a, Model-based CO2 reconstruction21 and relevant warm (orange bands) and cold (blue bands) climatic periods. b, Inferred GMSL and ice volume from Mallorcan POS are shown as black markers (age uncertainties are 2?; the GMSL of the marker corresponds to the mode; uncertainties are 16th and 84th percentiles). The sample code for each POS is indicated on the grey band between panels. Coloured curves show three different GMSL reconstructions (uncertainties on GMSL curves are 1? ). See Methods for the derivation of the GMSL curves. PCO, Pliocene Climatic Optimum.

Reconstructing the evolution of sea level during past warmer epochs such as the Pliocene provides insight into the response of sea level and ice sheets to prolonged warming1. Although estimates of the global mean sea level (GMSL) during this time do exist, they vary by several tens of metres2,3,4, hindering the assessment of past and future ice-sheet stability. Here we show that during the mid-Piacenzian Warm Period, which was on average two to three degrees Celsius warmer than the pre-industrial period5, the GMSL was about 16.2 metres higher than today owing to global ice-volume changes, and around 17.4 metres when thermal expansion of the oceans is included. During the even warmer Pliocene Climatic Optimum (about four degrees Celsius warmer than pre-industrial levels)6, our results show that the GMSL was 23.5 metres above the present level, with an additional 1.6 metres from thermal expansion. We provide six GMSL data points, ranging from 4.39 to 3.27 million years ago, that are based on phreatic overgrowths on speleothems from the western Mediterranean (Mallorca, Spain). This record is unique owing to its clear relationship to sea level, its reliable U–Pb ages and its long timespan, which allows us to quantify uncertainties on potential uplift.

Accurate predictions of future sea-level change hinge on our understanding of how ice sheets respond to changes in global temperature. To understand ice-sheet stability under prolonged warming (such as if the current level of temperature increase continues), we can use reconstructed sea level during past periods when Earth’s climate was warmer than today1. The Pliocene epoch (5.33 to 2.58 million years ago, Ma) was the most recent extended global warm period immediately preceding the inception of the high-magnitude glacial/interglacial variations of the Pleistocene8. The mid-Piacenzian Warm Period (MPWP), an interval during the Late Pliocene (3.264 to 3.025 Ma), has been used as an analogue for future anthropogenic warming since atmospheric CO2 conditions were comparable to present-day values (~400 ppm)9 and estimated global mean temperatures were elevated by 2–3?°C relative to the pre-industrial period5.

Oxygen isotope ratios from benthic foraminifera10 paired with deep ocean temperature estimates have been used to approximate ice-volume-equivalent GMSL changes over the Pliocene11,12. While invaluable, these approaches are limited by uncertainties in the methodology and a number of factors (for example, post-burial diagenesis, long-term changes in seawater chemistry and salinity) that are poorly constrained and may bias the sea-level estimates3. Field mapping of palaeoshorelines has been a complementary approach...

Here we present Pliocene sea-level data from Coves d’Artà in the western Mediterranean (Mallorca, Spain; Fig. 1a, b) that are based on U–Pb absolute-dated phreatic overgrowths on speleothems (POS). POS offer several important benefits over other Pliocene sea-level indicators since they store all information needed for a meaningful sea-level index point: (1) precise spatial geographic positioning, (2) accurate elevation, (3) clear indicative meaning (their growth covers the full tidal range, thus having an explicit relation to past sea level; see Methods), and (4) an absolute age (since the crystalline aragonite/calcite often contains suitable uranium concentrations for robust dating19). POS are primarily precipitated in caves, at the water/air interface as CO2 degasses from brackish cave pools. The water table in these caves is, and was in the past, coincident with sea level, given that the caves are at most 300 m away from the coast and the karst topography is low. Six POS levels have been identified at elevations from 22.6 to 31.8 m.a.p.s.l. (uncertainties in the elevation measurement and the indicative range are less than 1 m; Fig. 1c, Table 1)...

a, Map showing Mallorca (red circle) in the western Mediterranean. b, Location of Coves d’Artà on the island. c, Longitudinal profile through the lower section of the cave showing the present-day elevation and ages of the six POS horizons and the sampling sites. d, POS at three elevations within the Infern Room with close-up views (insets). Maps (a, b) are available under CC Public Domain License from https://pixabay.com/illustrations/map-europe-world-earth-continent-2672639/ and https://pixabay.com/illustrations/mallorca-map-land-country-europe-968363/, respectively.

Contribution of different corrections (GIA, uplift and thermal expansion) and uncertainties when inferring the GMSL from the POS elevation (this breakdown is for AR-03; see Extended Data Fig. 8 for all POS). Probability density function of the POS elevation with consecutive corrections for the measurement and indicative range leading to an estimate of local sea level (LSL; blue), GIA (orange), long-term uplift (purple curve) and thermal expansion (yellow). We choose the mode (solid black line) as the best estimate and the 16th and 84th percentiles (dashed black line) as the uncertainty range.

Given that global average temperatures during the MPWP were 2–3?°C higher than pre-industrial values5 and CO2 concentration was 400 ppm (ref. 9), our results indicate that an ice volume equivalent to a GMSL change of 16.2 m.a.p.s.l. (5.6–19.2 m.a.p.s.l.) may eventually melt (over hundreds to thousands of years) if future temperatures stabilize at that level of warming. Given present-day melt patterns26, this sea-level rise is likely to be sourced from a collapse of both Greenland and the West Antarctic ice sheets. A temperature increase to 4?°C above pre-industrial levels is comparable to conditions during the Pliocene Climatic Optimum6 with a GMSL estimate of 23.5 m.a.p.s.l. (9.0–26.7 m.a.p.s.l.), which indicates further ice melt if temperatures stabilize at this higher level. Thermal expansion is expected to cause additional sea-level rise in these scenarios.

High-entropy alloys are a class of materials that contain five or more elements in near-equiatomic proportions1,2. Their unconventional compositions and chemical structures hold promise for achieving unprecedented combinations of mechanical properties3,4,5,6,7,8. Rational design of such alloys hinges on an understanding of the composition–structure–property relationships in a near-infinite compositional space9,10. Here we use atomic-resolution chemical mapping to reveal the element distribution of the widely studied face-centred cubic CrMnFeCoNi Cantor alloy2 and of a new face-centred cubic alloy, CrFeCoNiPd. In the Cantor alloy, the distribution of the five constituent elements is relatively random and uniform. By contrast, in the CrFeCoNiPd alloy, in which the palladium atoms have a markedly different atomic size and electronegativity from the other elements, the homogeneity decreases considerably; all five elements tend to show greater aggregation, with a wavelength of incipient concentration waves11,12 as small as 1 to 3 nanometres. The resulting nanoscale alternating tensile and compressive strain fields lead to considerable resistance to dislocation glide...

...These deformation mechanisms in the CrFeCoNiPd alloy, which differ markedly from those in the Cantor alloy and other face-centred cubic high-entropy alloys, are promoted by pronounced fluctuations in composition and an increase in stacking-fault energy, leading to higher yield strength without compromising strain hardening and tensile ductility. Mapping atomic-scale element distributions opens opportunities for understanding chemical structures and thus providing a basis for tuning composition and atomic configurations to obtain outstanding mechanical properties.

In principle, high-entropy alloys (HEAs) should form a single phase with what has been presumed to be a random solid solution1. Some HEAs, in particular the CrCoNi-based systems, display exceptional mechanical performance, including high strength, ductility and toughness, particularly at low temperatures3,5, making them potentially attractive materials for many structural applications. These special characteristics have been attributed to factors that include high entropy, sluggish diffusion and severe lattice distortion13, issues that are related to the degree of randomness of the solid solution. A fundamental question is whether such solid solutions with multiple principal elements involve unconventional atomic structures or elemental distributions, such as local chemical ordering or clustering, that could affect the defect behaviour and thus enhance mechanical properties. Most theoretical descriptions of solid solutions in HEAs assume that they comprise a random distribution of different atomic species. However, some simulations and more limited experimental results14,15,16,17,18 suggest that local variations in chemical composition or even short-range order may exist in HEAs. All five elements in the most studied HEA alloy, CrMnFeCoNi, belong to the first row of transition metals in the Periodic Table, with similar atomic size and electronegativity (a measure of tendency to form intermediate compounds instead of primary solid solutions19,20)

a, HAADF image of atomic structure, taken with the [110] zone axis, and corresponding EDS maps for individual elements of Cr, Mn, Fe, Co and Ni. b, Line profiles of atomic fraction of individual elements taken from the respective EDS maps in a; each line profile represents the distribution of an element in a (11¯1) plane projected along the [110] beam direction. c, Plots of pair correlation function S(r) of individual elements against concentration wavelength r; S(r) is shifted by C¯2, where C¯ denotes the average atomic fraction of the corresponding element. d, Magnification of local regions in a (all to same scale), showing small groups of neighbouring atomic columns with similar brightness. e, Comparison of the local concentration distribution of individual elements for the same region, showing that an Ni-poor region is filled by more Fe and Co than Cr and Mn.

a, HAADF image of atomic structure, taken with the [110] zone axis, and corresponding EDS maps for individual elements of Cr, Fe, Co, Ni and Pd. b, Line profiles of atomic fraction of individual elements taken from respective EDS maps in a; each line profile represents the distribution of an element in a (11¯1) plane projected along the [110] beam direction. c, Plots of pair correlation functions S(r) of individual elements against concentration wavelength r; S(r) is shifted by C¯2, where C¯ denotes the average atomic fraction of the corresponding element. d, Comparison of the local concentration distribution of individual elements for the same region, showing no obvious preference for specific neighbours.

a, HAADF image taken with the ⟨110⟩ zone axis, showing the atomic structure of a 60° full dislocation, with the Burgers vector b of 12[110]. This 60° dislocation is dissociated into a 30° partial and a 90° partial. The distance between the two partials—that is, the stacking fault width—is as small as about 1 nm. b, TEM image taken during in situ TEM straining experiments, showing a dislocation array. Some of the moving dislocation lines exhibit widely separated leading and trailing partials (marked by yellow arrows), showing the temporary pinning of one of the partials. c, TEM images showing the sluggish motion of dislocations in a pile-up, where the leading dislocation was obstructed by a strong obstacle. d, TEM images at an early time (left image) and a late time (right image) showing massive cross-slip everywhere in the dislocation pile-up. e, TEM image showing the activation of new slip systems due to the interaction of intersecting slip bands. Green and yellow arrows respectively indicate the primary and secondary dislocation slip bands. f, Post-mortem TEM images showing dislocation microstructures in large-scale samples at the early stage of plastic deformation (left), as well as at the late stage of plastic deformation (right) with an applied large strain of about 30%, where dislocation interactions and multiplication are complex, resulting in a high dislocation density.

a, Uniaxial tensile stress–strain curves measured at room temperature (293 K) and at liquid nitrogen temperature (77 K) for CrFeCoNiPd (marked as Pd HEA) and CrMnFeCoNi (marked as Mn HEA) with an average grain size of about 130 ?m. b, Same as a except for an average grain size of about 5 ?m. c, Comparison of yield strength between the CrFeCoNiPd alloy and other related HEAs (see the yield strength data in Extended Data Table 1), which have the pure fcc phase or combined fcc and hexagonal close-packed (hcp) phases. d, Comparison of atomic strain distribution between the CrFeCoNiPd and CrMnFeCoNi alloys, based on HAADF image and corresponding maps of horizontal normal strain (?xx), vertical normal strain (?yy) and shear strain (?xy).

The rectangular bars were homogenized in vacuo at 1,200?°C for 24 h before rolling into 1.8-mm-thick plates at room temperature. Single fcc phase was obtained and confirmed by X-ray diffraction (Extended Data Fig. 1). Equiaxed grain microstructures with average grain sizes of about 130 ?m and 5 ?m were obtained by recrystallizing at 1,150?°C for 1 h and 20 min in vacuum, respectively.

Lithium-ion batteries (LIBs) have emerged as a remarkable power source for consumer electronics, electric and hybrid electric vehicles, and large-scale energy storage, owing to their high energy and power density.(1?3) The fast-growing electric-vehicle market has strongly boosted the application of LIBs, resulting in the inevitable generation of a large amount of spent LIBs each year. It is estimated that the mass of discarded LIBs will exceed 11 million tons by 2030.(4) Discarded LIBs have the dual attributes of a resource and an environmental hazard: On the one hand, they are rich in valuable metals, such as lithium and cobalt; on the other hand, they contain organic substances harmful to the environment. Recycling of discarded LIBs will not only address the problem of resource shortages but will also avoid environmental pollution.(5)

State-of-the-art techniques for recycling of spent LIBs have been reviewed in several studies.(2,6?10) Generally speaking, existing recovery strategies can be divided into pyrometallurgical, direct generation, and hydrometallurgical processes. Pyrometallurgical processes are usually undertaken at high temperature, resulting in relatively high energy consumption.(11?13) The alloyed product and residue require further processing to obtain high-purity products.(14) Direct generation has stringent requirements for the purity of the feed and is not appropriate for recycling materials containing large amounts of impurities.(15) In contrast, hydrometallurgical methods have obvious advantages, including high recovery efficiency of valuable metals, mild reaction conditions, and environmentally friendly conditions. These advantages make hydrometallurgical methods a preferable and promising approach for disposing of spent LIBs; however, the inorganic(16?19) and organic(14,20?22) acids used in hydrometallurgical processing may cause secondary pollution, such as acidic waste waters and production of toxic gases (Cl2, SO3, and NOx) during the leaching process, which are threats to the environment. Therefore, developing a method to reduce acid consumption, decrease the emissions of the secondary pollution, and increase recovery efficiencies of metals is critical. A salt-assisted calcination method has recently been proposed that avoids the above issues and increases recycling efficiency.(23)

Salt-assisted calcination has become widely used as a metal-recovery method for waste materials due to its high reactivity, high volatility, low melting point, and high solubilities of salts.(23?30) In addition, the solid salt agents used are generally environmentally friendly, and the process has simple operation and low-cost equipment.(28) For example, Liu et al.(31) reported a vacuum chlorinating process for simultaneous sulfur fixation and lead recovery from spent lead-acid batteries using calcium chloride (CaCl2) and silicon dioxide (SiO2) as reagents. Dang et al.(23) proposed a chlorination roasting method to recycle lithium from a pyrometallurgical slag using three chlorine donors; namely, NaCl, AlCl3, and CaCl2. These findings showed that it is possible to recover metals through salt-assisted roasting, but the reaction temperature of these solid salt fluxing agents is quite high.

In this work, we developed a combination of low-temperature ammonium salt roasting and water leaching to efficiently and environmentally recover valuable metals from spent LIBs. NH4Cl, a nontoxic and noncorrosive solid chlorinating agent, can be decomposed to NH3 and HCl gases above 230 °C; (32) the introduction of NH4Cl as a chlorinating agent into the calcination process can therefore destroy the crystal structure of LiCoO2 and enable precipitation of Li and Co. Owing to the high solubilities of the chloride salts, the metals are recovered by water leaching after chlorination. The effects of various parameters on recovery efficiencies of Co and Li, including calcination temperature, time, and mass ratio of LiCoO2 to NH4Cl, were systematically investigated. The mechanism underlying the low-temperature molten-salt-assisted recovery process is discussed in detail. The method was proven by efficiently recovering valuable metals from LiMn2O4 and LiCo1/3Mn1/3Ni1/3O2 spent LIBs. This study presents an environmentally friendly and efficient method for recovery of metals from mixed cathode materials of spent LIBs.

Figure 1. Flowsheet of the proposed recycling.

Figure 2. Relationship between ?G and temperature for different reactions.

Figure 3. Effects of (a) temperature, (b) LiCoO2/NH4Cl mass ratio, and (c) roasting time on the leaching efficiency of Li and Co.

Figure 5. Possible reaction pathway of the cotreatment process and SEM patterns of different stages: (a) NH4Cl, (b) LiCoO2, (c) mixed raw materials, and (d) sample after roasting.

Figure 6. XRD pattern of recycled CoC2O4·2H2O and Li2CO3.

Figure 7. Leaching efficiency of different samples.

Based on the above discussions, Li and Co can be recovered as chlorides from LiCoO2 powders by roasting with NH4Cl. To verify the feasibility of recovering these metals from other LIBs, wastes from LiMn2O4 and LiCo1/3Mn1/3Ni1/3O2 batteries were also examined. Spent cathode materials were prepared using the pretreatment process described above. Similar to the LiCoO2 material, LiMn2O4 and LiCo1/3Mn1/3Ni1/3O2 materials are metal oxides. The crystal structures of LiMn2O4 and LiCo1/3Mn1/3Ni1/3O2 are spinel and layered, respectively. The main metal element ingredients of waste LiMn2O4 and LiCo1/3Mn1/3Ni1/3O2 materials are shown in Table S1. The waste cathode powders were then completely reacted with NH4Cl by roasting under optimized calcination conditions. The metal elements were recovered by water leaching. Using this process, 97.99% Li and 95.27% Co were recovered from the spent LiMn2O4 material, while the leaching efficiencies of Li, Ni, Co, and Mn reached 94.95%, 92.87%, 91.59%, and 90.91%, respectively, from the spent LiCo1/3Mn1/3Ni1/3O2 battery material (Figure 7). After leaching, Ni2+, Co2+, and Mn2+ in the leachate can also be recovered and separated from lithium by the coprecipitation method. The coprecipitation reaction can use oxalic acid, sodium hydroxide and ammonia–water, and sodium carbonate as the precipitating reagents.(42?45)

A closed-loop and environmentally friendly process, including salt roasting, water leaching, and precipitation, was developed to recover valuable metals from spent LIBs. Using hydrogen chloride gas produced by the decomposition of NH4Cl as an acid source, Li and Co were released from LiCoO2. The optimal calcination conditions for the recycling process were 350 °C, a mass ratio of LiCoO2/NH4Cl 1:2, and a 20 min reaction time. Leaching efficiencies of Li and Co were 99.18% and 99.3%, respectively, using water leaching at a solid–liquid ratio of 100 g/L. Based on thermodynamic analysis and characterization conducted by XRD and XPS, a reaction mechanism was proposed. More than 90% of the Li and Co from spent LiNi1/3Co1/3Mn1/3O2 and LiMn2O4 cathode materials could be recovered by this novel process. In summary, this study presents a promising technology for metal recovery from spent LIBs.

Because limited battery storage capability decreases the feasibility of electrical engines in long-distance transport and goods shipping(1) and the inefficiency of natural gas storage limits natural gas engine capabilities in long-distance haulage,(2) combustion engines are likely to be used for the foreseeable future. However, as the world drives for cleaner, renewable energy and fossil fuels become more difficult and expensive to extract, replacements for diesel fuel are currently being explored. Created through the transesterification of lipids into fatty acid methyl esters,(3) biodiesel is gaining popularity as a renewable, sustainable fuel because of its ability to directly replace diesel fuel in many engines.(4) However, as biodiesel usage is predicted to increase worldwide,(5,6) concerns have been raised over the health impact of exposure to its exhaust emissions.(7)

Most previous studies comparing mineral diesel and biodiesel combustion have found that biodiesel exhaust contains more toxic gases such as nitrogen oxides and a greater proportion of smaller particles which, when inhaled, penetrate deeper into the lungs.(3,4,8,9) Despite the potentially more toxic effects of biodiesel exhaust, most studies comparing biodiesel with commercial mineral diesel rather focus on fuel economy and engine wear or the physicochemical differences between the exhausts.(4,9) Few studies compare the health effects with exhaust exposure.(7,10,11) Such studies primarily use the Ames mutagenic assay(12,13) or immortalized cell lines(8,14) and the majority of the studies only focus on the cytotoxic and mutagenic potential of the particulate matter, ignoring the effects of the gaseous components of the exhaust entirely.(7,15) Particle concentrations are also rarely relevant to real-world exposure levels, often being far too concentrated to simulate a realistic dosage.(15) In addition, in in vitro-based studies, the cell lines used are not always human, or even derived from respiratory tissues.(3,16) This brings into question their relevance in human exposure studies where the main exposure route through inhalation of the exhaust means that the respiratory epithelium is among the first tissue exposed and thus likely to be among the most effected. Immortalized cell lines also negate genetic variability and are limited in how accurately they can model normal human tissues.(17)

If exhaust is typically inhaled, health complications can occur in the respiratory,(18,19) circulatory,(20) and immune systems.(21) Of concern, inhalation of ultrafine exhaust particles has been correlated with exacerbation of childhood asthma,(22) and associations between air pollution from major roads and decreased lung function in children have been identified.(23,24) This suggests that children may be at greater risk from adverse health effects caused by exhaust exposure. This is unsurprising as children breathe faster than adults and have higher ventilation to lung surface area/body weight ratios,(25) meaning that over the same period of time, they are exposed to a larger dosage of exhaust than adults.(25,26) In addition, the respiratory and immune systems of children are still developing and insults, such as exposure to large concentrations of exhaust, are known to have lifelong consequences.(23,27,28) Despite this, the effect of exposure to biodiesel exhaust has not yet been studied in children.

Because of the paucity of information in this setting, we tested the hypothesis that the soy biodiesel exhaust would contain a greater proportion of ultrafine particles and more oxides of nitrogen and thus exposure would result in more pronounced effects on the airway epithelium. To test this, we exposed primary human airway epithelial cells from young healthy volunteers to whole exhaust from a diesel engine fueled by either pure mineral diesel, a 20% blend of soy biodiesel with mineral diesel, or pure soy biodiesel. Physicochemical exhaust properties were recorded and 24 h post exposure, cells were analyzed for a variety of health effect end points.

Figure 1. Combustion gas analysis from the diluted exhaust of the three different fuel types: (a) oxygen concentration, (b) carbon monoxide concentration, (c) carbon dioxide concentration, (d) nitrogen monoxide concentration, (e) nitrogen dioxide concentration, and (f) sulfur dioxide concentration. Measurements were taken every 10 min for 4 h (* = p value < 0.05, ** = p value < 0.01, *** = p value < 0.01, **** = p value < 0.001). Figure 1a,c shows concentration measurements as a percentage; all other figures show concentration in parts per million (ppm).

Figure 2. Particle size spectra for all three fuels (* = p value < 0.05, ** = p value < 0.01) for the (a) 1, (b) 2, and (c) 4 h time points. Data were analyzed using total particle number concentration values for each fuel and time point. The dotted line indicates the particle size of 23 nm. Within fuels, particle size spectra are significantly different between the 1 h and the 2 and 4 h time points (p < 0.001). Both B100 and B20 show peaks around the ultrafine particle size of 100 nm, which is absent in the ULSD exhaust.

Figure 3. (a) Cell viability measurements 24 h after exposure using Annexin V staining. All results are normalized to control measurements (dotted line). The mean viability measurements for the 1, 2, and 4 h time points are 79.9 ± 11.5, 97.7 ± 7.9, and 102.2 ± 6.1% for B100, 94.1 ± 7.7, 95.6 ± 8.9, and 100.5 ± 6.4% for B20, and 99.1 ± 4.8, 99.6 ± 8.5, and 102.9 ± 5.4% for ULS, respectively. (b) Percentage of cell death via necrotic mechanisms 24 h after exposure. Asterisk symbols on legend indicate the significance between fuels (* = p value < 0.05, ** = p value < 0.01, **** = p value < 0.0001). Superscripts on x-axis indicate significant differences across time. A superscript of “A” indicates the significant increase to a superscript of “B” (a) p < 0.001 and p < 0.0001 for 1 vs 2 and 4 h, respectively (b) p < 0.05. Boxplots indicate the spread of data, and median value is marked by the horizontal line inside the box.

Figure 4. Measured cytokine release for all fuels and times for 11 cytokines released above the limit of detection. (a–k) in order: Mip-1?, IL-1?, IL-1RA, IL-6, IL-8, VEGF, G-CSF, GM-CSF, TNF-?, IP-10, and RANTES. A significant difference in the release between fuels is indicated on the legend of each graph (* = p value < 0.05). On the x-axis of each graph, a superscript of A indicates a significant increase to a superscript of B between time points (p < 0.05). Boxplots indicate the spread of data, and median value is marked by the horizontal line inside the box.

The results of this study show that the exposure to mineral diesel, pure soy biodiesel, or a 20% blend of soy biodiesel in mineral diesel induced airway epithelial cell death, increased the percentage of necrotic cell death mechanisms and increased the release of immune modulating cytokines compared to control cells. Exhaust characteristics varied significantly between all three fuel types, with B100 containing significantly higher levels of respiratory irritants including NO2, CO, CO2, and ultrafine particulate matter at a smaller median particle size, in comparison to both B20 and ULSD. The B20 exhaust contained significantly higher levels of NO in comparison to both B100 and ULSD and more particles than ULSD. Correspondingly, the B100 exhaust was significantly more toxic than both B20 and ULSD, resulting in a higher percentage of cell death and the increased release of the largest number of cytokines, particularly in the first hour of exposure. The B20 exhaust was second most toxic with significantly more cell death than ULSD. In contrast, ULSD exposure resulted in a higher release of cytokines than the B20 exposure, suggesting that mineral diesel is more immunogenic. Thus, exposure to the exhaust of all three fuels resulted in toxic effects on human airway epithelial cells associated with the exposure effects of a complex mixture of both gaseous and particulate matter components, displaying why it is vital that exhaust exposure studies use whole exhaust when assessing potential exposure health effects...

Wind energy technologies are often regarded as an important enabler in many low-carbon scenarios, such as the International Energy Agency (IEA)’s Sustainable Development Scenario(1) and the global emission mitigation pathways in the Intergovernmental Panel on Climate Change (IPCC)’s 1.5 °C special report.(2) However, transitioning toward a low-carbon society, where large amounts of renewable energy infrastructure are urgently needed, requires vast amounts of metals and minerals.(3) Such resource implications(4?7) of energy transition and consequent supply security(8?11) and embodied environmental impacts(12,13) have gained increasing attention in recent years.

For example, Denmark, a pioneer in developing commercial wind power since the 1970s’ oil crisis, has built up an energy system of which already about 48% of electricity is from wind in 2017.(14,15) The intermittent yet abundant wind energy in Denmark will continue to play a major role for achieving the Danish government’s ambition to have a “100% renewable” energy system by 2050.(16,17) Understanding potential resource supply bottlenecks, reliance on foreign mineral resources, and secondary materials provision is, therefore, an important and timely topic for both the Danish wind energy sector and Denmark’s energy and climate policy.

Construction and maintenance of wind power systems needs large quantities of raw materials mainly due to large-scale deployment of wind turbines and infrastructure on land or at sea.(18) In particular, two rare earth elements (neodymium and dysprosium) mainly used in permanent magnets have raised special concerns in the wind energy sector(10,19,20) due to overconcentration of rare earth’s supply in China,(21) sustainability of upstream mining and production processes,(22) and complexity of wind turbines’ supply chain.(23) Moreover, the wind energy sector also faces increasing challenges in both meeting future demands for several base metals (e.g., copper used in transmission(18)) and managing mounting end-of-life (EoL) materials (e.g., glass fiber in blades(24?26)) arising from decommissioned wind turbines.

A variety of methods have been used to translate wind energy scenarios into material demand. If the annual newly installed capacity of wind turbines is given, its associated material demand is often directly determined by material intensity per capacity unit.(5,8,9,27?30) If annual installed capacity is not given, its associated material demand can be derived from a life cycle assessment (LCA)-based input–output method,(31) economic model,(32) or dynamic material flow analysis (MFA) model.(6,11,20,25,30,33?36) The dynamic MFA model has been increasingly used to explore material requirements of wind energy provisioning on a global scale,(6,11,30,33,34) country scale (e.g., the US,(20) France,(25) and Germany(35)), or country scale with a regional resolution.(36) The principle of mass balance constitutes the foundation of any MFA, so that the annual newly installed capacity (“inflow”) and annual decommissioned capacity (“outflow”) of wind turbines are driven by their lifetime and the expansion and replacement of the installed wind power capacity (“stock”),(20,36) which has also been widely used in other anthropogenic stock studies.(37)

However, the current practice of modeling raw material requirement or secondary material availability in different wind energy technologies generally overlooks the hierarchical, layered characteristics of wind power systems. This is important because materials embedded in a technology system are usually distributed in its subsystem or subcomponents with varying compositions and recycling potentials.(38,39) In the case of wind power systems, materials employed in a wind turbine are distributed in its subcomponents such as rotor, tower, and nacelle, and their mass is largely determined by the turbine size (e.g., rotor diameter or hub height) and capacity.(13,40) These constraining factors and their leverages on the sustainability and resilience of the wind energy provisioning should be fully examined. Such information would enable wind turbine manufacturers, material suppliers, recyclers, end users, and policy makers to plan their material-related policies with a comprehensive understanding on a range of important aspects related to wind energy provisioning, such as secondary material supply, technological development, and material efficiency.

Here, we developed a component-by-component and stock-driven prospective MFA model to characterize material requirements and secondary material potentials of different Danish wind energy development scenarios. Based on two datasets that cover a range of microengineering parameters (e.g., capacity, rotor diameter, hub height, rotor weight, nacelle weight, and tower weight) of wind turbines installed in Denmark and worldwide, we established empirical regressions among these parameters in order to address the size scaling effects of wind turbines.

Figure 1. Stock-driven modeling framework for material demand of the Danish wind energy system at the component level. Elec. & Ca.: electronics and cables. EoL: end-of-life. MW: megawatt. PM: permanent magnet.

Figure 2. In-use capacities of Danish wind power systems (onshore and offshore) (a) from 1977 to 2017 and (b) from 2018 to 2050 in the hydrogen, IDA, wind, fossil, biomass, and biomass+ scenarios. Note: for onshore capacity scenarios, the lines of the wind, biomass, biomass+, hydrogen, and IDA scenarios are overlaid by the line of fossil scenario, because they use the same target value.

Figure 3. Empirical regressions among engineering parameters of wind turbines, (a) between capacity and rotor diameter; (b) between capacity and hub height; (c) between rotor diameter and rotor weight; (d) between rotor diameter and nacelle weight; and (e) between the square of rotor diameter multiplied by hub height and the tower weight. D: rotor diameter; H: hub height. Sample size: Danish Master Data Register of Wind Turbines (n = 9450) and The Wind Power (n = 1451).

Figure 4. Newly installed wind power capacity (for expansion and replacement) and decommissioned capacity from 2018 to 2050 in the hydrogen, IDA, wind, fossil, biomass, and biomass+ scenarios.

3.2. Material Requirements and Potential Secondary Materials Supply

Figure 5 assembles the results of material requirements (inflows) and potential secondary materials supply (outflows) during 2018–2050 under the six scenarios. Several key observations on the trends of inflows and outflows are detailed below.

•The inflows of bulk materials (concrete, steel, cast iron, nonferrous metals, polymer materials, and fiberglass) under the hydrogen, IDA, and wind scenarios will increase by 413.31, 211.91, and 328.83%, respectively. Meanwhile, the outflows of bulk materials will increase by 52.90, 49.86, and 33.15%, respectively. On the contrary, the inflows of bulk materials will increase at a slower rate under the fossil and biomass scenarios or fall slightly under the biomass+ scenario. Meanwhile, the outflows of bulk materials will decrease by 23.71, 15.98, and 37.76%, respectively.

•The inflow of neodymium under the hydrogen, IDA, and wind scenarios will climb to 14.50, 12.36, and 11.15 tonne year–1, respectively. Meanwhile, the outflow of neodymium will swell to 5.64, 5.71, and 4.98 tonne year–1, respectively. On the contrary, the inflow of neodymium will decrease at first and increase to 3.78 and 4.28 tonne year–1 under the fossil and biomass scenarios, respectively, or decrease to 2.46 tonne year–1 under the biomass+ scenario; meanwhile, the outflow of neodymium will climb up and stabilize at a certain level under the fossil (3.07 tonne year–1), biomass (3.34 tonne year–1), and biomass+ (2.60 tonne year–1) scenarios.

•A similar trend is observed in the inflow and outflow of dysprosium. The inflow of dysprosium under the hydrogen, IDA, and wind scenarios will eventually climb to 1.73, 1.48, and 1.33 tonne year–1, respectively. Meanwhile, the outflow of dysprosium will simultaneously grow to 0.67, 0.68, and 0.59 tonne year–1, respectively. On the contrary, the inflow of dysprosium will decrease at first and increase to 0.451 and 0.51 tonne year–1 under the fossil and biomass scenarios, respectively, or decrease to 0.29 tonne year–1 under the biomass+ scenario; meanwhile, the outflow of dysprosium will climb up and stabilize at a certain level under the fossil (0.37 tonne year–1), biomass (0.40 tonne year–1), and biomass+ (0.31 tonne year–1) scenarios.

•The aforementioned observations indicate that, in the case of both bulk materials and critical materials, the gap between their inflow and outflow will be enlarged under the hydrogen, IDA, and wind scenarios, and it will still be enlarged but to a lesser degree under the fossil, biomass, and biomass+ scenarios.

Figure 5. Material requirements (inflows) for newly installed capacity and potential secondary materials supply (outflows) from decommissioned capacity from 2018 to 2050 in the hydrogen, IDA, wind, fossil, biomass, and biomass+ scenarios. Note: positive numbers represent inflows and negatives represent outflows. Nd: neodymium. Dy: dysprosium.

Evidently, Denmark’s wind energy sector would be exposed to high supply risk if the country is transitioning toward a wind powered economy in all 100% renewable energy scenarios. To demonstrate the imbalance between material requirements and potential secondary material supply, as well as its dynamics over time, we propose an indicator “circularity potential”, which is defined by the ratio of outflow to inflow. This indicator measures not only the material supply risk that the wind energy sector is exposed to but also to what extent the secondary material supply can potentially mitigate the material supply risk. We could observe that the “circularity potential” of both bulk materials and critical materials in the fossil, biomass, and biomass+ scenarios is consistently higher than that in the hydrogen, IDA, and wind scenarios because in-use capacities in the former scenarios will remain stable or only slightly increase and decommissioned capacities will gradually rise. It can be observed that the “circularity potential” of critical materials (neodymium and dysprosium) under the hydrogen, IDA, and wind scenarios will increase from 0.24, 0.17, and 0.26%, peak at 45.5, 54.30, and 51.31%, and fall to 38.91, 46.23, and 44.68%, respectively. On the contrary, the “circularity potential” of critical materials under the fossil, biomass, and biomass+ scenarios will climb from 0.34, 0.23, and 0.28 to 81.27, 77.89, and 105.55%, respectively. The consistently higher “circularity potential” of critical materials is explained by two factors: mounting secondary supply from decommissioned wind turbines and less material intensities of new turbines (see Table S3 in the Supporting Information).

Although Denmark is sending its wastes abroad (e.g., Germany, Turkey, Sweden, Spain, or China),(49) the “circularity potential” can help understand to what extent circular economy strategies reduce raw material requirement in Denmark if the country is to stipulate extended producer responsibility (EPR) policies expanded across national borders.(50)

It should be noted that harnessing secondary materials in decommissioned wind turbines, as identified in the “circularity potential”, depends on many other socioeconomic and technological factors as well. For example, a wide range of circular economy measures on fiberglass were identified in a prior study,(51) such as reuse, resize, recycle, recovery, and conversion. However, commercial applications of secondary fiberglass are extremely limited because of its low value and complex composition, the lack of material composition documentation, the long transportation distance, and the underdevelopment of EPR regulations. Another typical example is the currently negligible recycling of neodymium and dysprosium,(52) because their recycling technologies are still in their infancy. Reuse of a permanent magnet seems to be a better option, but the size and materials specifications of the permanent magnets available from decommissioned wind turbines might not fit future wind turbine design.

Figure 6. (a) Impacts of increasing market share on annual neodymium flows from 2018 to 2050 in the hydrogen, IDA, wind, fossil, biomass, and biomass+ scenarios; (b) impacts of lifetime extension on cumulative neodymium flows from 2018 to 2050 in the hydrogen, IDA, wind, fossil, biomass, and biomass+ scenarios; and (c) impacts of uncertainties in parameters of lifetime function and coefficients of empirical regressions on fiberglass flows under the wind scenario. Dashed lines represent the baseline values of inflows and outflows; solid lines represent the simulation outputs of the one-factor-at-a-time sensitivity analysis on 22 parameters or coefficients.

Using Denmark as an example, we presented a prospective model that incorporates the microengineering parameters, delivering a comprehensive assessment of the material demand and secondary supply potentials of wind energy development. Our results signaled that Denmark’s ambitious transition toward 100% renewable energy will be facing increasing challenges of material provision and EoL management in the next decades. We believe unlocking the material-energy-emission nexus, as we show in this study, can eventually help understand the synergies and trade-offs of relevant resource, energy, and climate strategies and inform governmental and industry policy making in future renewable energy and climate transition.

The steel industry accounts for 30% of global industrial greenhouse gas emissions (GHG).(1) The Intergovernmental Panel on Climate Change (IPCC) recommends an overall 40–70% reduction in GHG emissions from 2010 levels by 2050.(2) However, with current best steel-making practices already approaching thermodynamic limits, even deployment of cutting-edge production technologies will not be enough for the steel industry to meet the IPCC’s emission targets.(3,4)

Realization that steel production must decrease if emission targets are to be achieved has helped lead to new research areas under the banners of “material efficiency”(5) and “circular economy”,(6) both aimed at reducing emission-intensive material production. Researchers in these new areas require a detailed material map in order to identify opportunities.

Unlike in the developing world, U.S. per capita steel stocks plateaued around 1980. The stock saturation level has been estimated at 9.1–14.3 t/capita.(7?9) Per capita stocks are expected to saturate in much of the developing world to a level similar to those in the U.S. by the late 21st century.(10,11) Therefore, the derived U.S. consumption pattern may represent a population-scaled surrogate model of the future global state.

Figure 2. Formally reconciled U.S. flow of iron and steel (including embedded alloying elements) in 2014. Drawn using eSankey software.(59)

Figure 3. Low-resolution U.S. steel map for 2014. U.S. population in 2014: 318.6 million.

U.S. scrap sent to landfill and export (34 Mt) exceeds carbon-intensive pig iron production (24 Mt) and intermediate steel product imports (29 Mt). On the face of it, increased domestic recycling could help to displace these carbon-intensive steel sources. However, a technical barrier to realizing this opportunity is contamination of EOL scrap with tramp elements, of which copper is the primary concern.(63) Daehn et al.(34) showed that copper contamination does not currently constrain global recycling rates. Their study highlighted that construction products, in particular, rebars (?0.4 wt %Cu), act as impurity sinks. In contrast, the cold-rolled sheet used mainly in transport applications has the strictest impurity requirements (?0.06 wt %Cu). The U.S. has a relatively large end-use transport sector—26% of final consumption in the U.S. (Figure 2) versus 13% globally(12)—and a small construction sector (38 vs 55% globally(12)). Moreover, the new steel map shows that just 21% of U.S. construction demand is for impurity-tolerant rebar, compared to 28% globally.(12) A smaller construction sector with less rebar means a smaller sink for scrap contaminants.

Yttrium belongs to the rare earth element (REE) group, also composed of lanthanides and scandium. REEs are necessary for the development of modern technologies, and specifically, yttrium has important applications, for instance, in fluorescent lamps as phosphors,(1) and in the aircraft industry, used in the thermal barrier coatings for jet engines.(2) The increasing demand for REEs and their low worldwide supply have led to considering REEs as critical raw materials, boosting searches for alternative resources, such as recycling used stocks or identifying new geological sources of these elements. Because the REE concentrations in acid mine drainage (AMD) are from one to two orders of magnitude higher than the average concentrations in natural waters,(3) it may be possible to perform secondary REE recovery from precipitates from AMD neutralization in passive remediation systems. These active systems were developed to minimize the environmental impacts of AMD and they are used worldwide.(4,5) However, due to the high water content, sludge storage has substantial operational costs and environmental concerns.(6,7) In contrast, passive remediation systems, which have been developed extensively in recent decades,(8?11) allow the AMD neutralization generating lower amounts of solid waste precipitates. Ayora et al. documented nearly complete aqueous REE retention in two laboratory columns, simulating a disperse alkaline substrate (DAS), a passive treatment already implemented in the field, for two highly acidic AMDs (SW Spain).(12,13) The REEs were scavenged by basaluminite, a mineral precipitated in the columns, which also presented Y enrichment due to the higher yttrium concentration with respect to the rest of REEs in the two treated AMDs. Basaluminite, an aluminum oxyhydroxysulfate (Al4(SO4)(OH)10·5H2O), precipitates in acidic environments as a consequence of the natural attenuation of the AMD when mixed with more alkaline waters, or due to the induced neutralization of the acid waters, when the solution pH reaches ?4.(14) Basaluminite is considered a nanomineral, with a short-range order, around 1 nm of coherent domain size, which is described as layers of Al-octahedra with structural point defects and with sulfate groups as outer-sphere complexes between the Al layers.(15)

Similarly to the REE uptake by basaluminite in DAS treatments, Gammons et al. reported the precipitation of hydrous aluminum oxides accompanied by a decrease in REE concentration from AMD when mixed with natural water.(16) Recently, the scavenging of REEs by basaluminite precipitates has been described as a sorption mechanism.(17) AMD is characterized to contain high loads of dissolved sulfate and the affinity of REEs to form aqueous species with sulfate is very high, the MSO4+ aqueous complex being more abundant in AMD solutions.(18) Sorption of dissolved REEs from sulfate-rich waters onto basaluminite is thus described as the sorption of the MSO4+ aqueous complex via ligand exchange with a surface site of basaluminite, forming a monodentate surface complex with the Al-octahedron as one proton is released.(17) Here, a structural description of the aqueous YSO4+ complex and of the local environment of the surface complex formed upon adsorption onto basaluminite are reported...

...The objective of this study is to elucidate the structure of Y adsorbed onto basaluminite. Its chemical similarities with HREEs allow us to assume similar structural configuration for this subgroup. Moreover, this element was one of the most concentrated in waste samples allowing performing X-ray absorption spectroscopy experiments. Since the YSO4+ aqueous complex is adsorbed onto the mineral,(17) a previous characterization of the geometry of the aqueous complex has been carried out. Finally, a quantification of Y-species in basaluminite solids precipitated from AMD treatments has been performed. Structural studies were performed using EXAFS and pair distribution function (PDF) analyses of aqueous and solid samples combined with ab initio molecular dynamics (AIMD) simulations of the aqueous YSO4+ complexes.

Two hypotheses are used to investigate the local structure of the aqueous YSO4+ ion pair: (1) an outer-sphere complex, with water located between Y3+ and SO42–, and (2) an inner-sphere complex. In the latter case, two more hypotheses must be considered: (a) a monodentate complex, with one oxygen atom shared between the sulfate and the first coordination sphere of Y3+, and (b) a bidentate complex, with two oxygen atoms shared between the yttrium hydration sphere and the sulfate group.

Once the structure of the aqueous solution is fully described, different hypotheses have been considered to interpret the YSO4+ surface complexation onto the Al-oxyhydroxysulfate:

Figure 1. (A) Top: Experimental PDFs of the YSO4-sol and Y-sol samples. Bottom: Simulated (AIMD) PDF (YSO4-calc) and partial PDFs of a YSO4+ aqueous complex. (B) Fourier-filtered signal from 1.8 to 4.2 Å for the EXAFS data. (C) EXAFS Fourier transform (FT) amplitude functions of the YSO4-sol sample. Black lines: experimental; red lines: fits. Simulated (AIMD) PDF and partial PDFs have been multiplied for visualization purposes: YSO4-calc (3×), Y–S (5×), and Y–O and S–O (2×). Dashed lines indicate the position of the Y–O, Y–S, and S–O bonds in the YSO4-sol sample.

Figure 2. PDFs of basaluminite coprecipitated in the presence of Y (B-Ycop) and pure basaluminite (B-pure) and differential PDF (d-PDF). The d-PDF spectrum has been amplified (3×) for visualization purposes.

Figure 3. k3-Weighted EXAFS (A) and FT amplitude functions (B) for four waste samples from column treatments, W-MR-C1-3, W-MR-C1-4, W-Alm-C3-8, and W-Alm-C3-9 (upper part); solid standards (basaluminite sorbed with YSO4 (B-YSO4), basaluminite sorbed and coprecipitated with Y: B-Yads and B-Ycop, respectively); and aqueous solution (free ion and sulfate complex: Y-sol and YSO4-sol, respectively) (bottom part). The dashed lines in the EXAFS signals of the column samples represent LCF with B-Yads (basaluminite with sorbed yttrium) and YSO4-sol (solution of Y with SO4) standards as the most representative references (results in Table 3). The arrows indicate a frequency present in the solid standards.

Figure 4. Atomistic representations of the three models of YSO4 aqueous complexes adsorbed on the basaluminite–water interface. The different atomic positions of YSO4 to octahedral-Al are used to fit the EXAFS signal of the B-YSO4 sample. The three models show different inner-sphere surface complexes: (A) monodentate, (B) bidentate mononuclear, and (C) bidentate binuclear.

Figure 5. (A) k3-Weighted EXAFS spectra at the Y K-edge of the basaluminite with YSO4 sorbed (B-YSO4-ads reference) and (B) its Fourier transform amplitude. The experimental and fitted curves are shown in black and red, respectively.

The YSO4+ aqueous species has been characterized combining PDF analyses of aqueous solutions and AIMD simulations, confirming the formation of an inner-sphere Y–SO4 ion pair with a monodentate configuration, with a Y–S interatomic distance of 3.5 Å. Results from the thermodynamic sorption model describe REE sorption onto basaluminite via sorption of aqueous REESO4+. The use of an atomistic model using this positively charged ion yields the best results for the EXAFS fitting of Y sorbed on basaluminite. However, the EXAFS technique cannot confirm the presence of YSO4+ sorbed onto basaluminite by itself, due to the low sensitivity to discern between Al and S neighbors. However, the EXAFS fitting, together with the PDF, can confirm the strong interaction and the formation of inner-sphere surface complexes of Y on basaluminite precipitates, via ligand exchange with AlO6 units of its structure. EXAFS analyses of column waste samples show that most of the Y is retained as the same inner-sphere sorbed species, YSO4+, with a low proportion of YSO4+ in the outer-sphere configuration.

The description of the local structure of yttrium sorbed onto the basaluminite surface provided here complements the atomic configuration studies of other trace metals, such as As and Se elements in this mineral.(44) The chemical similarity between yttrium and other HREEs (from Tb to Lu) suggests that similar environments could be present for the other elements of the same group. This fact has important environmental consequences, as the HREE would be strongly sorbed, forming inner sphere complexes, which could result in their long-term immobilization at least until the host phase is dissolved or re-precipitated. A key question emerges about the long-term stability of the complex, particularly with an increase in the solution pH.

During pyrolysis, heat alone can simultaneously cleave several chemical bonds in polymeric materials.1?3 This method is advantageous for the treatment of mixtures that cannot be physically separated and recycled.4,5 Therefore, in this study, we focused on cellulose/plastic mixtures. Synergistic interactions were investigated during fast co-pyrolysis at 500 °C of binary mixtures of cellulose with plastics, such as polyethylene (PE), polypropylene (PP), polystyrene (PS), polyvinyl chloride (PVC), and polyethylene terephthalate (PET). PS addition increased the yield of levoglucosan (LG) from cellulose by 3.5 times; PVC catalyzed LG dehydration to produce levoglucosenone; and cellulose enhanced the production of gaseous and liquid aliphatic hydrocarbons from PE and PP and liquid aromatic hydrocarbons from PS. These synergistic effects facilitate the elucidation of the complemental co-pyrolysis mechanism, allowing for the prediction of pyrolysis products and maximization of the recovery of useful fuel and chemical feedstock from cellulose and plastic composite materials. Recently, lightweight reinforced resins containing cellulose-rich plant fibers have received research attention as construction and automotive materials.6,7 The forecasted production of natural fiber composites in the European Union (EU) in 2020 has almost quadrupled that in 2012.

Pyrolytic interactions, often called synergies, during the copyrolysis of lignocellulosic biomass and plastic are of widely recognized importance...

...Because these plastics decompose through radical chain mechanisms,12?14 radical interactions during copyrolysis have recently been studied by employing an electron spin resonance spectrometer featuring a novel heating unit.11 However, the hydrogen exchange ability and the radical interaction mechanism depend upon the plastic type, and the influence of various plastics on cellulose pyrolysis and vice versa remain unclear. The pyrolysis of other polymers, such as PVC and PET, progresses via radical and ionic reactions...

...Thus, herein, we investigated the impact of the five most commonly used plastics, viz., PE, PP, PS, PVC, and PET, on cellulose pyrolysis and vice versa by employing the yield difference factor (YD; eq 1) and the difference from the estimated weight fraction [Di (wt %); eq 2]. Pyrolysis of each sample and co-pyrolysis of binary mixtures of cellulose and each plastic (1:1 weight ratio) were carried out at 500 °C in a horizontal quartz tube reactor (Figure S1 of the Supporting Information). Detailed experimental information has been summarized in the Supporting Information...

Molten salt Reactor (MSR) is one of six reactors proposed by the Generation IV International Forum [1]. There are graphite, molten fluoride salt, and Nickel-based alloy, such as Hastelloy N alloy, in MSR [2]. The Molten-Salt Reactor Experiment (MSRE) per-formed in the 1960s revealed that Hastelloy N alloy and graphite are compatible with molten fluoride salt [2]. However, the permeation of molten fluoride salt into porous graphite is a critical issue, which can create local hot spots where the damage rate would be increased [2]. Silicon carbide (SiC) coating can be used to protect graphite moderator against the permeation of molten fluoride salt [3]. SiC ceramic and SiC fiber reinforced SiC ceramic composite (SiCf/SiC) are potential materials for MSRs due to their superior high-temperature properties, corrosion resistance, and irradiation resistance. SiCf/SiC composite is also one of candidate materials for shut down rod channel liners, core support plates, ex-core control blade guides and wetted refueling mechanism in MSRs [4,5].Experiments performed at Oak Ridge National Laboratory revealed that SiC is not obviously corroded by molten FLiNaK (46.5 mol% LiF,11.5 mol% NaF, and 42 mol% KF) salt [6]. However, impurities may cause the corrosion of SiC in molten fluoride salt. There are intrinsic impurities from molten fluoride salt and impurities from the corrosion of Hastelloy N alloy. Oxygen is common impurity in molten fluoride salt. SiC may react with oxygen in salt to form oxide that can be corroded by molten fluoride salt [7,8]. Fluoride anions can cause the breakage of Si O Si bonds and the formation of Si F bonds [7]. Oxides of SiC may react with molten FLiNaK salt to formSiF4, K2SiF6, Na2SiF6, [SiO4]4?, [Si2O5]2?, [SiO3F]3?, [SiO2F2]2?, andSi4O7F2[7–9]. The corrosion of SiC may be aggravated because of an increase in the concentration of metal corrosion products in salts induced by the corrosion of Hastelloy N alloy. Hastelloy N alloy is nickel-based alloy and is composed of 15–17 wt% Mo, 6–8 wt% Cr, 4–6 wt% Fe...

...Molten FLiNaK salt is one of candidate coolants in MSR due to its chemical and radiolytic stability [28]. In the whole MSR system, the salt-containing piping and equipment are composed of Hastelloy N alloy. The corrosion of Hastelloy N alloy can increase the concentration of metal impurities in salt. The effect of corrosion products of Hastelloy N alloy on the corrosion of SiC should be investigated. Ni is the base element of Hastelloy N alloy. Cr in Hastelloy N alloy can be easily corroded. Therefore, we studied the effect of Hastelloy N alloy and its corresponding corrosion products, CrF2and NiF2...

Fig 1. Raman spectrum of CVD SiC before and after corrosion in FLiNaK salt at 700 °C for 15 days.

Fig. 2. Cross-section EDS line-scan spectrum of Hastelloy N alloy after corrosion with SiC in FLiNaK salt at 700 °C for 15 days.

Fig. 3. (a) Cross-section SEM image of Hastelloy N alloy after corrosion with SiC in FLiNaK salt at 700 °C for 15 days, (b) GI–XRD pattern of Hastelloy N alloy before and after corrosion with SiC in FLiNaK salt at 700 °C for 15 days.

Fig. 4. Surface SEM image of SiCf/SiC composite with CVD SiC coating, (a) and (b) is before experiment, (c) and (d) is after corrosion in FLiNaK salt with NiF2 at 700 °C for 45 days. The inset of (a) is GI-XRD pattern with a fixed grazing angle of incidence at 0.3°.

Fig. 5. Raman spectrum of pure FLiNaK salt and FLiNaK salt mixed with NiF2 after corrosion of SiC at 700 °C for 45 days.

Fig. 6. (a) Surface SEM image of CVD SiC after corrosion with Hastelloy N alloy in FLiNaK salt at 700 °C for 15 days, (b) GI–XRD pattern of CVD SiC before and after corrosion with Hastelloy N alloy in FLiNaK salt at 700 °C for 15 days, (c) surface SEM image and (d) GI–XRD pattern of CVD SiC after corrosion in FLiNaK salt mixed with CrF2 at 700 °C for 15 days.

4. Conclusions

The effect of Hastelloy N alloy and its corresponding corrosion products on the corrosion of SiC in molten FLiNaK salt was investigated by SEM/EDS, GI–XRD, and GD–MS. Results reveal that Hastelloy N alloy and the corresponding corrosion products can induce the corrosion of SiC in FLiNaK salt. Three factors affect the corrosion of SiC.

(1) Ni in Hastelloy N alloy can trigger the corrosion of SiC. Ni can react with Si in salt to form NiSi and Ni31Si12on the surface of Hastelloy N alloy. Silicide deposited on the surface of alloy can drive the dissolution of Si from SiC into salt and alloy.

(2) The corrosion product, NiF2can drive the corrosion of SiC. NiF2can make SiC with a thickness of ?50 _m almost disappear after experiment for 45 days. Thus, the Si content in salt increased up to 0.5 wt%. Raman spectrum indicates that Si in salt is in the form of [SiF6]2?.

(3) The corrosion product, CrF2, can induce the corrosion of SiC. CrF2 can react with SiC to form Cr7C3 and CrC on SiC, and then induce the dissolution of element Si from SiC into salt.