Robust incorporation in multi-donor patches created using atomic-precision advanced manufacturing

Atomic-precision advanced manufacturing enables the placement of dopant atoms within $\pm$1 lattice site in crystalline Si. However, it has recently been shown that reaction kinetics can introduce uncertainty in whether a single donor will incorporate at all in a minimal 3-dimer lithographic window. In this work, we explore the combined impact of lithographic variation and stochastic kinetics on P incorporation as the size of such a window is increased. We augment a kinetic model for PH$_3$ dissociation leading to P incorporation on Si(100)-2$\times$1 to include barriers for reactions across distinct dimer rows. Using this model, we demonstrate that even for a window consisting of 2$\times$3 silicon dimers, the probability that at least one donor incorporates is nearly unity. We also examine the impact of size of the lithographic window, finding that the incorporation fraction saturates to $\delta$-layer like coverage as the circumference-to-area ratio approaches zero. We predict that this incorporation fraction depends strongly on the dosage of the precursor, and that the standard deviation of the number of incorporations scales as $\sim \sqrt{n}$, as would be expected for a series of largely independent incorporation events. Finally, we characterize an array of experimentally prepared multi-donor lithographic windows and use our kinetic model to study variability due to the observed lithographic roughness, predicting a negligible impact on incorporation statistics. We find good agreement between our model and the inferred incorporation in these windows from scanning tunneling microscope measurements, indicating the robustness of atomic-precision advanced manufacturing to errors in patterning for multi-donor patches.