Climate and Air Quality Impact of Using Ammonia as an Alternative Shipping Fuel

As carbon-free fuel, ammonia has been proposed as an alternative fuel to facilitate maritime decarbonization. Deployment of ammonia-powered ships is proposed as soon as 2024. However, NO x , NH 3 and N 2 O from ammonia combustion could impact air quality and climate. In this study, we assess whether and under what conditions switching to ammonia fuel might affect climate and air quality. We use a bottom–up approach combining ammonia engine experiment results and ship track data to estimate global tailpipe NO x , NH 3 and N 2 O emissions from ammonia-powered ships with two possible engine technologies (NH 3 –H 2 (high NO x , low NH 3 emissions) vs pure NH 3 (low NO x , very high NH 3 emissions) combustion) under three emission regulation scenarios (with corresponding assumptions in emission control technologies), and simulate their air quality impacts using GEOS–Chem High Performance global chemical transport model. We find that the tailpipe N 2 O emissions from ammonia-powered ships have climate impacts equivalent to 5.8% of current shipping CO 2 emissions. Globally, switching to NH 3 –H 2 engines avoids 16,900 mortalities from PM 2.5 and 16,200 mortalities from O 3 annually, while the unburnt NH 3 emissions (82.0 Tg NH 3 yr -1 ) from pure NH 3 engines could lead to 668,100 additional mortalities from PM 2.5 annually under current legislation. Requiring NH 3 scrubbing within current Emission Control Areas leads to smaller improvements in PM 2.5 -related mortalities (22,100 avoided mortalities for NH 3 –H 2 and 623,900 additional mortalities for pure NH 3 annually), while extending both Tier III NO x standard and NH 3 scrubbing requirements globally leads to larger improvement in PM 2.5 -related mortalities associated with a switch to ammonia-powered ships (66,500 avoided mortalities for NH 3 –H 2 and 1,200 additional mortalities for pure NH 3 annually). Our findings suggest that while switching to ammonia fuel would reduce tailpipe greenhouse gas emissions from shipping, stringent ammonia emission control is required to mitigate the potential adverse effects on air quality.


Introduction
Maritime shipping burns fossil fuels in large diesel engines for energy (propulsion, heat, and electricity), which leads to emissions of CO2 and air pollutants.The main air pollutants emitted by the maritime transport sector include SOx (º SO2 + SO4 2-), NOx (º NO + NO2), nonmethane volatile organic compound (NMVOC), CO and carbonaceous aerosols.These are either components or precursors of particulate matter (PM) and ozone (O3).Exposure to PM, particularly the fine PM (aerodynamic diameter < 2.5 µm, named PM2.5)that can reach deep inside the respiratory tract, is estimated to have caused 3.7 -4.8 million deaths in 2015 by increasing the risk of cardiopulmonary and cerebrovascular diseases (Cohen et al 2017).O3 exposure exerts oxidative stress on the respiratory tract (Nuvolone et al 2018), which also leads to increased risk of cardiopulmonary diseases, and therefore another 1.04 -1.24 millions of respiratory deaths in 2010 globally (Malley et al 2017).Shipping emissions are estimated to account for 2.7% of global energy-related CO2 emissions and caused an estimated 84800 -103000 annual premature deaths from PM2.5 exposure globally in 2015 (Zhang et al 2021b), and account for up to 14 and 25% of PM2.5 concentration over East Asia and Mediterranean area, respectively (Contini and Merico 2021).
The International Maritime Organization (IMO) has outlined a goal of reducing greenhouse gas (GHG) emissions from international shipping by at least 40% by 2030 compared to the 2008 level (International Maritime Organization 2018).The uses of alternative fuels (e.g.NH3, H2, methanol) and other energy solutions (e.g.electrification) are essential for reaching such a decarbonization goal (Balcombe et al 2019).NH3 is one of the main candidates for alternative maritime fuels, and could represent up to 43% of the energy mix of shipping in 2050 (IRENA 2021).Since NH3 is mainly manufactured with H2 and N2 through the Haber-Bosch Process, the carbon footprint of NH3 production can be reduced by carbon capture (blue NH3), or using renewable energy for N2 and H2 production and the synthesis process (green NH3) (Valera-Medina et al 2021).Wolfram et al (2022) and Bertagni et al (2023) summarized scientific concerns about the potential environmental impacts of using NH3 as a marine fuel.NH3 combustion may generate additional NOx and N2O compared to other fuels (Hinokuma and Sato 2021).NH3 emission is one of the major source of global PM2.5 pollution (e.g.Gu et al 2021) by neutralizing H2SO4 and HNO3 in the atmosphere (Jacob 1999).Heo et al (2016) find that NH3 emission leads to much higher PM2.5 mortality costs per ton ($23000 -66000) than SO2 ($14000 -24000) and NOx ($3800 -14000) in the United States.These show the potential danger of uncontrolled NH3 emission via worsening PM2.5 air quality.Emitted NOx and NH3 would then deposit to Earth's surface, causing damages to ecosystems (e.g.soil acidification and eutrophication) and may lead to additional emission of N2O, which is a potent greenhouse gas and contributes to stratospheric ozone depletion.
Here, we explore the possible ranges of air quality and climate impacts of transitioning from using fossil fuels to ammonia as the major shipping fuel under different technologies and policies, aiming to highlight the opportunities and challenges of ammonia combustion as a strategy to decarbonize maritime transport.

Method
We use a bottom-up approach to estimate the global NOx, NH3 and N2O emissions from converting the entire fleet into NH3-powered ships as a function of engine technologies, emission control strategies and policy under 6 scenarios, using results from ammonia engine experiments and ship Automatic Identification System (AIS) data.We then simulate the associated changes in O3 and PM2.5 air quality using a global 3-D chemical transport model (GEOS-Chem High Performance).Finally, we estimate the impacts of simulated changes in O3 and PM2.5 on public health (expressed in annual premature mortalities) using concentration functions derived from epidemiological studies.

Scenario Name
Emission control inside current ECA  1. Description of the engine technology and policy scenarios considered in this study.SCR refers to Selective Catalytic Reduction (assumed to be 90% effective), which converts NOx and NH3 into N2 in 1:1 ratio under ideal conditions.NH3 scrubbing is assumed to remove 95% of NH3 slip after SCR.
In all scenarios, we apply an AIS-based shipping emission model (Zhang et al 2019) to estimate the global spatially-resolved pollutant and GHG emissions for every ship track in 2015 following the technology and policy assumptions of each scenario.The emission model calculates ship emissions as a function of engine power demand, ship specifications, emission factors (EF) and activity time.Missing entries in ship specifications are filled based on the lengths and capacities of the associated ships.
Table 1 shows the scenario design of our study.We choose the emission scenario with 0.5% cap on fuel sulphur content from Zhang et al (2021b) as our baseline.The "post-2020 NOx baseline" scenario imposes the most stringent IMO NOx emissions (Tier III) limit on top of baseline scenario, which represents the emissions from fossil fuel powered ships if all of them DO NOT CITE, COPY OR DISTRIBUTE DO NOT CITE, COPY OR DISTRIBUTE D R A F T were retrofitted to follow IMO emission standards for newly-built ships.6 counterfactual scenarios are designed to examine the possible range of air quality outcomes from total conversion to ammonia-powered ships given the possible engine technologies (and therefore emission management strategies) and emission regulations (current legislation versus additional NH3 emission regulations).Reduction (SCR).The derivations of EF and load dependences for the two types of engines, and a discussion about the uncertainty in engine technologies are given as Supplemental Information.
Given the uncertainty in ammonia engine designs, the engine technology scenarios do not intent to realistically replicate how ammonia combustion would be implemented on ships.Rather, the two engine technologies considered in our study reflects two extremes of, and therefore provide bounding scenarios for NOx and NH3 emission management approaches: 1) with pure NH3 engine having low NOx (currently regulated) and very high NH3 (currently unregulated) emissions, versus 2) NH3-H2 engine that strictly maintains the NOx/NH3 ratio to allow SCR to simultaneously control both pollutants.
We consider three policy scenarios.The first ("2020") follows the IMO regulations as of 2020.The untreated NOx EF are 32.7 g/kWh for NH3-H2 and 7.08 g/kWh for pure NH3 engines following the load corrections prescribed by IMO (International Maritime Organization 2008) (fig.1).Current IMO guidelines (International Maritime Organization 2017) cap NOx EF for new vessels at 7.7 -14.4 g/kWh (Tier II limit) when operating outside the Emission Control Area (ECA, mostly includes North America and United States Caribbean Sea as of 2020, and additionally Baltic Sea and North Sea in 2021) and 2 -3.4 g/kWh (Tier III limit) within ECA, depending on the engines' rated speed.Compliance with such a guideline would require SCR that can remove 90% of NOx to operate globally for NH3-H2 and within ECA only for pure NH3 engines.The second ("NH3_ECA_LIMIT") assumes that additional NH3 scrubbing requirements (assumed to be 95% effective from available technology) (Melse and Ogink 2005, Van der Heyden et al 2015, Boero et al 2023) are implemented within ECA for both types of engines, while the third ("GLOB_LIM") extends Tier III NOx compliance and NH3 scrubbing requirements to the whole globe.

Atmospheric Chemistry Modeling
We use version 13.

Health Outcome
We estimate the impacts of air quality changes on public health using the global gridded population data at 30 arc-second resolution from the Gridded Population of the World version 4.11 (Center for International Earth Science Information Network -CIESIN -Columbia University 2018).Country-level age distribution and baseline mortality rates are provided by the World Health Organization (WHO) (WHO 2018).We estimate the risk of relative mortality from chronic O3 and PM2.5 exposure under the baseline (RRbase) and each alternative scenario i (RRi) for every age group.The change in the annual mortality for scenario i (ΔMorti) due to some disease for that age group is then calculated for each grid cell as: where Mortbase is the number of mortalities due to that disease in 2016.The relative risk is calculated by comparing the simulated exposure-relevant concentration under scenario i to that under the baseline scenario using an appropriate concentration response function (CRF).We use a log-linear CRF for O3 from Turner et al (2016), which estimate a 12% increase (95% confidence interval (CI): 8.0 -16%) in respiratory mortality per 10 ppb increase in annual mean maximum daily 8-hour average (MDA8) O3 concentration.For PM2.5 we estimate RR for noncommunicable diseases and lower respiratory infections using the age-specific non-linear CRFs from the Global Exposure Mortality Model (Burnett et al 2018).
We estimate the median and 95% confidence interval of changes in mortalities due to O3 and PM2.5 for each scenario by performing 1,000 random draws of the CRF parameters in a paired Monte-Carlo simulation.Table 2 shows the modelled global annual shipping emissions of NOx, NH3 and GHG under different scenarios, and Figure 2 shows the spatial distribution of NOx emissions.Under current regulations ("2020"), ammonia-powered ships have lower NOx emissions (4.4 Tg NOx/yr for NH3-H2 and 6.9 Tg NOx/yr for pure NH3).Such comparison mostly reflects regulatory rather than technological differences, since the older ships in the baseline scenario do not follow the newer and more stringent (Tier II or Tier III) NOx regulations, while all newly built ammoniapowered ships abide the Tier II regulation outside ECA and Tier III regulations within ECA.To comply with Tier II NOx regulations, SCR is required for the NH3-H2 engine while no NOx control is needed for the pure NH3 engine.This leads to higher total post-treatment NOx emissions from pure NH3 engines than that from NH3-H2 engines, despite pure NH3 engines has  Figure 3 shows the spatial distribution of modelled NH3 emissions under different technology and policy scenarios.Under current regulations ("2020"), switching to NH3-H2 engines leads to 2.5 Tg/yr NH3 emissions, while switching to pure NH3 engines leads to NH3 emissions (82.0 Tg/yr) that are 32.8 times higher than that from NH3-H2 engines.For pure NH3 engines, SCR can only remove 7% of NH3 from engine exhaust, leading to high tailpipe NH3 emissions.In the "NH3_ECA_LIM" scenario, which requires NH3 scrubbing over ECA (mostly North American coast and northern Europe), global NH3 emissions reduce by 12% for both NH3-H2 (2.2 Tg/yr) and pure NH3 (71.7 Tg/yr ) engines.In the "GLOB_LIM" scenario, with both SCR and NH3 scrubbing are required globally, NH3 emissions fall to 0.1 Tg/yr for NH3-H2 engines and 3.9 Tg/yr for pure NH3 engines.
Table 2 also shows the long-lived GHG emissions from each scenario, given as the equivalent amount of CO2 (CO2,e) in terms of 100-year Global Warming Potential (GWP100) using a conversion factor of 273 from N2O emission to CO2,e (Smith et al 2021).CO2,e from the baseline scenario does not include GHG other than CO2 (mainly CH4 and N2O), which contribute to less than 3% of global shipping CO2,e during 2013 -2015 (Olmer et al 2017).We find that the tailpipe CO2,e from the ammonia-powered fleet is 5.8% of that from the current fossil-fuelpowered fleet.Our analysis (see Supplemental Information) also shows that the "secondary N2O emissions" from reactive nitrogen deposition (Wolfram et al 2022) is not a problem for NH3-H2 engine as the total reactive nitrogen emissions are lower than current fleets.For pure NH3 engine, the net climate effects from nitrogen deposition are likely to be smaller than reduction in tailpipe GHG emissions (817.2Tg CO2,e/yr) from switching to ammonia-powered ships, showing the potential of blue and green ammonia as a climate-friendly shipping fuel, though considerable uncertainties exist on how CO2 uptake and N2O emissions respond to nitrogen deposition.This In addition, we find substantial sensitivity of O3 response to assumptions in ship plume chemistry (mainly NOx lifetime, see Supplemental Material), which could be a major source of uncertainties.This shows the importance of understanding the plume chemistry of NH3 ship in capturing the O3 response.) for all ammonia-powered ships scenarios.

Scenario
Table 3 shows the changes in annual global mortality attributable to O3 (DMO3) and PM2.5 (DMPM2.5)for each scenario.We estimate that current shipping emissions leads to 87,400 and 16,900 mortalities from PM2.5 and O3, respectively.The lower NOx emissions from ammoniapowered ships provide significant O3 air quality benefit, reducing annual O3-related mortality by 12,600 to 73,100.Despite the lack of primary PM (BC, OC) and secondary PM precursors (SO2, NMVOC) emissions other than NOx and NH3, ammonia-powered ships lead to worse DMPM2.5 (-22,100 to +668,100) than fossil fuel powered ships with similar NOx regulation ("Post-2020 NOx Baseline", -46,200) except the scenario with lowest NH3 emissions ([NH3-H2]GLOB_LIM), -66,500).This highlights the importance of NH3 as a PM2.5 precursor in coastal environment, and therefore minimizing tailpipe NH3 emission to mitigate the negative air quality impacts from ammonia-powered ships.
Under currently legislation ("2020"), switching to NH3-H2 engine reduces annual global mortalities from PM2.5 (16,900) and O3 (16,200) in comparable magnitudes.While providing substantial benefits from reducing O3-related mortality (-73,100) .5-relatedmortality (+668,100).Since current ECA are mostly over North America and northern Europe, additional NH3 emissions control over current ECA ("NH3_ECA_LIM") only provides marginal benefits in terms of PM2.5-related mortalities (5,200 (31%) for NH3-H2 engines and 44,200 (7%) for pure NH3 engines) since most of the increases in PM2.5 occur overs East Asia, North Africa, Southeast Asia and Mediterranean region.In contrast, when both Tier III NOx and NH3 emission controls are extended globally ("GLOB_LIM"), the negative impacts of pure NH3 drivetrains on PM2.5 (1,200 additional mortalities) can be mitigated to a level that could be offset by the benefits on O3 (22,400 avoided mortalities).For NH3-H2 engines, the low NH3 emissions, and therefore global reduction in PM2.5 level, lead to substantial reduction in PM2.5-related mortalities (-66,500) equivalent to 79% of that from current shipping emissions.

Discussion
Using blue and green NH3 to facilitate decarbonization of maritime transport has been gaining traction among the industry, while concerns have been raised about the consequences (e.g.secondary N2O emissions, air pollution, eutrophication, soil acidification) of such large additional reactive nitrogen production and emission into the Earth System (Baessler et al 2019, Wolfram et al 2022).Despite the uncertainties in the drivetrain design, fuel mix, emission factors and plume chemistry of ammonia-powered ships as they are not yet deployed in real world, an early evaluation using currently available information can provide information to help stakeholders identify the potential climate and air quality issues and formulate mitigation measures.
We combine results from engine experiments and ship activity data to estimate the possible GHG and air pollutant emissions and impacts from ammonia-powered ships.We find that the GWP attributable to tailpipe N2O emissions from ammonia-powered fleet is a small fraction (5.8%) of that of the current fleet.Our findings confirm the potential of blue and green NH3 as a climate-friendly shipping fuel.However, the impacts of large reactive nitrogen deposition over land ecosystems on GHG balance remain highly uncertain.
We find that the public health impacts of switching from fossil fuel to ammonia depends largely on the technology and policy choices.If tuned to balance NOx and NH3 concentration from engine exhaust to allow simultaneous reduction of NOx and NH3 emissions using welloptimized exhaust post-treatment systems with highly efficient combustion modes, deployment of ammonia combustion technology can lead to net health benefits by reducing both O3 and PM2.5 levels.If the engines are tuned to have lower NOx emissions than NH3-H2 combustion, which is more compatible with current NOx-focused regulatory framework, the unburnt NH3 emission, if unmitigated, can lead to large increases in PM2.5, and consequently 668,100 additional global PM2.5-related mortalities annually.Imposing NH3 emission regulation over current ECA only mitigates 7% of the increases in annual PM2.5-related mortalities from pure NH3 engines, since the largest negative impacts are expected over East Asia, which is not Our study assumes total conversion to ammonia-powered ships, while in reality ammonia-powered ships will operate alongside SOx-emitting fossil fuel powered ships, which would increase the sensitivity of PM2.5 to NH3 emissions.This shows the urgency of updating shipping emission regulations in anticipation of the real-world deployment of ammoniapowered-ships. Particularly, given the availability of effective (> 95%) NH3 removal strategies, priority should be given towards developing and enforcing working NH3 emission regulations.More stringent control of SOx and NOx emissions, which is foreseeable in the future, could be another viable strategy to reduce the PM2.5 formation from unburnt NH3 emissions (Bauer et al 2016).
The practicality and efficacy of SCR for ammonia engines remain highly uncertain.The lack of sulfur and particulate poisoning of catalyst, and not requiring a separate NH3 source to operate could potentially lead to cheaper SCR operation since catalyst and urea recharge are estimated to account for at least 61% of the total cost of SCR ownership and operation (Zhang et al 2021a).However, NH3 combustion generates more H2O than diesel combustions (see Supplemental Information), which limits the efficacy of SCR (Kuta et al 2023, Xiang et al 2024).Excessive tailpipe N2O emissions can result from mistuned SCR and ammonia oxidation systems (Yates et al 2005), which could potentially offset the climate benefits.Optimizing the SCR systems for ammonia engines is crucial to limiting their potential air quality and climate impacts.
Our study shows the feasibility of NH3 to be a climate-friendly shipping fuel despite the concern of tailpipe N2O emission, and highlights the adverse effects of unburnt NH3 emissions on PM2.5 air quality, which can be mitigated by emission control measures feasible under current technology.Apart from tailpipe emissions, NH3 leakages also occur over the whole value chain (e.g.production, distribution, bunkering, fueling) (Bertagni et al 2023), which can deteriorate the PM2.5 air quality over localities near the NH3 supply chain if unabated (Rathod et al 2023).Development and enforcement of new NH3 emission regulations is critical for ammonia-powered ships to provide positive impact on air quality and prevent negative impacts from excessive nitrogen deposition, alongside reducing GHG emissions.

Figure 1 .
Figure 1.Load-corrected NH3 and NOx emission factors (EF) of pure NH3 and NH3-H2 engines, as a function of emission control strategy.Red bar ("Engine") refers to EF from completely untreated engine exhaust.Blue (Post-SCR) and green bars (Post-SCR + NH3 Scrubbing) refer to EF after implementations of emission control measures.SCR and NH3 scrubbing are done sequentially.Red dotted lines indicate IMO NOx regulations for slow engine speed (<130 rpm), which is typical for large engine.We consider the emissions from ammonia-powered ships with two types of engine technologies.The first type of engine technology considered is pure NH3 combustion (Mounaïm-Rousselle et al 2022).The second type ("NH3-H2" ) is proposed by Imhoff et al (2021) based on the experimental data from Lhuillier et al (2020).Part of the NH3 is transferred to a catalytic NH3 cracker to generate H2 to improve combustion.This balances NH3 and NOx concentration in engine exhaust, allowing both NOx and NH3 emissions to be controlled by Selective Catalytic 4.1 of the GEOS-Chem High Performance model (GCHP, https://doi.org/10.5281/zenodo.4429193)(Martin et al 2022, Eastham et al 2018) to simulate the response of O3 and PM2.5 to pollutant emission changes in each scenario through resolving the chemistry, transport, emission and deposition of relevant chemical species.The model is driven by the Modern-Era Retrospective analysis for Research and Application (MERRA-2) assimilated meteorological fields (Gelaro et al 2017).The model is run at a horizontal resolution of ~200km in cubed-sphere configuration (C48) from 1 st Oct 2018 to 31 st Dec 2019, with the first 3 months of output discarded as spin-up.O3 is simulated from a coupled O3-NOx-VOCs-CO-halogenaerosols chemical mechanism (Sherwen et al 2016).Anthropogenic emissions are from Community Emission Data System (Hoesly et al 2018) except the shipping sector.Biogenic VOCs, soil NOx and sea salt aerosol emissions follow Weng et al (2020) and dust emissions follow Meng et al (2021).Re-emissions of deposited NOx and NH3 are not considered.Formation of secondary inorganic aerosols are simulated by the ISORROPIA II, which considers thermodynamic equilibrium of the NH4 + -Na + -SO4 2− -NO3 − -Cl − -H2O (Fountoukis and Nenes 2007).PM2.5 concentrations are derived by summing the mass of its constituents at standard conditions to align with the sampling standard used by the United States Environmental Protection Agency (Latimer and Martin 2019).Ship plume chemistry is parameterized by the DO NOT CITE, COPY OR DISTRIBUTE DO NOT CITE, COPY OR DISTRIBUTE D R A F T PARANOX scheme (Vinken et al 2011).Model evaluation is provided as Supplemental Information.

Figure 3 .
Figure 3. Spatial pattern of annual total NH3 emissions (kg m -2 yr -1 ) under different scenarios Figure 4. Changes in annual mean MDA8 O3 concentration (DO3, ppb) for different ammoniapowered ship scenarios Figure 4 shows the modelled global changes in annual mean MDA8 O3 due to converting current fleet to ammonia-powered ships with different technology and policy options.Generally, the lower NOx emissions from ammonia-powered ships reduce annual mean MDA8 O3.Under all scenarios, global population-weighted average MDA8 O3 decreases (-0.27 ppb for [NH3-H2]2020, -1.13 ppb for [Pure NH3]2020, -0.37 ppbv for [Pure NH3]GLOB_LIM).The greatest reductions in population-weighted O3 are simulated over coastal and island nations (e.g.1.5 to 1.9 ppb for Sri Lanka and Djibouti, 1.4 to 2.2 ppb for Panama, 1.4 to 1.7 ppb for Jamaica).However, over highly NOx-saturated coasts near northern China, northern Europe, and Persian Gulf, local increases in surface O3 are simulated, especially under the scenarios with greater NOx reductions ([NH3-H2]2020 and [Pure NH3]GLOB_LIM).Over North Sea, the NOx-saturation leads to further increases in MDA8 O3 as NOx emissions become lower, increasing the populationweighted O3 from 1 ppb under [Pure NH3]2020 to up to 1.5 ppb under [Pure NH3]GLOB_LIM over the Netherlands.Over East Asia, population-weighted MDA8 O3 decreases by 2.4 ppb under the scenario with least NOx reduction ([Pure NH3]2020), but increases by 0.2 ppb under [Pure NH3]GLOB_LIM and [NH3-H2]2020 as NOx emissions become lower.This shows the importance of local chemical environment in controlling the response of O3 pollution to marine NOx control.
, switching to pure NH3 engines DO NOT CITE, COPY OR DISTRIBUTE DO NOT CITE, COPY OR DISTRIBUTE any ECAs.Extending stringent control of NOx and NH3 emissions to the globe provides substantial air quality benefits.

Table 2
Modelled global total nitrogen-based air pollutants (in Tg/yr) and GHG emissions (in Tg CO2,e/yr) from different scenarios.CO2,e (equivalent amount of CO2 in terms of 100-year Global Warming Potential) is calculated as CO2 emissions + (N2O emissions ´ 273).