Carbon‐Negative Ammonia Production from the Air

Abstract

Ammonia holds paramount importance as a fundamental chemical commodity for large‐scale production of fertilisers and hydrogen carriers. The conventional Haber–Bosch process relies heavily on fossil fuels, making ammonia synthesis a significant contributor to greenhouse gas emissions. Here, we show a carbon‐negative ammonia synthesis process that not only produces ammonia from the air but also directly captures the atmospheric CO₂ (DAC). This process uses lithium as a mediator to cleave the nitrogen gas via a nitridation reaction at ambient pressure below 80 °C. Exposing the nitride to moisture naturally yields ammonia. The co‐produced lithium hydroxide intermediate allows for spontaneous absorption of CO₂ with an exceptional capacity of 18.4 mmol CO₂/g. The resultant lithium carbonate is electrolysed at ambient conditions to recycle the metallic lithium and release high‐purity gaseous CO₂. This proof‐of‐concept of ammonia production from air coupled with DAC paves a new way for the development of sustainable and negative‐emission technologies.

Our work demonstrates a carbon‐negative ammonia synthesis process that not only produces NH₃ from air but also directly captures the atmospheric CO₂ (DAC), which achieves low‐cost DAC process with exceptional CO₂ capacity and free ammonia production. This paves a new way in the development of sustainable and negative‐emission technologies.

Introduction

Current chemical engineering processes that generate significant CO₂ emissions necessitate critical re‐engineering for future technological growth and sustainable adoption. Ammonia, a cornerstone of the global economy, serves vital roles as a fertiliser feedstock, an industrial and household chemical, and a chemical precursor, with additional potential as a future carbon‐free fuel and hydrogen storage molecule.^{[ 1} , ^{2 ]} Approximately 170 million metric tonnes of ammonia are produced worldwide annually, with production projected to increase by 2.3% per year.^{[ 3 ]} Despite ammonia being widely used and considered a potential zero‐carbon molecule critical to the second chemical revolution, the current industry large‐scale ammonia production is the Haber‐Bosch process operating at high pressure and high temperature, requires fossil fuel stock as a feasible energy source and on‐site hydrogen sources through the hydrocarbon reforming process. Ammonia production consumes 1%–2% of the world's annual energy supply and is responsible for more than 1.6% of global carbon emissions.^{[ 4 ]}

To date, numerous innovative methods and strategies have been employed to produce ammonia more economically and in an environmentally friendly manner.^{[ 5 ]} While extensive works have been focused on the electrochemical synthesis of ammonia directly from air/N₂ and water, until now, the Faraday efficiency (FE) in aqueous systems remains mostly limited to around 30%.^{[ 6} , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ^{16 ]} Some other recent studies have explored the use of Li as a mediator, offering heightened reactivity with nitrogen under standard atmospheric conditions to break the triple bond of nitrogen and transform it into nitride. Additionally, recent research works demonstrate that achieving high Faraday efficiency in the lithium‐mediated nitrogen reduction reaction (Li‐NRR) system involves the use of additional high‐cost chemicals.^{[ 17} , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ^{28 ]} For example, alcohol is frequently employed in the electrolyte as a sacrificial proton carrier, which is continuously consumed and supplied by the electrochemical oxidation of H₂ to produce alcohol.

On the other hand, given the pressing need to reduce the greenhouse gas CO₂ emissions in every aspect of anthropogenic activities,^{[ 29} , ^{30 ]} negative emission technologies play a pivotal role in slowing down or even reversing the concentration of atmospheric CO₂ and enabling the sustainable production of carbon source essential to the chemical industry when fossil fuels phase out in the projected future. Direct air capture of CO₂ (DAC) represents a collection of negative emission technologies based on various separation principles.^{[ 31} , ³² , ³³ , ^{34 ]} One major challenge of DAC is the high energy required for the regeneration of the carbon capture agent. Very recently, DAC plus integrated catalytic reduction of CO₂ has gained much attention due to the ability to waive the regeneration step.^{[ 35} , ³⁶ , ^{37 ]} However, how to integrate DAC with chemical engineering processes, particularly ammonia synthesis, remains largely unexplored to date, posing a significant opportunity/challenge in revolutionising process engineering.

In this study, we successfully unite an innovative ammonia (NH₃) synthesis process with DAC to achieve carbon‐negative ammonia production from the air. This advanced process features a low energy and low waste‐generating chemical‐electrochemical Li‐assisted cycle. The cycle stages entail ammonia synthesis from the air and DAC, followed by Li regeneration. After the nitridation of Li, the produced Li₃N naturally reacted with moisture in the ambient air, generating ammonia and lithium hydroxide (LiOH). Then the ammonia is recovered through absorption into liquid water. Subsequently, LiOH efficiently captured CO₂ directly from the air (∼400 ppm), forming lithium carbonate (Li₂CO₃). The products underwent acidification to liberate CO₂ and yield the necessary electrolyte for the Li metal recovery process. Finally, lithium ions (Li⁺) were electrochemically converted to metallic lithium (Li⁰), and the acid was regenerated, thus bringing the cycle to a close loop.

Given that the moderate energy requirements for chemical reactions and electrochemical recovery at near or room temperature can be met through renewable energy sources and that the DAC stage effectively removes CO₂ from the atmosphere, the overall process operates as a carbon net‐negative system. This methodology represents a significant advancement in negative emission technology for the sustainable production of ammonia. Meanwhile, our results suggest that the integration of multiple chemical engineering processes is a promising, effective, and energy‐efficient strategy to address the long‐standing problem in the chemical synthesis processes.

Results and Discussion

As shown in schematic Figure 1, we demonstrated a three‐stage process of Li‐assisted chemical loop including: i) NH₃ synthesis. Li pre‐activation and consequent nitrogen fixation (Equation 1), and the hydrolysis of Li₃N with the moisture in the air to produce ammonia and an alkaline material LiOH (Equation 2). ii) DAC. Absorption of atmospheric CO₂ by the alkalines (Equation 3) followed by acid decarbonisation to produce highly concentrated CO₂ (Equation. 4). iii) Li regeneration by Li electrodeposition in a hybrid aqueous/Li‐ion conducting glass‐ceramics (LICGC)/organic electrolytic cell together with acid reproduction (Equations 5 and 6). The overall reaction process can be summarised in Equation 7, with air components including N₂, H₂O, and CO₂ as the only feedstock, pure CO₂, O₂ and NH₃ as the main products, and renewable power as the energy input, demonstrating a chemical loop with Li serving as an integrated catalyst that is not consumed.

Figure 1.

The schematic diagram of the three‐stage cyclic process. The overall process entails ambient pressure lithium‐assisted integrated production of NH₃, DAC of CO₂, and Li regeneration, respectively. Stage 1 demonstrates the synthesis and production of NH₃. Stage 2 shows DAC of CO₂ producing Li₂CO₃. Stage 3 elucidates Li recovery. To close the loop, Li is recovered through an electrodeposition regeneration electrochemical reaction. Stage 1 includes a nitrogen fixation reaction temperature of 80 °C, with all other stages occurring at room temperature and ambient pressure. The moderate energy requirements of the stages feature sustainable processes compatible with renewable energy sources. This characteristic makes it well‐suited for decentralised production and transportation.

The reactions in each step consist of the following equations:

• Stage 1:

  $$6\mathrm{L}{\mathrm{i}}_{\text{(pre-activated)}}+{\mathrm{N}}_2\to \text{2L}{\mathrm{i}}_3\mathrm{N}$$ (1)

  $$2\mathrm{L}{\mathrm{i}}_3\mathrm{N}+6{\mathrm{H}}_2{\mathrm{O}}_{(\text{from air})}\to 6\text{LiOH}+2\mathrm{N}{\mathrm{H}}_3$$ (2)
• Stage 2:

  $$6\text{LiOH}+3\mathrm{C}{\mathrm{O}}_{2(400\text{ppm from air})}\to 3\mathrm{L}{\mathrm{i}}_2\mathrm{C}{\mathrm{O}}_3+3{\mathrm{H}}_2\mathrm{O}$$ (3)

  $$3\mathrm{L}{\mathrm{i}}_2\mathrm{C}{\mathrm{O}}_3+3{\mathrm{H}}_2\mathrm{S}{\mathrm{O}}_4\to 3\mathrm{L}{\mathrm{i}}_2\mathrm{S}{\mathrm{O}}_4+3{\mathrm{H}}_2\mathrm{O}+3\mathrm{C}{\mathrm{O}}_{2\text{(pure)}}$$ (4)
• Stage 3:

  $$\text{Cathode}:6{\mathrm{Li}}^{+}+6{\mathrm{e}}^{-}\to 6\text{Li},\text{in}1\mathrm{M}\text{LiCl}{\mathrm{O}}_4/\text{PC}$$ (5)

  $$\text{Anode}:3{\mathrm{H}}_2\mathrm{O}-6{\mathrm{e}}^{-}\to 3/2{\mathrm{O}}_2+6{\mathrm{H}}^{+},\text{in}1\mathrm{M}\mathrm{L}{\mathrm{i}}_2{\mathrm{SO}}_4/{\mathrm{H}}_2\mathrm{O}$$ (6)
• Overall:

  $$\begin{array}{cc}& {\mathrm{N}}_2+3{\mathrm{H}}_2{\mathrm{O}}_{(\text{from air})}+3\mathrm{C}{\mathrm{O}}_{2(400\text{ppm from air})}\hfill \\ & \to 3\mathrm{C}{\mathrm{O}}_{2(\text{pure})}+3/2{\mathrm{O}}_2+2\mathrm{N}{\mathrm{H}}_3\hfill \end{array}$$ (7)

Stage 1. NH₃ Synthesis

Bulk Li remains stable under dry nitrogen for several months. As shown in Figure S1a, Li plates show a negligible colour change under dry N₂ after 2 months. Interestingly, this constraint can be eliminated with pre‐exposure of the Li surface to moisture. This pre‐activation mechanism has been demonstrated to enhance the post‐nitridation reaction of Li metal^{[ 38 ]} (Figure S2). Therefore, activated Li can be used effectively during the nitridation reaction, and the nitridation can be accomplished under low temperatures (<100 °C). Without incorporating the pre‐activation step, the Li surface remained shiny under the nitridation experimental condition (Figure S1b). We incorporated this mechanism as an essential “pre‐activation” step prior to the nitridation reaction step with control for the moisture exposure to Li (more detail in Supplementary information–Preparation of the pre‐activated lithium samples section). The controlled dosing of moisture uses N₂ as the carrier gas (Figure S3). The duration of the pre‐activation step is then optimised according to the nitrogen fixation reaction, as elucidated in Figure 2a,b,d. The final products were reacted with H₂O to measure the ammonia concentration and calculate the nitrogen fixation reaction specified as conversion efficiency (Figure 2a,b,d and Equation 1). With a pre‐activation of 12 h, a maximum conversion efficiency of 99.9% was achieved. This result indicates that the controlled introduction of moisture consumes less than 0.1% of lithium (based on the mass balance of NH₃ and Li). The quantitative measurements were performed using the amount of generated NH₃ measured by nuclear magnetic resonance (NMR) spectroscopy and confirmed with Ultraviolet‐Visible (UV–vis) Spectroscopy (Figures S4 and S5).

Figure 2.

Pre‐activation, nitrogen fixation on Li plate and the reaction between moisture and Li₃N. a) Li₃N yield under different pre‐activated times, all samples are then treated with N₂ for 24 h at 80 °C after the pre‐activation. b) Li₃N yield with different pre‐activated times and N₂ treatment time of 3 h and c) 24 h at 80 °C. d) Conversion efficiency of commercial Li₃N to ammonia under 2.5 L min⁻¹ air flow and different relative humidities with the duration of 24 h. e) schematic representation of Li surface pre‐activation process by controlled moisture diffusion under N₂ atmosphere. Optical photos show the surface of Li plate before, after pre‐activation and after nitridation. f) the exothermic hydrolysis reaction of Li₃N, releases NH₃ expanding the syringe volume. The NH₃ is then dissolved and captured in H₂O (phenolphthalein added as pH indicator), consequently contracting the syringe volume (Videos S1 and S2, detailed information in Experimental procedures and Supporting Information Notes 2).

To analyse the conversion efficiency and kinetics of preactivated Li to Li₃N, the preactivated Li was exposed to N₂ with different durations, up to 24 h, which was used to calculate the conversion rate of Li to Li₃N. Accordingly, the optimised pre‐activation time of 12 h was obtained (Figure 2a). To analyse the conversion kinetics of Li to Li₃N, the pre‐activated Li metal was consequently exposed to N₂ at 80 °C for different durations of 3 and 24 h, achieving ∼94.4% and ∼100% conversion rate of Li to Li₃N (Figure 2b,c), respectively. Common practices for Li₃N synthesis from Li and nitrogen involve an operating temperature range of 180–250 °C. ^{[ 39} , ^{40 ]} However, processing melted lithium poses safety concerns, and it would be desirable to operate the reaction of lithium with nitrogen at a temperature below its melting point. The process demonstrated here provides a pathway to pre‐activate Li with moisture, leading to an energy‐effective process on the nitridation reaction, which can be achieved with nearly 100% conversion efficiency working under lower temperatures (80 °C) and ambient pressure.

To obtain NH₃ from Li₃N, liquid water was reacted with Li₃N with ease. The hydrolysis reaction is spontaneously, leading to the rapid and substantial liberation of NH₃ gas (Figure 2f; Videos S1 and S2, Supporting Information Notes 2). Li₃N features high reactivity with air moisture under ambient conditions, producing LiOH and releasing NH₃. Therefore, the moisture directly from the air is utilised as the reactant to release NH₃ from Li₃N (Figures 1 and 2c), producing LiOH. The produced LiOH naturally captures and removes CO₂ from the air, resulting in the formation of Li₂CO₃. Meanwhile, the generated NH₃ is efficiently captured through absorption into liquid water, enabling its separation from the carrier air. Thus, NH₃ synthesis in air can be coupled with DAC technology, wherein Li₃N reacts with air moisture to release NH₃, and the resulting LiOH products capture CO₂ from the air. In addition, direct exposure to air moisture is also superior to adding liquid water to Li₃N due to the highly exothermic reaction and the high water‐solubility for both Li⁺ ions and NH₃, which makes the separation and recycling processes challenging. Hence, the devised moderate condition for producing ammonia allows for controlled NH₃ production of while capturing CO₂.

Figures 2d and S6 shows the conversion rate for commercial and synthesised Li₃N from Li₃N to NH₃ under varying relative humidities. An increase in humidity clearly results in a higher conversion rate. Enhanced reaction kinetics are especially shown for Li₃N with increased adsorbed water molecules.^{[ 41 ]} Complete conversion of Li₃N to LiOH (Figures 2c and 3a) is also consistent with findings from previous reports that Li₃N is unstable under high humidity conditions.^{[ 42 ]}

Figure 3.

Li₃N to NH₃ conversion rate in air moisture and the characterisation of Li₃N. a) The conversion rate from different preparation methods of Li₃N to NH₃, the data shown as the means ± SD (n = 3). b) The XRD results for Li₃N prepared using different methods. c) The composition of Li₃N obtained from different preparation methods, elucidating the proportion of α‐Li₃N and β‐Li₃N. (purchased: commercial lithium nitride; as‐synthesised: as‐synthesised lithium nitride without ball milling and ground by mortar and pestle; bm: as‐synthesised sample including ball milling with agate balls; bmss: as‐synthesised sample including ball milling with stainless steel balls.) d) N1s XPS spectra for as‐synthesised Li₃N and e) N1s XPS spectra for the bmss Li₃N sample. The bmss sample underwent partial conversion to NH₃ with an appearance of a small N─H peak. This is due to the sample instability in the air during transportation and handling for XPS measurements.

The majority of previous works have focused on commercial Li₃N to study its moisture stability, while the stability difference between different Li₃N crystal structures and morphologies in response to moisture has remained under‐explored. Figure 3a shows the conversion kinetics of Li₃N to NH₃ using different Li₃N synthesis methods. The as‐synthesised and bm Li₃N resulted in a relatively low conversion rate of ∼60% while the bmss Li₃N achieve a conversion rate comparable to that of the purchased Li₃N with over 99% conversion to ammonia within 24 h. The bm and bmss Li₃N was processed using ball‐milling with agate and stainless‐steel balls from as‐synthesised Li₃N, respectively. The ammonia conversion kinetics of Li₃N obtained from different routes show significant differences, as presented in Figure 3a. This difference may be attributed to the change in particle size and crystal structure. The powder particle sizes were characterised using scanning electron microscope (SEM) image processing. Considering consistent particle size between the bm and bmss samples (Figure S7), a relatively similar reaction kinetics was anticipated. However, the bm kinetics anomaly features significantly lower Li₃N to ammonia kinetics than that of the bmss sample. Understanding the mechanism of the hydrolysis of Li₃N is crucial for improving the kinetics and, consequently, the reaction yield. Therefore, we further investigated the crystal structure and reaction mechanism of different Li₃N samples (Figures S7 and S8). After analysing ∼70 Li₃N specimen high‐resolution crystal grains by high‐resolution transmission electron microscopy (HRTEM) (Figures S8 and S9), the proportion of the beta phase was found to be higher than the alpha phase in the bmss sample than that of the as‐synthesized sample. This is in agreement with the XRD obtained ratio for different phase of Li₃N presented in Figure 3b,c. The as‐synthesised and bmss samples were also characterised by XPS (Figure 3d,e). The peak located at 398 eV consists of the N1s obtained from Li₃N.^{[ 43 ]} Figure 3a shows that the sample with a higher beta phase ratio exhibits higher reactivity. Because the samples were momentarily exposed to air prior to sample loading under vacuum for XPS characterisation, the samples were partially decomposed to form an intermediate N─H bond located at 403 eV. This confirms the hydrolysis of β‐Li₃N with air moisture featuring high kinetics (Figure 3a).

We conducted density function theory (DFT) calculations to gain fundamental insight into this hydrolysis reaction mechanism and explored experimental observations of reaction kinetics differences between our samples. The DFT calculations were employed to investigate the hydrolysis reaction kinetics. X‐ray diffraction (XRD) (Figures 3b and S10) identified crystal facet orientation characteristics, predominantly exposing the (001) and (100) α‐Li₃N surfaces while exposing (001), (100), and (101) β‐Li₃N surfaces. These low‐index surface facets may terminate with distinct configurations of Li(1), Li(2), and N atoms and exhibit different stability and reactivity. Therefore, we calculated these surface energies to identify the relatively stable termination among these low‐index surfaces (Details in Supporting Information Experimental procedures section). Previous studies have suggested that Li(1) forms a Li layer, whereas Li(2) forms a Li‐N layer with N.^{[ 44 ]} Refining our approach due to spatial complexities in β‐Li₃N, we identified Li(2) and N atoms within the same plane, forming three equivalent Li─N bonds (Figure S10a). Figure S10b illustrates all possible surface terminations among these surfaces. Figure 4a shows the surface energies of various α‐Li₃N and β‐Li₃N surfaces as a function of nitrogen chemical potential, where lower values of surface energy indicate higher stability. Under the experimental conditions marked by the solid black line, the comparatively stable surface terminations have been identified as follows: α‐Li₃N (001) with Li(1) termination, α‐Li₃N (100) with Li(2) termination, β‐Li₃N (001) with Li(1) termination, β‐Li₃N (100) with Li(2) termination, and β‐Li₃N (101) with N termination (Figure S11). In the following, the term “stable surface” refers to surface terminations that exhibit the lowest surface energy among those considered for α‐Li₃N and β‐Li₃N in this study.

Figure 4.

Density Functional Theory (DFT) calculation about the reactivity of α/β‐Li₃N in H₂O decomposition. a) Calculated surface energy (γ) of various α/β‐Li₃N surfaces plotted against the nitrogen chemical potential (µ_N) variation. These surfaces correspond to those identified in XRD analyses. The black solid line represents the experimental conditions of 1 standard atmosphere (atm) pressure at a temperature of 298.15 K. b) The energy profile of H₂O adsorption and decomposition on the most stable α/β‐Li₃N surfaces at 1 atm at 298.15 K. The selected most energetically favourable surfaces are shown in Figure S10. The ground state for each energy profile is the unadsorbed state. The black rectangle highlights the decomposed H.

We further explored the adsorption and decomposition of H₂O on these stable surfaces to simulate the interaction of Li₃N with atmospheric H₂O. Figure 4b displays the configurations and energy profiles for H₂O adsorption and decomposition on these stable surfaces. It is important to note that two types of the β‐Li₃N (100) surface showed very close surface energies (differing by only 3 meV Å⁻²). Therefore, we also analysed the H₂O adsorption and decomposition on the β‐Li₃N (100) surface with Li(1) termination (Figure S12). As depicted in Figure 4b, H₂O adsorption and decomposition occur as two separate processes on the α‐Li₃N (001) surface, releases energy of 1.8 eV. In contrast, on the other surfaces, H₂O decomposes spontaneously upon adsorption, accompanied by a decrease in energy. For the comparatively stable (101) β‐Li₃N surfaces, the energy reductions are more pronounced, up to 3.41 eV. The less stable β‐Li₃N (100) surface also exhibits spontaneous H₂O decomposition upon adsorption, with an energy decrease of 3.93 eV (Figure S12).

To elucidate the observed spontaneous decomposition and significant energy reductions during H₂O interaction with various surfaces, we conducted a Bader charge analysis to assess the chemical reactivity of Li atoms on these surfaces. Table S1 summarises the electron loss from Li(1) and Li(2) atoms on the five surfaces before H₂O adsorption, with the electron count of an isolated neutral Li atom serving as the reference for comparison. The Li atoms on the α‐Li₃N (001) surface have lower electron loss, indicating weaker reactivity. On the other four surfaces, Li atoms lose more electrons and acquire a highly positive charges, suggesting a high reactivity that could promote the spontaneous decomposition of H₂O on these four surfaces.

Our calculations indicate that the reaction between Li₃N and H₂O on all examined surfaces leads to energy release, facilitating the formation of LiOH and NH₃. However, the catalytic reactivity of different surfaces significantly influences the reaction rates. Specifically, the relatively stable surfaces of the β‐phase exhibit superior activity in the decomposition of Li₃N via H₂O compared to the α‐phase surfaces. Among the α‐phase surfaces, only the α‐001 surface supports the spontaneous hydrolysis of Li₃N. Overall, the hydrolysis on β‐phase surfaces results in a greater release of energy than on α‐phase surfaces. This is consistent with the accelerated reaction rates of β‐phase surfaces observed in experiments.

Stage 2. DAC

In addition to the ammonia production reaction that occurs naturally in air, solid LiOH is produced simultaneously (Equation 2). Strikingly, as the ammonia reaction proceeded, the LiOH products reacted with CO₂ directly from the air, forming Li₂CO₃ (Equation 3, detailed in Experimental procedures). XRD results shown in Figure 5a elucidate peaks corresponding to Li₂CO₃, confirming the conversion of LiOH. A control experiment (Figure 5a) also confirms that purchased LiOH naturally converts to Li₂CO₃ during exposure to air for 24 h with slower kinetics compared to that of experimentally obtained LiOH. Subsequently, CO₂ can be released by adding H₂SO₄, which produces an aqueous electrolyte containing Li ions (Equation 4). Moreover, the H₂SO₄ used in this step is recycled from the anode reaction in Stage 3 (Schematic Figure 1), indicating the design of an environmentally friendly process for CO₂ capture and release. The released CO₂ gas is collected and analysed using GC equipment, which contains 99.9% CO₂ (Figure S13).

Figure 5.

CO₂ desorption and Li regeneration. a) XRD measurements of the reaction products between Li₃N and moisture air, including reference purchased powder XRD of Li₂CO₃ and control experiment for purchased LiOH. b) The comparison of DAC CO₂ uptake in this work and various typical and recently developed DAC agents. c) Schematic of the drainage gas collection process, in the Li regeneration cell, the acidic electrolyte is generated at the anode and subsequently injected into a flask contains Li₂CO₃, the product of DAC. The release of CO₂ gas is indicated by the formation of bubbles at the bottom of the flask. The CO₂ is collected via the drainage gas method, as shown by the direction of the arrows, and finally captured as pure CO₂. d) Regenerated Li was used to reproduce ammonia in the next chemical loop cycle. The produced ammonia was measured to calculate the efficiency of the Li regeneration in the previous step. The bar chart shows the calculated realistic Li regeneration efficiency when it was recycled for ammonia synthesis. The steady‐state current was used for FE calculation, and initial SEI‐formation‐induced transient states were excluded. The regeneration efficiency varied with current densities, as shown. e) SEM image of the cathode electrode surface after 1 h, with current densities of 2 mA cm⁻² and f) 1 mA cm⁻². g) N 1s XPS peak after nitridation of recycled Li and (h) C 1s XPS spectra from the cathode electrode surface after electrolysis showing the development of characteristic peaks of solid electrolyte interphase (SEI).

Meanwhile, the material featured a relatively high weight capacity for DAC. A calculated CO₂ uptake of 18.4 mmol CO₂ g⁻¹ was achieved at a productivity of 1.53 mmol h⁻¹ g⁻¹ (Supporting Information Notes 3), considering LiOH as the DAC material. Figures 5b and S14; Table S2 illustrate the CO₂ uptake for various typical and recently developed DAC technologies, including chemical absorption with alkaline or amino acid‐based aqueous solution, amino polymer with silica or resin and adsorption with Metal‐Organic Frameworks (MOFs).^{[ 33} , ³⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ , ⁵⁴ , ^{55 ]} Compared to the high‐performance aqueous alkaline DAC systems of Ca(OH)₂ with 13.5 mmol CO₂ g⁻¹, the synthesised material shows more than 36% enhancement of CO₂ adsorption. Moreover, compared with alkaline and amines as benchmarks of DAC, the method presented does not require aqueous media, supporting materials, or elevated temperatures for CO₂ capture and release, respectively.

Stage 3. Lithium Regeneration

To close the cycle, the mediated Li that was used for the NH₃ synthesis and DAC was regenerated on the cathode surface through the electrochemical conversion of Li⁺ ions to metallic Li. The catholyte is composed of 1M LiClO₄ in propylene carbonate (PC) solvent. At the anode, the reaction is engineered and devised so that the oxygen evolution reaction acts as an electron donor reaction while also generating sulfuric acid due to the production of proton at the anode. The experimentally produced H₂SO₄ subsequently was used to release CO₂, which in turn produced Li₂SO₄, serving as a lithium salt in anolyte (Figure 5c). This developed green chemical looping method for the regeneration of acid and anolyte are demonstrated in Figure 1 with a blue circular arrow. To prevent the mixing of deposited Li with aqueous anolyte, the transport of produced proton to catholyte and the free transport of lithium ions from anolyte to catholyte, Lithium‐Ion Conducting Glass‐Ceramics (LICGC) is utilised as a separator (Figure S15).

The electrochemical reactions are shown below:

$$\text{Cathode reaction}:{\text{Li}}^{+}+{\mathrm{e}}^{-}\to \text{Li}$$ (8)

$$\text{Anode reaction}:{\mathrm{H}}_2\mathrm{O}-2{\mathrm{e}}^{-}\to 2{\mathrm{H}}^{+}+1/2{\mathrm{O}}_2$$ (9)

$$\text{Overall reaction}:2{\text{Li}}^{+}+{\mathrm{H}}_2\mathrm{O}\to 2{\mathrm{H}}^{+}+1/2{\mathrm{O}}_2+2\text{Li}$$ (10)

The electrochemical performance of this cell was evaluated by LSV (Figure S16), revealing a linear curve when the voltage is greater than −5.3 V. This is consistent with the literature, indicating the high and dominant ohmic resistance from the ceramic membrane.^{[ 5 ]} The LSV was initiated based on the theoretical value of −3.86 V, considering the oxygen evolution potential at 0.82 V in aqueous Li₂SO₄ and with reduction potential of Li ions at −3.04 V.

Determining the FE of this reaction is challenging due to the high reactivity of newly deposited Li with air, rendering the quantification of metallic Li a complex process. One possible quantification approach is to utilise regenerated Li for the subsequent step of nitridation and ammonia release processes. Then by the quantification of ammonia, the amount of regenerated Li can be calculated, providing an FE representing the realistic efficiency of the designed chemical loop. Figure 5d illustrates the FE calculated under different current densities. The presented method for Li regeneration FE and current densities is comparable with highly regarded literature reports, yielding an FE of 44.4%.^{[ 8} , ²² , ²⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸ , ^{59 ]} This includes technologies involving nitrogen reduction and hydrogen uptake to produce NH₃. Our work presents an alternative to the use of H₂, which typically requires water splitting or fossil fuels for its production.

SEM analysis of the deposited Li on the Cu plate reveals a highly formless structure with complex layers of dendrites while the current exceeds 2 mA cm⁻² (Figure 5e). This indicates that the dendrite formation is challenging to avoid under high current density when using LiClO₄/PC as the cathode electrolyte for lithium deposition. The high surface area of SEI and dendrites consumes “live” Li and decreases the efficiency. When the current density lowered to 1 mA cm⁻², the surface of cathode exhibited a densely packed arrangement of uniform spheres or cylinders (Figure 5g). This surface configuration, where Li forms a denser SEI with reduced surface area, enhances the Faradaic efficiency. However, when current densities below 1 mA cm⁻² are applied, voids are observed on the electrode surface, indicating suboptimal experimental conditions (Figure S18). The morphology from our electrochemical deposition of lithium metals is consistent with the literature.^{[ 60} , ⁶¹ , ⁶² , ^{63 ]} XPS performed on the cathode surfaces (Figure 5g,h) after THF cleaning and after the lithium nitration process confirmed the presence of Li₃N and a corresponding SEI containing Li₂CO₃ and ROCO₂Li.^{[ 5 ]} The N1s peaks in the XPS spectrum were consistent with the Li₃N XPS patterns, confirming the Li₃N formation on the cathode as a precursor of ammonia. The presence of SEI limits the Li regeneration efficiency. The Li recovery efficiency is also confirmed through inductively coupled plasma‐optical emission spectroscopy (ICP‐OES) (Figure S19). The results show that all remaining Li‐containing compounds on the cathode surface after the reaction can be almost completely recycled by dissolving them in aqueous solution and using them as an anode electrolyte in the cycled loop.

Specifically, the synthesis rate of NH₃ was quantified. By keeping 1 kg of Li element in the loop, an overall production rate of approximately 150 kg of NH₃ per annum can be achieved (Figure S20). The synthesis rate of Li₃N was determined to be 3.2 mmol h⁻¹ g_Li ⁻¹ while the subsequent NH₃ release rate was 1.2 mmol h⁻¹ g_Li3N ⁻¹ (Supporting Information Notes 3). Compared with recent reports on NH₃ synthesis from N₂ in aqueous systems, this rate is considerably high (Table S3). Additionally, the calculated regeneration rate of metallic lithium in Stage 3 was 4.5 µmol h⁻¹ cm⁻². Notably, the reaction rates for nitridation, NH₃ release, and direct air capture (DAC) are all within the same order of magnitude. To achieve a rate‐matching electrolysis step, the electrode surface area would need to be scaled to approximately 1000 cm² (roughly equivalent to the area of an A4 sheet) per gram of lithium.

In our devised green ammonia production process, all the reaction steps except Li recycling are spontaneous and performed at or near room temperature. Remarkably, sulfuric acid is produced during the Li recycling process, which can be reused in CO₂ desorption and consequently as an electrolyte in the electrodeposition of Li. Alkaline LiOH is also generated as a byproduct in the loop with exceptional CO₂ uptake for the DAC process. The DAC process does not require additional energy for absorption and regeneration reactions, respectively (Figure 6a).^{[ 64 ]} In contrast, conventional DAC processes are energy intensive, necessitating significant energy for the regeneration of adsorbents or absorbents (e.g., by heating up to decompose and release CO₂ or via pressure swing to desorption). The average energy consumption of DAC is approximately 10 GJ per ton of CO₂.^{[ 65} , ^{66 ]} In our scenario, 3.4 tons of CO₂ are captured per ton of NH₃ produced (Figure 6b), which indicates an energy saving of ∼34 GJ per ton of NH₃ produced when coupled with DAC (Supporting Information Notes 4). Furthermore, with the potential improvements in the Faraday efficiency and DAC yield, we calculated the gross energy consumption and the net energy consumption when incorporating CCS (Figure 6c,d).^{[ 67 ]} The combined energy consumption for the integrated NH₃ synthesis and DAC is lower than that for independent DAC on the basis of per unit mass of CO₂ captured (Figure 6d). This result suggests our novel carbon negative ammonia synthesis technology can be also considered as a low‐cost DAC process with free ammonia production, promising transformational pathway for industrial applications.

Figure 6.

Energy analysis for NH₃ production and carbon capture and storage (CCS). a) Energy profiles of the three‐stage Li‐assisted chemical loop. For each stage of the chemical looping process, the enthalpy is calculated based on the experimentally derived standard enthalpy of formation. The key Li compounds are labelled for each step, the dashed arrows indicate other compounds involved in the loop, the substances in blue represent chemicals supplied externally to the loop and the pure CO₂ and NH₃ in red font represent the products harvested from the loop. A more detailed figure is included in Figure S21. b) CO₂ emissions of different NH₃ production processes. c) Gross energy cost for different NH₃ production processes. d) Net energy footprint including CCS to cover the CO₂ emission for different NH₃ production processes. (Commercial: current commercial NH₃ production in average, others are the best available technology; NG: natural gas; SMR: steam methane reforming; ATR: auto‐thermal reforming; Theo. Max.: the theoretical maximum CO₂ capture amount and maximum efficiency in this work.)

Conclusion

In conclusion, we successfully developed an innovative NH₃ synthesis process with a combination of the DAC process to achieve carbon‐negative NH₃ production from the air. Briefly, we produce NH₃ and pure CO₂ from the air and achieve lithium source recycling using the electrochemical lithium deposition methodology. This work demonstrated an excellent ammonia production rate of 4.5 µmol h⁻¹ cm⁻² and recorded high direct air CO₂ uptake of 18.4 mmol CO₂ g⁻¹. In this newly developed carbon‐negative ammonia production methodology, Li is used as a nitrogen carrier to synthesize ammonia from N₂. The direct moisture adsorption and reaction energies on different surface facets of produced α/β‐Li₃N are fundamentally investigated, shedding light on enhanced kinetics observed from experimental hydrolysis of Li₃N to produce NH₃. All these chemical processes operate at room temperature and ambient pressure except the lithium nitridation, which was performed at 80 °C. The devised process demonstrates how the exorbitant synthesis conditions required for ammonia synthesis can be met under moderate conditions with a net negative CO₂ emission. Compared with emerging electrochemical approaches for NH₃ synthesis, the demonstrated process overcomes the competition of hydrogen evolution reaction, avoids the consumption of liquid water or sacrificial agents, and provides an effective pathway for DAC, which paves the way for NH₃ synthesis from carbon neutral to carbon negative.

Supporting Information

The authors have cited additional references within the Supporting Information.^{[ 17} , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ³³ , ³⁴ , ⁴⁵ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ , ⁶⁶ , ⁶⁷ , ⁶⁸ , ⁶⁹ , ⁷⁰ , ⁷¹ , ⁷² , ⁷³ , ⁷⁴ , ⁷⁵ , ⁷⁶ , ⁷⁷ , ⁷⁸ , ⁷⁹ , ⁸⁰ , ⁸¹ , ⁸² , ⁸³ , ⁸⁴ , ⁸⁵ , ⁸⁶ , ⁸⁷ , ⁸⁸ , ⁸⁹ , ⁹⁰ , ⁹¹ , ^{92 ]}

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Supporting Information

Supporting Information

Wang D., Yan X., Guo J., Qu L., Wu C., Liu J. Z., Zavabeti A., Li G. K., Angew. Chem. Int. Ed. 2025, 64, e202423934. 10.1002/anie.202423934

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.