Abstract The reduction of neoclassical energy transport in stellarators has traditionally focussed on optimizing magnetic fields for small values of ‘effective helical ripple’ — ϵ eff , the geometric factor associated with electron 1 / ν transport—and relying on the radial electric field, E r , needed to maintain ambipolarity in the plasma, to simultaneously diminish ion energy losses to a tolerable level. As one must generally expect E r < 0 , such a strategy has a drawback for reactor operation, however, as negative values of E r tend to hinder the exhaust of helium ash, and this will become intolerable if it results in excessive fuel dilution. Theoretically, one can show that the neoclassical transport of low-Z impurities depends critically on the ratio L 11 e / L 11 i , where the L 11 σ are the neoclassical particle diffusion coefficients of the bulk-plasma electrons ( σ = e ) and ions ( σ = i ). Increasing the value of this ratio is shown here to counteract impurity retention, but maintaining good confinement of the bulk species requires this to be achieved by decreasing L 11 i rather than increasing L 11 e , implying the need for more than just minimization of ϵ eff in reactor design. To assess the prospects of such an endeavor, a predictive 1-D transport code is used here to determine the range of L 11 e / L 11 i values which arises in simulations of conventionally optimized reactor candidates. This range is found to be considerable, and includes examples with L 11 e / L 11 i values large enough to provide neoclassical temperature screening of the helium ash and even to flip the sign of the neoclassical convective velocity from inward- to outward-directed. More intriguingly, the strong reduction of L 11 i in such cases can also lead to the appearance of E r > 0 (a so-called ‘electron root’) within the core of plasmas having central densities as large as 2 × 10 20 m−3. Having E r > 0 in the plasma core produces the ideal situation in which all thermodynamic forces aid in the exhaust of helium ash, although this benefit is tempered by the small values of L 11 z which accompany an electron root (where σ = z denotes helium ash). Means are discussed for improving the results presented here, so that avoiding helium-ash retention can be explicitly targeted in future reactor optimizations.