Low-temperature ammonia synthesis on electron-rich [RuH6] catalytic centers

Ammonia is a central vector in sustainable global growth, but the usage of fossil feedstocks and centralized Haber-Bosch synthesis conditions causes >1.4% of the global anthropogenic CO2 emissions. While nitrogenase enzymes convert atmospheric N2 to ammonia at ambient conditions, even the most active manmade inorganic catalysts fail due to low activity and parasitic hydrogen evolution at low temperatures. Here, we show the [RuH6] catalytic center in ternary ruthenium complex hydrides (Li4RuH6 and Ba2RuH6) activate N2 preferentially and avoid hydrogen over-saturation at low temperatures and near ambient pressure by delicately balancing H2 chemisorption and N2 activation. The active [RuH6] catalytic center is capable of achieving an unprecedented yield at low temperatures via a shift in the rate-determining reaction intermediates and transition states, where the reaction orders in hydrogen and ammonia change dramatically. Temperature-dependent atomic-scale understanding of this unique mechanism is obtained with synchronized experimental and density functional theory investigations.


Introduction
Ammonia is critical to our food production ecosystem 1,2 and the single most produced polluting chemical (~170 million tons per year) [3][4][5] , while also holding the potential to become one of the most promising carbon-free and low-cost long term energy carrier 6,7 . The industrial Haber-Bosch (H-B) process employs a Fe-based catalyst, fossil-fuel sourced H2, and requires harsh operating conditions (typically 673-723 K and 100-300 bar of pressure). The large-scale and centralized H-B process accounts for nearly 2% of the world's consumption of fossil fuels 8 , and consequently over 1.44% of the global anthropogenic CO2 emissions 5 . The development of smallscale processes that rely on renewable electricity as an energy source to sustainably produce the H2 feedstock would thus be transformative in several ways. It would provide critical technological support towards the audacious goal of carbon-free growth and ensuring the green transition. Two indispensable targets would be reached simultaneously, where renewable energy penetration is arduous -food production and clean mobility 4,9 . A decentralized, low CAPEX NH3 synthesis process targeted at emerging markets with significant future population growth needs would also support the core U.N. sustainability goals.
While direct electrochemical ammonia production represents the Holy Grail, the documented yields remain very far from any kind of commercialization 10 . The discovery of efficient heterogeneous or homogeneous catalysts that exhibit high activity under mild conditions would thus be a key enabler for the decentralized production of green ammonia. For industrial ammonia synthesis, it is widely recognized that Ru-based catalysts work better than Fe-based catalysts under milder reaction conditions 11,12 . However, the high activation energy for direct N2 dissociation and the severe poisoning effect of hydrogen on conventional Ru metal catalyst renders efficient NH3 synthesis under lower temperatures (< 623 K) and lower pressures (< 50 bar) unattainable 13 . Therefore, many attempts to develop new catalysts for efficiently catalyzing N2+H2 to NH3 under mild conditions [14][15][16][17] . Recently discovered, a new class of ammonia catalysts -the ternary ruthenium complex hydrides 18 is a breakthrough in this endeavor. The ternary ruthenium hydride's ability to efficiently synthesize NH3 at <10 bar and < 573 K conditions lies in the unique chemistry of the coordination complex and the alkali (alkaline earth) metal framework, facilitating a catalytic mechanism bridging homogeneous and heterogeneous concepts, which are clearly distinct from the Ru metal catalyst. For ternary Ru complex hydride catalysts, Ru is in an ionic state, and N2 undergoes non-dissociative hydrogenolysis over the hydride(H -)-rich and electron-rich [RuH6] 4complex with the aid of the surrounding Li or Ba cations. The dynamic and synergistic engagement of all the components of the ternary hydrides creates a reaction path with a narrow energy span and leads to ammonia production with superior activities.
In this paper, we present the reaction mechanism facilitating ternary ruthenium complex hydrides to successfully produce NH3 at low temperature (448 K ≤ T ≤ 573 K) by selective N2 activation and escaping H2 over-saturation. This work shows the unique ability of the [RuH6] 4catalytic center in the ternary ruthenium hydride to shift its rate-determining intermediate states and transition states of the N2+H2 to NH3 reaction path in response to the lowering of the reaction temperature, which brings a significant change in reaction order of hydrogen and ammonia. This variation in the kinetics as a function of operating conditions (temperature, reactant partial pressures, etc.) is not a common phenomenon in catalysis but is observed in some cases [19][20][21][22][23] . Nonetheless, the mechanistic details behind it are seldom investigated, especially at an atomic level. Here, we achieve this via seamless integration of experimental and computational techniques to reveal the temperature-dependent catalytic process. Our finding discloses that an electron-rich active center with a comparable affinity towards N2 and H2 are critical for mild-condition ammonia catalysis. The thorough fundamental understanding developed in this study can be further used to design new low-temperature ammonia catalysts with better performance and has the potential to drive the green ammonia technology into a new direction. Figure 1 shows that the [RuH6] catalytic center in Ru complex hydride catalysts (Li4RuH6 and Ba2RuH6) not only outperforms the B5 site of Ru metal catalysts under the same working conditions but also produces NH3 at low temperatures. As discussed below, two inherent properties of the [RuH6] catalytic center are critical towards the observed outstanding activity at low temperatures: (a) its selectivity for chemisorbing N2 over H2 and (b) a self-adjusting mechanism of avoiding hydrogen over-saturation sustaining the N2 to NH3 conversion cycle.    The chemisorption H2 on the Li4RuH6 active surface here can be expressed as: ) and becomes part of the lattice and creates two [RuH7] in the process (see Figure S2). The transformation of the chemisorbed H2 to lattice H can be written as:

Results and discussion
Meawhile, the competitive chemisorption of N2 on the Li4RuH6 active surface is: The calculated binding free energies show that the preferential adsorption of N2 over H2 at the [RuH6]* active center is further enhanced at lower temperatures ( Figure 3A). This feature is critical for the understanding of the catalytic mechanism, particularly when combined with the kinetics of dissociative hydrogen chemisorption and hydrogen transfer over the [RuH6] centers. Figure 3B Figure S3). The lack of hydrogen poisoning effect allows enhanced ammonia production at a higher hydrogen partial pressure on Li4RuH6.
The chemisorption of H2 on the ternary hydride active surface has unique fingerprints ( Figure 3B). The chemisorption of H2 happens through a physisorbed transition state TS0-I, with a negligible barrier of 0.07 eV ( Figure S4). The H from the chemisorbed H2 participates in on-site scrambling with the lattice H on the [RuH6] active center. The on-site scrambling of the hydrogen has an insignificant activation energy of 0.04 eV ( Figure S5). Experimentally, we observe a minor reversible adsorption/desorption of H2 in the temperature range of 373 K to 473 K in the temperature programmed desorption (TPD) profile with no trace of net LiH, Ru powder, or a mixture of LiH and Ru ( Figure S6), which reinforces the observation of the chemisorbed nature of the H2 adsorption on the Li4RuH6 active surface. After charging with D2, the detection of the mixed HD signal in the TPD profile strengthens the conclusion of the on-site scrambling of D from chemisorbed D2 with lattice H (inserted plot of Figure 3B). A more robust signal of H2/HD than D2 in the TPD profile points out the magnitude of the on-site scrambling of the chemisorbed D2 with the lattice H.
Our study shows that the N2+H2 to NH3 conversion cycle on the Li4RuH6 catalyst surface happens through 13 different surface states (states 0-12). Visualization of the NH3 formation mechanism on the Li4RuH6 catalyst surface with intermediate states is provided in Figure  4. Here The overall chemical reaction in one catalytic cycle on Li4RuH6 catalyst surface is: The path shows a series of well-balanced and moderate activation energies -all with Ea ≤ 0.82 eV (see Table S1).
For better understanding the low temperature reaction pathway, the variations of reaction energetics as a function of temperature are explored and then analyzed by applying the energetic span model, in which the turnover frequency (TOF) determining transition state (TDTS) and TOF determining intermediate state (TDI) that maximize the energy span determine the rates and kinetics of the catalytic cycle 24,25 . The energetic span approximation 25 of the exothermal catalytic cycle to calculate turn over frequency ( ) from the energetic span (δE) of the free energy path: In this model, the free energy of TOF-determining transition state (∆ ), TOF-determining intermediate state (∆ ) and the free energy of reaction (∆ ) defines δE: The activation enthalpy ( ∆ ) for the catalytic path is backcalculated from the and the entropy correction ( ∆ ): Figure 5 (and Figure S7) shows the development of the free energy path of the catalytic cycle of N2+H2 to NH3 on the Li4RuH6 catalyst surface with a lowering of reaction temperature (448 K ≤ T ≤ 573 K).
The change in temperature shifts the TDI and TDTS of the catalytic cycle, with an inflection temperature being at 498 K. Experimentally, the Arrhenius plot for ammonia synthesis ( Figure 6A) locates this inflection point around 523 K. In addition, all measured kinetic parameters for ternary hydride catalyst (Li4RuH6/MgO) are temperature-dependent ( Figure 6), indicating the complex temperature-dependent switching of rate-determining states (i.e. TDI and TDTS).

Figure 5
The evolution of the free energy path with the lowering of temperature shifts the TDI and the TDTS of the energetic span model, and the inflection point is at 498 K of Li4RuH6 catalyst surface. For the lower temperature range (< 498 K), the TDI and TDTS are the initial/final state (state-0) and the transition state of the third hydrogenation (TS3-4) on the surface, respectively. In the higher temperature range (≥ 498 K), the TDI and the TDTS are the second H2(g) adsorption (state-12) and the transition state of the first H2 (g) adsorption (TS4-5) on the surface respectively. With further increase in temperature, state-9 comes energetically closer to state-12. At 573 K, the TDI and the TDTS are state-9 and TS4-5, respectively. The inserted plot shows the schematic presentation of the shift in TDI/TDTS. The TDI and TDTS at different temperature ranges are marked by the square-and circle-symbols, respectively, in the free energy plot and the inserted plot. We applied a fixed total pressure of 1 bar (N2:H2=1:3) to generate the free energy paths and barriers using density functional theory based free energy estimations and nudged elastic band method.
As shown in Figure 5, the TDTS moves from the transition state 4-5 (TS4-5) to TS3-4 as the temperature goes lower than the inflection point 498 K. Meanwhile, the TDI shifts from state 12 to state 0. For clarity, the inserted plot in Figure 5 presents a schematic view of the shift in TDI/TDTS with temperature. There might be one inflection for each change in TDI/TDTS, which we can not resolve due to their proximity. Another essential feature in the catalytic path is the energy difference between state 12 and state 9. At 573 K, the free energies of states 9 and 12 are similar, and they are both likely candidates for the TDI. The energy difference between state 12 and state 9 increases with decreasing temperature, and the state 12 is TDI in the range 498 K < T < 573 K. The theoretically derived activation enthalpy (ΔHa) and TOF, and experimentally derived apparent activation energy (Eapp) and TOF are listed in Table S2. The value of ΔHa at 448 K is 98.2 kJ mol -1 . While at 573 K, with state 9 as TDI, the ΔHa is calculated to be 72.4 kJ mol -1 . An increase in temperature lowers the activation enthalpy and increases the TOF, agreeing well with the trends observed experimentally. The apparent activation energy for Li4RuH6 catalyst determined by Arrhenius plot is Eapp = 71.2 kJ/mol at temperatures higher than 523 K, and significantly increased value of 102.8 kJ/mol at temperatures below 523 K ( Figure  6A). In contrast, there is no change in Eapp and other kinetic parameters for conventional Ru metal catalyst (Ru/MgO) in a wide temperature range (498-648 K). For Li4RuH6 catalyst, the energetic span and the TOF vary continuously with temperature. The temperature dependent TDI and TDTS modifications follow the entropy of intermediates and transition states. The entropy of a state is strongly affected by the adsorption/desorption of gas molecules. Such changes in the TDI or TDTS will tend to affect both ΔHa and the reaction order in gas molecules. This is beautifully captured by the analysis of the reaction order of NH3, N2, and H2 for Li4RuH6 ( Figure 6B-D We have been able to achieve a precise temperature resolved atomic-scale understanding of the reaction mechanism at the [RuH6] catalytic center, its unique thermodynamics, and kinetic aspects that enable exceptional low-temperature activity. These scientific insights need to be exploited towards optimizing complex transition metal hydrides as ammonia catalysts as well as exploring a newer class of materials that can replicate the behavior of [RuH6] catalytic center in the pursuit of renewables powered decentralized room temperature/pressure ammonia synthesis.

Conflicts of interest
The authors declare no competing financial interest