In this PhD thesis, the desulphurisation in 21st century iron- and steelmaking is investigated. The current state of the art in sulphur removal in ironmaking and oxygen steelmaking is discussed (Part I of this thesis) and optimisation of the hot metal desulphurisation (HMD) slag, which is an important aspect of present day desulphurisation, is investigated (Part II of this thesis). Furthermore, since the steelmaking industry will change significantly as a result of the global climate change mitigation, desulphurisation of hot metal from HIsarna, a new low-CO2 ironmaking process, is studied (Part III of this thesis). Finally, an overview of the main conclusions of this work, as well as an outlook about desulphurisation in iron- and steelmaking for the coming decades, based on the research presented in this thesis, is given in Part IV of this thesis. Sulphur is an unwanted impurity in steel that lowers the formability and weldability of steel and it makes steel more brittle. Therefore, steelmakers try to limit the concentration of sulphur in the steel. In 2021, globally roughly two third of the steel is produced via the BF-BOF steelmaking process, where iron ore is reduced by carbon (coal and coke) in the blast furnace (BF), and the hot metal from the BF is refined in the basic oxygen furnace (BOF, or converter). Sulphur can be removed at different process steps in the steelmaking process chain, like at the HMD, at the converter, or at the secondary metallurgy processes. Because of the low oxygen activity in hot metal, sulphur is most efficiently removed at the HMD process. In Chapter 2, the different sulphur removal steps in the steelmaking process chain are discussed. Here also the different HMD processes that are globally being used are discussed. The two most important HMD processes are the co-injection process (where desulphurising reagents, typically Mg and CaO or CaC2, are injected into the hot metal) and the Kanbara reactor (KR; where calcium-based reagents are mixed through the hot metal with an impellor). Currently, the co-injection process is globally the most commonly used process of the two and it is dominant in Europe and North America. Typically, magnesium and lime are used as reagents in the co-injection HMD process. Magnesium dissolves in the hot metal and reacts with the dissolved sulphur to form solid MgS. Although some lime directly reacts with the sulphur, its main task is to react with the MgS to form the more stable CaS, which moves to the slag phase. When the reagent injection is finished, the slag is removed with a skimmer. During the removal of the slag, some iron is lost with it. The amount of iron lost per heat is typically 0.5-2.5 wt% of the total hot metal weight, which is a major cost for the HMD process. In Chapter 3 it is explained that iron loss is governed by two mechanisms: colloidal loss (iron present in the slag in a colloidal form, which is removed together with the slag) and entrainment loss (iron being entrained with the slag during the slag removal). Entrainment loss can be minimised by optimising skimming conditions like an experienced operator, a clean skimmer paddle and a well-controlled skimmer. Colloidal loss can be minimised by decreasing the apparent viscosity of the slag, which under typical HMD conditions means that the solid fraction of the slag should be minimised. This can be achieved by either increasing the slag temperature (in practice minimising the temperature loss) or by decreasing the slag’s basicity, which lowers the melting temperature of the slag. Furthermore, in Chapter 3 it is shown that the HMD slag also needs a B2 basicity (ratio of the concentrations of CaO and SiO2) of at least 1.1 and enough lime to convert all present sulphur to CaS, in order to have a sufficient sulphur removal capacity. The optimal HMD slag has a B2 basicity high enough to allow all the removed sulphur to stay in the slag (sufficient sulphur removal capacity), but its basicity is low enough to keep the slag’s melting temperature below the actual temperature of the slag (typically 1300-1450 °C), ensuring a mostly liquid slag, resulting in a low colloidal loss. In Chapter 4, these findings are evaluated and supported with a Monte Carlo simulation based on thermodynamic data from FactSage, melting point and viscosity measurements with artificial HMD slags and plant data analysis. The temperature of the slag has the strongest influence on the colloidal loss and total iron loss, where a lower temperature leads to a slag with a higher solid fraction and, thus, a higher iron loss. From the typical HMD slag components, MgO has the largest influence on the slag’s melting temperature, where a higher concentration of MgO leads to a higher melting temperature (thus to a higher iron loss). In an industrial setting, it is difficult to increase the temperature of the slag (which is typically 1300-1450 °C). Also, it is difficult to influence the HMD slag composition, because 60-80 wt% of the HMD slag is carryover slag from the BF (changing that would require changing the BF process) and the rest is determined by the reagents injected to remove a certain amount of sulphur (resulting in a certain amount of CaO, MgO and CaS being added to the slag). A more practical method to change the HMD slag composition for a lower viscosity is to add a slag modifier. In Chapter 5, fly ash and nepheline syenite are investigated as suitable slag modifiers for the HMD process. Fly ash contains SiO2 and Al2O3 and decreases the basicity of the slag and, thus, its melting temperature. Nephelene syenite contains SiO2 and Al2O3 as well, but it also contains Na2O, which is a basic network modifier that decreases the slag’s viscosity. Melting point and viscosity experiments with synthetic HMD slags show that both fly ash and nepheline syenite are viable slag modifiers and are a good alternative to the fluoride-based slag modifiers, which are common in industry. Fluoride-based slag modifiers lower the slag’s melting temperature and the viscosity of the liquid fraction. However, fluoride leads to health and environment issues and it decreases the desulphurisation efficiency of magnesium as well. As a result of the global climate change mitigation, the steel industry has to lower its CO2 emission. One new process that can contribute to a lower CO2 footprint of the steelmaking industry is the HIsarna process, which is being developed at Tata Steel in IJmuiden, the Netherlands. Like a BF, HIsarna produces hot metal, but with a 20 % lower CO2 emission. Even an 80 % lower CO2 emission can be achieved when using carbon capture and storage or usage, due to the concentrated CO2 off gas. Compared to a BF, HIsarna produces hot metal with a lower temperature and with lower carbon, manganese and phosphorus concentrations. HIsarna hot metal contains almost no silicon and titanium. However, compared to a BF, HIsarna produces hot metal with roughly 3-4 times more sulphur (typical sulphur concentration in hot metal is around 0.1 wt%). This means that for HIsarna hot metal more sulphur needs to be removed compared to typical BF hot metal. The consequences for desulphurisation of HIsarna hot metal are discussed in Part III of this thesis. Typically, due to cooling, hot metal from a BF is supersaturated in carbon by the time it arrives at the HMD. This carbon supersaturation leads to graphite formation, also known as kish. The formed graphite flakes can form a layer between the slag and the hot metal. Earlier research suggested that this graphite layer could hamper the HMD process, as it would block MgS formed in the hot metal, thus it cannot reach the slag phase and form the more stable CaS. This would result in a lower desulphurisation efficiency. Since HIsarna hot metal contains less carbon, this effect could be smaller or even non-existent for desulphurisation of HIsarna hot metal. However, as is explained in Chapter 6, this hampering effect of precipitated graphite on the efficiency of the HMD process is very small, for both HIsarna and BF hot metal. Analysis of plant data shows that there is only a small correlation between expected graphite formation and HMD efficiency. Only for heats with a low initial sulphur concentration (below 225 ppm sulphur) showed a significant correlation. This means that there is no significant benefit for desulphurisation of low-carbon HIsarna hot metal, compared to carbon-saturated BF hot metal. The lower temperature and the higher initial sulphur concentration of HIsarna hot metal, compared to BF hot metal, do influence the HMD process, as is discussed in Chapter 7. A literature study, a thermodynamic analysis with FactSage and plant data analysis show that the lower temperature and higher initial sulphur concentration lead to a lower specific magnesium consumption. The lower temperature, typically 50 °C colder than BF hot metal, thermodynamically favours the desulphurisation reaction with magnesium. The higher initial sulphur concentration leads to a higher sulphur activity, which enhances the desulphurisation reactions. However, it should be noted that the higher efficiency caused by the initial sulphur concentration is only valid for the surplus of sulphur, compared to BF hot metal. The total amount of sulphur that has to be removed is still 3-4 times higher for HIsarna hot metal than for BF hot metal. Therefore, the total magnesium consumption for desulphurisation of HIsarna hot metal is higher as well. This also means that desulphurisation of HIsarna hot metal will take longer than desulphurisation of BF hot metal, which could lead to the HMD becoming the bottleneck in a steel plant. It is estimated that the oxygen concentration in HIsarna hot metal (~6 ppm) is roughly 5-10 times higher than in BF hot metal (0.5-1 ppm). A high oxygen concentration leads to a lower desulphurisation efficiency. However, since the oxygen concentration in HIsarna hot metal is still low, the expected extra magnesium consumption as a result of the higher oxygen concentration is limited to about 2-3 kg for a 300 t heat. The absence of silicon and titanium in the hot metal will not influence the efficiency of a magnesium-based HMD process. However, a lime-based HMD process, like KR, will have a lower efficiency, since silicon reacts with the oxygen in lime as the calcium reacts with sulphur to form CaS. Furthermore, since HIsarna produces hot metal without slag, an alternative for the carryover slag from the BF needs to be found. A slag based only on the injected magnesium and lime and the formed CaS would be solid at HMD temperatures (see Chapter 3). Therefore, the use of acidic slag additions (like SiO2 and Al2O3) is required, to keep the slag liquid and minimise the iron loss. Still, the higher slag volumes as a result of more sulphur that has to be removed will lead to a higher iron loss, compared to desulphurisation of typical BF hot metal. Because the HIsarna produces, and taps, hot metal continuously with a constant composition and temperature, it is ideal for continuous hot metal desulphurisation. Therefore, at Tata Steel in IJmuiden, the Netherlands, a new continuous hot metal desulphurisation (CHMD) process is being developed. In Chapter 8, this novel CHMD process is introduced. The CHMD process is based on the magnesium-lime co-injection HMD process. It uses several reactors in series (process simulations suggest three reactors in series are required to desulphurise typical HIsarna hot metal to typical post-HMD sulphur concentrations), to limit the total reactor volume. The desulphurisation efficiency of the process is increased by optimising the reactor dimensions to a height to diameter ratio of 5:1, whereas a typical hot metal ladle, used for the batch HMD, has a height to diameter ratio of 1.5:1. It is expected that this leads to a reduction in reagent consumption of ~20 %. Furthermore, the continuous nature of the process allows for a foxhole-type slag skimming (separating slag and hot metal by their density difference), which will lead to an estimated 60 % lower total iron loss, compared to the skimming method of the batch HMD process (a remote-controlled skimmer arm, raking off the slag). Based on the cost estimation for iron loss and reagent costs, the cost for desulphurising one tonne hot metal with the CHMD process will be approximately € 2 lower than with the state-of-the-art batch HMD process. However, according to the current calculations, the residence time of the hot metal in the CHMD is 3-4 times longer than in a batch HMD process, leading to a higher temperature loss and, possibly, a higher CO2 footprint, as a lower temperature allows for less scrap being charged at the converter. Given the already lower temperature of HIsarna hot metal, compared to BF hot metal, this is an issue that needs to be solved before the CHMD process can be used in industry. Currently the development of the CHMD process is still in the conceptual design phase. The changes in the global steel industry as a result of the climate change mitigation will not stop after 2030. Finally, in 2050, the steel industry should be CO2-neutral. In Chapter 10, an outlook is given for the expected changes in the steel industry between now and 2050 and its impact on sulphur removal in iron- and steelmaking. The amount of carbon used to reduce iron ore will gradually decrease and so will the demand for HMD, as the carbon sources coal and coke are the largest source of sulphur in hot metal. However, it is unlikely that carbon can be fully replaced by hydrogen or electricity. It is expected that in 2050 still a significant amount of steel will be produced via carbon-utilising smelting processes, like HIsarna, in combination with carbon capture and usage. Besides, scrap contains sulphur that needs to be removed as well. It is expected that the share of scrap as a source of iron in the steelmaking industry will increase in the coming decades. Therefore, steel desulphurisation will remain necessary in every steel plant and hot metal desulphurisation will be required as well for the carbon-utilising plants.