Dynamic Friction Unraveled by Observing an Unexpected Intermediate State in Controlled Molecular Manipulation

The pervasive phenomenon of friction has been studied at the nanoscale via a controlled manipulation of single atoms and molecules with a metallic tip, which enabled a precise determination of the static friction force necessary to initiate motion. However, little is known about the atomic dynamics during manipulation. Here, we reveal the complete manipulation process of a CO molecule on a Cu(110) surface at low temperatures using a combination of noncontact atomic force microscopy and density functional theory simulations. We found that an intermediate state, inaccessible for the far-tip position, is enabled in the reaction pathway for the close-tip position, which is crucial to understanding the manipulation process, including dynamic friction. Our results show how friction forces can be controlled and optimized, facilitating new fundamental insights for tribology.

Figure 1

(a) Schematic of CO manipulation on Cu(110) between top and bridge sites by a metallic tip over the top site. The thin arrows indicate the reaction processes of CO for tip approach and retract. (b) Same as (a), but for CO manipulation from top to neighboring top site when the tip approach and retract occurs for a tip position laterally closer to the neighboring top site. (c) Definition of lateral tip positions along Cu $[\overline{1}10]$ ($d=255\mathrm{pm}$): $x/d=0$ (top site) and $x/d=0.63$ (outside of bridge). (d) Measured frequency shift ($\mathrm{\Delta }f$) as a function of vertical tip position ($z_{\mathrm{l}}$) for the tip-on-top-site case ($x/d=0$). The $\mathrm{\Delta }f$ curves for tip approach and retraction are depicted by red solid and dotted lines, respectively. The $\mathrm{\Delta }f$ curve for the tip on Cu is also depicted by a gray line. (e) Same as (d), but for the lateral tip position $x/d=0.63$ (outside of bridge). In the inset, typical STM images of CO before and after the manipulation are shown. (f) Measured potential energy between the tip and CO up to the point of manipulation. (g) Energy dissipation per cycle of the vertically oscillating tip, where both cases of tip approach and retract are depicted.

Figure 2

(a) Theoretical potential energy between the ${\mathrm{Cu}}_{11}$ tip [inset of (a)] on $x/d=0.3$ and CO on Cu(110) [inset of (b)] as a function of the vertical tip position ($z_{\mathrm{cal}}$). The cases for CO on top (T), bridge (B), and neighboring top (NT) sites are depicted by black, red, and blue lines, respectively. (b) Same as (a) but for the tip over $x/d=0.8$. (c) Calculated reaction pathways for CO from one top site (0.0) to the next top site (1.0) via bridge configuration (0.5) for a fixed ${\mathrm{Cu}}_{11}$ tip over $x/d=0.3$. The cross marks correspond to CO in top (T), bridge (B), and neighboring top (NT) configurations, respectively. (d) Same as (c) but for the tip over $x/d=0.8$.

Figure 3

(a) Calculated potential energy between the ${\mathrm{Cu}}_{11}$ tip and CO as a function of lateral tip position (at a constant tip height of 170 pm) for CO in top (T), bridge (B), and neighboring top (NT) configurations. The thick path represents the CO configuration during the lateral tip scan. (b)–(d) Same as (a) but for tip heights: (b) 145 pm, (c) 125 pm, and (d) 105 pm. (e) Calculated CO configuration as a function of tip height and lateral position during the forward tip scan. The gray, light red, and light blue regions represent CO in top (T), bridge (B), and neighboring top (NT) sites, respectively. The thick lines indicate the transition points. (f) Energy dissipation per cycle of the vertically oscillating tip, experimentally measured during the lateral CO dragging. In (e) and (f), the onset of dragging is indicated by the dashed green line.