The setup used in this bachelor project is a Createc Low Temperature Scanning Tunneling Microscope developed in 1996 [9]. The STM has very recently been acquired by the group and therefore much effort has gone into implementing the system, which was the primary task of this bachelor project. The STM is capable of working at liquid nitrogen and liquid He temperatures at ultra-high vacuum (UHV). This provides very low thermal drifts and reduces thermal fluctuations, which is essential when performing STS
measurements.
3.1 The STM
Figure 3.1 depicts the STM unit which is based on a Besocke “beetle” design [9]. In experimental setup the STM unit is surrounded by two radiation shields in order to keep the STM thermally stable during low temperature measurements. The outer shield is cooled by liquid 𝑁 to 77K, while the inner shield can be cooled down to 4K by liquid He. As can be seen on the image, the wires are very thin to avoid heat conductance to the STM and thereby destabilize it thermally. The basic STM parts are the following:
1) Scanner tube
2) Piezoelectric moters for approach
3) Tip
4) Suspension connecters 5) Sample position 6) Sample connections
1. Scanner tube: The scanner tube is responsible for movement of the tip during the scan. It is made from a piezoelectric material, which allows very accurate movement by applying a voltage to it. The tube is coated by a metal both inside and outside and divided into a large central electrode and two upper and lower electrodes. The outer electrodes enable the piezoelectric material to move in the xy-plane which is needed for raster scanning the surface. The inner electrode is for movement in the z direction.
Figure 3.1: STM unit of the Besocke "beetle" design
6 2. Piezoelectric scanner tubes for approaching: These three scanner tubes are responsible for coarse approach of the tip to the sample. The scanner tubes carry sapphire balls on which the central scanner tube rests. The approach is done by applying a saw tooth wave to the piezos and thereby performing a “stick and slip” movement, i.e. the ring carrying the tip is rotated during the slow rise of the voltage after which the sapphire balls slip back to the original position during the fast drop of the voltage. This cycle is repeated many times until the tip approaches the sample. Tip movement in the xy-plane is done by applying an appropriate voltage to the scanner tubes such that all of them are shifted in the same direction.
3. Tip: The tip is in our case made of either tungsten or a PtIr alloy. It is attached to the piezo material by a magnet and can easily be exchanged without opening the chamber. The tungsten tip is made by etching.
To clean the tip from impurities after the etching it is annealed in vacuum, using electron bombardment from a filament ring mounted on the sample garage.
4. Suspension springs: To reduce mechanical vibrations, the STM is suspended by three helical stainless-steel springs which are mounted on the He cryostat. The springs are often very soft to make the largest mismatch between the mechanical modes of the springs and the high resonance frequency of the stiff STM unit.
5. Sample position: The sample is loaded into the STM by pulling the connections (6) down and transferring the sample holder (Fig. 3.2) into the available space. When the sample is in position, (6) is pulled up to release the STM and lock the sample holder into position.
3.2 Sample holder
Figure 3.2 shows the sample holder. The sample (1) is mounted on a ceramic heater and tightened by a molybdenum cap. The sample temperature is monitored by a thermocouple (2) mounted on the Mo cap
and shows the temperature difference between the sample holder and the sample. For this reason, the temperature cannot be monitored during cooling, since the sample and sample holder are cooled simultaneously to the same temperature. The connections for the thermocouple and the heater are located on (3). This is also where the bias is applied during the scan. The holder is grabbed by the manipulator in the groove (4).
Figure3.2: Sample holder for LT STM
7
3.3 Instruments on the chamber
The Low Temperature setup is imaged in figure 3.3. The chamber consists of two main chambers and a load lock. The preparation chamber is where samples are cleaned by annealing and, for metals, sputtering. This is also where the hydrogen source and molecular doser are mounted. The STM chamber is of course where measurements are performed and is not really used for anything else although it would be possible to mount a doser on this chamber to do in situ dosing.
1)-4) Manipulator used for
(1) Manipulator is used to transfer sample between the preparation and the STM chamber. The sample position can be adjusted in x and y directions by the micrometer screw gauges (2) and (3) respectively, in z-direction on the handle below the manipulator (not shown) and rotated around the z-axis on (4). Sample cleaning and preparation, e.g. sputter/ annealing, dosing molecules or atomic hydrogen is done while the sample is held in the manipulator. Since the bearings in the rotating part of the manipulator give rise to vacuum leaks, this part is at all times pumped to keep the chamber pressure in the UHV region.
(9) Sputter gun is used for cleaning the sample. Neon or Argon atoms are ionized, accelerated through a potential ranging from 1𝑘𝑉 − 3𝑘𝑉 and bombarded onto the sample to remove dirt. After the sputtering, the sample is usually annealed to reconstruct the surface. Sputtering is primarily used on metals since the atoms are quite mobile and therefore the surface is easy to reconstruct.
(5) Ion gauge is used for pressure measurements in the range of 10 𝑚𝑏𝑎𝑟 − 10 𝑚𝑏𝑎𝑟. A hot filament emits electrons which are attracted towards a positively biased metal grid. On the path from the filament to the grid, electrons hit rest gas molecules and ionize them. The ions are collected at a wire in the middle of the grid. The measured ion current is then related to the pressure. The experimental setup is equipped with three of these gauges for separate reading of the pressure of different parts of the setup– the preparation chamber, the STM chamber and the load lock.
Figure 3.3: The Low Temperature STM experimental setup
8 (10) Hydrogen source is of the Juelich design [10]. It consists of a tungsten capillary which is heated to around 2100K. A flow of hydrogen molecules at a feeding pressure of ~10 𝑚𝑏𝑎𝑟 is passed through the capillary to crack the molecule into hydrogen atoms, which are then dosed to the sample surface. The exposure can be controlled by a shutter.
(12) Cryostat is used for cooling the STM to cryogenic temperatures. The cryostat consists of an inner and outer dewar, which are in mechanical contact to the inner and outer radiation shields of STM, respectively.
The outer shield is always cooled by liquid 𝑁 while the inner can be cooled by either liquid 𝑁 or liquid He.
When using He, the STM can be cooled down to 4,2K at which the thermal fluctuations are minimal and thus the surface is very stable.
(8) Load lock enables loading samples without venting the chamber. In the load lock the sample is loaded onto a transfer stick at ambient conditions. Afterwards the load lock is pumped down for a few hours until high vacuum is reached. The valve separating the load lock and the preparation chamber can then be opened without compromising the vacuum too much.
3.4 Pumping of the chamber
In order to achieve UHV, several different pumps are needed. The chamber is equipped with a roughing pump as well as turbomolecular pumps, ion pumps and sublimation pumps:
Roughing pump: Uses a rotating arm to push the gas out through an oil-filled compartment. Is capable of pumping the chamber to around 10 𝑚𝑏𝑎𝑟, which is necessary since the turbomolecular pump does not work at ambient pressure and after pump down to keep the pressure behind the turbomolecular pump.
Turbomolecular pump: This pump is built of a fast spinning turbine rotor. The principle of operation is that gas molecules can be given a momentum in a desired direction by repeated collisions with the rotor blades.
The molecules at the inlet hit the rotor and are sent towards the exhaust, where the roughing pump is connected, thus lowering the pressure. UHV can be obtained using the turbomolecular pump but ion pumps are used on most chambers due to better efficiency of removing some gasses. The Low
Temperature chamber has two turbomolecular pumps, one on the preparation chamber and one on the load lock. The first is primarily used to keep the pressure during sputtering and dosing where the ion pump is turned off while the last is used for sample loading, differential pumping of the sputter gun, and pumping of the hydrogen source and molecular doser.
Ion pump: Ionizes gas molecules by a plasma discharge using a large potential of 7𝑘𝑉 to accelerate the ions into a solid from which they cannot escape. The ion pump does not function above pressures higher than 10 − 10 mbar and hence the turbo pump is needed at higher pressures. Ion pumps are employed on most chambers because it has high pumping efficiency for different atomic masses than the turbomolecular pump. Also it is often kept running during STM measurements which would be impossible with the
turbomolecular pump due to a very high mechanical noise level.
Titanium sublimation pump: Works by sublimating reactive Ti atoms from a filament which is heated by applying a large current. The sublimated Ti atoms are deposited to the chamber walls where rest gas molecules can stick to the atoms. It is only necessary to run the sublimation pump occasionally, but after
9 some time the Ti coating is no longer active and a new layer is needed. A major advantage of the
sublimation pump is that hydrogen easily sticks since this is usually hard to pump out with other types of pumps.
3.5 Noise measurements of the STM setup
Since STM is a very delicate experimental technique, high levels of noise are not tolerated. Because of pumps and electronics connected to the chamber, they all can contribute to the noise, which can spoil the measurements. This will be investigates in this section.
Figure 3.4: Spectrum with everything turned on
Figure 3.4 shows a frequency spectrum of the noise in tunneling current for a situation where both the turbomolecular- and ionpumps are turned on. As seen in the figure there are two frequency ranges with large level of noise intensity. The root mean square of the noise level is about 48 meV. To identify partial contribution from different sources, the pumps will be turned off one after each other while measuring the frequency spectrum of the tunneling current.
Figure 1.5: Ion pump turned off
Figure 3.5 depicts the situation where only ion pumps are turned off. All the frequencies have been reduced drastically except for the peak at 1𝑘𝐻z. That frequency corresponds to the spinning frequency of the turbomolecular pumps. The root mean square (RMS) of the noise has been reduced substantially from 48𝑚𝑉 to 16𝑚𝑉. The spectrum, however, still shows two smaller peaks at ca. 0.8𝑘𝐻𝑧 and 5 peaks between
10 1𝑘𝐻𝑧 and 2.5𝑘𝐻𝑧. These I cannot account for, but it is probably electronic noise, since they are found at relatively high frequencies.
Figure 3.6a shows the contribution to the noise from the ion pumps only. In that case the sharp peak at 1kHz from the turbopumps has disappeared. The broad range of frequencies seen in the spectra is due to a high voltage (HV) generator for the ion pump. If this is switched off the noise level is reduced from ~39 meV to ~ 9meV, figure 3.6b. However, as figure 3.6b shows it is not only contributions of the HV generator to the noise which is seen in figure 3.6a. Even when HV generator has been turned off, a significant noise level is apparent. To find out the source, the ion pump controller itself was un-plugged from the power socket, which had indeed a quite large effect. The RMS was reduced by almost a factor 3 to 3.6𝑚𝑉 which is an acceptable noise level, figure3. 6c.
Figure 3.6: Ion pump: a) ion pump running, b) ion pump turned off c) Controller turned off
These results show that a lot of electronic noise can be eliminated by turning off the ion pump controller when scanning for atomic resolution. This result is rather unexpected since electronics should not contribute that much to the noise.
11