Where can a Trappist-1 planetary system be produced?

We study the evolution of protoplanetary discs that would have been precursors of a Trappist-1-like system under the action of accretion and external photoevaporation in different radiation environments. Dust grains swiftly grow above the critical size below which they are entrained in the photoevaporative wind, so although gas is continually depleted, dust is resilient to photoevaporation after only a short time. This means that the ratio of the mass in solids (dust plus planetary) to the mass in gas rises steadily over time. Dust is still stripped early on, and the initial disc mass required to produce the observed 4 M_⊕ of Trappist-1 planets is high. For example, assuming a Fatuzzo & Adams distribution of UV fields, typical initial disc masses have to be >30 per cent the stellar (which are still Toomre Q stable) for the majority of similar mass M dwarfs to be viable hosts of the Trappist-1 planets. Even in the case of the lowest UV environments observed, there is a strong loss of dust due to photoevaporation at early times from the weakly bound outer regions of the disc. This minimum level of dust loss is a factor of 2 higher than that which would be lost by accretion on to the star during 10 Myr of evolution. Consequently, even in these least irradiated environments, discs that are viable Trappist-1 precursors need to be initially massive (>10 per cent of the stellar mass).