STM Lab 5-1 Physics 212
5. Scanning Tunneling Microscopy Lab
Suzanne Amador Kane 3/13/06 (parts of this manual were adapted from the Burleigh Instruments ISTM manual, and the Physics 407 Lab manual from
University of Wisconsin)
In this lab, you will learn about the principles behind the operation of the scanning tunneling microscope, the first of many modern ―scanning probe microscopies‖ that have opened up the wonders of surface nanoscale imaging for scientists. You will see how the basic quantum mechanical principles of tunneling are utilized in the operation of this instrument, how tunneling is used to create a surface ―topogram‖ which allows the height of the surfaces of conductors to be imaged at the atomic-scale, and how one can use this information to take quantitative measurements on surfaces. You will learn how to operate our Nanosurf EasyScan STM (an instructional STM capable of atomic resolution) and take images of a gold-coated nanoscale grid, the surface of graphite and other samples with nanometer-scale features. You will use surface analysis tools to measure the dimensions of the nanogrid (and calibrate your STM) and the bond angles and lengths for graphite. If you have time, you can also use mathematical image processing methods to process the images to reduce noise and to extract useful information.
We will review six important topics in this lab to understand how STM works. While we are studying these topics in the context of STM, they all are of great general utility and interest for experimental science:
1) How quantum mechanical tunneling works in STM
2) How to control very small displacements using piezoelectric transducers
3) How to use feedback to control tunneling currents
4) How to vibrationally isolate sensitive systems
5) How to collect data electronically
6) How to image process STM data to extract useful information
2) Background on Tunneling and the STM
The quantum mechanical phenomenon of tunneling is described in texts such as Griffiths, Introduction to Quantum Mechanics (section 8-2, pp. 320-325), Eisberg and Resnick Quantum Physics of Atoms, Molecules, Solids, Nuclei and Particles (pp. 199-209) and
Modern Physics by Bernstein, Fishbane and Gasiorowicz (pp. 203-218). You should reread the relevant sections of your textbook if you are not familiar with them at this point.
We first consider the case of a massive particle such as an electron which travels along a one-dimensional path from a region with potential energy V=0 to one with potential energy V = V = constant (this is known as a step potential). (Fig. 1(a)) The electron o
wave is totally reflected from the interface, yet unlike a classical particle, the electron has
STM Lab 5-2 Physics 212
a finite probability of being found in the classically forbidden region where E < V. This o
is because its wavefunction decays to zero exponentially over a distance determined by V and E. (Fig. 1(b)) o
Figure 1. (a) Potential energy function for a step potential and corresponding wavefunction (b). Reproduced from Eisberg and Resnick.
Now, consider the case where the potential energy only equals Vo over a distance a, after
which it drops back down to V=0. This case, known as a barrier potential, is illustrated in Fig. 2(a). Now, the wavefunction will not in general have decayed to zero when it reaches the other side of the potential energy barrier. :
Figure 2. (a) Potential energy function for a barrier potential and corresponding wavefunction (b). Reproduced from Eisberg and Resnick.
The net result is that the electron wavefunction has nonzero amplitude with probability amplitude T (for transmission) on the other side of the barrier, with approximate dependence:
T ， exp( - 2 k a) Eq.1
where 1/k is a measure of the distance over which the exponentially varying wavefunction decays within the barrier. (Here we have kept only the dominant exponential variation of T; see the recommended texts for a full equation for T.) It is
determined by the values of the particles total energy, E, and the potential energy, V(x),
within the barrier by:
2m；？k！V？E Eq.2 o2;
This means that if the electron wavefunction describes a situation in which an electron is incident from the left, it has a probability of either being reflected from the barrier or
STM Lab 5-3 Physics 212
being transmitted, even though it must pass through a classically forbidden region to do so. It is as though a tennis ball thrown against your dorm room wall suddenly disappears from your room and reappears on the other side!
The electrons within an electrical conductor (such as a metal or suitably prepared semiconductor) are in states well described by a free particle wavefunction. As a result, when two conductors are brought very close together yet still separated by an insulating barrier (such as an air gap or layer of insulating oxide), electrons can still flow between them by tunneling. If an electrical circuit is completed between the two conductors, this flow of electrons can be sustained and measured as an electrical current. Just as the transmission coefficient, T, has an exponential dependence on distance, so does the tunneling current depend exponentially upon separation between the two conductors. This is the situation in many common lab settings. If you join two pieces of wires by twisting them together or by sticking them into a breadboard, you often are relying on efficient tunneling across the small gap between them to complete your circuit. This is because you often have thin layers of insulating metal oxides coating the surfaces of copper wires.
This is also what happens in STM. There, one conductor is the very sharp tip of a metal such as tungsten or platinum (with a small 10% admixture of iridium to improve its stiffness). These materials are chosen because you can use them to produce STM tips that have very sharp protrusions ending in only one or a few atoms (if you are lucky!) Imagine that you get a tip in which one atoms protrudes beyond the others by a few Angstroms, as shown in Fig. 3
Figure 3. The STM tip (at top) narrows down to a very sharp point at which one (or a very few) atoms protrude by atomic dimensions. The sample to be imaged is shown at the bottom. From Wisconsin ISTM manual.
STM Lab 5-4 Physics 212
The sample to be imaged is shown at the bottom of Fig. 3. The sample must be approximately flat and itself electrically conducting (or at least a semiconductor). Now the distance, a, between conductors is the tip-sample separation between the bottom-most tip atom and the atoms most close to it on the sample. Now, assume there is some way to bring the tip and sample to a separation of several Angstroms. The tunneling current between them varies exponentially with tip-sample separation, a. This allows us to see
why we can get away with assuming that just the one protruding atom contributes to the tunneling current. The tiny extra distance between the sample and the other tip atoms leads to an enormous reduction in their tunneling currents, due to this strong exponential drop-off. As a result, in the discussion to follow, we will assume that the tunneling current arises only from the one protruding atom.
There are two ways in which STM tunneling is more complicated than the barrier potential discussed above. First, the potential energies of the electrons in the tip and samples may differ. This corresponds in the simplest case to different workfunctions (like the workfunctions discussed in the photoelectric effect) for the different conductors. Second, in order to keep the tunneling current flowing, one must apply an electrical ―bias‖ voltage between the tip and sample. This results in a bias electric field being applied across the gap between the tip and sample, and this modifies the potential energy function the electron experiences. (Fig. 4)
Figure 4 From Wisconsin ISTM manual
Instead of a constant, flat-topped barrier, a sloping potential energy barrier results, with:
V(x) = W – e;： x, Eq.3
where e = electronic charge, W = work function for one of the metals, x = distance across
the gap, and ： = electric field from the applied bias voltage. To compute the new
STM Lab 5-5 Physics 212
transmission coefficient, T, which is proportional to the tunneling current, one would compute instead:
??m2?，；？；？ Eq.4 T，？dxVx？Eexp22)?，;(?
using the equation above for V(x). This detailed relation still yields a tip-sample tunneling current vs. distance that varies approximately exponentially. (Fig. 5)
Figure 5. From Wisconsin ISTM manual
SUMMARY: We expect our tunneling current from the STM: to be virtually zero for large tip-sample separations; to be dominated by tunneling currents from only the bottom-most atom for nanometer scale separations; to vary exponentially with tip-sample separation; and to depend upon tip-sample bias voltage.
3) STM Operation
The EasyScan STM works in one of two modes for imaging the surfaces of samples. You can see how these work in a good video at the website:
STM Lab 5-6 Physics 212
http://www.iap.tuwien.ac.at/www/surface/STM_Gallery/stm_schematic.html The easiest
to understand is Constant Height mode. (Also, see the EasyScan Manual page 30. Note
that your laboratory notebook contains two EasyScan manuals: one for the instrument and one for the software. Unless we specify the software manual, you should assume that references to the manual refer to the main instrument manual.) In this mode, one simply scans the tip in the plane of the sample, left and right, while holding the height of the tip constant. The tip’s motions are controlled by a cylinder of piezoelectric material. Such materials have the properties that they respond to an applied voltage by changing their dimensions. One polarity of voltage results in a shortening of the piezoelectric, while the opposite polarity induces an expansion. By varying the voltage on a piezoelectric, displacements at the sub-Angstrom range can be achieved reproducibly. The position of the STM tip can thus be finely controlled both in the plane of the sample (the x-y plane) and in the z-direction.
The tip-sample spacing varies as the tip is scanned horizontally over the sample, because the surface has atomic-level peaks and valleys due to its atomic structure, and so the tunneling current also varies. So, if one measures the tunneling current I as a function of t
in-plane location (x,y), one can ―map out‖ the topography with atomic precision; high current corresponds to a raised area of the sample. The act of repeatedly scanning x and y back-and-forth to yield an image is also called rastering. (Fig. 6(c)). This yields a map
of (x,y,I) from which an image of the sample’s surface can be made. Since one cannot t
plot in three dimensions, two methods are used to create such surface plots. Either colors (or shades of gray) are used to indicate current (a scale bar is conventionally printed by the side of the image to indicate correspondences between currents and shadings) or a computer reconstruction of the surface is generated and the image is viewed from an angle to indicate its 3D structure. (Fig. 6)
Figure 6 Different methods of represent STM measurements. (a) Burleigh instruments image of the surface of graphite, in which gray-scale (shades of gray)
STM Lab 5-7 Physics 212
are used to indicate height within the plane; (b) IBM image of a “quantum corral” (ring of atoms binding a surface electron) in which computer 3D reconstructions are used to indicate surface structure. (c) Cartoon of how rastering works to accumulate images, and how the information collected is used to make up images of the sort shown in (a) and (b). (IBM STM Gallery
Current mode/constant height mode is a good method for imaging atomic scale structure and you will use it to image the surface of graphite at atomic resolution. However, if you try to use constant height mode to image structures with bumps large compared to the tip-sample spacing, you will hit your tip on the sample and damage it! You can avoid this problem by imaging instead in Constant Current mode. (See the EasyScan Manual
page 30.) In this mode, feedback is used to fix the current at a constant target value (called the Reference Current) as the piezoelectric is used to scan the tip back-and-forth in-plane. If the sample surface is higher at one point than another, the tunneling current goes up. The feedback circuit responds by retracting the tip so the tunneling current is restored to the reference value. If the sample surface is lower at one point, the tunneling current decreases, and the feedback circuit and uses the piezos to lower the tip until the tunneling current returns to the reference value. In this mode, the position of the tip
(rather than tunneling current) is recorded to yield the map of the surfaces’ topography as (x,y, tip z). This interplay between measurement (of tunneling current) and regulated control (of tip height, which regulates the tunneling current) is an instance of negative feedback. It’s negative feedback because a greater tunneling current results in a tip
displacement that reduces the tunneling current. In other words, the motion of the tip always opposes the change measured in the tunneling current.
As a practical matter, you will begin imaging by using Constant Current mode, then—if
you wish to see atomic scale details—switch over to Constant Height mode after you
have found a flat region to safely scan this way.
Understanding the Easy Scan STM Controller Electronics
The EasyScan STM is controlled by a computer attached to a controller box that allows you to set and monitor the tunneling current, bias voltage, scan range and feedback controls. While these controls are described in the manual, which follows this introduction, it will help if you have an overview first. All of this is controlled in software only (no external knobs to turn!) within the Easy Scan program. You should have an icon for this program on your lab setup. First, be sure you have the EasyScan power supply turned OFF. This is so you can use the software in simulation mode to
understand how it works. When you double-click on the EasyScan icon on your desktop to start the program, it will say that it cannot find the controller box, and ask if you wish to run in simulation mode. Answer ―Yes‖, and proceed.
The X-Range setting determines how wide a range you scan over (how large the sample area scanned is). You will use a point and click technique to determine where on the
sample you scan. To start your scans, it is helpful to click ―Full‖ on the top menu. This starts the scans at the largest scan range, 560 nm. You can then zoom in using
STM Lab 5-8 Physics 212
instructions in the manual in section pp. 27-28. The Z-Range setting determines over how wide a range of z-values are recorded. This is necessary because the height values are stored as an 8-bit number. An 8-bit number can only represent at most 28 values = 256 separate distinct z-values. When you use a large Z-Range, you can measure a wider range of z-values, but your smallest distinguishable step in z is large. A small Z-Range allows finer measurements of z, but your tip can go off-scale more easily by measuring z-values too high or too low to ―fit‖ into the measurement range.
You can also set a few options relating to the samples overall tilt. It may not look tilted, but at the submicron scale, it’s impossible to avoid a little bit of a tilt. This may be so
large as to make it hard to see your actual sample’s features. A section in the manual (pp.
25-27) tells you how to correct for sample tilt by subtracting out an average slope in the data.
Nanosurf suggests standard working values for the reference tunneling current, called the
SetPoint of 1.00 nA. This is the average value the system tries to maintain in constant current mode. You will see a working display of the actual measured current as you scan.
Since the STM must scan in order to image, this current will vary in time in either mode. A feedback circuit is used to regulate the current at a constant value (in constant current mode).
Their suggestion for GapVoltage (the tip-sample voltage, also called the Bias Voltage) is 0.05 Volts. This voltage corresponds to the tip-to-sample voltage in our earlier discussion.
Your STM’s feedback circuit uses a common design known as PID (for proportional/ integral/differential.) You will find such feedback circuits commonly used in applications such as temperature control where a physical quantity (temperature, humidity, current, etc.) needs to be regulated by using a physical device (heater, dehumidifier, piezotube, etc.) to adjust its value to a set, desired level. Your STM has several ways of achieving this feedback. You will be using recommended values of these parameters determined by trial-and-error, but here is a short justification of what these different settings do. The proportional setting (P-Gain, default value 13) is useful if you are trying to correct for currents far from the set current value. The correction voltage applied to the piezos to correct the current is equal to a gain factor (set by the P-Gain in software) multiplied by (I = (actual current) – (reference current). If (I is large, the
correction voltage is large, and the sign of the correction voltage varies with the sign of (I, as it should. However, just proportional gain by itself becomes less effective as the current approaches that of its set point, since then (I ~ 0 and no information is available
to adjust the piezo for good control. In this case, it’s useful to move the piezos based on an electronically filtered version of the (I signal. (If you remember how low and high
pass filters work from electronics, that will form a good model for what happens next.) The Integral setting basically has the piezo voltage respond to an integrated version of the (；, while the Differential setting (not used here in the EasyScan STM) changes the piezo voltage based on a differentiated version of (；，;;；In filter, terms, the integral mode
responds to low frequencies, while the high pass responds to high frequencies in the
STM Lab 5-9 Physics 212
current signal.) The time constant over which the current difference signal is integrated is set by the Integrator I-Gain (default value 13) software control. The net effect is to give information about the time behavior of the way the current difference (；;changes, to
enable the circuit to more effectively maintain the current at its set value. Since a differentiated signal responds strongly to high frequency fluctuations the most, and noise introduces unimportant high frequency fluctuations, the Nanosurf EasyScan STM uses only integration. The integrator can give a time-integrated feedback to the piezo which helps it track larger-scale outlines of the surface, which helps in tracing out longer range scans with more surface topography.
We have found that the default settings of the proportional and integral gain do well for the gold nanogrid imaging in constant current mode. However, for moving to atomic-scale imaging, you may wish to start with these values, then reduce the I-Gain to around 2 (to preserve some tracking to avoid crashing the tip) and the P-gain to 0 (to avoid having the tip simply move to track the atomic scale features.) However you may find other values work better for your samples at the atomic-scale; for instance, sometimes it works well to start with the defaults and gradually lower the gains to see what gives the best atomic-scale image quality.
4 The STM Experiments
4.1 Experiment A: Running the STM in Simulation Mode
You will first get acquainted with your software by leaving the STM controller power supply turned off, and starting the software in Simulation Mode. This is done automatically if you startup without power to the STM. Double-click on the EasyScan E-line icon to start the software, and indicate you wish to run it in Simulation Mode when a window appears saying ―No connection to microscope!‖ by clicking on ―Start Simulation‖. Follow the instruction in the manual starting on page 21. (ignore the instructions about mounting tips and samples for now!) Do all of the exercises up to page 30. AT NO TIME WILL YOU BE TOUCHING THE STM DURING THESE
EXERCISES! USE THEM TO FAMILIARIZE YOURSELF WITH THE SOFTWARE!
ONCE YOU HAVE FINISHED, WITH YOUR
INSTRUCTOR’S PERMISSION, PROCEED TO THE NEXT
4.2 Experiment B: The Gold-coated Nanogrid
Nanogrid: STM Magnification Experiment
To attain atomic resolution for gold, the STM signal has to be particularly low in noise. The very nature of the metal means that the electrons will be strongly delocalized between the atoms and there will only be small variations in the electron density with atomic position. The periodic modulations are typically on the order of 0.1 Å, so one should not expect to image gold atoms in normal room conditions. In addition, gold surfaces are ordinarily fairly rough on atomic dimensions, unless they have been freshly
STM Lab 5-10 Physics 212
annealed—a process whereby one heats the gold surface close to its melting point, then allows it to cool so as to leave atomically-flat regions. The purpose of this experiment is instead to introduce the concepts of tunneling and the extremely delocalized nature of electrons defined by the metallic state.
The first sample to examine is the gold-coated nanogrid. This sample is described by its manufacturer as a grid made by injection-molded plastic with a spacing of 160 nm. This sample illustrates imaging in Constant Current mode and allows you to study the level of magnification possible with the STM and imaging of (relatively) larger samples. We have previously collected some good images of the grating for you to use and analyze later if for some reason you cannot get good STM images of your gratings; for now, procede to the next step, setting up your STM for imaging.
Getting set up to image & imaging the Gold Nanogrid
Get your instructor to give you a tour of the actual STM before you begin. You will be following instructions in the EasyScan Manual (attached). The apparatus consists of four pieces of equipment:
1) The STM itself
2) An electronics controller box that measures and sets tunneling current, sets the
bias voltage, sets the piezo voltages and provides the feedback.
3) A computer that communicates with the electronic control box to run experiments,
assemble STM images, and allow analysis and storage of the images.
4) A vibration isolation chamber--in this case, a granite block with shock-absorbing
IMPORTANT NOTE: Many components inside the STM are very static
sensitive. Put on the grounding wrist strap before you handle the STM!!! Remember—the piezo voltages are in the hundreds of volts—treat these
electronics with care and respect!
A few comments on vibration isolation: The very best modern vibration isolation technologies employ special tables that sense any acceleration and counteract it. This is called active (feedback-based) noise reduction. These devices can do an excellent job at suppressing vibrational noise from the tens of Hertz (Hz) range to very high frequencies, and we use them in our atomic force microscopy (AFM) facility. Non-active noise suppression methods are cheaper and simpler. They always entail two components: moving resonant frequencies toward very low frequencies far away from the scanning frequencies of the device, and using an energy absorbing mechanism to damp vibrational energies at higher frequencies. Your EasyScan STM uses such a simple but effective method for suppressing vibrational noise: 1) The entire device is mounted on a block of granite. This material is chosen both for its rigidity (so it does not flex and itself vibrate during scans) and because of its density. Its large mass means that the resonant frequencies of the device will be moved toward lower frequencies, which are less likely to affect the device’s operation. To see why this is so, recall the relationship between