Heavy-duty engines are the hardest part of road, rail and marine transport to decarbonise. Their duty cycles demand large amounts of stored energy, and the weight and charging time of batteries become limiting as range and payload grow. Attention has therefore turned to fuels that can be made from renewable electricity and burned in engines broadly similar to those in service. Hydrogen is the simplest such fuel, but it is costly to store and move. Ammonia, a compound of nitrogen and hydrogen, offers a way of carrying hydrogen in a form that liquefies under modest pressure or mild refrigeration (Zamfirescu and Dincer, 2008; Valera-Medina et al., 2018).

Ammonia also has an industrial history that hydrogen lacks. It has been synthesised by the Haber-Bosch process for more than a century, chiefly for fertiliser, and that industry has shaped world agriculture (Smil, 2001; Erisman et al., 2008). Pipelines, tanks, ships and handling codes already exist. Production today relies largely on hydrogen from natural gas and is carbon intensive, so ammonia is a low-carbon fuel only if the hydrogen comes from electrolysis or if the carbon dioxide is captured, a transition whose technical routes are set out by MacFarlane et al. (2020).

Whether ammonia can then be burned efficiently and cleanly in a compression ignition engine is a separate question, and it is the subject of this review. The compression ignition engine is the natural target because it dominates heavy-duty service and because its efficiency is the benchmark any replacement must approach.

The article is a narrative review of published work: the technical papers from the first period of interest in the 1960s, the engine studies that followed renewed interest from about 2008, the combustion science reviews that summarise flame and kinetic data, and earlier reviews of ammonia engines (Dimitriou and Javaid, 2020; Chiong et al., 2021; Kurien and Mittal, 2022). No formal search protocol was applied and no claim of completeness is made. Spark ignition studies are drawn on only where they clarify matters common to both engine types.

At atmospheric pressure ammonia boils at about minus 33 °C, and at ordinary ambient temperatures it remains liquid under a pressure of roughly ten bar, conditions similar to those for liquefied petroleum gas. Its lower heating value is less than half that of diesel fuel per unit mass, so tanks must be larger and injected quantities greater. Because ammonia needs much less air per unit mass than a hydrocarbon, the energy content of a stoichiometric cylinder charge is nonetheless close to that of conventional fuels, and power density need not suffer greatly (Kobayashi et al., 2019).

The combustion properties are less accommodating. The autoignition temperature of ammonia is far above that of diesel fuel, its flammability range in air is narrow, and its laminar burning velocity is a small fraction of that of common hydrocarbons, as flame measurements over a range of pressures confirm (Hayakawa et al., 2015). Its latent heat of vaporisation is several times that of diesel, which cools the charge strongly when liquid ammonia is injected and further delays ignition. An engine that relies on compression alone to ignite ammonia therefore faces a fuel that is hard to light and, once lit, slow to burn. Figure 1 traces how these properties constrain the choices made in combustion, emissions control and vehicle design.

Ammonia has been burned in engines before. Kroch (1945) described motor buses operated in Belgium during the Second World War on ammonia with coal gas as a combustion promoter, when diesel fuel was unobtainable. Systematic research came in the 1960s, when the United States Army examined fuels that could be synthesised in the field. Starkman et al. (1966) analysed ammonia in spark ignition engines, and two studies addressed compression ignition directly.

Gray et al. (1966) found that neat ammonia could be made to ignite by compression only at compression ratios far beyond normal diesel practice, and that at ordinary ratios it needed a pilot injection of a more ignitable fuel or some other ignition aid. Pearsall and Garabedian (1967) reached the same conclusion and reported satisfactory running with a diesel pilot, as well as with spark ignition in a converted diesel engine. These papers established the framework that later work has refined: ammonia in a compression ignition engine is, in practice, a dual-fuel problem.

The strategies studied since differ in how the ammonia is introduced and in what ignites it. Table 1 summarises the main approaches, which are discussed below.

An ammonia engine emits no carbon dioxide from the ammonia itself, but the nitrogen in the fuel opens reaction paths that a hydrocarbon engine lacks. Nitrogen oxides form not only by the thermal route from atmospheric nitrogen but also from the oxidation of ammonia and its fragments, through chemistry described in the classic account of Miller and Bowman (1989) and updated by Glarborg et al. (2018). The same chemistry works in reverse: ammonia and its radicals reduce nitric oxide to nitrogen within a certain temperature window, which is the basis of selective non-catalytic reduction. Engine-out nitrogen oxides from ammonia combustion therefore depend on mixture strength and temperature history in a less straightforward way than in diesel combustion.

Nitrous oxide deserves separate mention. It is an intermediate of fuel-nitrogen chemistry that survives where the gases cool before it can decompose, as happens in lean, low-temperature regions and near walls. It is a greenhouse gas far more potent than carbon dioxide per unit mass and is now the most important ozone-depleting emission from human activity (Ravishankara et al., 2009). Even small exhaust concentrations can therefore cancel a meaningful part of the climate benefit of removing carbon from the fuel. Studies that report carbon dioxide alone give an incomplete account, and nitrous oxide should be measured as a matter of routine.

Heavy-duty diesel engines already carry selective catalytic reduction systems that use ammonia, released from urea solution, to convert nitrogen oxides to nitrogen and water (Koebel et al., 2000). An ammonia-fuelled engine has the reductant on board and often in the exhaust already, so the same catalyst technology is the obvious means of dealing with nitrogen oxides and, with an oxidation stage downstream, with slip. The balance between ammonia and nitrogen oxides in the exhaust varies with load, however, and catalysts that remove nitrous oxide at exhaust temperatures are less mature.

Ammonia attacks copper, brass and zinc and some elastomers, which rules out materials common in diesel fuel systems, and its interaction with lubricating oil over long service is not well documented. It is toxic and has a sharp odour detectable at low concentrations, which gives warning but does not remove the hazard of a large release in an enclosed space. Industrial handling practice is well developed (Valera-Medina et al., 2018). Carrying that practice over to vehicles serviced in ordinary workshops, or to ships with crew quarters near the machinery, is a task for regulation as much as for engineering.

Three gaps stand out. The first is the absence of durability evidence: published engine tests are short, and little is known about injector wear, deposit formation, oil degradation or catalyst ageing over the service life expected of a heavy-duty engine. The second is transient behaviour. Dual-fuel and direct-injection results come almost entirely from steady operating points, whereas slip and nitrous oxide are likely to be worst during cold starts and load changes. The third is the pilot fuel. Diesel pilots leave a residual carbon emission, and how small a pilot of renewable origin can be made while still giving reliable ignition across the operating range has not been settled.

A broader question concerns the comparison with alternatives. Ammonia competes with hydrogen, methanol and batteries, each with its own losses between renewable electricity and the wheels or propeller. The engine literature can inform that comparison only if efficiency and all nitrogen-containing emissions are reported on a common basis.

The literature shows that ammonia can be burned in compression ignition engines and that it cannot yet be burned alone under practical conditions. A pilot or promoter fuel is required, and the choice among diesel, dimethyl ether and hydrogen, together with the choice between premixing and direct injection of the ammonia, determines efficiency, slip and nitrogen oxide formation. The carbon benefit is real but must be weighed against nitrous oxide and unburned ammonia, both of which demand aftertreatment designed for the purpose. Durability, transient emissions and safe handling outside industrial settings are the areas in which evidence is thinnest, and progress there will decide how far ammonia moves from the test bed into heavy-duty service.

To be declared by the author(s) before publication.

To be declared by the author(s) before publication.