Hydrogen in steel production: what is happening in Europe – part two


The steel industry accounts for 4% of all the CO2 emissions in Europe and 22% of the industrial carbon emissions in Europe. Hydrogen is one way of reducing emissions in the steel sector, either as an auxiliary reducing agent in a blast furnace (H2-BF), which was covered in part 1 of this series, or as the sole reducing agent (H2-DRI) in a process known as direct reduction of iron, which is the topic of this article.  

Using hydrogen in the Direct Reduction – Electric Arc Furnace route (DR-EAF) 

The figure below shows the process of steelmaking using the DR-EAF route. Iron ore is reduced with hydrogen while in a solid state, hence the name direct reduction, to produce direct reduced iron (DRI) called sponge iron. Sponge iron is then fed into an EAF, where electrodes generate a current to melt the sponge iron to produce steel. Some carbon is needed so that steel can be produced. This carbon can come from pulverized coal, biomethane or other biogenic carbon sources. This, combined with electrode consumption, adds up to ±53 kgCO2 per tonne of steel – if the electricity used is not fully renewable, its carbon intensity should also be added to that figure. 

In the EAF process, some scrap can also be added to reduce the need for iron ore. 

R article

DR-EAF route with green hydrogen, either using a shaft furnace or a fluidized bed reactor, where hydrogen is the sole reducing agent [1]. Source: Roland Berger 2020 

It should be noted that direct reduction is not a new technique and has been around at a commercial scale since the late 1960s, though not with pure hydrogen. There is only one DRI plant in Europe, owned by ArcelorMittal and located in Hamburg, producing 0.6 Mt of steel per year [2]. Even though this method of production reduces emissions compared to coal steel making, it still results in emissions to the atmosphere. For DRI to be carbon neutral, a carbon neutral feedstock (such as hydrogen produced with renewable electricity) must be used in the steel making process. 

There are also some other ways DRI can be used: 

  • DRI can be coupled to a Basic Oxygen Furnace (BOF) via a melting unit 

Fourteen planned DRI project were found in Europe (table 1). The projects are in varying stages ranging from being at a planning stage to operating a pilot plant. By the mid-2020s, Europe will have several cities with commercial DRI production if the steel companies meet their goals.  

Table 1. DRI projects planned in Europe. 

Steel producerLocationCurrent status
ArcelorMittal Hamburg, Germany Demonstration plant by 2023. Target for commercial operation 2025.
ArcelorMittal Dunkirk, France Feasibility study
ArcelorMittal Taranto, Italy Planning stage
ArcelorMittal Eisenhuttenstadt, Germany Online in 2026 (pilot plant)
ArcelorMittal Bremen, Germany Online in 2026 (large-scale)
VoestalpineLeoben (Donawitz), AustriaCommissioning in Q2 of 2021
Salzgitter AG Salzgitter, Germany Demonstration plant ordered 12/2020. Scheduled to go online 1st half of 2022.
Salzgitter AG Wilhelmshaven, GermanyFeasibility study
SSABGällivare-Oxelösund, Sweden Pilot plant, market production 2026
LKABKiruna-Malmberget-Svappavaara, Sweden First DRI plant in Malmberget in 2029 [3]
Thyssenkrupp Duisburg, Germany First production in DRI plant 2025
Liberty Galati, Romania DRI plant installed between 2023-2025
Liberty Dunkirk, France Feasibility study ongoing
H2 Green Steel Boden-Luleå, Sweden Large scale production by 2024

Use of DRI 

Most projects plan to connect their DRI plant to an EAF from the start (ArcelorMittal [4], SSAB [5], Liberty in Romania, H2 Green Steel). However, some projects include different pathways: 

  • Voestalpine: it is not clear whether the produced DRI will be used in an EAF directly. But given that Voestalpine built a DRI plant in Texas to produce hot briquetted iron for the blast furnaces in Linz and Leoben, it seems likely that it will use HBI in its blast furnaces and later transition to an EAF.  
  • Salzgitter AG intends to use DRI both in BFs and EAFs and gradually shift to use only DRI-EAF, which they expect to result in emission reductions of up to 95% 
  • Thyssenkrupp is planning to use HBI with a BF first, followed by DRI in combination with a melting unit which allows them to keep their BOF’s. 
  • Liberty’s project in Dunkirk intends to use DRI both in an EAF and to produce hot briquetted iron for a blast furnace [6].  

Even though companies have made various claims about the climate change mitigation potential of these projects, their final climate impacts will depend on factors such as the carbon intensity of the feedstock used for DRI (e.g., electricity used to produce hydrogen), the full lifecycle of the product (e.g., transport emissions of shipping iron from overseas) and other emissions coming from residual fossil fuel use (e.g., remaining coal use in a blast furnace or electricity use for EAFs).  

Production capacities 

Specific initial annual production targets are given for nine projects. Combined, they account for 20.45 Mt/year, which is about eight times the capacity of a large DRI plant. It would significantly increase European DRI capacity.  However, it still constitutes only a minor fraction of EU steel production, which was 159 Mt in 2019 

capacity dri projects 2

Emission reductions by DRI 

For evaluating the climate impact of DRI, assessing the carbon intensity of the reducing agent is key (table 2) 

Table 2: DRI projects in Europe and their reducing agents. 

Steel producerLocationReducing agent
ArcelorMittal Hamburg, Germany Grey hydrogen
ArcelorMittal Dunkirk, France Natural gas
ArcelorMittal Taranto, Italy Unknown
ArcelorMittal Eisenhuttenstadt, Germany Hydrogen from electrolysis and pyrolysis of natural gas
ArcelorMittal Bremen, Germany Natural gas --> hydrogen from electrolysis
VoestalpineLeoben (Donawitz), AustriaHydrogen [8]
Salzgitter AG Salzgitter, Germany Natural gas and hydrogen
Salzgitter AG Wilhelmshaven, GermanyNatural gas --> hydrogen from electrolysis
SSABGällivare-Oxelösund, Sweden Hydrogen from electrolysis [9]
LKABKiruna-Malmberget-Svappavaara, Sweden Hydrogen likely from electrolysis
Thyssenkrupp Duisburg, Germany Natural gas
Liberty Galati, Romania Natural gas
Liberty Dunkirk, France Hydrogen and natural gas --> hydrogen from electrolysis
H2 Green Steel Boden-Luleå, Sweden Hydrogen from electrolysis

Currently, the availability of green hydrogen is limited. Most of the projects intend to use natural gas or grey hydrogen initially, saying they can use green hydrogen, but that it is not available yet in sufficient quantities and at affordable prices. This “hydrogen-ready” argument is often cited by industry. ArcelorMittal also mentions the use of blue hydrogen until enough green hydrogen is ready.  

Liberty Dunkirk and Salzgitter AG have concrete plans of constructing an electrolyser giving credibility to their claims of switching from natural gas to hydrogen. However, as said in the previous article, electrolysis can only produce low-carbon hydrogen if it is powered by renewable electricity. While Salzgitter AG have plans to use wind power for their electrolyser, the electricity source for Liberty’s project in Dunkirk is unknown. The grid emission intensity of France in 2018 was the second lowest in the EU, only Sweden did better. This would result in a relatively low hydrogen carbon footprint of 175 kgCO2/ton of DRI [10]. 

Emissions do not only depend on how DRI is produced, but also whether it is used in an EAF or a BF: 

  • A recent project in Japan demonstrated the potential of using DRI in a BF, reducing emissions by 20%, while Voestalpine reported 5% emission reductions. In both cases, the DRI used was produced using hydrogen and carbon monoxide from natural gas. However, the blast furnace cannot fully switch to a renewable fuel such as hydrogen from renewable electricity, which is why it ultimately cannot be carbon neutral.  
  • Natural gas-based DRI-EAF (± 0.95 tCO2/tSteel) reduces emissions by ±42% compared to BF-BOF (±1.65 tCO2/tSteel) [11]. While that is a reduction in emissions compared to the worst performing process (coal-based steal making), driving the emissions down to zero will depend on the use of hydrogen produced only with renewable electricity.  
  • When using hydrogen from electrolysis, emission savings can be as high as 95% in the case of green hydrogen. For the DRI production to be green, hydrogen needs to be generated by renewable electricity. For the amount of DRI mentioned earlier (20.45 Mt), approximately 66 TWh per year would be needed [12]. For comparison, that is 28% of Germany’s renewable capacity of 2019. 

It is clear that we need more action to increase the share of renewables and call out “hydrogen-ready” plants that are using grey hydrogen or natural gas. While there are some promising projects, it seems unlikely that DRI production with green hydrogen will account for 100% of EU steel production in 2050. There is a race for producing the first truly climate-neutral steel and Sweden looks set to win it. 


  1. While the fluidized bed reactor does not require a pellet plant, thereby reducing emissions, it is less developed and requires higher investments than when using a shaft furnace (Roland Berger, 2020).  
  2. The main reason why DR is mostly used outside of Europe is the high price of natural gas in Europe, which is reformed to produce carbon monoxide and hydrogen, which is then used in DRI. Around 75% of global hydrogen is produced this way. 
  3. LKAB wants to fully switch to sponge iron in the next 20 years. It’s not possible to say at this moment how and where that sponge iron will be used.  
  4. There is a lot of uncertainty around the Taranto plant. The plant is very polluting but provides thousands of jobs, and its management has been a difficult balancing act by the Italian government. It is likely recovery funds will be used to clean up the plant including plans for a DRI-EAF installation. Decisions still have to be made. 
  5. While SSAB started pilot plant trials with hydrogen-based DRI in 2020, these are for test campaigns and evaluations and are not on a commercial scale (1t/hr discontinuous production). Therefore, SSAB is considered to start with DR-EAF in 2026 
  6. The HBI refers to any surplus produced in Dunkirk. It will be used in Liberty’s Ostrava and Galati integrated steelworks and the Dillinger and Saarstahl plants in Germany. 
  7. Hamburg (0.1 Mt), Leoben (0.25 Mt), Wilhelmshaven (2 Mt), Oxelösund/Gällivare (2.7 Mt), Duisburg (2.4 Mt), Galati (2.5 Mt), Dunkirk Liberty (2 Mt), Bremen and Eisenhuttenstadt (3.5 Mt) and Boden-Luleå (5 Mt). The added capacity by 2030 at Gällivare will be for SSAB and others. The pilot plant at Salzgitter is not included in the figure due to its low production capacity of 100 kg/hour or 0.000876 Mt/year if operation would be continuous throughout the year. The number for Duisburg is a minimum value as the second DRI plant for 2030 will be larger than the first one of 1.2 Mt capacity. 
  8. How the hydrogen is produced is unknown. 
  9. Some of the testing will use natural gas initially, then hydrogen from electrolysis. 
  10. Grid emission intensity of 0.054 kgCO2/kWh x 60 kWh/kgH2 x 53.93 kgH2/tDRI = 174.73 kgCO2/tDRI. The value 53.93 results from using a conversion factor to convert from 600 Nm³ to kg hydrogen. It is a theoretical value and could be higher in reality 
  11. (1-0.95/1.65) x 100% = 42%.  Specific CO2-emissions were given in ranges. 0.95 and 1.65 reflect the middle of these ranges of natural gas-based DRI and BF-BOF respectively. 
  12. 60 kWh/kgH2 x 53.93 kgH2/tDRI x 20.45 Mt of DRI = 66.2 TWh