Enhancing CO₂ Mineralization Rate and Extent of Iron and Steel Slag via Grinding

Abstract

Roughly 10% of the CO₂ emissions from iron and steel making are attributable to the direct release of CO₂ from the thermal decomposition of carbonates to produce flux, mainly CaO, used for impurity removal. Notably, these direct emissions remain even if carbon-based steelmaking is replaced by hydrogen-based steelmaking. After removing impurities from the molten metal, this flux becomes the solid waste product called ‘slag’, a primarily Ca-silicate material. The transformation of slag back into carbonates is thermodynamically spontaneous with negative ΔG in the ambient environment, meaning that ~10% of the CO₂ emissions from iron and steel making could be negated if equipment and methods were developed to support CO₂ mineralization. However, the rate of CO₂ mineralization using slag is slowed by several environmental, geometric, and processing factors. We leverage an experimentally verified model of CO₂ mineralization to determine how to efficiently accelerate the process. Increasing the crystallinity of slag, increasing the relative humidity, and reducing the grain size of slag particles provide the greatest increase in CO₂ mineralization rate at the lowest energy penalty. Increasing the concentration of CO₂ and the temperature provide only modest increases in the CO₂ mineralization rate while incurring a substantial energy penalty. For steelmaking slags, CO₂ mineralization represents low-hanging fruit as the current reuse pathways are low value. For ironmaking slag, replacing the production of amorphous slag for the cement industry with the production of crystalline slag for CO₂ mineralization becomes financially preferable when a carbon price/tax exceeds 67.40 USD/t-CO₂.

1. Introduction

The iron and steelmaking industry (ISM) is a major anthropogenic source of CO₂ emissions and is considered one of the more difficult industries to decarbonize due to the chemical necessity for coking carbon and high temperature heat.¹⁾ Recently, declarations of net zero CO₂ steel production by replacing coking coal with so-called ‘green H₂’ have become in vogue.^2,3) However, even idealized H₂-based steelmaking involves CO₂ emissions from fossil carbon due to the thermal decomposition of CaCO₃ to create CaO (termed ‘flux’ in ISM). Flux is necessary to remove impurities such as silica and alumina to produce high quality iron and steel. In the modern integrated blast furnace (BF) and Basic Oxygen Furnace (BOF) process, the direct CO₂ emissions from thermal decomposition of carbonates accounts for ~10% of total CO₂ emissions, or ~268 Mt-CO₂/y in 2019.⁴⁾

Currently, the majority of BF slag is quenched and finely ground to form an amorphous product sold to the cement industry called ‘ground granulated blast furnace slag’ (ggbs).⁵⁾ The cement industry uses ggbs as a partial replacement for cement (so-called ‘supplemental cementitious material’: SCM), reducing the quantity of cement that they produce.⁶⁾

Unlike BF slag, the slag from the BOF (and the electric arc furnace: EAF) steelmaking process is typically not suitable for conversion to a SCM. Steelmaking slag is generally repurposed into less valuable products such as road base and ballast.⁵⁾ Compared to ironmaking slag, steelmaking slags are typically richer in Ca (and Mg), indicating a larger embodied direct CO₂ emission from the decomposition of fossil carbonates.

Previously, we have demonstrated the gas-solid reaction and stabilization of CO₂ into solid carbonates (CO₂ mineralization) for pure minerals, natural rocks, and slag from commercial ISM facilities.^4,7,8) For process optimization and cost considerations, it is necessary to examine the impact of potential operating conditions on the mineralization rate and energy consumption. We evaluate the impact of slag crystallinity, slag particle size, CO₂ concentration, relative humidity, and temperature on reaction rate and extent. We further analyze the idealized energy consumption of altering each operating condition to assess its relative merit. Finally, we compare the financial incentive to use BF slag for CO₂ mineralization instead of as ggbs as a function of a price on CO₂ emissions.

2. Theory

2.1. Prediction of CO₂ Mineralization Rate and Extent

For the common minerals contained in slag at the particle diameters (d) generally considered for CO₂ mineralization (~10’s–100’s μm), the rate-limiting step of CO₂ mineralization is typically diffusion of calcium and magnesium ions through the product layer. In such cases, CO₂ mineralization can be modeled by the Unreacted Core Model (UCM) per Eq. (1), where t is the mineralization time in s, D is the product layer diffusivity in m²/s, ρ_Ca,Mg is the density of Ca and Mg in slag in mol/m³, C_CO2 is the concentration of CO₂ in the gas phase in mol/m³, r is the radius of the slag particle in m, and δ is the mineralization depth in m. For minerals with very rapid product layer diffusion and/or a very slow CO₂ mineralization reaction, it is possible for the mineralization reaction kinetics to become the rate limiting step. In such situations, the standard UCM would predict erroneously rapid CO₂ mineralization. The UCM is modified in Eq. (2) to calculate CO₂ mineralization time as a function of both the CO₂ mineralization reaction and the product layer diffusion (RDM). In this 1D model, the reaction rate constant is given by k in m/s and δ is replaced by the unitless reaction extent (ξ) per Eq. (3).   

$$t=\frac{1}{D}\left(\frac{{\rho }_{Ca,Mg}}{C_{C{O}_2}}\right)\left(\frac{r^2}{6}\right)\left[1-3{\left(\frac{r-\delta }{r}\right)}^2+2{\left(\frac{r-\delta }{r}\right)}^3\right]$$(1)

  

$$\begin{array}{l}t=\frac{1}{D}\left(\frac{{\rho }_{Ca,Mg}}{C_{C{O}_2}}\right)\left(\frac{r^2}{6}\right)\left[1-3{\left(1-\xi \right)}^{\frac{2}{3}}+2\left(1-\xi \right)\right]\\ \hspace{0.5em}+\frac{1}{k}\left(\frac{{\rho }_{Ca,Mg}}{C_{C{O}_2}}\right)r\left[1-3{\left(1-\xi \right)}^{\frac{1}{3}}\right]\end{array}$$(2)

  

$$\xi =1-\frac{{\left(r-\delta \right)}^3}{r^3}$$(3)

The d at which the mineralization reaction switches from the UCM-regime to the RDM-regime (d_transition) is dependent on the k and D. Several minerals common to ISM slags are plotted in Fig. 1. Calcium silicates and oxides transition to the RDM-regime when ground to d <~1–10 μm. Magnesium silicates remain in the UCM-regime until d ~1 nm. Amorphous compounds remain in the UCM-regime to much smaller d than their crystalline counterparts due both to a reduction in D and an increase in k. Since the transition into the RDM-regime indicates a decreasing return (i.e., acceleration of CO₂ mineralization) on energy investment (i.e., grinding), d_transition provides a rule-of-thumb target grinding extent for efficient and fast CO₂ mineralization.

Fig. 1.

Transition from product layer diffusion-controlled (UCM) to product layer diffusion and reaction rate-controlled (RDM) regime as a function of particle diameter. D from 27); k from 15,16) and in-house calculations. (Online version in color.)

2.1.1. Calculation Details

A few technical points are of quantitative importance for applied CO₂ mineralization. Both UCM and RDM assume spherical slag particles with smooth surfaces; non-sphericity and surface roughness can be accounted for, but they primarily act to shorten only the initial mineralization time.⁴⁾ Also note that the full particle size distribution of a ground slag sample must be considered to obtain an accurate prediction of mineralization rate and extent.⁹⁾

Lab-synthesized, pure mineral D have been previously published, but slag is a mixture of minerals.⁴⁾ The overall diffusivity of slag (D_slag) is estimated based on the individual mineral diffusivities and volume concentrations of each as described by Effective Medium Theory (EMT). This only applies when individual mineral grains are significantly smaller (<1/100) than the slag particle. When particle sizes approach the grain size, each mineral’s D, t, and ξ should be calculated separately. As grain size depends on solidification and cooling rate of slag, the application range of EMT must be evaluated on a case-by-case basis.

In some cases, CO₂ reactive compounds (e.g., Ca₂SiO₄) can be completely surrounded by unreactive compounds (e.g., SiO₂, Al₂O₃), leading to incomplete CO₂ mineralization extent regardless of the reaction time. This phenomenon is termed ‘mineral locking’ (ML) for polycrystalline materials and has been described in detail elsewhere.⁴⁾ The inhibitory effects of mineral locking are established in Fig. 2. A severe form of mineral locking occurs when a non-reactive amorphous phase forms in the intergranular spaces before growing crystal phases contact one another. In this case, the maximum CO₂ mineralization extent is limited to reactive crystalline grains on the exterior of the particle.

Fig. 2.

The maximum CO₂ mineralization extent as a function of reactive content and grinding extent for polycrystalline slag (‘mineral locking’). (Online version in color.)

2.2. Variation of Product Layer Diffusivity with Humidity

The D is not only mineral-dependent, it also varies with the relative humidity of the gas phase (RH), with higher RH acting to increase D. The literature contains five decades of experimental, empirical, and theoretical work supporting the acceleratory influence of water vapor on CO₂ mineralization.^{10,11,12,13,14)} As shown in Fig. 3, using the methods of Myers et al. 2019, increasing the RH leads to an exponential increase of D for crystalline CaSiO₃. Simple fitting suggests that each 10% reduction in RH reduces D by ~3.8 for CaSiO₃. Equivalent tests for MgO results in a ~4.1 times reduction in D for every 10% reduction in RH. The acceleratory effect despite the inherently different structures of these compounds and their product layers suggests that this phenomenon may be somewhat consistent across crystalline compounds. At sufficiently low relative humidity (i.e., RH ≤ 30–50%), CO₂ mineralization appears to halt at the surface layer in agreement with theoretical modelling.¹²⁾

Fig. 3.

The increase in D as a function of RH for CaSiO₃ and MgO measured in this study using the methods of Myers et al. 2019. (Online version in color.)

2.3. Variation of Reaction Rate

When a mineral is ground small enough to enter the RDM-regime, increasing the k of Eq. (2) would accelerate the CO₂ mineralization rate. Per the Arrhenius equation ($k=A{e}^{-E_a/RT}$), an increase in reaction temperature provides an increase in k. For activation energies of common ISM compounds, increasing the temperature from 30°C to 95°C provides an increase in k of ~2–20.^15,16)

The k can also be modified by alteration of the activation energy (E_a). The reaction rate of an amorphous compound (k_amorphous) is generally higher than a chemically equivalent crystalline compound as dislocations in the crystalline structure result in an increase in the molar Gibbs free energy (ΔG_d).¹⁷⁾ This increase reduces the E_a, leading to a reaction rate increase per Eq. (4).   

$$\frac{k_{amorphous}}{k}=ex{p}^{\left(\frac{\mathrm{\Delta }{G}_d}{RT}\right)}$$(4)

Since ΔG_d is a function of dislocation density, it is maximized when a material becomes fully amorphous. An example of the connection between k and amorphous content of ISM slag is the regulatory requirement that ggbs be ≥67% glassy content—in many cases ≥95%.¹⁸⁾ This regulation was developed from user experience around the pozzolanic nature of ggbs (i.e., not from theoretical underpinnings).¹⁹⁾ The theoretical increase in reaction rate of ggbs increases rapidly around the 67% amorphous value; the rate increase at 60%, 70%, 80%, 90%, and 95% amorphous content is 0.3%, 3%, ×1.3, ×12.1, and ×890, respectively. As will be discussed in Section 2.4, extended fine grinding at low pressures can also generate amorphous compounds, generally up to ~80–90%.²⁰⁾ For the range of 90–95% amorphous content (representative of most ggbs), the average ΔG_d is ~1–50 kJ/mol. These values equate to an increase of k of ~10¹–10³. It is important to remember that these rate increases are only applicable to the initial reactions between the surface of minerals and CO₂. As CO₂ mineralization reactions proceed, the product layer that forms at the surface of particles imposes a resistance to ion diffusion. This phenomenon is well illustrated by the lack of CO₂ mineralization of ggbs in concrete even after decades of service when other cement phases have largely converted to carbonates.²¹⁾

The ‘reactive’ nature of ggbs and its lack of substantial CO₂ mineralization can be understood by examining crystalline and amorphous Ca₂SiO₄ in Fig. 1. An increase in k causes a vertical translation of a mineral until reaching the UCM-regime. Once in the UCM-regime, further increase in k has no acceleratory effect because mass transfer through the product layer is the rate-limiting mechanism. Amorphization not only increases k, it also decreases D generally by ~10^3–5 (i.e., translation to the left in Fig. 1). The combined effect of amorphization is to move a mineral deeper into the UCM-regime. Though increasing k via amorphization will accelerate the initial CO₂ mineralization reactions, it will also increase the time to reach high CO₂ mineralization extents by decreasing of D.

2.4. Variation of Crystallinity

2.4.1. Amorphization via Grinding

Fine grinding—especially at low pressures—can convert crystalline minerals into amorphous compounds.²²⁾ The d at which grinding inevitably acts to induce amorphization instead of generating new surface area (d_{amorphization}) is calculated by Eq. (5). In Eq. (5), K_IC is the mineral-dependent critical stress intensity for crack propagation in Pa·m^1/2, ν is the Poisson ratio, and σ_p is the maximum impact stress in Pa experienced by the particle (K_IC and ν for minerals common to ISM slags can be found in Myers et al. 2019).   

$$d_{amorphization}=\frac{\pi }{2}{\left(\frac{K_{IC}}{2{\sigma }_p\nu }\right)}^2$$(5)

The σ_p can be calculated based on the device type and operating parameters, but typical values for industrial grinding equipment are on the order of 1–10 GPa while laboratory equipment (e.g., ball mills) are typically well below 1 GPa. The d_{amorphization} as a function of σ_p for several minerals common to ISM slag is shown in Fig. 4. Even with industrial grinding equipment d_{amorphization} is generally 0.1–10 μm.

Fig. 4.

The particle diameter at which grinding causes amorphization of crystalline minerals as a function of the compressive pressure of grinding. (Online version in color.)

When viewed with Fig. 1, it is unlikely that most minerals in ISM slag can be pushed deep into the RDM-regime without undergoing amorphization. On the other hand, well-designed current-generation comminution equipment should be able to produce slag of d_transition without substantial amorphization of crystalline compounds.

2.4.2. Crystallization via Slow Solidification

An additional method by which crystallinity is controlled is the solidification and cooling rate applied to molten slag. Rapid solidification is purposefully applied to blast furnace slag, typically using high pressure water jets, to produce amorphous ggbs. Slowing down the cooling and solidification rate of slag ensures that crystalline phases are formed, with slower rates generating larger crystals with higher purity.²³⁾

2.5. Variation of Slag Chemical Composition

From Eq. (2), decreasing ρ_Ca,Mg reduces the time to complete CO₂ mineralization. The total content of Ca and Mg in slag is determined by furnace operations and product quality (i.e., iron and steel). The solidification and cooling processes applied to slag can induce local compositional heterogeneity through the formation of distinct mineral compounds (i.e., local alteration of ρ_Ca,Mg). However, the goal of the current process is to efficiently maximize CO₂ emissions reduction. Since the total CO₂ mineralization potential (ψ: kg-CO₂/kg-slag) is determined by the Ca and Mg content of slag, reducing the ρ_Ca,Mg of some compounds necessarily increases the ρ_Ca,Mg of other compounds. Therefore, there is no method to leverage ρ_Ca,Mg to increase mineralization rate without decreasing mineralization extent.

2.6. Variation of CO₂ Concentration

Equation (2) shows that mineralization rate increases linearly with an increase in C_CO2. The most common method to increase C_CO2 is through some form of CO₂ capture and separation, generating a higher purity CO₂ stream from a lower purity source. Though a wide variety of CO₂ capture and separation technologies exist, they all can generate high concentration, high purity CO₂; thus, no technology-specific analysis is performed.

In addition to increasing the percentage of CO₂ in the gas stream, the CO₂ concentration can be controlled to some extent through control of the temperature (T) and pressure (P) of the gas. Though the van der Waals equation of state should be used since gas-solid CO₂ mineralization involves CO₂, water vapor, and O₂, the ideal gas law (C = P/RT; R: gas constant) provides a more intuitive theoretical basis for how to increase C_CO2.²⁴⁾ Increasing P or decreasing T provide a linear increase in CO₂ concentration. Since gas-solid CO₂ mineralization is limited to temperatures above the freezing point of water (273.15 K), temperature reduction can only provide a slight increase in CO₂ concentration and so is not considered further. Despite high P and/or T CO₂ mineralization is common in the literature, there is insufficient evidence to assess whether the D determined at atmospheric P and T is applicable different conditions. No modification of P is considered in this analysis, modification of T is limited to 95°C.

2.7. Energy Penalty of Grinding Slag

As evidenced by Eq. (2), grinding is a powerful lever to accelerate CO₂ mineralization—at least until d_{amorphization}. In principle, the minimum energy to grind a material is equivalent to the quantity of newly generated surface area (ΔSA) multiplied by the material-specific surface energy (γ). The underlying mineral structure can be used to calculate the mineral-specific γ as has been done for most minerals common to slag.⁴⁾ To determine the maximum efficiency of compressive grinding (η_limit) requires consideration of the volumetric distribution of strain in a material and the geometry of the fracture surface. When such realities are accounted for, η_limit for most minerals is ~7.5% ± 2.5%, dictated primarily by the mineral’s ν.²⁵⁾ Real-world efficiencies (η_actual) for fine grinding are typically ~1–2% with efficiency losses due to non-productive impacts, heat generation, and sound generation. The energy for grinding is calculated using Eq. (6), where η is η_limit for the theoretical minimum and η_actual to approximate current technology. The value γ_slag is the volumetric average of γ based on slag mineral composition.   

$$E_{grind}=\frac{{\gamma }_{slag}\mathrm{\Delta }SA}{\eta }$$(6)

2.8. Energy Penalty of Increasing CO₂ Concentration

The energy to increase the CO₂ concentration (E_concentrate) was calculated by the Gibbs free energy of de-mixing as described in Eq. (7).²⁶⁾ In Eq. (7), n is the number of moles, χ is the mole percentage, η is the efficiency, subscript i indicates the gas type (i.e., CO₂ and non-CO₂ gases), and the other subscripts indicate the gas stream (source: flue gas or air; mineralize: concentrated CO₂; emit: CO₂-depleted gas). The η was set to either 1 (for the thermodynamic limit) or based on the concentration-dependent, empirical ‘2nd-law efficiency’.²⁶⁾ Note that the CO₂ capture extent and purity were set such that the emitted gas stream has a CO₂ concentration roughly equivalent to atmospheric (i.e., 420 ppm). This allows an ISM operator to achieve zero CO₂ steelmaking if the other flue gases are treated.   

$$\begin{array}{l}{E}_{concentrate}=\\ \frac{RT}{\eta }\left(\sum n_{i}ln{\left({\chi }_i\right)}_{mineralize}+\sum n_{i}ln{\left({\chi }_i\right)}_{emit}-\sum n_{i}ln{\left({\chi }_i\right)}_{source}\right)\end{array}$$(7)

3. Methods

As the model described in Section 2 has previously been experimentally verified for a variety slags and operating conditions, here we focus on leveraging the model to optimize slag processing and CO₂ mineralization conditions.

3.1. Slag Composition

Four slags with the properties listed in Table 1 are used for calculations: amorphous BF slag (ggbs), crystalline BF slag (BF), crystalline BOF slag (BOF), and crystalline EAF slag (EAF). The mineral compositions of the crystalline slags were determined using FactSage^® with industry-average slag compositions. A very slow cooling rate was assumed such that the equilibrium composition at the solidus temperature could be used. Both D_slag and k_slag are volume-average values based on the slag composition. When grinding below the application range of EMT, D_slag and k_slag are replaced with mineral-specific D and k with calculations performed separately for all minerals in a slag.

Table 1. Properties of prototypical slags used in analysis; all values are mass% unless otherwise noted.

	ggbs§	BF§	BOF§	EAF§
ψ (kg-CO2/kg)	0.408	0.408	0.462	0.354
Dslag (m2/s)	1.2×10−18	9.1×10−16	3.7×10−13	1.5×10−13
kslag (m/s)‡	4.5×10−3	2.4×10−7	5.9×10−5	2.1×10−5
γslag (J/m2)	2.5	3.5	3.9	3.6
cP (J/kg·K)	726	1214	1213	1223
ρ (kg/m3)	2906	2902	3825	4217
ηlimit (grinding)	7.73%	7.76%	7.85%	7.96%
Crystallinity	0%	100%	100%	100%
Ca2MgSi2O7	–	56.8%	–	–
Ca2SiO4	–	–	27.1%	46.5%
Ca2Al2SiO7	–	35.1%	–	–
FeO	–	–	–	34.3%
CaO	–	–	24.0%	–
Ca2Fe2O5	–	–	20.2%	–
Fe2SiO4	–	–	16.1%	–
CaAl2O4	–	–	–	10.2%
MgO	–	–	6.4%	8.2%
CaAl2Si2O8	–	4.7%	–	–
CaSiO3	–	3.3%	–	–
Fe	–	–	3.1%	0.8%
Al2O3	–	–	3.0%	–

§:  elemental composition based on industry averages per 27);

‡:  calculated as the volume average of k for each slag

3.2. Particle Size Distribution

The PSD for each polycrystalline slag that would maximizing the volume of slag near d_transition without grinding below d_{amorphization} (assuming σ_p=5 GPa) is considered the baseline. Coincidentally, the baseline for polycrystalline slags (i.e., volumetric PSD peak diameter ~7 μm) is very similar to the PSD of ggbs (i.e., volumetric PSD peak diameter ~5 μm) required to meet the regulatorily-required Blaine surface area criteria (>4500 cm²/g).¹⁸⁾ Therefore, the same PSD (i.e., that of ggbs) was used for all slags (termed: ‘Standard PSD’). The ‘Overground PSD’ was set such that the volumetric PSD peak diameter was ~1/10 of ‘Standard PSD’. The ‘Large PSD’ was set to a volumetric PSD peak diameter of ~×10 the ‘Standard PSD’. All PSD are provided in Fig. 5.

Fig. 5.

Particle size distribution classes used for all analyses. Displayed Blaine specific surface areas are for ggbs. (Online version in color.)

For calculation of the energy penalty of grinding-based CO₂ mineralization acceleration, the ‘Large PSD’ was used for the baseline mineralization rate but the energy penalty was calculated assuming an initially unground state.

3.3. Gas Conditions

The baseline CO₂ concentration was assumed to be 7.2 mol% to match the weighted average of the flue gas from the lime furnace, reheating furnace, and sinter furnace of an integrated BF-BOF plant (Fig. 6).²⁸⁾ Mixing these three flue gases provides roughly the correct amount of CO₂ to fully regenerate the fossil carbonates decomposed in ISM. The water vapor content of this merged gas stream (0.090 kg-H₂O/kg-dry gas) is sufficient to achieve a RH of 100% at the baseline operating temperature of 30°C.

Fig. 6.

Flue gas streams available at an integrated BF-BOF plant showing the benefit to CO₂ capture systems if low volume, low CO₂ concentration streams are treated via CO₂ mineralization. (Online version in color.)

Treating these flue gases via CO₂ mineralization has the added benefit of increasing the CO₂ concentration and decreasing the volume of gas sent to CO₂ capture and storage as seen by comparing the white and red bubbles in Fig. 6 (e.g., to reach net-zero CO₂ steelmaking). Lime production flue gas is sent to CO₂ mineralization instead of the higher volume, lower CO₂ concentration coke production flue gas because the coke production flue gas contains ~11% of an ISM facility’s CO₂. To fully mineralize this CO₂ would consume all the available slag. As a result, the flue gas sent to CCS would increase in volume (from 78% to 85% of total), increase in the number of streams (from 3 to 5), and decrease in CO₂ concentration (from 24.3% to 20.7%). Note that coke production flue gas is exhausted from the coke oven batteries and is not to be confused with coke oven gas (COG) which is used as fuel in other ISM subprocesses.

4. Results

4.1. CO₂ Mineralization Rate and Extent

The expected CO₂ mineralization extent over time for the four slag types with various degrees of grinding is shown in Fig. 7. The relative timeframes involved using the ‘Standard PSD’ for each slag are particularly noteworthy. While BOF slag (Fig. 7(c)) and EAF slag (Fig. 7(d)) require only a few days to near complete CO₂ mineralization, BF slag (Fig. 7(b)) requires months, and ggbs (Fig. 7(a)) requires decades. For the crystalline slags, the reduction in D for ‘Overground PSD’ (via amorphization) decreases the CO₂ mineralization rate despite the increase in SA and k. This effect is not observed for ggbs as the starting material is already fully amorphous. As the evaluated slags all contain reactive content >20 vol%, ML is not observed and CO₂ mineralization extent eventually reaches 100%.

Fig. 7.

CO₂ mineralization extent over time as a function of particle size distribution for a) ggbs, b) BF slag, c) BOF slag, and d) EAF slag. All reactions set to 30°C, 101.325 kPa, 100% relative humidity, and gas phase CO₂ concentration of 7.2 mol%. Note the change in x-axis magnitude. (Online version in color.)

The expected CO₂ mineralization extent over time for the slags at 7.2% and 100% CO₂ is shown in Fig. 8. For all scenarios, the ‘Standard PSD’ was used. Altering the CO₂ concentration provides a linear change to mineralization rate for all slags. This supports the concept of using concentrated CO₂ gas streams when available. Nevertheless, increasing C_CO2 is not a panacea for accelerating CO₂ mineralization. This fact is clearly demonstrated by the barely registerable change in CO₂ mineralization extent of ggbs even when increasing CO₂ to 100%.

Fig. 8.

CO₂ mineralization extent over time for the prototypical slags when reacted at a CO₂ concentration of 7% and 100%. All reactions use the ‘Baseline’ PSD with reactions set to 30°C, 101.325 kPa, and 100% RH. (Online version in color.)

4.2. Energy Penalty of CO₂ Mineralization Acceleration

The energy penalty for accelerating CO₂ mineralization via grinding is provided in Fig. 9. The theoretical limit refers to a perfectly efficient operation (i.e., η = 1 of Eq. (6)). An empirical efficiency of 2% is used to indicate likely performance with current technology. The baseline mineralization rate by which the acceleration factor was calculated is the ‘Large PSD’ for each slag. Energy consumption is calculated assuming an initially unground state. For the crystalline slags, acceleration of CO₂ mineralization is reduced as the minerals are converted from crystalline to amorphous compounds. This effect is less severe for BF slag as the difference in D between crystalline and amorphous phases is smaller than that of BOF and EAF slag. As the σ_p experienced during grinding largely determines the onset of amorphization, greater CO₂ mineralization acceleration could be achieved using higher pressure comminution techniques. The inherently amorphous nature of ggbs means that grinding monotonically increases the CO₂ mineralization rate. The reader is cautioned that the large acceleration factor for ggbs in Fig. 9 is relative to a very slow mineralization rate as indicated in Fig. 7(a).

Fig. 9.

Energy penalty of accelerating CO₂ mineralization via grinding in terms of the theoretical efficiency limit and an empirical efficiency of 2%. Acceleration rate factors are relative to ‘Large PSD’. Energy consumption is calculated from an unground state. (Online version in color.)

The energy penalty for accelerating CO₂ mineralization via increasing CO₂ concentration is provided in Fig. 10. Both the theoretical minimum energy penalty (i.e., η = 1 of Eq. (7)) and the empirical, concentration-dependent 2nd-law efficiency are shown for a range of CO₂ concentrations. The decrease in separation efficiency that comes with a decrease in CO₂ concentrations leads to DAC incurring a large energy burden in practice, regardless of the specific process applied (e.g., amine scrubbing, alkali scrubbing, solid sorbent).²⁹⁾ For higher CO₂ concentration streams, the acceleratory effect is limited by the upper bound of 101.325 kPa and 100% CO₂.

Fig. 10.

Energy penalty of accelerating CO₂ mineralization via increasing the CO₂ concentration of different CO₂ sources in terms of the thermodynamic limit and the empirical 2nd-law efficiency. (Online version in color.)

5. Discussion

With net-zero greenhouse gas pledges being made for 2050–2070 by politicians in the major ISM countries, the hard-to-abate industries will need to move beyond efficiency improvement of existing systems. While ISM emissions are typically thought of as being based entirely on fossil fuel use, the use of fossil carbonates is responsible for ~10% of total CO₂ released—this portion is even larger for electric arc furnaces. The regeneration of calcium and magnesium carbonates by reacting steelmaking slags with ISM flue gas represents a win-win for the industry. The story for ironmaking slags is less clear. Currently, both the cement industry and ISM industry claim that ggbs reduces their CO₂ emissions, an underappreciated source carbon leakage through double counting. The sale of ggbs provides revenue to ISM on the order of ~27.50 USD/t, equating to ~8.70 USD/t-iron.³⁰⁾ Since the market prices for pig iron is ~266 USD/t, the sale of ggbs constitutes ~3% of the total revenue from ironmaking activities. Given the extremely thin profit margins of the global ISM industry, ggbs is an important revenue stream.³¹⁾ However, with the spread of carbon taxes/markets, the embodied CO₂ emissions of ironmaking slag (0.408 t-CO₂/t-slag, or 0.129 t-CO₂/t-iron) will become financially important. When the carbon tax/price exceeds 67.40 USD/t-CO₂ (a value already exceeded in the EU and US), the ‘value’ of mineralizing CO₂ using slag will exceed the revenue gained from sale of ggbs. Note that since the cement industry currently claims the CO₂ emissions reductions from using ggbs as a SCM, it is unlikely that the ISM industry can seize those same emissions reductions without the value of ggbs decreasing in kind.

6. Conclusion

Methods to accelerate the mineralization of CO₂ using iron and steelmaking slags were evaluated in terms of the potential magnitude of acceleration and energy efficiency. Analysis indicated that diffusion through the product layer is the primary rate-limiting factor in almost all instances. As such, producing and maintaining crystalline compounds is of central importance to achieving rapid CO₂ mineralization because of their higher diffusion coefficients (typically ×10³–10⁵) relative to chemically equivalent amorphous compounds. Crystalline slags can be generated by eliminating the rapid cooling processes used to produce ggbs for the cement industry. As slow cooling eliminates several energy-intensive processing steps associated with ggbs production, this activity comes at a net energy savings for ISM.

In addition to crystallinity, the diffusivity through the product layer is empirically very sensitive to the relative humidity of the gas phase. Higher relative humidity generates faster CO₂ mineralization rates across a wide variety of compounds. Luckily, the various flue gases available at an ISM plant contain sufficient water vapor to maintain saturated water vapor conditions at the temperatures used in direct, gas-solid CO₂ mineralization (i.e., ~30°C).

After ensuring a high diffusivity through the product layer, grinding provides the greatest lever by which to increase the CO₂ mineralization rate. However, care must be taken when grinding slag not to induce amorphization; in general, grinding of slag should be limited to ~1–10 μm and high-pressure grinding equipment should be used. Excessive grinding can reduce the overall CO₂ mineralization rate due to the reduction in product diffusivity that comes with amorphization. With proper care, grinding can increase the mineralization rate by a factor of ~10³ relative to slag sized ~50 μm. Grinding is a particularly energy efficient method, with ~1000× acceleration achieved at ~100 MJ/t-CO₂ in the theoretically most efficient processes, and ~500 MJ/t-CO₂ using modern comminution machinery.

The least effective and least efficient methods to increase the CO₂ mineralization rate are to increase the temperature and/or CO₂ concentration. Elevated temperatures increase the CO₂ mineralization reaction rate, but since this is rarely the rate-limiting phenomenon it has no acceleratory effective on the overall CO₂ mineralization rate. While increasing the CO₂ concentration provides a linear increase in the CO₂ mineralization rate, its potential effect is limited by the difference between available ISM flue gas concentrations (i.e., 4.6–19.4%) and 100% CO₂. Increase the CO₂ concentration also comes at a high energy penalty due to thermodynamics inherent in gas separation and purification. Using a mixture of flue gases from ISM, the minimum energy requirement to accelerate CO₂ mineralization by 10× is ~150 MJ/t-CO₂. Empirical gas separation efficiencies suggest 10× acceleration will require closer to ~350 MJ/t-CO₂. For the same energy expenditure current grinding technology would increase the CO₂ mineralization rate by ~500×.

Acknowledgement

Work at the Lawrence Livermore National Laboratory was performed under the auspices of the U.S. DOE under Contract DE-AC52-07NA27344.

References

• 1)   J. C.  Minx,  W. F.  Lamb,  R. M.  Andrew,  J. G.  Canadell,  M.  Crippa,  N.  Döbbeling,  P. M.  Forster,  D.  Guizzardi,  J.  Olivier,  G. P.  Peters,  J.  Pongratz,  A.  Reisinger,  M.  Rigby,  M.  Saunois,  S. J.  Smith,  E.  Solazzo and  H.  Tian: Earth Syst. Sci. Data, 13 (2021), 5213. https://doi.org/10.5194/essd-13-5213-2021
• 2)  ArcelorMittal: ArcelorMittal Sestao to become the world’s first full-scale zero carbon-emissions steel plant, https://corporate.arcelormittal.com/media/press-releases/arcelormittal-sestao-to-become-the-world-s-first-full-scale-zero-carbon-emissions-steel-plant, (accessed 2022-06-18).
• 3)  SSAB: SSAB plans a new Nordic production system and to bring forward the green transition, https://www.ssab.com/news/2022/01/ssab-plans-a-new-nordic-production-system-and-to-bring-forward-the-green-transition, (accessed 2022-06-18).
• 4)   C. A.  Myers,  T.  Nakagaki and  K.  Akutsu: Int. J. Greenh. Gas Control, 87 (2019), 100. https://doi.org/10.1016/j.ijggc.2019.05.021
• 5)   K.  Horii,  T.  Kato,  K.  Sugahara,  N.  Tsutsumi and  Y.  Kitano: Nippon Steel Sumitomo Met. Tech. Rep., (2015), No. 109, 5.
• 6)   J.  Lehne and  F.  Preston: Making Concrete Change: Innovation in Low-carbon Cement and Concrete, Chatham House, London, (2018), 1.
• 7)   C.  Myers and  T.  Nakagaki: Environ. Res. Lett., 15 (2020), 124018. https://doi.org/10.1088/1748-9326/abc217
• 8)   C. A.  Myers: Proc. Goldschmidt2021, European Association of Geochemistry, Aubiere, (2021), 4352. https://doi.org/10.7185/gold2021.4352
• 9)   C.  Myers and  T.  Nakagaki: ISIJ Int., 58 (2018), 211. https://doi.org/10.2355/isijinternational.ISIJINT-2017-478
• 10)   R. L.  Berger and  W. A.  Klemm: Cem. Concr. Res., 2 (1972), 647. https://doi.org/10.1016/0008-8846(72)90002-6
• 11)   F.  Larachi,  I.  Daldoul and  G.  Beaudoin: Geochim. Cosmochim. Acta, 74 (2010), 3051. https://doi.org/10.1016/j.gca.2010.03.007
• 12)   R. C.  Longo,  K.  Cho,  P.  Brüner,  A.  Welle,  A.  Gerdes and  P.  Thissen: ACS Appl. Mater. Interfaces, 7 (2015), 4706. https://doi.org/10.1021/am508313g
• 13)   A.  Funk and  H. R.  Trettin: Ind. Eng. Chem. Res., 52 (2013), 2168. https://doi.org/10.1021/ie302972k
• 14)   Y.  Tao,  Y.  Mu,  W.  Zhang and  F.  Wang: Front. Mater., 7 (2020), 299. https://doi.org/10.3389/fmats.2020.00299
• 15)   W.  Ashraf and  J.  Olek: Cem. Concr. Compos., 93 (2018), 85. https://doi.org/10.1016/j.cemconcomp.2018.07.004
• 16)   C. J.  Goodbrake,  J. F.  Young and  R. L.  Berger: J. Am. Ceram. Soc., 62 (1979), 168. https://doi.org/10.1111/j.1151-2916.1979.tb19046.x
• 17)   D.  Tromans and  J. A.  Meech: Miner. Eng., 14 (2001), 1359. https://doi.org/10.1016/S0892-6875(01)00151-0
• 18)   O.  Aßbrock,  W.  Brameshuber,  K.  Droll,  E.  Lang,  H.  Motz,  C.  Müller,  R.  Pierkes,  L.  Urbonas and  E.  Siebel: Ground Granulated Blast Furnace Slag (GGBS) as a Concrete Additive, The Federal Association of the German Ready-Mixed Concrete Industry, Dusseldorf, (2007), 1.
• 19)   R.  Walker and  S.  Pavía: Mater. Struct., 44 (2011), 1139. https://doi.org/10.1617/s11527-010-9689-2
• 20)   S.  Atashin,  J. Z.  Wen and  R. A.  Varin: J. Alloy. Compd., 618 (2015), 555. https://doi.org/10.1016/j.jallcom.2014.08.142
• 21)   R.  Taylor,  I. G.  Richardson and  R. M. D.  Brydson: Cem. Concr. Res., 40 (2010), 971. https://doi.org/10.1016/j.cemconres.2010.02.012
• 22)   N.  Romero Sarcos,  D.  Hart,  H.  Bornhöft,  A.  Ehrenberg and  J.  Deubener: J. Non-Cryst. Solids, 556 (2021), 120557. https://doi.org/10.1016/j.jnoncrysol.2020.120557
• 23)   M.  Gautier,  J.  Poirier,  F.  Bodénan,  G.  Franceschini and  E.  Véron: Int. J. Miner. Process., 123 (2013), 94. https://doi.org/10.1016/j.minpro.2013.05.002
• 24)   P.  Fishbane,  S.  Gasiorowicz and  S.  Thornton: Physics for Scientists and Engineers, 3rd ed., Pearson Prentice Hall, Hoboken, NJ, (2005), 1.
• 25)   D.  Tromans: Miner. Eng., 21 (2008), 613. https://doi.org/10.1016/j.mineng.2007.12.003
• 26)   J.  Wilcox: Carbon Capture, Springer Science & Business Media, New York, (2012), 1.
• 27)   C. A.  Myers: Ph.D. thesis, Waseda University, (2019), http://hdl.handle.net/2065/00063313, (accessed 2022-02-10).
• 28)   S.  Santos: Iron and steel ccs study (techno-economics integrated steel mill), IEAGHG, https://ieaghg.org/docs/General_Docs/Reports/2013-04.pdf, (accessed 2022-06-18).
• 29)   F.  Sabatino,  A.  Grimm,  F.  Gallucci,  M.  van Sint Annaland,  G. J.  Kramer and  M.  Gazzani: Joule, 5 (2021), 2047. https://doi.org/10.1016/j.joule.2021.05.023
• 30)  U.S. Geological Survey: Mineral commodity summaries 2020, (2020), https://doi.org/10.3133/mcs2020, (accessed 2022-06-18).
• 31)   C.  Bataille: Low and Zero Emissions in the Steel and Cement Industries: Barriers, Technologies and Policies, OECD Green Growth Papers, OECD Publishing, Paris, (2020), 1.