The issue of hydrogen emissions from energy infrastructure has enormous climate implications but is vastly understudied. The few studies that have attempted to estimate H₂ emissions from the hydrogen economy (Kammen et al., 2003; Schultz et al., 2003; Tromp et al., 2003; Warwick et al., 2004; Bond et al., 2011), hydrogen systems (e.g., H₂ fuel cell vehicles (Colella et al., 2005), refueling stations (Genovese et al., 2019)), or a specific value chain component (e.g., liquid H₂ [LH₂] handling (Petitpas, 2018)) have served as a basis for similar studies that assess the effect of H₂ deployment on the atmosphere and the climate (Prather, 2003; Jacobson et al., 2005; Jacobson, 2008; Wuebbles et al., 2010). Further studies have estimated H₂ emissions from each value chain component considering previous estimates. This mini-review summarizes these initial attempts to estimate H₂ emissions from hydrogen value chains. Notably, we synthesize the studies by van Ruijven et al. (2011), Arrigoni and Diaz (2022), Cooper et al. (2022), Frazer-Nash (2022), and Fan et al. (2022), which are the sole studies that provide hydrogen emissions estimates for total and per-component value chains. Because of the limited number of publications on hydrogen emissions, which are well-known among the community in this field, no specific strategy was employed to find literature on this topic.

Van Ruijven et al. (2011) assumed low and high H₂ emissions scenarios based on previously estimated H₂ emission rates (Schultz et al., 2003; Colella et al., 2005; Wuebbles et al., 2010). Arrigoni and Diaz (2022) published a technical report based on a 2022 workshop focused on the environmental impacts of hydrogen emissions, which includes the H₂ emissions estimates presented by Air Liquide for today and 2030. However, the underlying assumptions/data leading to their reported estimates are not detailed. Cooper et al. (2022) estimated the potential emissions of H₂ supply chains in the UK using natural gas supply chain emissions as a proxy. Similarly, Frazer-Nash Consultancy (2022) used natural gas emission estimates and conventional fluid mechanics to predict H₂ emissions in a 2050 UK hydrogen economy using a probabilistic model with 50% and 99% confidence intervals. Lastly, Fan et al. (2022) estimated the emissions of different value chain components in 2050 for low- and high-emission cases. For their estimates, they used published emission rates and extrapolated from comparisons with similar technologies (e.g., Alvarez et al., 2012; Hormaza Mejia and Brouwer, 2018; Panfilov, 2016; Petitpas, 2018; Shen et al., 2021; Xia et al., 2019).

Hydrogen emissions from hydrogen systems described in the literature can be divided into intentional and unintentional emissions or operational and fugitive emissions (see Supplementary Figure S1). The intentional emissions comprise operational purging and venting (Arrigoni and Diaz, 2022; Cooper et al., 2022; Fan et al., 2022; Frazer-Nash Consultancy, 2022). Most unintentional emissions are assumed to occur because H₂ is the smallest molecule and has low molecular weight, high diffusivity, and low viscosity (Völkl and Alefeld, 1975; Fichtner and Idrissova, 2009), making it hard to contain. Therefore, H₂ leaks more easily than other molecules managed (e.g., compressed, liquefied), transported, and stored. Accordingly, the reported unintentional and fugitive emissions include leakage from pipework (pipes, valves, joints, traps, assets, seals, etc.) and equipment, diffusion from pipelines, and permeation from where it is stored (e.g., compression or liquefaction tanks, salt caverns, and depleted oil and gas fields). Other unintentional emissions, such as residual H₂ in exhaust streams and natural boil-off from liquefied H₂, derive from operational procedures.

H₂ emissions are likely to occur throughout the value chain, from production, conversion, transport, and storage to end-use applications. Figure 1 and Supplementary Table S1 summarize the type of potential H₂ emissions described in the literature for value chain components. The primary potential H₂ emissions attributed to H₂ production are leakage through pipework and equipment and operational purging for maintenance, removal of impurities, and during fault conditions. Additionally, for blue H₂ production, there may be residual H₂ in the CO₂ stream. Similarly, for green H₂ production, the vented O₂ stream may carry residual H₂ (due to H₂ cross-over through the membrane between the electrodes). While venting of process gases is assumed negligible in new blue H₂ plants (gases will be sent to flare), venting during startup (venting air within the cathode) and shutdown (to remove moisture and any explosive gas mixture from the system) has been identified as a potential source of H₂ emissions in green H₂ production (Arrigoni and Diaz, 2022; Frazer-Nash Consultancy, 2022).

Regarding H₂ conversion, potential H₂ emissions reported for compression and liquefaction are mainly due to leakage through pipework and equipment (e.g., engine slip and compressor leakage through seals). Planned process venting, pipeline venting (e.g., for maintenance), and purging during startup and shutdown have also been identified as potential emissions sources during compression (Cooper et al., 2022; Frazer-Nash Consultancy, 2022).

As for transportation, potential H₂ emissions attributed to transmission and distribution pipelines include pipework and equipment leakage, venting, and diffusion (Arrigoni and Diaz, 2022; Cooper et al., 2022; Frazer-Nash Consultancy, 2022). For the transport of gaseous H₂ on road tube trailers, leakage from fittings and valves and purging of the trailer hose have been reported as primary potential emissions sources. Boil-off is the main potential source reported for LH₂ trucking. For LH₂ handling (e.g., loading/unloading), leakage through equipment and venting, in addition to boil-off, have also been reported (Petitpas, 2018; Arrigoni and Diaz, 2022; Cooper et al., 2022; Frazer-Nash Consultancy, 2022).

The primary potential H₂ emissions from H₂ storage (above ground as a compressed gas or a cryogenic liquid and below ground in salt caverns and depleted oil and gas fields) include natural permeation and leakages through equipment due to pressurization and depressurization. For aboveground LH₂ storage, there are additional identified potential emissions from boil-off and venting when boil-off leads to pressures deemed the safe maximum or releases to avoid getting close to critical pressures. In addition, for underground storage, venting and purging at surface processing plants due to annual maintenance, shutdown, and fault conditions (emergency shutdown) are other potential sources of emissions (Cooper et al., 2022; Fan et al., 2022; Frazer-Nash Consultancy, 2022).

The H₂ emissions from the last component of the value chain, end uses, will vary depending on the application (e.g., transport, industrial, power generation with fuel cells, internal combustion engines, or gas turbines). However, emissions from leakages through equipment and purging or venting have been identified as potential sources of emissions across end-uses (Figure 1 and Supplementary Table S1).

To provide a comprehensive overview of potential H₂ emissions from each value chain component, we compiled published H₂ emission rates and summarized them in Figure 2 and Supplementary Table S2. In the absence of direct measurements, these H₂ emission rates were estimated through assumptions, calculations via proxies, laboratory experiments, and theory-based models (simulations). Many of these estimates are based on previously published H₂ emissions assumptions and estimates from natural gas systems.

The synthesized references report emissions rates as a percentage of hydrogen produced, managed, or used in the respective supply chain component. All studies report ranges except Arrigoni and Diaz (2022), who provide a single estimate for today and 2030. Moreover, while van Ruijven et al. (2011) and Cooper et al. (2022) estimate current H₂ emissions, Frazer-Nash Consultancy (2022) and Fan et al. (2022) predict H₂ emissions within three decades. It is also important to emphasize that Cooper et al. (2022) and Frazer-Nash’s estimates are specific to UK hydrogen systems. Consequently, due to these differences and the different methodologies used to estimate emissions rates, it is evident in Figure 2 that there is a very wide range of estimates for each value chain component.

For H₂ production, estimated emissions rates range between 0.5% and 1.0% for gray H₂, 0.0%–1.5% for blue H₂, and 0.03%–9.2% for green H₂. Blue H₂ production may have a higher leakage risk than gray H₂ due to additional separation processes (Fan et al., 2022). Green H₂ production may have higher emissions than blue and other pathways (e.g., bioH₂) due to the higher emission rates associated with electrolysis compared to steam methane reforming and biomass gasification (Cooper et al., 2022).

Estimated emissions rates associated with LH₂ have the largest ranges; 0.15%–10.0% for liquefaction, 2.0%–13.2% for LH₂ trucking, 2.0%–20% for LH₂ handling, and 2.0%–15.0% for LH₂ refueling stations. While some studies report emissions specific for liquefaction (Arrigoni and Diaz, 2022; Cooper et al., 2022), LH₂ trucking (Fan et al., 2022; Frazer-Nash Consultancy, 2022) or refueling (Petitpas, 2018; Arrigoni and Diaz, 2022), others mention LH₂ handling in general (Petitpas, 2018; Arrigoni and Diaz, 2022).

Estimated emissions rates associated with gaseous H₂ transport, storage, and use also vary widely; 0.02%–5.0% for transmission pipelines, 0.0003%–5.0% for distribution pipelines, 0.3%–2.3% for transportation on road tube trailers, 2.8%–6.5% for storage above ground, and 0.25%–3.0% for gas refueling stations (including the estimations for gaseous H₂-powered vehicles by (Colella et al., 2005)). H₂ emissions rates estimated for common end-uses include 0.01%–3.0% for power generation, 1.0%–2.3% for ships and ports, and 3.0% for aviation (Figure 2). It should be noted that some of these emissions rates include emissions from two value-chain components (see details in Supplementary Table S2).

Overall, H₂ transport is predicted to generate the most emissions (Frazer-Nash Consultancy, 2022), especially transporting and handling LH₂, which implies the need to manage boil-off (Sherif et al., 1997; Petitpas, 2018). Our survey of published emissions estimates makes it clear that more robust data is required to have confidence in the H₂ emissions rates for each value chain or its components.

Without empirical data, very little confidence should be placed in total value chain emissions, as they are expected to be highly dependent on the configuration of the pathway and existing technology. To assess the environmental impacts of a potential hydrogen economy, several studies have estimated total H₂ emissions rates between 0.2% and 20% from a diversity of value chains (e.g., Bond et al., 2011; Cooper et al., 2022; Warwick et al., 2004; Wuebbles et al., 2010) (Supplementary Table S3). Specifically, van Ruijven et al. (2011) estimated total value chain H₂ emissions from 0.2% to 10% based on a per-component configuration. However, this range does not include emissions from production and compression and components of storage and end-use applications (Supplementary Table S2). Comparatively, Frazer-Nash includes components of production, transportation, storage, and several end-use applications but estimates total H₂ emissions rates of 0.96%, at 50% confidence, and 1.50%, at 99% confidence. Tromp et al. (2003) estimate total H₂ emissions rates between 10% and 20%, mainly due to H₂ transport. However, Schultz et al. (2003) indicate that only for extreme individual cases, like uncontrolled evaporation from LH₂ storage tanks, would 10%–20% emissions rates be possible. Further studies on the climate impacts of H₂ have assumed a range of 1%–10% of total value chain emissions citing previous studies (Derwent et al., 2020; 2001; Prather, 2003; Paulot et al., 2021; Hauglustaine et al., 2022; Ocko and Hamburg, 2022; Warwick et al., 2022).