Lateral manipulation of single iron adatoms by means of combined atomic force and scanning tunneling microscopy using CO-terminated tips

CO-terminated tips currently provide the best spatial resolution obtainable in atomic force microscopy. Due to their chemical inertness, they allow us to probe interactions dominated by Pauli repulsion. The small size and inertness of the oxygen front atom yields unprecedented resolution of organic molecules, metal clusters, and surfaces. We study the capability of CO-terminated tips to laterally manipulate single iron adatoms on the Cu(111) surface with combined atomic force and scanning tunneling microscopy at 7 K. We find that even a slight asymmetry of the tip results in a distortion of the lateral force field. In addition, the influence of the tilt of the CO tip on the lateral force field is reversed compared to the use of a monoatomic metal tip which we can attribute to the inverted dipole moment of a CO tip with respect to a metal tip. Moreover, we demonstrate atom-by-atom assembly of iron clusters with CO tips while using the high-resolution capability of the CO tips in between to determine the arrangement of the individual iron atoms within the cluster. In all conducted experiments using CO tips within this study, the CO was never changed or lost from the tip's apex.

Figure 1

(a) Selected raw $\mathrm{\Delta }f\left(x_{\mathrm{A}},z_{\mathrm{i}}\right)$ traces ($z_{1,\mathrm{Cu}}=571$ pm, $z_{2,\mathrm{Cu}}=291$ pm, $z_{3,\mathrm{Cu}}=211$ pm, $z_{4,\mathrm{Cu}}=186$ pm) acquired with a monoatomic metal tip along the ${\vec{x}}_{\mathrm{A}}$ direction: The $z$ values refer to the distances to point contact on the Cu(111) surface [33, 36] (see also Appendix pp2). $x_{\mathrm{A}}=0$ pm indicates the center of the iron adatom before manipulation. It is manipulated at $x_{\mathrm{man}}>0$ pm, indicated by the vertical green dotted line. The gray dotted trace is the $\mathrm{\Delta }f\left(x_{\mathrm{A}}\right)$ curve after manipulation. The inset shows a sketch of a Cu(111) surface with an iron adatom adsorbed on a fcc hollow site in top view. The vectors ${\vec{x}}_{\mathrm{A}}$, ${\vec{x}}_{\mathrm{B}}$, and ${\vec{x}}_{\mathrm{C}}$ indicate the directions to three next-neighbor fcc hollow sites. (b) Selected $\mathrm{\Delta }f\left(x_{\mathrm{A}},z_{\mathrm{i}}\right)$ curves ($z_{1,\mathrm{CO}}=560$ pm, $z_{2,\mathrm{CO}}=340$ pm, $z_{3,\mathrm{CO}}=220$ pm, $z_{4,\mathrm{CO}}=205$ pm) by using a CO tip. The iron adatom is manipulated before the tip has crossed the center of the adatom at $x_{\mathrm{man}}<0$ pm. The complete $\mathrm{\Delta }f\left(x_{\mathrm{A}}\right)$ data sets of (a) and (b) are shown in Appendix pp1. (c) Short-range potential $U_{\mathrm{SR}}\left(x_{\mathrm{A}},z\right)$ between the metal tip and the iron adatom computed from the $\mathrm{\Delta }f\left(x_{\mathrm{A}}\right)$ data shown in (a). The interaction between the adatom and the metal tip is purely attractive. (d) The short-range interaction potential corresponding to the $\mathrm{\Delta }f\left(x_{\mathrm{A}}\right)$ data shown in (b) acquired with a CO tip is also entirely attractive. (e) and (f) Selected lines of the corresponding lateral forces $F_{x,\mathrm{SR}}\left(x_{\mathrm{A}}\right)$. Darker lines refer to closer tip-sample distances. The absolute lateral force increases with decreasing tip-sample distance $z$. Before the deconvolution processes of $\mathrm{\Delta }f$, a Gaussian filter ($\sigma =16$ pm) was applied.

Figure 2

Conductance traces $\langle G\rangle$ acquired during two separate lateral manipulation experiments of a single iron adatom using a monoatomic metal tip (blue, top) and a CO tip (orange, bottom) are shown, respectively. The manipulation direction was chosen such that an asymmetric lateral force profile was observed for both tips. The tip-sample distances were reduced until the lateral force thresholds for lateral manipulation was overcome on both sides of the adatom. In both cases, the adatom hopped in the first place towards the tip while the tip was approaching the adatom from the left side (lateral position change from site 0 to site 1 at the leftmost vertical blue and orange dotted lines), and afterwards followed the tip twice to the right side, indicated by sharp changes in $\langle G\rangle$ marked by vertical dotted lines (lateral position change from site 1 to site 2 and site 2 to site 3).

Figure 3

(a) The Constant-height $\mathrm{\Delta }f$ image acquired above a CO molecule (COFI image) shows a tilted monoatomic metal tip [30, 38]. The scale bars in (a) and (b) correspond to 300 pm. (b)–(d) Tilt of the monoatomic metal tip along the three high-symmetry directions of the Cu(111) surface. (e) COFI image of the monoatomic metal tip from (a) with a CO molecule attached to its apex. Due to symmetry reasons, the CO molecule adsorbs to the tip's apex as illustrated in (f)–(h). (i) Lateral force acting at the moment of manipulation against manipulation direction. The tilt of the tips introduces an asymmetry in the force profiles. In case of the monoatomic metal tip, the iron adatom is at first manipulated when the tilted part of the tip is facing away from the adatom [as sketched in (j)]. For CO tips, the initial lateral manipulation of the iron adatom takes place when the tilted part of the tip is facing towards the adatom [as sketched in (k)]. In both cases, the adatom can be manipulated in the opposite directions as well by reducing the tip-sample distance further (see Fig. 2).

Figure 4

Sketch of the (a) monoatomic metal tip and the (b) CO tip used for the analytical model. Copper atoms are used to model the metal tip since the tip was repeatedly poked into the clean Cu(111) surface during tip shaping. (c) and (d) Positions of the point charges of the monoatomic metal tip and the CO tip. The parameters were set to $q_{\mathrm{Cu}}=+0.13e$ [36], $d_{\mathrm{Cu}}=135$ pm [36], ${\mathit{off}}_{\mathrm{Cu}}=0$ pm [36], $q_{\mathrm{CO}}=-0.03e$ [36], ${\mathit{off}}_{\mathrm{CO}}=-100$ pm [20], and $q_{\mathrm{Fe}}=+0.1e$. (e) A monoatomic tip exhibits a conductance of $G=G_0=\frac{2e^2}{h}$ at a shell-shell-distance between the tip's apex atom and an atom of the topmost layer of the Cu(111) surface of $\mathrm{\Delta }{z}_{\mathrm{ss}}=172$ pm [41]. (f) The core-core distance between the iron atom and the topmost copper layer ($\mathrm{\Delta }{z}_{\mathrm{Fe},\mathrm{Cu}}=186$ pm) [42], the atomic radius of copper (135 pm) and the closest approach in experiment $z_{4,\mathrm{Cu}}$ [see Fig. 1] yield the closest core-core distance between the apex atom and the iron adatom in experiment of $z_{\mathrm{model},\mathrm{Cu}}=442$ pm. (g) The same considerations for the CO tip, with the atomic radius of oxygen (60 pm), the experimental closest approach $z_{4,\mathrm{CO}}$ [see Fig. 1] and obeying the $z$-distance relation between monoatomic metal tips and CO-functionalized monoatomic metal tips [36] (Appendix pp2), result in a core-core distance between the tip's oxygen atom and the iron adatom of $z_{\mathrm{model},\mathrm{CO}}=386$ pm. The lateral force $F_x$ acting due to pure vdW (blue dotted curve), pure ES (red dotted curve), and vdW+ES interaction (black solid curve) between the iron atom and (h) the monoatomic metal tip and (i) the CO tip are shown, respectively ($\alpha ={20}^{\circ }$).

Figure 5

The sum of the two extrema of the lateral force curve $\mathrm{\Delta }{F}_x=\max \left(F_x\right)+\min \left(F_x\right)$ is plotted for the vdW + ES model as a function of the parameters (a) $\alpha$, (b) $z_{\mathrm{model},\mathrm{Cu}}$, (c) ${\mathit{off}}_{\mathrm{Cu}}$, (d) $q_{\mathrm{Cu}}$, (e) $q_{\mathrm{Fe}}$, and (f) $d_{\mathrm{Cu}}$ in case of the monoatomic metal tip. The influence of $\alpha$, $z_{\mathrm{model},\mathrm{CO}}$, ${\mathit{off}}_{\mathrm{Cu}}$, $q_{\mathrm{Cu}}$, $q_{\mathrm{Fe}}$, $q_{\mathrm{CO}}$, and ${\mathit{off}}_{\mathrm{CO}}$ onto the sum of the two extrema of the lateral force curve $\mathrm{\Delta }{F}_x=\max \left(F_x\right)+\min \left(F_x\right)$ is depicted in (g)–(m) in case of the CO tip. The constant parameters were set to $d_{\mathrm{Cu}}=135$ pm [36], ${\mathit{off}}_{\mathrm{Cu}}=0$ pm [36], ${\mathit{off}}_{\mathrm{CO}}=-100$ pm [20], $z_{\mathrm{model},\mathrm{Cu}}=442$ pm, $z_{\mathrm{model},\mathrm{CO}}=386$ pm, and $\alpha ={20}^{\circ }$, respectively.

Figure 6

(a)–(e) Topographic images of a sample area with five iron atoms acquired in constant-current mode ($V_{\mathrm{tip}}=-10$ mV, $\langle I\rangle =10$ pA). By performing controlled lateral manipulation using a CO tip, a cluster consisting of five individual iron adatoms is formed atom by atom. The cluster is atomically resolved in AFM after each lateral manipulation event: The insets of each panel depict the $\mathrm{\Delta }f$ image acquired in (a) constant-height mode and (b)–(e) constant-current mode, respectively ($V_{\mathrm{tip}}=-10\mathrm{mV}$, $\langle I\rangle =300\mathrm{pA}$, same color scales).

Figure 7

(a) and (b) $I\left(z=\mathrm{const}.\right)$ and (e) and (f) simultaneously acquired $\mathrm{\Delta }f\left(z=\mathrm{const}.\right)$ images above a single CO molecule before and after the assembly of the pentamer shown in Fig. 6, respectively. (d) The differences $I{\left(z\right)}_{\mathrm{CO},\mathrm{before}}-I{\left(z\right)}_{\mathrm{CO},\mathrm{after}}$ and $I{\left(z\right)}_{\mathrm{Cu}\left(111\right),\mathrm{before}}-I{\left(z\right)}_{\mathrm{Cu}\left(111\right),\mathrm{after}}$ are within the noise level of the STM amplifier. (h) The differences $\mathrm{\Delta }f{\left(z\right)}_{\mathrm{CO},\mathrm{before}}-\mathrm{\Delta }f{\left(z\right)}_{\mathrm{CO},\mathrm{after}}$ and $\mathrm{\Delta }f{\left(z\right)}_{\mathrm{Cu}\left(111\right),\mathrm{before}}-\mathrm{\Delta }f{\left(z\right)}_{\mathrm{Cu}\left(111\right),\mathrm{after}}$ of the $\mathrm{\Delta }f\left(z\right)$ curves depicted in (g) show only a small deviation of less than 0.12 Hz which can be attributed to a tiny offset in the tips' lateral positions of the before and after spectra with respect to each other.

Figure 8

(a) and (b) Full raw data sets of the tunneling current $I$ in the backward scanning direction simultaneously acquired with the $\mathrm{\Delta }f$ linescans in Figs. 1 and 1, respectively. A bias voltage of 1 mV was applied to the tip. In both cases, the maxima of the tunneling current curves after the manipulation $I\left(x,z_{\mathrm{man}}\right)$ is larger than the previous current curves $I\left(x,z_{\mathrm{man}}+5\mathrm{pm}\right)$ because the tips are closer by 5 pm. (c) and (d) Full raw data sets corresponding to the $\mathrm{\Delta }f$ data in Figs. 1 and 1 along the $+{\vec{x}}_{\mathrm{A}}$ direction, respectively.

Figure 9

Along the manipulation direction ${\vec{x}}_{\mathrm{B}}$ [see inset of Fig. 1] in which the monoatomic metal tip shown in Fig. 3 is more tilted, the lateral force profiles are more asymmetric with respect to the center of the iron adatom at $x_{\mathrm{B}}=0$ pm compared to the lateral forces along the ${\vec{x}}_{\mathrm{A}}$ direction depicted in Fig. 1. In this case, the iron adatom is manipulated after the tip has passed the center of the iron adatom since the lateral forces are higher in absolute value for tip positions $x_{\mathrm{B}}>0$ pm. The green dotted line shows the tip position in the moment of lateral manipulation.

Figure 10

Sketch of a measurement of a single metal atom (copper or iron) adsorbed on the Cu(111) surface imaged with a CO tip in constant height in (a) side and (b) top view, respectively. (c) Sketch of a COFI measurement with a tilted monoatomic metal tip. The green torus indicates the repulsive torus resolved in previous experiments [22]. (d) COFI image of Fig. 3. (e) The repulsive sickle is interpreted as a torus segment of the front most atom of the tip's apex imaged with the CO molecule adsorbed on the surface. The white scale bar resembles a length of 300 pm.

Figure 11

(a)–(d) The vertical short-range force in four different tip-sample distances. (e) The vertical short-range force in the plane defined by the blue arrow and the $z$ axis. Due to symmetry arguments, the adsorption of the CO molecule to the metal tip's apex as sketched in (f) is more likely than an adsorption as sketched in (g). (h)–(k) The short-range lateral force in the $y$ direction. All $z$ distances are given with respect to point contact $G_0^{\mathrm{CO}}={\left(404497\mathrm{\Omega }\right)}^{-1}$ for CO tips [36]. The scale bars in this figure represent a length of 300 pm.

Figure 12

(a) The experimentally observed lateral force acting between a monoatomic metal tip and a single CO molecule on the surface is higher in absolute value when the tilted part of the tip is facing the CO molecule (see red curve along $x_{\mathrm{B}}$ in Fig. 2(g) of Ref. [38]). (b) The analytical model yields reversed asymmetries for a negative charge $q_{\mathrm{CO},\mathrm{surface}}=-0.03e$ (red curve) versus a positive charge $q_{\mathrm{CO},\mathrm{surface}}=+0.03e$ (blue curve) on the oxygen atom. (c) The charges $q_{\mathrm{CO},\mathrm{surface}}=\pm 0.03e$ were placed in a distance of ${\mathit{off}}_{\mathrm{CO},\mathrm{surface}}=100\mathrm{pm}$ with respect to the core of the oxygen atom. The tilt angle of the tip was set to $\alpha ={20}^{\circ }$.

Figure 13

(a)–(e) Topographic images of five different pentamers which were all created by controlled lateral manipulation using CO tips. The two pentamers shown in (a) and (b) were created and imaged with the same CO tip. The pentamer depicted in (c) was built up and imaged with another CO tip. The pentamers depicted in (d) and (e) were created and imaged with a third CO tip. (f)–(j) Simultaneously acquired frequency shift images which resolve the individual atoms within the clusters. (k)–(o) Frequency shift images of (f)–(j) after post processing in order to enhance the contrast [Gaussian filter ($\sigma =10\mathrm{pm}$), Laplace filter, Gaussian filter ($\sigma =10\mathrm{pm}$)]. All observed pentamers adsorb in an in-plane geometry. In all images of this figure, the tunneling current setpoint was set to 300 pA, except for the cluster shown in the middle column where a tunneling current setpoint of 400 pA was used. The bias voltage was set to $-10$ mV and all images consist of 256 pixel by 256 pixel.