Production of metals stands for 40% of all industrial greenhouse gas emissions, 10% of the global energy consumption, 3.2 billion tonnes of minerals mined, and several billion tonnes of by-products every year. Therefore, metals must become more sustainable. A circular economy model does not work, because market demand exceeds the available scrap currently by about two-thirds. Even under optimal conditions, at least one-third of the metals will also in the future come from primary production, creating huge emissions. Although the influence of metals on global warming has been discussed with respect to mitigation strategies and socio-economic factors, the fundamental materials science to make the metallurgical sector more sustainable has been less addressed. This may be attributed to the fact that the field of sustainable metals describes a global challenge, but not yet a homogeneous research field. However, the sheer magnitude of this challenge and its huge environmental effects, caused by more than 2 billion tonnes of metals produced every year, make its sustainability an essential research topic not only from a technological point of view but also from a basic materials research perspective. Therefore, this paper aims to identify and discuss the most pressing scientific bottleneck questions and key mechanisms, considering metal synthesis from primary (minerals), secondary (scrap), and tertiary (re-mined) sources as well as the energy-intensive downstream processing. Focus is placed on materials science aspects, particularly on those that help reduce CO2 emissions, and less on process engineering or economy. The paper does not describe the devastating influence of metal-related greenhouse gas emissions on climate, but scientific approaches how to solve this problem, through research that can render metallurgy fossil-free. The content is considering only direct measures to metallurgical sustainability (production) and not indirect measures that materials leverage through their properties (strength, weight, longevity, functionality).