Understanding the impact of aircraft speed heterogeneity on air traffic operation is crucial for airspace design and air traffic flow management. Speed heterogeneity (or aircraft mix) is recognized as a causal factor for complexity in air traffic operations, through qualitative or statistical analysis. Quantitative metrics on how it affects current and future Trajectory-Based Operation (TBO) is lacking, however. In this paper, we present an in-depth investigation of the impact of speed heterogeneity, by defining air traffic robustness at microscopic (aircraft pair) and macroscopic (air traffic flow) levels in nominal situations, derived mathematically and validated through fast-time computer simulations. A human-in-the-loop study follows investigating six 4D en route operation scenarios, where operators were instructed to resolve a large disturbance in a sector, using a novel interface. Results confirm the negative impact of speed heterogeneity on air traffic controller performance in terms of flow efficiency and workload. The mechanism of such impact is substantiated through analyzing several speed-based robustness metrics. Although the simulated traffic scenarios have similar baseline robustness, those with mixed speeds lead to significantly lower robustness and operational performance. This emphasizes the need to incorporate speed heterogeneity in robustness evaluations of air traffic control in current and future TBO environments.