Atom Dynamics in Atom Manipulation

Manipulation of single atoms with the scanning tunneling microscope is made possible through the controlled and tunable interaction between the atoms at the end of the STM probe tip and the single atom that is being manipulated. In the STM tunneling junction used for atom manipulation, a host of interactions that depend on the electric potentials, the tunneling current, and tip-adatom proximity effects that come into play in the atom manipulation process. Understanding these interactions and their optimization is central to understanding the atom manipulation process and is required for the efficient and reliable atom manipulation needed for large-scale construction of atomic scale devices using Autonomous Atom Assembly.

Manipulated atom imaging, an "atom based metrology". Imaging with the STM places the probe tip sufficiently far from the sample so that tip-sample interactions do not initiate atom movement, and hence one can image a stationary single atom, as show in Fig. 1 . In order to move the Co atom seen in Fig. 1, we turn on a the tip-adatom interaction by bringing the tip closer to the adatom by means of adjusting the tunneling junction resistance (the junction resistance is set by the ratio of the tunneling voltage to the tunneling current). It is useful to view the interaction as creating a highly localized potential well that traps the adatom under the tip. A great deal of information about the adatom motion can be learned from recording the tip height trace during the manipulation process. A new type of image, we call a "manipulated atom image" is obtained by repeating this tip height trace measurement by rastering the tip over the surface, as in the normal STM topographic mode, except now with the adatom trapped in the moving tip-induced potential well. Details on a fraction of a lattice constant are observed in the manipulated atom image, as shown in Fig. 2. Our understanding of the manipulated atom image follows from the dynamics of the adatom motion hopping between fcc and hcp sites, as illustrated in Fig. 3C-D. The manipulated atom image for Co on Cu(111) can be thought of as a binding site image for the Co atoms, with the additional contrast changes between the fcc and hcp sites yielding information concerning the difference in the surface potential energy for these two sites. Using the motion of the Co atom to image the local binding sites and give information on the local surface potential is an example of a new class of measurements based on single atom transducers sensing local environments, which we refer to as an "atom based metrology".

Figure 1. STM topographic image of a single Co atom on Cu(111) shown in a light shaded view. Trrent 1 nA, sample bias -10 mV, T=2.3 K.

Figure 2. (A) Manipulated atom image of Co over Cu(111) surface. Tunnel current, 50 nA; sample bias, −5 mV; T = 4.3 K. The labels A, B, and C denote fcc, hcp, and top sites, respectively. (B) Tunnel current recorded during manipulated atom image going through the hcp and fcc sites as indicated by the horizontal line in (A). The arrow shows the increased noise in the tunnel current corresponding to the position of the hcp site.

Figure 3. (A) Top view of the Cu(111) surface with the Co adatom shown in its natural fcc binding site. (B) Schematic potential well for the Co atom in fcc and hcp sites: blue curve, native potential well, no tip- Co interaction; green curve, tip-induced potential well; red curve, native potential with added tip-induced potential. The potential at the hcp site increases in depth because of the increase in tip-Co interaction as the tip-Co distance decreases. The tip-induced potential well over the hcp site causes the Co atom to switch between the fcc and hcp sites, producing discrete changes in the tunnel current. (C to E) Schematic of manipulated atom tip height trace. Initially, with the tip over the fcc site, the force on the Co atom is vertical and the tip images the Co atom. As the tip moves down the side of the Co atom, a lateral force develops (D). When the tip reaches the hcp site, the lateral force is large enough to induce the Co atom to hop to the hcp site (E). The green curve is the measured tip height trace from the manipulated atom image in Fig. 2A.

Listening to "Hip-Hop" atoms. Audio frequency components of the tunnel current can be used as a real-time diagnostic of the atom manipulation process. For Co on Cu(111), a periodic rasping sound is heard during the acquisition of a manipulated atom image. An examination of the tunneling current signal shows that this occurs when the Co atom is placed over the hcp site (see Fig. 2B). It is this increase in noise density that gives rise to the "rasping" sound. An examination of the tunneling current noise properties shows that the Co atom is switching between the favored fcc site and the meta-stable hcp site, and it is this switching phenomena that is responsible for the periodic rasping sound.

Insight into the atom dynamics is obtained from a quantitative measure of the noise properties of the tunneling current. To measure the current noise, we measure the time dependence of the current at the hcp site during a manipulated atom image. The tip is paused during the raster and the STM feedback is opened. The tunneling signal is then measured as a function of time for several values of bias voltage. Random telegraph noise (RTN) is seen these measurements, as shown in Fig. 4B and D . The RTN signal corresponds to the Co switching between the hcp site (the higher current signal) and nearby fcc sites (lower tunnel signal). When the tip position is near the center of the hcp site, the Co atom is observed to switch to all three neighboring fcc sites (see Fig. 4D). As the tip is biased towards one of the fcc sites, only two-states are observed in the RTN signal corresponding to the Co atom switching only to one fcc site (see Fig. 4B). In this way, an ideal two-state fluctuating system can be created based on the position of a single atom using the tip-adatom interaction, which allows the detailed study of the switching mechanisms.

Figure 4. (A) Co manipulated atom image on Cu(111). Tunnel current, 100 nA; sample bias, 11.0 mV; junction resistance, 110 kilohms; T = 2.3 K. (B and D) Tunnel current versus time measurements recorded at the positions indicated by the corresponding red spots in (A) near the hcp site. Sample bias, 3.3 mV. (C and E) The corresponding histograms of the current distributions. (F) Tunnel current versus \time measurement showing two-state random telegraph noise near the hcp site for Co on Cu(111) measurement at junction resistance of 120 kilohms. Sample bias, 8.4 mV; T = 2.3 K. (G) Corresponding histogram of the current distribution.

The switching data shows that the tip-Co interaction modifies the unperturbed surface potentials, as shown schematically in Fig. 3B. Because we observe the switching only with the tip at close tip-Co distances, we can conclude that the original potential well at the fcc site is deeper than the original well at the hcp site. The presence of the tip interaction deepens the potential well at the hcp site (Fig. 3B), which makes it favorable for the Co atom to switch back and forth at these temperatures. By analyzing the switching rates we can obtain a quantitative measure of these potential changes, and gain insight into the mechanisms for single atom switching.

Atom switching dynamics. Insight into the switching dynamics is obtained from analyzing the two-state RTN. As seen by comparing Figs. 4D and F, the switching rate is very dependent on bias voltage, or tunnel current. The RTN corresponding to the Co atom switching between the fcc and hcp sites can be described by a discrete two-state Markov process with an exponentially distributed residence time distribution. From an analysis of the residence time distributions we arrive at a switching rate for the Co atom to go from the fcc to hcp site, and vice-versa. Figure 5 shows these rates as a function of sample bias for fixed tip height set at 150 KO junction resistance. Two distinct regimes are evident in this atom switching data. At low bias below ±5 mV, the transfer rate is independent of bias or tunnel current (since bias and current are linearly related in this low bias regime). Above ~5 mV, the transfer rate shows a power-law like behavior spanning 4 orders of magnitude.

Figure 5. Transfer rate versus sample bias at constant tip height, obtained by measuring the distribution of residence time in the hcp and fcc states from two-state random telegraph noise in the tunnel current. Junction resistance 0 150 kilohms; T = 2.3 K. Rhcp, red circles; Rfcc, black squares. Solid red line shows a power-law fit to the initial threshold region; blue horizontal line shows the average transfer rate for the low-bias region for the hcp transfer rate.

A model for the atom switching at high tunnel currents is based on a vibrational heating model. In this model the atom overcomes the potential barrier through stepwise climbing of a vibrational ladder of the adatom-substrate bond excitations through a competition between gaining energy from inelastic tunneling electrons and losing energy to electron-hole pairs and phonons. The vibrational heating model qualitatively accounts for the transfer rate data in Fig. 5 for sample biases above ~ 5 mV. The model does not yield agreement for the transfer rate data below 5 mV, which is independent of voltage and current. Temperature dependent measurements also show the transfer rate to be independent of temperature in the range from 2 K to 4 K, which suggests an alternate mechanism for the switching rates observed at low biases. Quantum tunneling of the Co atom might seem improbable given the large atomic mass of Co. However, the narrow barrier and small barrier height yield quantum tunneling estimates which are comparable to the transfer rates seen in Fig. 5 for biases below ±5 mV, and therefore suggest that quantum tunneling of the Co atoms is observed in these measurements.

Tuning the potential landscape. The tip-adatom interaction can be highly controlled using the picometer resolution and stability that is inherent in the tip-sample distance in STM measurements at cryogenic temperatures. Measurements of the atom switching transfer rates allow a quantitative determination of the changes in the trapping potential of the tip. For these measurements, the probe tip is paused at the hcp site during an atom manipulated image, and the STM feedback loop is opened. The time dependence of the tunneling current is then recorded at fixed tunneling voltage for a series of tip-sample distances. Figure 6 shows the current distribution for two-state switching tunnel noise from such a measurement. At small tip-sample distance, the Co atom is completely trapped in the hcp site. As the tip-sample distance is increased and the tip-adatom interaction is weakened, the hcp population continuously decreases and the fcc population increases. Finally, at the larger distances, the Co atom resides in the fcc sites, completely reversing the relative binding energy of the hcp and fcc sites. From an analysis of the measured transfer rates during these measurements, we obtain a quantitative measure of the difference in ground state potentials of the hcp and fcc sites as a function of tip-sample distance.

Figure 6. Current distribution from two-state telegraph noise obtained near the hcp site during a Co manipulated atom measurement as a function of tip-sample distance. Zero tip-sample distance corresponds to the initial set point at junction resistance of 90 kilohms; Z = 0.35 Å corresponds to 180 kilohms. Sample bias, −5 mV; T = 2.3 K.

Online: March 2005
Last Updated: February 2008

egpwebmaster@nist.gov