**2. Simulation Method**

In this study, we focus on the (110)[001] cutting system of bcc Fe; this means that the top surface had a (110) orientation, and the cutting direction was along [001]. The tool edge was along the [1¯10] direction. We showed previously [21,22] that this cutting system allows for a simple quasi-two-dimensional understanding of the cutting process, since the major dislocation systems involved in chip formation—those with the Burgers vector *b* = <sup>1</sup> <sup>2</sup> [111]—have their dislocation lines aligned with the tool edge. In Appendix A, we illustrate the complexities that can arise for other cut systems.

The substrate was single crystalline with a (110) surface, while the cutting direction was along [001]. It was built from 1.72 × 106 Fe atoms, extended 611 Å along the cutting direction, and had a height of 404 Åand a thickness of 81 Å. Fe atoms interact with each other according to the Mendelev potential [38], which is known to describe the elastic and plastic properties of Fe, as well as the surface energies relevant for cleavage processes, faithfully [39–41].

The simulation system is displayed schematically in Figure 1. In order to prevent the substrate from any translation or rotation during the machining process, its bottom and the left boundary contained two fixed atom layers. The next two atom layers were thermostatted at a low temperature, <1 K, to ease detection of dislocations and other defects in the system. In the direction along the tool edge (the [1¯10] direction), periodic boundary conditions were applied, while the top and the right-hand side boundaries were free.

The tool was built from between (44–166) ×103 carbon atoms; it was rigid, i.e., no C atom could move with respect to the others. It is known from nanoindentation studies [42] that the atomistic surface structure of the tool may influence the nucleation of dislocations; we believe, however, that this effect lost its importance during the long cutting simulations performed in the present study. It had a clearance angle of 5◦ and a rounded edge with a curvature radius of *r* = 10 Å. The rake angle, *α*, of the tool was varied in this study between +45◦ and −45◦. Fe and C atoms interact purely repulsively with each other; their interaction is described by a Lennard–Jones potential [43] that was cut off at its minimum at 4.2 Å [20]. In all simulations presented here, the cutting depth amounted to *d* = 50 Å. The tool moved with a velocity of 20 m/s, and the total cutting length was 100 Å. This length might be considered as too small to provide steady-state cutting values, in particular for the negative rake angles; we therefore check this issue in Appendix B.

**Figure 1.** Schematics of the simulation setup showing important quantities characterizing the geometry of the cutting and of the chip. Quantities are defined in the text.

The open-source code LAMMPS [44] was used to perform the simulations. Dislocations were identified using the Crystal Analysis Tool (CAT) [45–47]. The atomistic configurations were visualized using the free software OVITO [48], while dislocations and surfaces were rendered by ParaView [49].

We note that previously, we simulated this cutting system with a rake angle of *α* = 15◦ [21,22]. Since the results were quite close to our new data for *α* = 0◦, we do not show snapshots for this case, but use the quantitative results of the forces and chip thickness in our discussion.
