ARTICLES NATURE CHEMISTRY DOL:10.1038/NCHEM.1476 from and nea ar to the Ru-C respective oxides Ru/C12A7:s s this in a be rega own that the (ig ection t the N.re orde that titute the Mgo of the reaction 12A7 the N propert Ku-l s&Tsndingcancatrbntedoncftaigntcectondo on order for +1,a valu rate-limiting step. H Method ad on Ru gu catalts for am in an tube 0.25℃m ion d by A7:6 d in a rated up to 1.5 MPa and w as H 140 Ru/C12AZy- dy-state c 0 Cs/ .1 to 1.3 MPa).T cid solutio ofN N and H med that am nia is n /C12A TEM (M- The pof the 12A in Cl with d over Ru ible Mgo.T g He (50 ml ine on ru duri h Ru/C12A7e (Fig.5)is of the of Ru h back specia RU-C a es are al ly on Ru es react ions ed as the bac ence spectra ad the 27 during the reaction erences NATURE CHEMISTRY ADVANCE ONLINE PUBLICATION wwwnature.com/naturechemistry 2012 Macmillan Publishers Limited All rights reserved. further by the electrons encaged in C12A7:e2. The peaks at 2,176 and 2,124 cm21 can be attributed to N2 species on Ru atoms far from and near to the Ru-C12A7:e2 interface, respectively39. To clarify the reaction mechanism for ammonia synthesis over Ru/C12A7:e2, the kinetics of ammonia production over the catalyst were investigated (Fig. 4b and Supplementary Table S3)41. It is well known that the reaction order for N2 is almost unity in the rate equation for ammonia synthesis over most conventional catalysts (Supplementary Table S3)11,20,41,42. Although the N2 reaction order of Ru/MgO shown in Supplementary Table S3 is in good agreement with the reported results, it was estimated to be 0.5 in the case of the Ru/C12A7:e2 catalyst, which indicates that the N2 dissociation over Ru/C12A7:e2 is much more efficient than those of conventional cat￾alysts. This finding can be attributed to an efficient electron donation from C12A7:e2, as shown in Fig. 4a. Figure 4b also shows that the reaction order for H2 is approximately þ1, a value distinct from other conventional Ru catalysts, which have reaction orders for H2 that are, in general, negative because H2-dissociative adsorption pre￾cedes that of N2 dissociation and hydrogen adatoms resist the effi- cient dissociative adsorption of N2 on a Ru surface11,20,21. This poisoning is enhanced with increasing total pressure, which limits Ru catalysts for ammonia synthesis, as mentioned above. The posi￾tive reaction order with Ru/C12A7:e2 means that Ru/C12A7:e2 is not poisoned by hydrogen. A plausible explanation for these unique characteristics of Ru/C12A7:e2 is attributed to the nano￾cages of C12A7:e2, which can capture hydrogen species as H2 ions26. Hydrogen adatoms on Ru readily spill over onto C12A7:e2 and are incorporated into the positively charged nanocages of C12A7:e2 as H2 ions, which prevents hydrogen adatoms from occu￾pying the Ru surface. Supplementary Table S4 summarizes the ammonia evolution from Ru/C12A7:e2 and Ru-Cs/MgO when heated in a pure N2 atmosphere after ammonia synthesis in the pres￾ence of N2 and H2. It was confirmed that ammonia is not adsorbed on both catalysts before exposure to pure N2 gas. Apparently, Ru/C12A7:e2 evolves ammonia, which means that hydrogen is incorporated into C12A7:e2 as H2 ions during ammonia synthesis and the H2 ions in C12A7:e2 react with nitrogen species formed by N2 dissociation, which results in ammonia formation. The amount of H2 ions encaged in the Ru/C12A7:e2 catalyst was estimated to be 7.6 × 1019 cm23 (that is, about 4% of the electrons in C12A7:e2 are replaced with H2 ions). However, no ammonia evolution was observed over Ru-Cs/MgO. These results verify that C12A7:e2 func￾tions as a reversible hydrogen-storage material and prevents hydro￾gen poisoning on Ru during ammonia synthesis. A possible mechanism for the ammonia synthesis reaction over the Ru/C12A7:e2 catalyst (Fig. 5) is proposed as follows. Electrons encaged in C12A7:e2 are transferred to the Ru metal, which causes a substantial lowering of the work function of Ru by raising the Fermi level. As a consequence, back donation from Ru with an excess electron density to the N2 p*-antibonding orbital is enhanced, especially near the Ru-C12A7:e2 interface, which leads to a weakening of the N;N bond of adsorbed N2. Although H2 molecules are also adsorbed dissociatively on Ru surfaces, hydro￾gen adatoms readily spill over onto C12A7:e2 and are incorporated into the nanocages as H2 ions. This incorporation prevents hydro￾gen adatoms from poisoning the Ru surface. The activated nitrogen species react with H2 ions entrapped by C12A7 nanocages or with hydrogen adatoms on Ru to form ammonia. In the case of the former, H2 ions leave electrons in the cages and the resulting hydro￾gen species react with the activated nitrogen near the Ru-C12A7:e2 interface. The electrons encaged in C12A7:e2 are used repeatedly during the reaction. Conclusions We found that the stable electride C12A7:e2 works as an excellent electron donor and reversible hydrogen-storage material to enhance the catalytic activity for ammonia synthesis, with a TOF an order of magnitude greater than those reported so far. This material is com￾posed of the abundant oxides CaO and Al2O3, and has the unique properties of a very low work function and high chemical and thermal stabilities. As this material may, in a sense, be regarded as the bulk crystal of Fþ centres of alkaline earth oxides, the surface concentration of electrons energetically available for injection to Ru is much higher than that from a conventional Fþ centre and the subnanometre-sized cages that constitute the crystal structure include the surface layer as a H2 reservoir, which suppresses the lowering of the reaction rate at high pressure. These characteristic properties make Ru-loaded C12A7:e2 an efficient catalyst for ammonia synthesis. These findings imply the potential for catalysts of chemical reactions in which the activation of an inert molecule, such as N2 and CO2, by electron injection is the rate-limiting step. Methods Preparation of Ru/C12A7:e2. C12A7:e2 powders were prepared by the reaction of C12A7, CaO.Al2O3 (CA) and Ca metal at 700–1100 8C. Ru loading was conducted using a solvent-free preparation method. Typically, C12A7:e2 powder and Ru3(CO)12 were sealed in an evacuated silica tube and heated under a temperature programme of 2 8C min21 up to 40 8C, held for one hour, 0.25 8C min21 up to 70 8C, held for one hour, 0.4 8C min21 up to 120 8C, held for one hour, 0.9 8C min21 up to 250 8C, hold for two hours and then cooled to ambient temperature. Ammonia synthesis reactions. These were conducted in a silica glass or a stainless￾steel flow set-up that operated up to 1.5 MPa and were supplied with an extra pure (.99.99995%) mixture of H2:N2 (3:1). Before the reactions, all the catalysts were treated in a stream of N2 þ 3H2 under 0.1 MPa using a temperature programme of heating to 400 8C in two hours and then standing at 400 8C for one hour. The concentration of ammonia in the stream that left the catalyst bed (0.2 g) was monitored under steady-state conditions of temperature (320–400 8C), gas flow rate (60 ml min21 ) and pressure (0.1 to 1.3 MPa). The ammonia produced was trapped in 5 mM sulfuric acid solution and the amount of NH4 þ generated in the solution was determined using an ion chromatograph (LC-2000 plus, Jasco) equipped with a conductivity detector. Characterization. TEM (JEM-2010F, Jeol) images of the samples were obtained to determine the microstructural characteristics. The Brunauer–Emmett–Teller specific surface areas of the samples were determined by the measurement of nitrogen adsorption–desorption isotherms at 2196 8C using an automatic gas-adsorption instrument (NOVA 4200e, Quantachrome) after evacuation of the samples at 150 8C. Ru content was determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES; ICPS-8100, Shimadzu). Ru dispersion was determined by CO-pulse chemisorption (BELCAT-A, BEL, Japan) at 50 8C using a He flow of 30 ml min21 and pulses of 0.09 ml (9.88% CO in He). Prior to the dispersion analysis, the catalyst was treated with flowing He (50 ml min21 ) at 400 8C for 15 minutes and then flowing H2 (50 ml min21 ) at 400 8C for 15 minutes. Adsorbed hydrogen atoms on the reduced catalysts were removed by purging with He (50 ml min21 ) at 400 8C for 15 minutes. To calculate the metal dispersion a stoichiometry of Ru/CO ¼ 1 was assumed. FT-IR spectra of adsorbed N2 were measured using a spectrometer (FT/IR-6100, Jasco) equipped with a mercury– cadmium–tellurium detector at a resolution of 4 cm21 . Samples were pressed into self-supported disks (20 mm diameter, about 20 mg). A disk was placed in a silica-glass cell equipped with NaCl windows and connected to a closed gas-circulation system to allow thermal adsorption–desorption experiments. The disk was heated under vacuum at 500 8C for one hour, and then treated with circulated H2 (27 kPa) with a liquid-nitrogen trap for one hour. The disk was then evacuated at the same temperature for one hour to remove the hydrogen, then exposed to 5 kPa of N2 to obtain infrared spectra for adsorbed N2. After the pretreatment, the disk was cooled to 2170 8C under vacuum to obtain a background spectrum. Pure N2 (99.99995%) and H2 (99.99999%) were supplied to the system through a liquid-nitrogen trap. Isotopic nitrogen (15N2, 98%) was used without purification. The infrared spectrum of the sample at 2170 8C prior to N2 adsorption was used as the background for difference spectra obtained by subtracting the backgrounds from the spectra of N2-adsorbed samples. In this experiment, 2 wt% Ru/C12A7:e2 with an electron concentration of 5.0 ×1020 cm23 , 2 wt% Ru/C12A7:O22 and 3 wt% Ru/Al2O3 were used for FT-IR measurements. Received 24 May 2012; accepted 6 September 2012; published online 21 October 2012 References 1. Mittasch, A. Early studies of multicomponent catalysts. Adv. Catal. 2, 81–104 (1950). ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1476 6 NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2012 Macmillan Publishers Limited. All rights reserved.