With 1.85 billion tons produced per year, steel is the most important material class in terms of volume and environmental impact. While steel is a sustainability enabler, for instance through lightweight design, magnetic devices, and efficient turbines, its primary production is not. For 3000 years, iron has been reduced from ores using carbon. Today 2.1 tons CO2 are produced per ton of steel, causing 30% of the global CO2 emissions in the manufacturing sector, which translates to 6.5% of the global CO2 emissions. These numbers qualify iron-and steel-making as the largest single industrial greenhouse gas emission source. The envisaged future industrial route to mitigate these CO2 emissions targets green hydrogen as a reductant. Although this reaction has been studied for decades, its kinetics is not well understood, particularly during the wüstite reduction step which is dramatically slower than the hematite reduction. Many rate-limiting factors of this reaction are set by the micro-and nanostructure as well as the local chemistry of the ores. Their quantification allows knowledge-driven ore preparation and process optimization to make the hydrogen-based reduction of iron ores commercially viable, enabling the required massive CO2 mitigation to ease global warming. Here, we report on a multi-scale structure and composition analysis of iron reduced from hematite with pure H2, reaching down to near-atomic scale. The microstructure after reduction is an aggregate of nearly pure iron crystals, containing inherited and acquired pores and cracks. Crucial to the reduction kinetics, we observe the formation of several types of lattice defects that accelerate mass transport inbound (hydrogen) and outbound (oxygen) as well as several chemical impurities within the Fe in the form of oxide islands that were not reduced. With this study, we aim to open the perspective in the field of carbon-neutral iron production from macroscopic processing towards a better understanding of the underlying microscopic reduction mechanisms and kinetics.