Nonmagnetic atom deposition

Figure 2.6: Ag(111) sample in heating stage viewed through win-dow into UHV preparation cham-ber. To form large terraces, the Ag(111) was annealed to 450 − 600°C by ∼2.8 A current for for 10-15 min with additional high voltage (1000 V) e-beam heating.

Sample temperature was measured by a stage-mounted thermocouple.

During some annealing cycles the manipulator was cooled with liquid nitrogen to prevent unwanted mate-rial degassing onto the sample.

dI/dV but rather resembled measurements of hydrogenated Ce on Ag(111) (figures 2.8 and 2.9) [99]. In these cases we suspect we measured inelastic tunneling events in a CoH complex formed by residual hydrogen gas that is difficult to remove with turbomolecular or ion pumps. If Co coverage was too high to build Ag atom corrals around a single Co, or10 V pulses applied with the STM tip at contaminated Co atoms did not reliably remove contamination so Kondo spectra could be measured, the Ag(111) was annealed to room temperature in the STM chamber transfer arm, which causes deposited Co atoms to sublimate or mobilize and cluster.

0 50 100 150

Figure 2.7: Ag(111) terrace with de-posited Co atoms (small bright contrast points). Clustering of Co atoms (or other abdsorbates) occurs at step edges.

Residual subsurface neon atoms from the sputtering process are visible as dark contrast circles. Imaging parameters:

1 V,500 pA.

.

2.6 Nonmagnetic atom deposition

We used Ag atoms to form corral walls from nonmagnetic atoms rather than Co atoms which might interact with the central Co atom via magnetic interaction through surface state electrons. Another option was to use CO molecules deposited by leaking CO gas

Figure 2.8: Likely inelastic tunneling events seen in dI/dV spectra on Co atom, before and after a 10 V pulse was applied on top of the atom with the STM tip. In inelastic tunneling processes, the tunneling electron loses energy to (or gains energy from) an additional tunneling channel, which can be a vibrational, translational, or spin degree of freedom. This additional channel causes a step-like feature in the tunneling spectrum [99].

into the STM, but we found these molecules to be highly mobile on Ag(111) which caused difficulties for stable scanning and manipulation.

Various options exist for depositing Ag atoms for manipulation. One method is via sublimation from a heated wire or rod. The following two options use the STM tip as a source for Ag atoms. The first is a method in which the tip is crashed ∼ 4 nm into the sample, creating a10 nmwide depression and scattering Ag atoms across the nearby surface [100]. One drawback to this method is subsequent Co atom deposition demands explicit time-consuming spectroscopic or topographic differentiation between Ag and de-posited Co atoms to construct corrals. It is also quite imprecise in placement of Ag atoms for manipulation.

A second option for atom deposition assumes the STM tip contains a practically infi-nite number of Ag atoms and uses it to deposit single atoms on the surface with gentle tip indentations. We follow a procedure whereby we measure tip extension in constant current mode, indent the tip, and measure tip extension again. The distance we indent the tip into the sample is increased in50 pm steps from6Å until the difference between tip height before and after indentation is > 20 pm [58]. Topography measurement checks whether the crash resulted in deposition of a single atom, dimer, trimer, or larger cluster.

Following this method we create ‘silver atom storage’ (figure 2.10) by depositing arrays of Ag atoms before manipulating them around a Co atom using a template.

2.6 Nonmagnetic atom deposition 23

0 1 2 3 4

100 nm 50 0 50 100

Bi as ( m V )

0.0 2.5 5.0 7.5 10.0 nm

0 2 4

6 0.00

0.05 0.10

nm

(a) (b)

Figure 2.9: (a): Atypical line spectrum over the diameter of quantum corral with r = 2.25 nmmade from 8 Ag atoms with a central Co atom showing inelastic tunneling events, in contrast to the spectrum in figure 3.2 which shows the expected Kondo resonance.

(b): Topography of the quantum corral. The horizontal line is the line over which STS spectra were taken. Spectra over the Co atom did not show the typical asymmetric Kondo resonance centered at low bias nearE ∼4 meV(i.e. figure 3.3) but rather a step indI/dV symmetric aroundE_F [99]. A potential cause of this difference is contamination of the Co atom with hydrogen, which creates an additional relaxation or excitation pathway for electron tunneling. Additional noise between the spectra are likely due to changes in the STM tip or changes in the inelastic tunneling pathway during a spectrum acquisition.

Applying large voltage (i.e.10 V) pulses with the STM on top of the Co atom occasionally reverted the state of the atom so that typical Kondo spectra could be measured, but results were inconsistent with respect to long-term stability of the ‘pure’ Kondo state of the atom.

0 10 20 30 40 50 nm

0 5 10 15 20 25 30

nm

0.0 0.2 0.4 0.6 0.8 1.0

Å

Figure 2.10: Constant current topography of Ag adatoms dropped from the STM tip on Ag(111). Atom topographies (bright contrast) have dark contrast to their right due to high scan speed and constant-current feedback; the DSP overshoots as it passes the atoms at high velocity trying to maintain constant current. The image is forward scan data recorded as the tip goes left to right. In ‘backward scan’ data, the depression is to the left. On the top from left to right we deposited: cluster, single atom, cluster, cluster, single atom, cluster, single atom, single atom. Some tip form operations create dark-contrast features;

three are in this image in the second row, for example. Scanning conditions1 V, 500 pA, scanning speed200 nm s⁻¹.