Laser-induced nitrogen fixation

For decarbonization of ammonia production in industry, alternative methods by exploiting renewable energy sources have recently been explored. Nonetheless, they still lack yield and efficiency to be industrially relevant. Here, we demonstrate an advanced approach of nitrogen fixation to synthesize ammonia at ambient conditions via laser–induced multiphoton dissociation of lithium oxide. Lithium oxide is dissociated under non–equilibrium multiphoton absorption and high temperatures under focused infrared light, and the generated zero–valent metal spontaneously fixes nitrogen and forms a lithium nitride, which upon subsequent hydrolysis generates ammonia. The highest ammonia yield rate of 30.9 micromoles per second per square centimeter is achieved at 25 °C and 1.0 bar nitrogen. This is two orders of magnitude higher than state–of–the–art ammonia synthesis at ambient conditions. The focused infrared light here is produced by a commercial simple CO2 laser, serving as a demonstration of potentially solar pumped lasers for nitrogen fixation and other high excitation chemistry. We anticipate such laser-involved technology will bring unprecedented opportunities to realize not only local ammonia production but also other new chemistries .


Detection of produced ammonia and performance evaluation
The generated ammonia was detected and quantified after the hydrolysis of Li3N in 100 mL of 0.05 M hydrochloric acid solution.The generated ammonia from other metal nitrides was detected and quantified after the hydrolysis of metal nitrides in 15 mL of 0.05 M hydrochloric acid solution.The surface of the titanium sheet was refreshed, and the titanium nitride coating remains due to the high solubility of Li3N in an acid solution.The ammonia in the 0.05 M hydrochloric acid solution was quantified by the indophenol blue method using the spectrophotometric analysis (fig S2) 1 , whose maximum absorbance occurred at a wavelength of 660 nm.The ammonia yield rate is given by ε =   (1)   where c is the measured NH3 concertation, V is the volume of the hydrochloric acid solution, t is the time of the laser-induced process and s is the working area of the laser in cm -2 .The correlated energy consumption is given by where p is the laser power used, t is the time of the laser-induced process, c is the measured NH3 concertation, V is the volume of the hydrochloric acid solution and M is the relative molecular mass of NH3.Considering the hydrolysis step is fast, in this work, only the lasertreated time was taken into account in calculating the yield rate.
For the isotope labeling experiment, 15 NH3 was quantified by 1 H NMR (400 MHz) with a recorded total of 1,000 transient scans.The spectra were collected in D2O with maleic acid as an internal standard.The scalar interaction between 1 H and 15 N in 15 NH4 + results in a splitting of the 1 H resonance into two symmetric signals, with a spacing of 73.6 Hz.
Further the DFT-D2 method 8 was used to include the van der Waals (vdW) corrections with our reparametrized C6 coefficients for metals 9 .The core electrons were represented by the Projected Augmented Wave (PAW) pseudopotentials 7 , while valence electrons were expanded in plane waves with a basis set cutoff energy of 450 eV.The converged lattice constants for the bulk fcc Li2O and Li were 4.45 Å and 4.10 Å respectively, bcc MgO was 4.12 Å, while for the hexagonal Li3N, they were a=3.47 Å, c=3.75 Å.
The Catkit package 10 was used to generate the Li2O surfaces, with a (111) surface of 63 atoms and 6 layers thickness and a (211) surface of 120 atoms and 9 layers thickness.A -centered k-point grid of 4x4x1 for (111) surface and 3x3x1 for (211) surface was employed which is generated through Monkhorst-Pack method with a reciprocal grid smaller than 0.03 Å -1 11 .All the simulations were spin polarized and a vacuum spacing of 15 Å was used in the z-direction.
A dipole correction 12 was applied to remove the artifacts due to asymmetry in the z-direction.
To determine the partial electron occupancies, the Gaussian smearing scheme was used with an energy smearing of 0.03 eV.The electronic step convergence was set to 10 -4 eV.In general, the structures were converged till the forces were converged to within 0.03 eV/Å.In a few cases, to avoid restructuring of the surface, the stopping criterion for the simulations was set to an energy difference between two ionic steps of less than 10 -3 eV.
For the estimation of the bulk formation energies Li2O and Li3N, the free energy of each species at 298 K was estimated using the following equation: ZPE, Cp and S were estimated using the thermal properties tags in PHONOPY for the solids 13-14 , while for the gases vaspkit 15 was used.The activation barriers were estimated using the Climbing Image Nudged Elastic Band (CI-NEB) method 16 developed by the Henkelman group.
The initial six images between the initial and final state were generated by the Image Dependent Pair Potential (IDPP) method 17 .The phonon density of states for the bulk and the surface were generated using the PHONOPY package in combination with VASP [13][14] .All the simulations have been uploaded to the iochem-BD database and can be accessed using the following link: https://doi.org/10.19061/iochem-bd-1-283.

Physicochemical characterization
Scanning electron microscopy was performed on a Zeiss LEO 1550 -Gemini system (acceleration voltage: 5 kV).EDX element maps were obtained at 10 kV acceleration voltage.Once all the oxides have been induced and excited within the laser's range, further increasing the laser power density has minimal impact on promoting subsequent nitride formation.
In addition, the production yield of ammonia is influenced by several factors.These include the laser power density, which determines the energy required to break the metal-oxygen bond, the scanning speed of the laser, which affects the overall process time including metal oxide dissociation, and the reaction time of the activated zero-valent metal with nitrogen.
Furthermore, the reactivity of the metal plays a crucial role in ammonia production.In essence, there exists an optimal scanning speed for a specific power density.Lower power densities necessitate longer reaction times, resulting in the higher ammonia yields.Conversely, higher power densities can achieve satisfactory results with shorter irradiation times, thereby consuming less energy and enhancing productivity.Therefore, multiple experimental parameters influencing each other lead to the results in Fig. S3: the slowest scanning speed (black, 0.17 mm s -1 ) shows the least ammonia production at 68 and 152 W cm -2 but the best or second to the best at 48, 91, 101, and 143 W cm -2 .
Supplementary Figure 4. 1 H NMR spectrum of the standard 15 NH4Cl solutions with 15 NH4 + concentration of 1 ppm.

Supplementary Note 1
Our experimental results suggest a reaction pathway that the metal-oxide bond dissociates with the formation of metal, which then reacts spontaneously with nitrogen and forms metal nitride.
In the Mg1s region and Al2p region of XPS spectra (fig.S17A), the peak signals attributed to zero-valent Mg and Al are detected, respectively.Besides, the EDX element maps of the laserinduced product obtained by using Al2O3 as a medium reveal a large amount of zero-valent Al there is only a weak overlap between the map corresponding to this element and others (fig.o C in the focused laser beam 18 .This TiN is very useful as it acts as a refractory passivation layer in further cycles.Note: It is essential to note that besides the breaking of metal-oxygen bonds plays a crucial role in the LINF process, the reactivity of zero-valent metals with nitrogen is another significant factor influencing nitride formation.Among these metals, lithium exhibits the highest level of reactivity and can even react with nitrogen at room temperature, leading to the formation of lithium nitride 28 .In contrast, the formation of zinc nitride necessitates the reaction of zinc with ammonia at temperatures exceeding 600 o C 29 .

XSupplementary Figure 1 .
-ray diffraction measurements were performed on a Bruker D8 Advance diffractometer equipped with a scintillation counter detector with CuKα radiation (λ = 0.15418 nm) applying a 2θ angle in the range 5-80°.1 H NMR spectra were recorded on Agilent 400 MHz. .The surface valence and chemical bonds of the samples were tested by X-ray photoelectron spectroscopy on an ESCALAB 250 spectrometer (Thermo Fisher Scientific) with monochromatic Al Kα radiation (1486.6 eV).The nitrogen content in synthesized TiN was measured using a LECO-TC500 ONH analyzer with a power of 5000 W (LECO Corporation, St. Joseph, MI).UV-vis absorption spectra were performed on a T70+ UV spectrometer (from PG instruments Ltd), which is a single split beam spectrophotometer available with a variable (0.5, 1, 2, 5nm) spectral bandwidth.Reaction cell: A photograph of a reaction cell for laser-induced nitrogen fixation.

Table 1 .
Comparison with other reported methods.The results incorporate the actual production of lithium under a current of 0.2 A for 1000 s.The calculation of the final ammonia yield takes into account the cumulative time of molten salt electrolysis, Li nitriding, and the conversion rate involved.£ In the study, the highest ammonia yield corresponds to a Li2O conversion of 11.2%.XRD and XPS analysis confirmed a reaction selectivity of nearly 100% for lithium oxide converting into lithium nitride, with no significant by-products detected.Consequently, the yield is approximately 11.2%. *

Table 2 .
List of bond dissociation energies of metal-oxygen bond Source: J. A. Dean, "Properties of Atoms, Radicals, and Bonds" in Lange's handbook of chemistry (McGRAW-HILL, INC., 1999)