3 /Ce4 redox pair) accompanied with the storage and release of oxygen when the temperature is higher than 150 C (ref. 44). At low temperature, the surface oxygen is less mobile and can be rmly held in the lattice matrix45,46. With these precedents in mind, the combination of the xed surface oxygen as Lewis base and surface defects as Lewis acid in ceria is similar to the sterically encumbered pairs of Lewis acids and bases for molecular FLP catalysts, if the surface Lewis acid and base are independent.

We choose ceria as another candidate to demonstrate the possibility to construct FLP sites due to its reversible Ce3 /Ce4 redox pair and rich surface chemistry38. Ceria has been reported to exhibit the catalytic activity for the selective semi-hydrogenation of alkynes through the H2 activation by homolytic or heterolytic pathway and subsequent hydrogenation4751. In general, a high temperature is required due to the inherent difculty of ceria for H2 activation. Heterolytic (between lattice

Ce4 and O2 ) pathway is considered as the pathway for H2 dissociation due to the relative low activation energy51. Hence,

it is not surprise to observe the decreased hydrogenation activity of ceria with the increased surface defects47,49. Incapability for hydrogenation of alkenes and restricted reaction conditions for semi-hydrogenation of alkynes indicate the low hydrogenation activity of ceria.

In this work, solid FLP sites on CeO2 surface, which deliver a very high catalytic activity for hydrogenation of alkenes and alkynes with a wide scope under mild conditions (T 100 C and

P(H2) 1.0 MPa), have been successfully created by regulating

their surface defects. The richness of surface defects is found to be the key for construction of a new surface Lewis acidic center by two adjacent reduced surface Ce atoms near an oxygen vacancy. The xed surface lattice oxygen as Lewis base and constructed Lewis acid have a large possibility to be close enough but independent due to the richness of surface defects and unique geometrical and electronic congurations, similar to the molecular-based FLPs. Porous nanorods of ceria (PN-CeO2)

with a large surface Ce3 fraction of 30.8% associated with a high concentration of oxygen vacancy have been demonstrated as the solid metal oxide FLP catalyst with a low activation barrier of 0.17 eV for H2 dissociation.

ResultsDesign strategy of solid FLP catalysts. Metal oxides possess the surface Lewis acid (metal cation) and Lewis base (oxygen anion), which form the Lewis acid-base adjuncts and deliver no FLP-like activity51,52. Fortunately, the surface defects induced by the reducible/oxidable valence states of metal cations associated with oxygen vacancy in metal oxides can create novel surface Lewis acidic sites and/or Lewis basic sites. The xed surface lattice atoms near the constructed surface defects in a close proximity may be unbonded, which can avoid the formation of classic Lewis acidbase adjuncts due to the rigid crystal lattice. Such a conguration may activate the small inert molecules such as H2 and CO2. Our hypothesis here is that the activation of inert molecules can be realized by rationally designed surface engineering and regulation on the structural defects of metal oxides to create novel FLP active sites. This possibility is illustrated by taking CeO2 as an example in view of both the geometrical and electronic congurations (Fig. 1).

Starting from the ideal CeO2(110) facet, the surface Ce and O form Lewis acidbase adjuncts (Fig. 1a). Two kinds of surface congurations, CeI-OI(aBd) and CeI-OIIc, are identied on the ideal CeO2(110) surface. CeI and OI(aBd) are adjacent and contiguous to each other, forming a classic Lewis acidbase adjunct, whereas the unbonded CeI and OIIc show a distance of4.60 , which may deliver FLP-like activity. However, the electronic interactions between CeI and its neighbouring OIb will block the function of CeI-OIIc pairs as shown in Fig. 1d. Therefore, the removal of oxygen atom directly linked with CeI is

the prerequisite to construct a pair of unbonded Lewis acid and base sites. When the OIb atom is removed, two surface Ce atoms (including CeI) are reduced and one oxygen vacancy is produced, as demonstrated by the Bader charge population analysis of surface Ce (Supplementary Table 1). After structural relaxation, the OIc near the oxygen vacancy immigrates to the middle of the

CeI and CeII, leading to CeI being surrounded by three adjacent oxygen atoms (Fig. 1b). In this case, three types of congurations can be found including CeI-OIa/Id, CeI-OIc and CeI-OIIc, of which

the CeI-OIa/Id and CeI-OIc form classic Lewis acidbase adjuncts.

The CeI-OIIc with a distance of 4.72 falls in the domain of solid FLPs to activate small molecules but is still hindered by OIc/Ia due

to the electronic interaction between CeI and OIc/Ia (Fig. 1e). When the second surface oxygen (OIa) is removed, two oxygen vacancies and four reduced Ce cations (including CeI and CeII)

2 NATURE COMMUNICATIONS | 8:15266 | DOI: 10.1038/ncomms15266 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15266 ARTICLE

a b c

OIa

OIa

OIc OIc OIc

OIb OIIc OIIc OIIc

OId OId OId

CeI CeI CeI

CeII CeII CeII

d(CeI,Ola) = 2.34 d(CeI,Ollc) = 4.72 d(CeI,Ollc) = 4.47

3.99

d e f

OId

CeI

CeII

OId

CeI

CeII

OIa OId

CeI

CeII

OIc OIIc

OIa

OIc

OIb OIIc

OIc OIIc

1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0

(eV)

Figure 1 | Schematic images of concept for design of solid frustrated Lewis pairs in CeO2 crystal structure. (a) Optimized structure of ideal CeO2(110). (b) Optimized structure of CeO2(110) with one oxygen vacancy. (c) Optimized structure of CeO2(110) with two adjacent oxygen vacancies. (d) Electron-density isosurface of ideal CeO2(110). (e) Electron-density isosurface of CeO2(110) with one oxygen vacancy. (f) Electron-density isosurface of CeO2(110)

with two oxygen vacancies. The electron-density isosurfaces are plotted at 0.01 e bohr 3. The colour bar represents the electrostatic potential scale.

are generated (Fig. 1c and Supplementary Table 1). The left oxygen OIc and OId will not migrate to the middle of two Ce cations. Thus, the reduced Ce cations (CeI and CeII) and surface lattice oxygen (OIIc) are independent Lewis acid and base, indicating that CeI-OIIc and CeII-OIIc may act as the solid FLPs

(Fig. 1c,f), similar to molecular FLPs. In addition, a special FLP-like active site of two adjacent reduced surface Ce sites (CeI and CeII) and lattice OIIc is constructed and denoted as (CeI,CeII)-OIIc hereafter. Compared with CeI-OIIc and CeII-OIIc congurations, (CeI,CeII)-OIIc with a shorter distance (3.99 ) and stronger charge contraction by two adjacent reduced surface Ce cations is likely to deliver a higher capability to activate small molecules. On the other hand, the larger concentration of oxygen vacancy also increases the electron density of oxygen atoms (Bader charge analysis, Supplementary Table 1), suggesting the higher ability of oxygen to donate electrons. Bearing the above discussion in mind, it is possible to realize all-solid FLP catalysts of CeO2 by regulating the surface defects.

Similar to the CeO2(110) surface, the solid FLPs can also be constructed by removing surface oxygen atoms of the CeO2(100) surface. The surface oxygen atoms locate in the rst atomic layer and surface cerium atoms sit in the second atomic layer on the ideal CeO2(100) surface (Supplementary Figs 1a and 14b). The

CeI-OIa pair forms a classic Lewis acidbase adjunct with a distance of only 2.19 . Although the CeI-OIc pair has a distance of 4.41 , it fails to be a FLP as the blocking of OIb next to CeI (Supplementary Fig. 1c). Removal of OIb can construct a FLP site, (CeI,CeII)-OIc with a distance of 3.95 (Supplementary Fig. 1b,d), similar to the solid FLP catalyst on the reduced CeO2(110) surface. However, the construction of surface FLP site cannot be realized by removing surface oxygen atoms on the CeO2(111) facet. Similar to the ideal

CeO2(110) surface, two types of Lewis acid-base pairs, that is, CeI-OIc and CeII-OIIc, are classic Lewis acid-base pairs on the ideal CeO2(111) surface. If one O atom (OIc in Supplementary Fig. 2a) is removed (Supplementary Fig. 2b), the OIIc atom is still surrounded by CeII and CeIII without the formation of FLP (Supplementary Fig. 2e). Similarly, when two O atoms (OIb and OId in Supplementary Fig. 2a) are removed (Supplementary Fig. 2c), OIa and

CeI also cannot form FLPs owing to the blocking by CeII* and CeIII* (Supplementary Fig. 2f). Therefore, the CeO2 nanostructures with the mainly exposed (110) and (100) facets may possess high catalytic hydrogenation activity rising from the formation of the solid FLPs.

Catalytic hydrogenation activity of PN-CeO2. To explore the possibility of metal oxides as all-solid FLP catalysts, PN-CeO2 with mainly exposed (110) and (100) facets, is selected as the model catalysts and hydrogenation of styrene is selected as a feature reaction. PN-CeO2 has been previously reported with a high surface Ce3 fraction (30.8%) and a large concentration of oxygen vacancy5356. Besides, the surface properties of PN-CeO2 can be effectively regulated by post-treatment with well preserved structural features, providing a platform to investigate the correlation between the hydrogenation activity of ceria and their surface properties.

PN-CeO2 was synthesized through a two-step hydrothermal synthetic approach, in which nonporous nanorods (Supplementary Fig. 3) with the mixed phases of Ce(OH)3 and CeO2 was obtained in the rst-step hydrothermal process at 100 C and porous nanorods with cubic uorite structure were synthesized by dehydration/oxidation of precursor nanorods at the second hydrothermal treatment at 160 C. Both bright-eld and dark-eld transmission electron microscopic (TEM) studies reveal a porous rod-like morphology with a dimension ofB8 60 nm (Fig. 2a,b). X-ray diffraction (XRD) analysis

conrms the precursor nanorods undergo the dehydration of the Ce(OH)3 content being transformed into CeO2 at 160 C (Supplementary Figs 3b and 4). High-resolution TEM (inset Fig. 2a) image of PN-CeO2 exhibits two kinds of lattice fringes of (220) and (200), which have respective inter-planar spacings of0.191 and 0.275 nm. Similar to previous reports, the nanorods preferentially grow along [110] direction and are enclosed by (220) and (200) planes57. Derived from dark-eld TEM images, the average pore size is 2.940.78 nm, which is similar as the average value of 2.2 nm obtained from the gas adsorption isotherms (Supplementary Fig. 5a,c,d). Surface area of PN-CeO2 determined by BrunauerEmmettTeller analysis is 122 m2 g 1 (Supplementary Fig. 5b).

Initially, optimization of the reaction conditions for hydrogenation reaction catalysed by PN-CeO2 is shown in

Supplementary Table 2. Despite the catalytic activity of PN-CeO2 had been observed at a low temperature of 60 C and low H2 pressure of 0.2 MPa, an unsatised styrene conversion of34.9% was delivered after 24 h reaction. Raising reaction temperature (100 C) and H2 pressure (1.0 MPa) realized the complete hydrogenation of styrene in 14 h successfully. Figure 2c shows time course of hydrogenation of styrene catalysed by

NATURE COMMUNICATIONS | 8:15266 | DOI: 10.1038/ncomms15266 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15266

a b

(220) 0.191 nm

(200)

0.275 nm

2 nm

5 nm

[110]

50 nm

30 nm

c d

100

Conversion (%)

80

60

40

20

0

Conversion (%)

100

80

60

40

20

0

0 3

Time (h)

6 9 12 1 2

Cycles

15

3 4 5

Figure 2 | Structural characterization and catalytic performance of PN-CeO2 catalysts for hydrogenation of styrene. (a) Bright-eld TEM image of PN-CeO2. Inset is the high resolution TEM image of PN-CeO2. (b) Dark-eld TEM image of PN-CeO2. High resolution TEM image of inset displays a pore size of 2.940.78 nm. (c) Time course of styrene conversion catalysed by PN-CeO2. (d) Recyclability of PN-CeO2 catalysts for hydrogenation of styrene. Reaction conditions: styrene (1.0 mmol), toluene (0.5 ml) and PN-CeO2 (20.0 mg) at 100 C and 1.0 MPa H2 pressure.

PN-CeO2. The reaction proceeded continuously to reach 98.2% conversion of styrene in 14 h and afford 499.9% chemoselectivity towards ethylbenzene.

After catalytic hydrogenation reaction, PN-CeO2 were recovered by centrifugation and washed with ethanol for three times. Before reuse, the spent PN-CeO2 catalysts were treated at 220 C for 4 h under the argon protection. Afterwards, PN-CeO2 can achieve ve-time recyclability without any obvious catalytic degradation (Fig. 2d). The unaltered morphology, unchanged phase and almost no changed surface Ce3 fraction of the spent PN-CeO2 illustrate their structural and catalytic stability during hydrogenation reactions (Supplementary Fig. 6).

Catalytic screening of various metal oxides. Derived from the previous studies on the semi-hydrogenation of alkynes by ceria47, PN-CeO2 with a high concentration of surface defects should inactive or even inert for hydrogenation of styrene. To understand the high catalytic activity of PN-CeO2 in the current work, the catalytic activity of various metal oxides including acidic oxides, basic oxides and transition metal oxides for hydrogenation of styrene were also investigated under 100 C and 1.0 MPa for 14 h. As presented in Fig. 3a, the typical acidic oxides (SiO2, Al2O3 and SnO2) delivered a very poor catalytic activity with styrene conversions below 1.4%. Poor catalytic activity of basic oxides (CaO, MgO and ZnO) was also observed. Meanwhile, the styrene conversions obtained from other common transition metal oxides, CuO, Fe2O3, NiO, MoO3 and

Co3O4, were below 0.8%, indicating their inertness for the hydrogenation of alkenes under the operated conditions.

Control experiments (Fig. 3a) indicated that the hydrogenation reaction of styrene could not happen in the absence of PN-CeO2,

indicating the nature catalytic ability of PN-CeO2. The major difference between PN-CeO2 and other metal oxides is the

coexistence of Lewis acidic and basic sites in high concentrations for PN-CeO2. Therefore, the survey of metal oxides for hydrogenation of styrene indicates that the absence of either acidic or basic sites on the surface of metal oxides results in near complete deactivation for the hydrogenation of alkenes.

To further conrm the necessity of the co-existence of Lewis acidic and basic sites of PN-CeO2 in current catalytic system, hydrogenation of styrene was performed in the presence of other molecular Lewis acid or base during the catalytic process. The Lewis base pyridine and Lewis acid pyrrole as common molecules are generally used to conrm the presence of Lewis acidic and basic sites for metal oxides, respectively. As a strong base pyridine (strong acid pyrrole) can be able to absorb on the Lewis acidic (Lewis basic) sites of PN-CeO2 immediately, when they are added to the reaction solution. Therefore, the FLPs can be effectively blocked by the pyridine and pyrrole molecule. As shown in Fig. 3b, addition of trace amount of either Lewis base pyridine or Lewis acid pyrrole completely terminated the hydrogenation reactions due to the blockage of surface Lewis acidic or basic sites by those small molecules, respectively, similar to that of molecular FLPs58,59. Trace amount of pyridine/pyrrole molecules were adsorbed on the surface Lewis acidic/basic sites of CeO2, leading to a low possibility for the formation of surface FLP sites in the aspect of population of adjacent surface Ce3 . The blockage of either Lewis basic sites or Lewis acidic sites of PN-CeO2 leads to the deactivated hydrogenation of alkenes, revealing that the coexistence of Lewis acidic and basic sites on the surface of PN-CeO2 are essential to activate HH bond and achieve high activity in the current system.

Role of surface defects of PN-CeO2 for hydrogenation. For the semi-hydrogenation of alkynes by ceria, hydrogen dissociative adsorption on surface oxygen leads to the formation of two

4 NATURE COMMUNICATIONS | 8:15266 | DOI: 10.1038/ncomms15266 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15266 ARTICLE

a

100 98.2

80

Conversion (%) Conversion (%) Conversion (%)

60

40

20

0

0.8

0.6 0.5 0.21 0.56 0.45 5 3.1 0.9 0.6 1.4 0.06

0 PN-CeO 2

CuO

Fe 2O 3

NiO

MoO 3

Co 3O 4

CaO

MgO

ZnO

SnO 2

Al 2O 3

SiO 2

No catalysts

b c

100

98.2

30

30.8

Lewis-base sites

H

Conversion (%) Ce 3+ fraction (%)

80

Lewis-acid sites

25

60

20

N N

15.7 16.7

9.3

15

14.5

9.2

40

10

20

5

0.5

0

5101

Concentration (mol l1 per gPN-CeO2)

5103 5101 5103

PN-CeO 2

0 NR-CeO 2

NC-CeO 2

NP-CeO 2NP-CeO 2-300

NP-CeO 2-500

d e

100

80

NC-CeO2 NP-CeO2

PN-CeO2-500 PN-CeO2-500

R2=0.998

100 PN-CeO2

80

PN-CeO2-300

60

NR-CeO2 PN-CeO2-300

60

PN-CeO2

1.3 0.02 0.3

40

40

20

20

1.8 1.3 2.7

34.1 39.4

98.2

0

0 0.016 0.06 0.085 0.121 0.137 0.295 50 60 70Adsorption quantity of CO2 (mol g1)

80 90

Total surface Ce3+ atom (mmol g1)

Figure 3 | Mechanism investigations on hydrogenation activity of PN-CeO2. (a) Catalytic activity of various metal oxides for hydrogenation of styrene. (b) Inuences of molecular Lewis-base or Lewis-acid on the catalytic activity of PN-CeO2 for hydrogenation of styrene. (c) Catalytic activity of nanoceria for hydrogenation of styrene. (d) Effects of the total surface Ce3 atom number for hydrogenation of styrene. (e) Correlation between conversion of styrene and adsorption quantity of CO2 on CeO2 catalyst. Reaction conditions: styrene (1.0 mmol), toluene (0.5 ml) and catalysts (20.0 mg) at 100 C and 1.0 MPa H2 pressure for 14 h.

surface OH groups and dissociation of CH bond of R C C H on the adjacent surface Ce and O atoms49. In those

studies, three points were outlined: (1) the obtained alkenes cannot be further reduced into alkanes; (2) experimental results suggest that the hydrogen activation process is the rate-limiting step due to the poor ability of ceria to cleave H H bond; (3) the

catalytic activity of ceria can be signicantly decreased in the presence of large amount of surface defects. Followed by the literature reports, PN-CeO2 with a high surface concentration of oxygen vacancy should deliver a poor catalytic activity for hydrogenation reactions. In fact, PN-CeO2 indeed shows a very high catalytic activity for styrene hydrogenation, suggesting a different catalytic mechanism for the current work. Besides, both alkenes and alkynes can be reduced by PN-CeO2 under much mild conditions (see Discussion below), also revealing an alternative catalytic pathway for hydrogen activation and subsequent hydrogenation.

To explore the roles of surface defects of ceria, hydrogenation of styrene was also performed for ceria nanoparticles (NP-CeO2)

with a surface Ce3 fraction of 9.3% (Supplementary Fig. 7), ceria nanocubes (NC-CeO2) with a surface Ce3 fraction of16.7% (Supplementary Fig. 8) and non-porous ceria nanorods

(NR-CeO2) with a surface Ce3 fraction of 15.7% (Supplementary Fig. 9). The conversions of styrene were only 1.3%, 1.8%

and 34.1% for NP-CeO2, NC-CeO2 and NR-CeO2 under the identical reaction conditions of those catalysed by PN-CeO2 for 14 h, respectively (Fig. 3d). The surface area of 98 m2 g 1 for

NR-CeO2 is close to that of PN-CeO2 of 122 m2 g 1 (Supplementary Table 3), indicating surface area is not the critical factor to determine their difference in their catalytic activity for hydrogenation of alkenes. The high consistence between the surface Ce3 fractions and the catalytic activity of ceria reveals that the hydrogenation activity of ceria is determined by the abundance of the surface defect sites.

As predicted from the theoretical calculations, it is also possible to create the surface FLP sites on CeO2(100) surface. However, the catalytic activity of NC-CeO2 was much lower than that of PN-CeO2. Considering the small surface area of NC-CeO2, we also performed the hydrogenation of styrene by 200 mg NC-CeO2 with the comparable surface area to that of 20 mg PN-CeO2 under the identical conditions. A very low styrene conversion of5.3% was observed, which could be attributed to the low concentration of surface defects of NC-CeO2 with a Ce3 fraction of 16.7%. The results further conrm that the

NATURE COMMUNICATIONS | 8:15266 | DOI: 10.1038/ncomms15266 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15266

concentration of surface defects is the main factor to construct large amount of surface FLP sites without regard to the morphology of CeO2 catalysts.

Hydrogenation performance of NR-CeO2 (20 mg) and NC-CeO2 (200 mg) catalysts suggested that the catalytic activity of FLP sites on CeO2(110) was higher than that of CeO2(100) with similar surface Ce3 fractions (Fig. 3c,d). It can be attributed to the different steric conguration of FLP sites on

CeO2(110) and CeO2(100). Both the acidic and basic sites of the constructed FLP sites on CeO2(110) are in the top atomic layer (Fig. 1c,f and Supplementary Fig. 14a), whereas the basic sites (O atoms) on CeO2(100) are in the top atomic layer with the second atomic layer of the acidic sites (Ce atoms) (Supplementary Figs 1d and 14b). Both two congurations can effectively activate hydrogen molecule. However, styrene can encounter less steric hindrance when reacting with the intermediate of heterolysis of H2 at the FLP sites on CeO2(110).

Besides, due to the high surface energy, oxygen atoms of

CeO2(100) is more prone to diffuse on surface than that of

CeO2(110)60. This diffuse of oxygen atoms (basic sites) can break the steric encumbered structures of FLP sites, leading to the decrease in activity of FLP sites or even the formation of classic acid-base adduct. Thus, the FLP sites on CeO2(100) are less stable than those on CeO2(110), indicating that the stability of FLP sites could be another factor in the decrease of activity on CeO2(100) surface.

The surface Ce3 fractions of PN-CeO2 were effectively reduced from 30.8% of as-synthesized catalysts to 14.5 and 9.2% (Fig. 3c and Supplementary Fig. 10), when PN-CeO2 was treated at 300 C (obtained PN-CeO2-300 catalysts) and 500 C (obtained PN-CeO2-500 catalysts) in air for 8 h, respectively. The calcinated PN-CeO2 preserved the porous morphology (Supplementary Fig. 11), as well as their surface areas (Supplementary Table 3). After calcination, the obtained PN-CeO2 catalysts were immediately used for hydrogenation of styrene under the same conditions. The 39.4% and 2.7% conversions of styrene were yielded with PN-CeO2-300 and

PN-CeO2-500, respectively. The decreased catalytic activity can only be attributed to their decreased concentration of surface oxygen vacancy due to the high temperature annealing at 300 and 500 C in air.

On the other hand, the PN-CeO2 was reduced under 10% H2/Ar at 200 C for 2 h and then transferred from the reducing atmosphere to the hydrogenation system immediately. The exposure of the reduced PN-CeO2 to air was controlled 5 min to minimize the re-oxidation of the reduced catalysts. The catalytic activity for reduced PN-CeO2 also enhanced from 69.1%

to 86.4% after 8 h reaction (Supplementary Fig. 12). The enhanced catalytic activity can be attributed to the increased concentration of surface oxygen vacancy, leading to an increased possibility to construct more surface FLP sites. Therefore, the parallel experiments further strengthen the evidence linking the concentration of surface oxygen vacancy of PN-CeO2 with their catalytic activity for hydrogenation of alkenes.

The decreased hydrogenation activity of PN-CeO2-300 and PN-CeO2-500 can be further revealed from the temperature-programmed hydrogen desorption (H2-TPD) tests. All PN-CeO2 catalysts were exposed to H2 at 50 C and the weakly physisorbed hydrogen was removed by ushing Ar for 20 min. A desorption peak at 190 C was observed for all PN-CeO2 catalysts (Supplementary Fig. 13), indicating a strong interaction between H2 and PN-CeO2. With the increase of calcination temperature, the decreased peak intensities demonstrated the reduced amount of activated hydrogen on PN-CeO2-300 and PN-CeO2-500.

Therefore, the decreased catalytic activity of the calcined PN-CeO2 was delivered for styrene hydrogenation.

Mechanism investigations. All control experiments strongly suggest that the coexistence of Lewis acidic and basic sites plays extremely important roles in activating hydrogen at mild conditions and realising their high catalytic activity for hydrogenation reactions. The results are totally different to the catalytic mechanism of the semi-hydrogenation of alkynes by ceria4749. In Fig. 1, we have illustrated that it is possible to create FLP active sites through regulations on the surface defects of CeO2(110). In general, the key features of organic molecules FLP catalysts are dened as the independent Lewis acid and base in the presence of sterically substituents on Lewis acid and base to preclude the formation of the classical Lewis acidbase adducts30,61. Addition of any other small Lewis acid or base molecules results in the formation of adducts and completely blocks the catalytic activity of FLP catalysts during the catalytic process9,12,28. On the basis of the known essentiality of molecular FLP catalysts and a strong correlation between the hydrogenation activity and surface properties of PN-CeO2 with coexistence of Lewis acid and base in current system, its logical to rationalize the PN-CeO2 as all-solid FLP catalysts. The analogous aspects of molecular FLP catalysts and PN-CeO2 are listed: (1) independent surface Lewis acidic and basic sites of PN-CeO2 (Fig. 1c); (2) close proximity (4B5 ) between surface Lewis acidic site and basic site for activation or splitting of small molecules; and(3) completely quenched hydrogenation activity with addition of other Lewis acids or bases.

To support the aforementioned conclusions, DFT calculations using the Vienna Ab-initio Simulation Package6264 were performed to investigate the adsorption and activation of hydrogen on the CeO2(110) surface. For ideal CeO2(110)

(Supplementary Fig. 14a), hydrogen molecule prefers to be adsorbed at the top site of CeI (Fig. 4a) or classic acidbase site (Fig. 4b) with weak adsorption energies of 0.06 eV, in

consistence with previous studies65,66. The HH bond length of the weakly adsorbed H2 is 0.75 and 0.76 for the two adsorption congurations, respectively. These values are close to that of gas phase H2 (0.75 ), suggesting that H2 is difcult to be activated on ideal CeO2(110). The Bader charge populations of the physisorbed H2 (Fig. 4a,b) further indicate that H2 is slightly polarized and hardly activated on ideal CeO2(110). With one oxygen vacancy constructed on CeO2(110), the adsorption and activation of H2 is still poor on both CeI-OIc (Fig. 4c, 0.06 eV)

and CeI-OIIc sites (Fig. 4d, 0.04 eV), which is further evidenced

by the long distance (B3.0 ) between H2 and CeO2(110) surface and the similar H H bond length of adsorbed H2 to that of

gas-phase H2. For the surface with two adjacent oxygen vacancies, H2 preferentially adsorbs at the created solid FLP site (CeI,CeII)-OIIc (Fig. 4f) rather than on the top site of the unsaturated Ce atom (Fig. 4e). The results reveal the easy activation of H2 molecule at the site of (CeI,CeII)-OIIc:(1) a relatively larger adsorption energy of H2 of 0.13 eV

(Fig. 4f). (2) A signicantly elongated H H bond length of

0.80 . (3) A shorter distance between H2 and the CeO2(110) than those in other adsorption congurations (Fig. 4a-e). The distance between one hydrogen atom of H2 and the base site of solid FLPs is only 1.89 . (4) A larger polarization of H H bond

due to the electrostatic interactions between the FLPs and H2, as indicated in the Bader charge of 0.16 e and 0.11 e for two

hydrogen atoms (Fig. 4f). The aforementioned aspects strongly suggest that the novel solid FLP site constructed by two Ce cations and one O anion is more effective in H2 adsorption and activation than the classic acidbase adjuncts.

To further conrm the high activity of such a constructed solid FLP site towards the activation of small molecules, the dissociative activation of H2 is investigated on the ideal

CeO2(110) and reduced CeO2(110) surface in the presence of

6 NATURE COMMUNICATIONS | 8:15266 | DOI: 10.1038/ncomms15266 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15266 ARTICLE

a c

e

Cel

Cel

0.75

Ola

Ola

Olc

0.75

0.04

0.03

+0.03

+0.03 +0.06

+0.11

0.76

0.07

0.16

Eads = 0.06 eV

Eads = 0.06 eV Eads = 0.04 eV

Reaction coordinate

Eads = 0.06 eV Eads = 0.06 eV

b

d

f

Cel

Cel Cel

Cel

Cell

Ollc

0.76

Olc Ollc

0.03 0.03

0.801.89

+0.02 +0.02

0.75

g

TS1

0.59 FS10.38

FS2 0.84

0.00

IS1 0.06

TS20.04 0.13

IS2

Eads = 0.13 eV

0

Energy (eV)

1

2

3

IS1 TS1

IS2 TS2

FS1

FS2

2.57

FS1

3.15

1.12

1.12

2.33

0.99 2.15

0.99

0.76

FS2

0.80

1.89 2.77

2.79 0.90

1.46

2.99

3.00

2.51 0.99

2.51

0.97

1.77

FS1 2.64

FS2

1.00

Figure 4 | The H2 adsorption and dissociation pathways on the surface of PN-CeO2. The optimized structures of H2 adsorption on (a) CeI atom of ideal CeO2(110); (b) CeI-OIa pair of ideal CeO2(110); (c) CeI-OIc pair of

CeO2(110) with one oxygen vacancy; (d) CeI-OIIc pair of CeO2(110) with one oxygen vacancy; (e) CeI atom of CeO2(110) with two oxygen vacancies;

and (f) (CeI,CeII)-OIIc pair of CeO2(110) with two oxygen vacancies in both top (inset) and side views. The adsorption energies (Eads), the bond

distances (in ) colored in black, and the Bader charge population (in e) colored in blue are also shown. (g) Energy proles for H2 dissociation on ideal CeO2(110) in black curve and CeO2(110) with two oxygen vacancies in red curves. The optimized structures of initial states (IS), transition states (TS) and nal states (FS) are labeled with bond distance (in ). The zero energy reference corresponds for the sum energy of H2 in the gas phase and the corresponding clean CeO2 surfaces.

(CeI,CeII)-OIIc. From the energy prole depicted by Fig. 4g (black curve), the dissociation of H2 on ideal CeO2(110) starts from a physisorbed state (IS1) and surpasses an activation barrier of 0.65 eV to reach a dissociated state (FS1). In FS1, the dissociated hydrogen bound to Ce site is meta-stable and readily diffuses to nearby surface oxygen, leading to the formation of a more stable FS1. The H2 dissociation on CeO2(110) with the solid FLP active sites experiences a similar process, which also starts from a physisorbed state. Noticeably, the dissociation barrier is only 0.17 eV on the constructed FLP site, which is0.48 eV lower than that on ideal CeO2(110). Moreover, the dissociation of H2 on solid FLPs (from IS2 to FS2) is exothermic by 0.71 eV in contrast to the endothermic energy (0.44 eV, from IS1 to FS1) on ideal CeO2(110). Both kinetic and thermodynamic results suggest that the dissociative activation of H2 is more favourable on the reduced CeO2(110) with active FLP sites. The dissociated hydrogen anchored at the Ce site (FS2) can immigrate to an adjacent oxygen to form FS2 with an exothermic energy of 0.93 eV. It is worth noting that FS2 is less stable than FS1, despite two hydrogen atoms are both adsorbed at surface oxygen in two congurations. The calculations suggest

that the OH bond in FS2 is weaker than that in FS1. The weakness of OH bond is attributed to the enhanced basicity of oxygen on the reduced CeO2(110), which is detrimental to the adsorption of H according to the rules of Lewis acid-base pairs on oxide surfaces67. Fortunately, the less stable OH bonds are expected to benet the subsequent hydrogenation reactions.

Derived from DFT calculations, CeO2 nanorods with a high concentration of surface oxygen defects deliver a high possibility to create surface (CeI,CeII)-OIIc FLP sites, yield an enhanced capability of dissociation of H2 and improve their catalytic activity for hydrogenation of styrene. The calculation results (Fig. 4) are highly consistent with experiments (Fig. 3), supporting the solid FLP-like activity of PN-CeO2 regulated and constructed by the surface oxygen defects as illustrated in Fig. 1. To further conrm this conclusion, the correlation between catalytic activity and the concentration of the surface oxygen vacancy of CeO2 were studied through the TPD of CO2 on the various PN-CeO2-T catalysts. The annealed PN-CeO2 at 300 and 500 C exhibited the identical morphological features to those of as-synthesized PN-CeO2 (Supplementary Fig. 11).

Importantly, the PN-CeO2-300 (120 m2 g 1) and PN-CeO2-500 (117 m2 g 1) showed similar surface areas to that of as-synthesized PN-CeO2 (122 m2 g 1). Thus, such a system provides a straightforward way to understand the correlations between the acidity/basicity of PN-CeO2 and their catalytic activity.

In previous reports, CO2 as a Lewis acid can bind with O2 surface ions forming carbonate species with various anchoring structures (bridged, bidentate, monodentate and polydentate)59. Furthermore, the presence of oxygen vacancy on CeO2 surface can make a strong interaction with CO2 and thus enhance its surface basicity58. The surface adsorption quantity of CO2 is the highest on PN-CeO2 (90.7 mmol g 1), followed by PN-CeO2-300(68.7 mmol g 1), and the least on PN-CeO2-500 (55.2 mmol g 1) from the CO2-TPD results (Supplementary Fig. 15 and Supplementary Table 4). The reduced surface basic sites mainly derived from the obviously decreased medium and strong basic sites, which indicated the decreased amount of surface oxygen vacancy. Meanwhile, the conversions of styrene showed a near linear decrease with the decreased adsorption quantity of CO2 on various PN-CeO2 catalysts (Fig. 3e, R2 0.998). Results

further demonstrate the importance of the surface abundance of defects in CeO2 catalysts to construct the surface FLP sites once more.

The correlation between the concentration of oxygen vacancy and their catalytic activity is further investigated with Raman spectra analysis. The concentration of oxygen vacancy is indexed by the ratio of the integrated peak area under the bands at 460 cm 1 (vibrational mode of CeO2 uorite structure) and 600 cm 1 (oxygen vacancy) of the Raman spectra of PN-CeO2 annealed at various temperatures (Supplementary Fig. 16)28. The decrease in the A600/A460 with the increased calcination

temperature indicates a decrease in the number of oxygen vacancy, as well as the surface Ce3 fraction. Thus, high temperature treatment on PN-CeO2 leads to the decrease in the number of FLPs sites. Therefore, it is not surprising to observe an obviously decreased conversion of styrene with PN-CeO2-300 and

PN-CeO2-500; even they have almost the same surface areas (Supplementary Table 3).

The conversion of styrene as a function of total surface Ce3 species (Fig. 3d) and the concentration of oxygen vacancy (Fig. 3e and Supplementary Fig. 16b) indicate that the abundance of surface defects is important for the construction of FLP active sites on PN-CeO2. The high concentration of surface defects introduces a high possibility to form the surface FLP sites as shown in Fig. 1c. Both theoretical calculations and experimental

NATURE COMMUNICATIONS | 8:15266 | DOI: 10.1038/ncomms15266 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15266

Table 1 | Scope of PN-CeO2 for catalytic hydrogenation of alkenes*.

Entry Substrate Product Pressure

(MPa)

Time(h)

Selectivity (%)

1 0.6 12 99.1 >99.9

2 0.6 12 98.1 >99.9

3 O O 1.0 14 99 >99.9

4 Br Br 1.0 20 98.5 >99.9

5 Cl Cl 1.0 20 79.5 >99.9

6 1.0 20 90.5 >99.9

7 1.0 20 56.3 >99.9

8 1.0 20 10.3 >99.9

9 1.0 20 8.9 >99.9

10 2.0 20 21.7 >99.9

O O

11 2.0 20 12.8 >99.9

Conversion (%)

O O

12 2.0 20 5.3 >99.9

HO HO

13 2.0 20 <1 >99.9

HO HO

GC, gas chromatography; MS, mass spectrometry.*Reaction condition: PN-CeO (20 mg), toluene (0.5 ml), substrate (1 mmol) at 100 C under various H pressures. The substrate conversion and product selectivity were determined by GC-MS and GC with m-xylene as the internal standard.

results illustrate that the FLP active sites in PN-CeO2 are created by regulations on their surface properties.

Hydrogenation scope of alkenes. The scope of hydrogenation reactions catalysed by PN-CeO2 was demonstrated in Table 1.

When the electron-donating groups of methyl (CH3), tertiary butyl (C(CH3)3) and methoxy (OCH3) were introduced in styrene, the reactant conversions of 99.1, 98.1 and 99% were yielded under 0.6, 0.6 and 1.0 MPa H2 pressure, respectively (Entry 1, 2 and 3). However, the catalytic hydrogenation activity of PN-CeO2 was obviously decreased in the presence of electron-drawing groups in the substrates. The introduction of Br and Cl afforded the conversions of 98.5% and 79.5% under 1.0 MPa H2 pressure for 20 h, respectively (Entry 4 and 5). It can be attributed to the decreased electronic density of vinyl group in the presence of electron-drawing groups in benzene ring. The hydrogenation of cyclooctene into cyclooctane reached a conversion of 90.5% under 1.0 MPa H2 pressure for 20 h (Entry 6). When the H at a or b site of vinyl is replace by other groups, the catalytic activity of PN-CeO2 for hydrogenation of vinyl is also decreased due to steric hindrance. Compared to56.3% conversion of a-methylstyrene under 1.0 MPa H2 for 20 h (Entry 7), trans-stibene and cis-stilbene only reached 10.3% and8.9% conversions under the same reaction conditions, respectively (Entry 8 and 9).

In light of both experimental and theoretical investigations, the introduction of oxygen vacancy is very important in the formation of FLP active sites for PN-CeO2. When alkenes include oxygen-containing groups, the blockage of the surface FLP active sites on PN-CeO2 may happen due to the strong adsorption of oxygen at the surface defects. Thus, the catalytic activity of PN-CeO2 for those molecules can be reduced or completely depressed. As shown in Table 1 (Entry 10 and 11), trans-cinnamaldehyde and allylacetone in the presence of formyl

or ketone groups only yielded very low reactant conversions of21.7% and 12.8%, respectively, despite a high chemoselectivity for olene products (499.9%). Owing to the strong interaction between OH or NO2 groups and surface defects of PN-CeO2, the activity of FLPs was shielded with 4-hydroxystyrene and 4-nitrostyrene, affording very low conversions (o6%, Entry 12 and 13). There results further support the solid FLP catalytic behaviour of PN-CeO2 for hydrogenation reactions.

Hydrogenation scope of alkynes. Ceria has shown the ability to selectively semi-hydrogenated alkynes into olens49. Herein, hydrogenation of various alkynyl substrates was also evaluated with PN-CeO2 (Table 2). For hydrogenation of phenylacetylene, the conversion was increased from 55.8 to 100% with the increase of H2 pressure from 1.0 to 3.0 MPa within 14 h (Entries 14 and 15). However, the selectivity towards styrene was decreased from 95.8 to 32.6% (Entries 14 and 15). By contrast, the selectivity of ethylbenzene was increased from 4.2 to67.4% (Entry 14 and 15). Only ethylbenzene was obtained with the extended reaction time (Entry 16). The presence of electron-donating methyl group in phenylacetylene yielded 100% conversion under 3.0 MPa H2 pressure but a poor chemoselectivity of 75.2% into 4-methylstyrene (Table 2, Entry17). For 4-methoxy phenylacetylene, similar catalytic behaviour was observed with a 84.7% selectivity of olen for the conversion of 100% under 3.0 MPa H2 pressure at 100 C (Entry 18).

Introducing the electron-drawing substituents on phenylacetylene such as Br and Cl, the catalytic activity was reduced, as the low conversions of 50% and 13.7% were yielded for 4-bromo phenylacetylene and 4-chloro phenylacetylene under a H2 pressure of 3.0 MPa and a temperature of 100 C for 14 h, respectively (Entries 19 and 20). For 4-phenyl-1-butyne, 38.9% conversion and 100% selectivity towards olen products were achieved under 3.0 MPa H2 pressure for 14 h reaction (Entry 21).

8 NATURE COMMUNICATIONS | 8:15266 | DOI: 10.1038/ncomms15266 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15266 ARTICLE

Table 2 | Scope of PN-CeO2 for catalytic hydrogenation of alkynes*.

14 1.0 14 55.8

Entry Substrate Pressure

(MPa)

Time(h)

Conversion (%)

Product

Selectivity (%)

95.8

4.2

15 3.0 14 100

32.6

67.4

16 3.0 24 100

-

>99.9

17 3.0 14 100

75.2

24.8

18

O

O

O

87.4

15.3

19 Br 100

Cl

20 Cl 100

21 100

22

3.0 14 100

3.0 14 50

3.0 14 13.7

3.0 14 38.9

3.0 20 13.8

Br

51.3

18.7

GC, gas chromatography; MS, mass spectrometry.*Reaction condition: PN-CeO (20 mg), toluene (0.5 ml), substrate (1 mmol) at 100 C under various H pressures. The substrate conversion and product selectivity were determined by GC-MS and GC with m-xylene as the internal standard.

The catalytic activity of PN-CeO2 was also decreased when the alkynes in the presence of the steric substituents. Under 3.0 MPa H2 pressure, hydrogenation of diphenylacetylene only afforded a 13.8% conversion after 20 h reaction (Entry 22). The products of hydrogenation were a mixture of trans-stilbene and cis-stilbene with the chemoselectivity of 51.3% and 48.7%, respectively (Entry 23).

In summary, we design all-solid metal oxide FLP catalysts through regulations on the surface properties of ceria. As demonstrated by both experimental and theoretical results, PN-CeO2 with a high concentration of surface defects exhibits the

FLP catalytic behaviour for the efcient hydrogenation of alkenes and alkynes with a wide scope. The high density of surface-bound defects is critical to generate the independent surface Lewis acidic and basic sites, construct the surface FLP pair of (CeI,CeII)-OIIc (Fig. 1c) and improve their capability for H2 activation with a low activation barrier of 0.17 eV. The capability to hydrogenate alkynes into alkanes further conrms the high hydrogenation activity of PN-CeO2. The current progress provides a new concept on the structurally correlated catalytic activity of ceria for hydrogenation reactions, which is important for the modern catalyst design. With the ease of preparation, such PN-CeO2 as

the active components in various hydrogenation reactions may be envisioned and so are their ultimate industrial applications.

Methods

Preparation of PN-CeO2 catalysts. Aqueous solutions of Ce(NO3)3 6H2O

(1.736 g in 10 ml of deionized water)53 and NaOH (19.2 g in 70 ml of deionized water) were mixed slowly. With continuous stirring, the mixture was aged for30 min before the reaction continued at 100 C for 24 h in a temperature-controlled electric oven. The reaction mixture was then cooled naturally to room temperature.

The products were collected by centrifugation, thoroughly washed with deionized water, and then air-dried at 60 C. Hydrothermal treatment of this precursor product at 160 C for 12 h afforded the formation of PN-CeO2.

Characterizations. The catalysts were characterized by powder XRD. The XRD patterns with diffraction intensity versus 2y were recorded in a Shimadzu X-ray diffractometer (Model 6000) using Cu Ka radiation. TEM studies were conducted on the Hitachi HT-7700 with an accelerating voltage of 120 kV. High-resolution and dark-eld TEM images were acquired from the Tecnai G2 F20 S-twin TEM at 200 kV. Surface area was measured by N2 physisorption (Micromeritics, ASAP 2020 HD88) based on BrunauerEmmetTeller method. X-ray photoelectron spectroscopy were acquired using a Thermo Electron Model K-Alpha with Al Ka as the excitation source.

Catalytic hydrogenation reactions. For a typical catalytic reaction, 1.0 mmol of substrate and 20.0 mg of catalysts were mixed in 0.5 ml of toluene. The reactions were performed in the autoclave charged with hydrogen with various pressures at the desired reaction temperature. The products were analysed by gas chromatographymass spectrometry and gas chromatography with m-xylene as the internal standard.

Data availability. Data supporting the ndings of this study are available within the article (and its Supplementary Information les) and from the corresponding author on reasonable request.

References

1. Higman, C. & Tam, S. Advances in coal gasication, hydrogenation, and gas treating for the production of chemicals and fuels. Chem. Rev. 114, 16731708 (2014).

2. Johnson, N. B., Lennon, I. C., Moran, P. H. & Ramsden, J. A. Industrial-scale synthesis and applications of asymmetric hydrogenation catalysts. Acc. Chem. Res. 40, 12911299 (2007).

3. Vispute, T. P., Zhang, H., Sanna, A., Xiao, R. & Huber, G. W. Renewable chemical commodity feedstocks from integrated catalytic processing of pyrolysis oils. Science 330, 12221227 (2010).

NATURE COMMUNICATIONS | 8:15266 | DOI: 10.1038/ncomms15266 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15266

4. Wang, W., Wang, S., Ma, X. & Gong, J. Recent advances in catalytic hydrogenation of carbon dioxide. Chem. Soc. Rev. 40, 37033727 (2011).

5. Stephan, D. W. & Gerhard, E. Frustrated Lewis pair chemistry: development and perspectives. Angew. Chem. Int. Ed. Engl. 54, 64006441 (2015).

6. Stephan, D. W. Frustrated Lewis pairs: from concept to catalysis. Acc. Chem. Res. 48, 306316 (2015).

7. Stephan, D. W. Frustrated Lewis pairs. J. Am. Chem. Soc. 137, 1001810032 (2015).

8. Morozova, V., Mayer, P. & Berionni, G. Scope and mechanisms of frustrated Lewis pair catalyzed hydrogenation reactions of electron-decient C C

double bonds. Angew. Chem. Int. Ed. Engl. 54, 1450814512 (2015).9. Mo, Z. et al. Facile reversibility by design: tuning small molecule capture and activation by single component frustrated Lewis pairs. J. Am. Chem. Soc. 137, 1222712230 (2015).

10. Stephan, D. W. Frustrated Lewis pairs: a metal-free landmark. Nat. Chem. 6, 952953 (2014).

11. Pereira, J. C. et al. Reaction of a bridged frustrated Lewis pair with nitric oxide: a kinetics study. J. Am. Chem. Soc. 136, 513519 (2014).

12. Voicu, D. et al. Microuidic studies of CO2 sequestration by frustrated Lewis pairs. J. Am. Chem. Soc. 136, 38753880 (2014).

13. Borger, J. E., Ehlers, A. W., Lutz, M., Slootweg, J. C. & Lammertsma, K. Stabilization and transfer of the transient [Mes*P4 ]- buttery anion using BPh3.

Angew. Chem. Int. Ed. Engl. 56, 613617 (2015).14. Chen, J. & Chen, E. Y. Elusive silane-alane complex [Si-HAl]: isolation, characterization, and multifaceted frustrated Lewis pair type catalysis. Angew. Chem. Int. Ed. Engl. 54, 68426846 (2015).

15. Chernichenko, K. et al. Intramolecular frustrated Lewis pair with the smallest boryl site: reversible H2 addition and kinetic analysis. Angew. Chem. Int. Ed.

Engl. 54, 17491753 (2015).16. Hansmann, M. M. et al. Cyclopropanation/carboboration reactions of enynes with B(C6F5)3. J. Am. Chem. Soc. 137, 1546915477 (2015).

17. Lgar, M.-A., Courtemanche, M.-A., Rochette,. & Fontaine, F.-G. Metal-free catalytic C-H bond activation and borylation of heteroarenes. Science 349, 513516 (2015).

18. Eisenberger, P., Bestvater, B. P., Keske, E. C. & Crudden, C. M. Hydrogenations at room temperature and atmospheric pressure with mesoionic carbene-stabilized borenium catalysts. Angew. Chem. Int. Ed. Engl. 54, 24672471 (2015).

19. Oestreich, M., Hermeke, J. & Mohr, J. A unied survey of Si-H and H-H bond activation catalysed by electron-decient boranes. Chem. Soc. Rev. 44, 22022220 (2015).

20. Chernichenko, K. et al. A frustrated-Lewis-pair approach to catalytic reduction of alkynes to cis-alkenes. Nat. Chem. 5, 718723 (2013).

21. Mahdi, T. & Stephan, D. W. Frustrated Lewis pair catalyzed hydroamination of terminal alkynes. Angew. Chem. Int. Ed. Engl. 52, 1241812421 (2013).

22. Courtemanche, M. A., Legare, M. A., Maron, L. & Fontaine, F. G. Reducing CO2 to methanol using frustrated Lewis pairs: on the mechanism of phosphineborane-mediated hydroboration of CO2. J. Am. Chem. Soc. 136, 1070810717 (2014).

23. Kalz, K. F., Brinkmeier, A., Dechert, S., Mata, R. A. & Meyer, F. Functional model for the [Fe] hydrogenase inspired by the frustrated Lewis pair concept.J. Am. Chem. Soc. 136, 1662616634 (2014).24. Lawrence, E. J., Oganesyan, V. S., Hughes, D. L., Ashley, A. E. & Wildgoose, G.G. An electrochemical study of frustrated Lewis pairs: a metal-free route to hydrogen oxidation. J. Am. Chem. Soc. 136, 60316036 (2014).25. Liedtke, R. et al. Frustrated Lewis pair modication by 1,1-carboboration: disclosure of a phosphine oxide triggered nitrogen monoxide addition toan intramolecular P/B frustrated Lewis pair. J. Am. Chem. Soc. 136, 90149027 (2014).

26. Lu, G. et al. Gold catalyzed hydrogenations of small imines and nitriles: enhanced reactivity of Au surface toward H2 via collaboration with a Lewis base. Chem. Sci. 5, 10821090 (2014).

27. Mahdi, T. & Stephan, D. W. Facile protocol for catalytic frustrated Lewis pair hydrogenation and reductive deoxygenation of ketones and aldehydes. Angew. Chem. Int. Ed. Engl. 54, 85118514 (2015).

28. Primo, A., Neatu, F., Florea, M., Parvulescu, V. & Garcia, H. Graphenes in the absence of metals as carbocatalysts for selective acetylene hydrogenation and alkene hydrogenation. Nat. Commun. 5, 5291 (2014).

29. Stephan, D. W. & Erker, G. Frustrated Lewis pairs: metal-free hydrogen activation and more. Angew. Chem. Int. Ed. Engl. 49, 4676 (2010).

30. Eros, G. et al. Expanding the scope of metal-free catalytic hydrogenation through frustrated Lewis pair design. Angew. Chem. Int. Ed. Engl. 49, 65596563 (2010).

31. Wang, X., Kehr, G., Daniliuc, C. G. & Erker, G. Internal adduct formationof active intramolecular C4-bridged frustrated phosphane/borane Lewis pairs.

J. Am. Chem. Soc. 136, 32933303 (2014).32. Lee, H. et al. Formation of frustrated Lewis pairs in Ptx-loaded zeolite NaY. Angew. Chem. Int. Ed. Engl. 54, 1308013084 (2015).

33. Ghuman, K. K. et al. Illuminating CO2 reduction on frustrated Lewis pair surfaces: investigating the role of surface hydroxides and oxygen vacancies on nanocrystalline In2O3-x(OH)y. Phys. Chem. Chem. Phys. 17, 1462314635 (2015).

34. Ghuman, K. K. et al. Photoexcited surface frustrated Lewis pairs for heterogeneous photocatalytic CO2 reduction. J. Am. Chem. Soc. 138, 12061214 (2016).

35. Ghuman, K. K. et al. Surface analogues of molecular frustrated Lewis pairs in heterogeneous CO2 hydrogenation catalysis. ACS Catal. 6, 57645770 (2016).

36. He, L. et al. Spatial separation of charge carriers in In2O3-x(OH)y nanocrystal superstructures for enhanced gas-phase photocatalytic activity. ACS Nano 10, 55785586 (2016).

37. Ghoussoub, M., Yadav, S., Ghuman, K. K., Ozin, G. A. & Singh, C. V. Metadynamics-biased ab initio molecular dynamics study of heterogeneous CO2 reduction via surface frustrated Lewis pairs. ACS Catal. 6, 71097117 (2016).

38. Vivier, L. & Duprez, D. Ceria-based solid catalysts for organic chemistry. ChemSusChem 3, 654678 (2010).

39. Wang, X., Liu, D., Li, J., Zhen, J. & Zhang, H. Clean synthesis of Cu2O@CeO2 core@shell nanocubes with highly active interface. NPG Asia Mater. 7, e158 (2015).

40. Paier, J., Penschke, C. & Sauer, J. Oxygen defects and surface chemistry of Ceria: quantum chemical studies compared to experiment. Chem. Rev. 113, 39493985 (2013).

41. Corma, A., Atienzar, P., Garcia, H. & Chane-Ching, J. Y. Hierarchically mesostructured doped CeO2 with potential for solar-cell use. Nat. Mater. 3, 394397 (2004).

42. Juarez, R., Concepcion, P., Corma, A. & Garcia, H. Ceria nanoparticles as heterogeneous catalyst for CO2 xation by omega-aminoalcohols. Chem.

Commun. 46, 41814183 (2010).43. Leyva-Prez, A., Cmbita-Merchn, D., Cabrero-Antonino, J. R., Al-Resayes, S.I. & Corma, A. Oxyhalogenation of activated arenes with nanocrystalline ceria. ACS Catal. 3, 250258 (2013).44. Tamura, M. & Tomishige, K. Redox properties of CeO2 at low temperature: the direct synthesis of imines from alcohol and amine. Angew. Chem. Int. Ed. Engl. 54, 864867 (2015).

45. Schmidt, P. O. et al. Spectroscopy using quantum logic. Science 309, 749752 (2005).

46. Esch, F. et al. Electron localization determines defect formation on ceria substrates. Science 309, 752755 (2005).

47. Vil, G., Colussi, S., Krumeich, F., Trovarelli, A. & Prez-Ramrez, J. Opposite face sensitivity of CeO2 in hydrogenation and oxidation catalysis. Angew.

Chem. Int. Ed. Engl. 126, 1226512268 (2014).48. Garca-Melchor, M., Bellarosa, L. & Lpez, N. Unique reaction path in heterogeneous catalysis: the concerted semi-hydrogenation of propyne to propene on CeO2. ACS Catal. 4, 40154020 (2014).

49. Vile, G., Bridier, B., Wichert, J. & Perez-Ramirez, J. Ceria in hydrogenation catalysis: high selectivity in the conversion of alkynes to olens. Angew. Chem. Int. Ed. Engl. 51, 86208623 (2012).

50. Carrasco, J. et al. Molecular-level understanding of CeO2 as a catalyst for partial alkyne hydrogenation. J. Phys. Chem. C 118, 53525360 (2014).

51. Garca-Melchor, M. & Lpez, N. Homolytic products from heterolytic paths in H2 dissociation on metal oxides: the example of CeO2. J. Phys. Chem. C 118, 1092110926 (2014).

52. Wischert, R., Laurent, P., Coperet, C., Delbecq, F. & Sautet, P. g-alumina: the essential and unexpected role of water for the structure, stability, and reactivity of "defect" sites. J. Am. Chem. Soc. 134, 1443014449 (2012).

53. Li, J. et al. Low pressure induced porous nanorods of ceria with high reducibility and large oxygen storage capacity: synthesis and catalytic applications. J. Mater. Chem. A 2, 1645916466 (2014).

54. Zhang, S., Li, J., Gao, W. & Qu, Y. Insights into the effects of surface properties of oxides on the catalytic activity of Pd for C-C coupling reactions. Nanoscale 7, 30163021 (2015).

55. Tian, Z. et al. Highly sensitive and robust peroxidase-like activity of porous nanorods of ceria and their application for breast cancer detection. Biomaterials 59, 116124 (2015).

56. Zhang, S. et al. Visible-light-activated Suzuki-Miyaura coupling reactions of aryl chlorides over the multifunctional Pd/Au/porous nanorods of CeO2 catalysts. ACS Catal. 5, 64816488 (2015).

57. Mai, H.-X. et al. Shape-selective synthesis and oxygen storage behavior of ceria nanopolyhedra, nanorods, and nanocubes. J. Phys. Chem. B 109, 2438024385 (2005).

58. Wu, Z., Mann, A. K. P., Li, M. & Overbury, S. H. Spectroscopic investigation of surface-dependent acidbase property of ceria nanoshapes. J. Phys. Chem. C 119, 73407350 (2015).

59. Binet, C., Daturi, M. & Lavalley, J.-C. IR study of polycrystalline ceria properties in oxidised and reduced states. Catal. Today 50, 207225 (1999).

10 NATURE COMMUNICATIONS | 8:15266 | DOI: 10.1038/ncomms15266 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15266 ARTICLE

60. Nolan, M., Grigoleit, S., Sayle, D. C., Parker, S. C. & Watson, G. W. Density functional theory studies of the structure and electronic structure of pure and defective low index surfaces of ceria. Surf. Sci. 576, 217229 (2005).

61. Welch, G. C., Juan, R. R. S., Masuda, J. D. & Stephan, D. W. Reversible, metal-free hydrogen activation. Science 314, 11241126 (2006).

62. Kresse, G. & Hafner, J. Ab initio molecular-dynamics simulation of the liquid-metal-amorphous-semiconductor transition in germanium. Phys. Rev. B 49, 1425114269 (1994).

63. Kresse, G. & Furthmller, J. Efciency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comp. Mater. Sci. 6, 1550 (1996).

64. Kresse, G. & Furthmller, J. Efcient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 1116911186 (1996).

65. Chen, H.-T., Choi, Y. M., Liu, M. & Lin, M. C. A theoretical study of surface reduction mechanisms of CeO2(111) and (110) by H2. ChemPhysChem 8, 849855 (2007).

66. Fernndez-Torre, D., Carrasco, J., Ganduglia-Pirovano, M. V. & Prez, R. Hydrogen activation, diffusion, and clustering on CeO2(111): A DFT U

study. J. Chem. Phys. 141, 014703 (2014).67. Metiu, H., Chrtien, S., Hu, Z., Li, B. & Sun, X. Chemistry of Lewis acidbase pairs on oxide surfaces. J. Phys. Chem. C 116, 1043910450 (2012).

Acknowledgements

We acknowledge the nancial support from a NSFC Grant 21201138, 21401148, 21603170, 91645203 and the National 1000-Plan programme. This work was also funded by the Fundamental Research Funds for the Central Universities under Grant xjj2013102, xjj2013043 and xjj2014064. We thank Professor Suitao Qi at Xian Jiaotong University for getting access to the software of Vienna Ab-initio Simulation Package. The calculations were performed by using supercomputers at Shanghai Supercomputing Center and National Supercomputing Center in Shenzhen.

Author contributions

S.Z., Z.Q.H., Y.M., C.R.C. and Y.Q. designed the studies and wrote the paper. S.Z., W.G., J.L., F.C. and L.L. performed most of the experiments. Z.Q.H. carried out the DFT calculations. S.Z., Z.Q.H., Y.M., C.R.C. and Y.Q. performed data analysis. 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 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: Zhang, S. et al. Solid frustrated-Lewis-pair catalysts constructed by regulations on surface defects of porous nanorods of CeO2. Nat. Commun. 8, 15266 doi: 10.1038/ncomms15266 (2017).

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) 2017

NATURE COMMUNICATIONS | 8:15266 | DOI: 10.1038/ncomms15266 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11