Nanotechnology: The Homebrew STM Page

The goal of this page is:
To provide a resource for those intrepid explorers who may want to try constructing their own STM. Any help in filling out this outline would be greatly appreciated. Send contributions/suggested changes to: Jim Rice

Index

• STM Theory
• STM Design
• STM Scanner
• Printed Circuit Board Incomplete.
• Digital-to-Analog/Analog-to-Digital (DAC/ADC) Board Incomplete.
• Control and Display Software Incomplete.
• Vacuum Systems
• Cryogenic Systems Incomplete.
• Samples
• References
• Supply Sources
• Scanning Probe Microscope Makers
• STM Technology-Related Links
• Nanotechnology Web Pages

STM Theory

• Tunneling Theory
• Piezoelectric Tube Theory
• Z Movement: Driver Voltages and Calculation of Position
• X, Y Movement: Driver Voltages and Calculation of Position
• Piezoelectric Nonlinearities and Scanning Error Correction
• Data Acquisition and Control Theory
  • Sampling and the Nyquist Frequency
• Spectroscopic Theory and Techniques

Tunneling Theory

Scanning Tunneling Microscopy (STM) tunneling theory is quite complex, but may be simplified greatly by several approximations. STM involves the tunneling of electrons through vacuum from the tip of the STM to the sample. For simplification, this can be considered as tunneling between two metallic electrodes, separated by a vacuum region. The potential in the vacuum region acts as a barrier to electrons. In this simplified form, one can apply the trivial solution of the Schrodinger equation, applied to a rectangular barrier:

Psi = exp{kx} **(note: kx can be either + or -)

The critical parameter in this solution is k. One finds that:

k = sqrt[2m(V_b - E)/(h²)]

where, V_b represents the barrier potential, E is the energy of the particular state, and h corresponds to the reduced Planck constant. In order to simplify the model, V_b can be substituted by its average value across the potential barrier; however, generally, V_b would not be constant. With this simplification, the system then becomes a rectangular potential barrier. Further analysis leads to the result that the tunneling current through the vacuum is proportional to:

exp{-2ks}

where, s is defined as the separation between the tip and the sample. This result turns out to be the key to STM. If an atomically sharp tip is used, the tunneling current from the first atom of the tip will be exponentially larger than that of the tip atoms which are slightly behind it. Considering general work function values, and substituting this for (V_b-E) in the equation for k, one finds that the tunneling current decreases about an order of magnitude for every angstrom of separation, s, between the tip and the sample.

Peizoelectric Tube Theory

In order to achieve scanning in the X and Y directions, the outer electrode of the tube is sectioned into quarters. The inner electrode is grounded. When a positive voltage is applied to one of the quartered electrodes, that segment of the tube contracts perpendicular to the electric field. This causes the whole tube to bend perpendicular to its axis. It is best to use push-pull voltages on opposite tube quadrants. A single voltage on one quadrant will result in tube bending in x or y, but also some undesirable net motion in the z-direction. Opposite polarity voltages of equal magnitude on opposite quadrants will bend the tube with no net z motion. Orthogonal x-y motion is obtained by controlling the voltages of the two electrode pairs in an additive vector fashion.

Motion of the tube in the Z direction is obtained by applying a voltage to all outside electrodes uniformly. A positive voltage causes a uniform contraction of the tube. A negative voltage causes an expansion. Important Note: Do not try to get increased range by applying an equal and opposite voltage to the inner electrode. This will tend to depolarize the piezo tube. The inner electrode should always be grounded, or at the negative pole, i.e., if your range is +-18 volts, the inner electrode should be held at a steady -18 volts.

Crystallographic conventions are used to describe the orientation of a poled piezoceramic and to characterize the performance parameters of the various piezoceramic meterials. The specific type of performance we are concerned with in a STM is:

d_a,b where:

d=Displacement, or the ratio of mechanical strain developed to electrical field applied, expressed as 10^-12 meters/V, or angstroms/volt.
a=Electrode orientation
b=Direction of induced strain

For a, the only value, for our application is:

3=the dimension from the positive to the negative electrode. In a tube piezoceramic, this is normally the thickness of the tube, as the electrodes normally coat the interior and exterior of the tube.

For b, the only value for our application is:

1=The direction of induced strain (movement) along the tube's length.

d_3,1 is the "piezoelectric coefficient" which, when you plug it into the Delta_L formula (below), tells you how much the tube will shorten (due to the negative sign) when a positive voltage is applied. D_3,1 = -1.73 angstroms/volt for PZT-5A type ceramic at 293 degrees Kelvin. It characterizes tube length changes with voltage.

This is an important parameter, and it does vary with temperature. The higher the temperature, the more the expansion or contraction. It doesn't vary much for PZT-5A, though: About 11% from 0-100 deg. Celsius, vs. over 30% for PZT-5H.

D_3,1 is an approximation.

Z Movement: Driver Voltages and Calculation of Position

Tube movement up and down is calculated using the formula:

Delta_L = d_3,1 * V * L / t

where Delta_L is angstroms, d_3,1 is -1.73 angstroms/volt for PZT-5A material, V=Voltage, L=Tube Length in mm, and t=wall thickness in mm.

This is the formula that you can use to calculate your total Z range (just plug in your total voltage range) and Z position from recorded Z voltages. Actual movement of the tube can vary from this calculated value up to plus or minus 20%, and can vary with temperature, tube age, etc. This is why a Look Up Table may be used to correct the image after scan, or (preferable) correct the scanning waveform to achieve an undistorted scan. Generally, this error increases with increasing driver voltages. It appears to be very small for voltages under 20.

Note that the Z movement range depends on only two things: How long your tube is, and how thick the walls are. The thinner the walls, the more the range. It doesn't depend on the radius of the tube. As an example, assuming a tube thickness of 0.022 inch (0.5588 mm), a tube length of 0.5 inch (12.7 mm) and using the formula just given, Delta_L is about -39 angstroms for +1 volt, and -707 angstroms for for +18 volts, or 1414 angstroms (141 nanometers) for +18V to -18V. This would be the movement of the z-tip if a -18 to +18 volt voltage ramp were applied to the inner tube, and represents the Z range. The negative sign of d_3,1 means that the tube would contract when a positive voltage is applied.

Z-positioning in the scanning, mode, uses a +- voltage range to vary the tip position to keep the tip within an angstrom or two of the sample surface, via feedback maintenance of a constant tunnelling current.

X, Y Movement: Driver Voltages and Calculation of Position

Calculation of x and y movement/volt is based on the following, assuming:
equal and opposite voltages applied to opposite quadrants,
and using the following formula for delta x,y for tubes with OD quadrant electrodes:

Delta_x = 0.90 * d_3,1 * V_x *( L)² / DT
Delta_y = 0.90 * d_3,1 * V_y * (L)² / DT

where V_x or V_y = applied voltage across one quadrant, L = tube length in mm(or cm), D=tube diameter at midpoint in wall thickness T= wall thickness in mm (or cm).

Note that as both the tube diameter and wall thickness get smaller, the X,Y range increases linearly, and as length increases, range increases as the square. Long, narrow tubes with thin walls have better X,Y range.

Assuming a tube diameter of 0.25 inches or 6.35mm,

=0.90*(-1.73Ang/V)*1 volt*(12.7mm)²/6.35mm x 0.56mm)
=-71 angstroms of movement/volt

For one angstrom change in the X and Y direction, a voltage change of +1volt/-71 angstroms movement=0.014 volt/angstrom would be required.

=71 Ang/volt times 36 volts total plus/minus range
=2556 Angstroms lateral (X,Y) range, or 255nm.

The Y sawtooth slowly increases from min to max either smoothly or ratcheting up one increment at the end of each scan line. Actually, there'd be two equal and opposite sawtooth waves, one from zero to +18 volts on one electrode quadrant, and one from zero to -18 volts on the opposite quadrant, while many triangle waves occur for X scanning. There is slow Y movement during X scanning, under the smooth waveform scheme, so the actual scan path is a kind of compact "Z" pattern back and forth across the sample rather than a rectangular raster pattern. Alternatively, the Y movement could be incremented by software at the end of each scan line, forming a regular raster pattern.

Actually, according to Chen, for the X scanning waveform it would be better to use a sine wave to avoid abrupt stops and starts to the X-Y piezo tube. This would make plotting the data somewhat more complicated, however.

Each display point sampled would contain the analog Z position, so we would need at a minimum a single 16 bit (preferably) analog-to-digital convertor (ADC).

Piezoelectric Nonlinearities and Scanning Error Correction

1. Hysteresis: This is the intrinsic non-linearity of the molecular dipoles in the ceramic tube. As the deflection of the tube increases, it takes more voltage than a linear increase of the initial deflection voltages. This effect, as mentioned, increases with voltage and deflection. Additionally, the return path (X-) of the scanner deviates from the extension path (X+). For this reason, although most commercial scanners permit data gathering on both X- and X+ legs, they are normally not integrated into a single image. Rather they are stored as two separate images. A homebrew STM scanner should be capable of acquiring data from either the X- or X+ leg, since image quality could be better in one direction, but it would probably be sufficient to be able to designate before the scan which direction to acquire data on.

2. Creep: After a driving voltage is applied to a scanner, the tube will move most of the way there quickly, then, over the next minute or two, will move an additional 5% or so in the direction of the original movement. This will interfere with the scan, of course. One way around this is to integrate a capacitor into the drive electronics which slowly discharges against this trend, keeping the tube more or less stationary.

You can determine X and Y position by either:

1. Measuring X and Y driver voltages (normally used), or
2. Recording the commanded X and Y driver voltages

Due to the two errors mentioned above, an uncorrected scan results in a distorted image. An error correcting scheme for the driver voltages may be employed to ensure accurate tip positioning and straight scanning. This would take the form, normally, of a look-up-table (LUT) of modified drive voltages. There would have to be a different LUT for each scan size and speed.

Data Acquisition and Control Theory

DAC Resolution: 12-bit vs. 16-bit

X and Y driver voltages may come from the control circuit's Digital-to-Analog convertors (DACs), driven by software. With 12-bit DAC/ADC accuracy, lateral accuracy of 36 volts/2048 steps=0.0176 volts. Now 0.0176 volts/71 Angstroms per volt = 1.24 angstroms per step. This is the granularity of your image with a 12-bit DAC/ADC. If you want better accuracy, you need a 16-bit DAC/ADC (64,000 steps). Since a typical atom is around 3 angstroms diameter, we do need 16-bit for atomic resolution. This would give us 36V/64000=.0006 volts/step=.04 angstrom positioning resolution.

Spectroscopic Theory and Techniques

It would be nice if our scope could be programmed to do the following:

1. High speed scanning I-V spectroscopy. A small high-frequency sinusoidal modulation voltage Vmod Wmod is superimposed on top of the constant DC bias voltage between sample and tip. The AC component of the tunnelling current is then measured with a lock-in amplifier, with the in-phase component directly giving dI/dV simultaneously with sample topography. The lock-in output is measured as a function of Vdc. The modulation frequency Wmod must be faster than the closed-loop bandwidth of the STM feedback system (typically 1-2 khz). If the frequency is too low, the feedback circuit will try to compansate by changing gap spacing. Optimal frequency is slightly above cutoff freq of feedback loop and can be found by identifying the frequency range where dI/dV is independent of frequency.

2. Stationary I-V spectroscopy. Local I-V measurements with constant tip separation: The feedback loop is interrupted to keep the tunneling gap constant. A voltage ramp is applied to the tunneling junction, and the tunneling current recorded as a function of bias voltage. This approach acquires complete curves of I vs V at intervals during slow scanning. A sample-and-hold circuit is installed in the feedback controller to gate the feedback control system on and off. A commercial sample-and-hold device is not good enough. The gating circuit usually consists of an electronic switch. The period of the waveform is approxomately 500 microsec, during which time approx 80 us is devoted to stabilising the sample-tip separation, 400 us devoted to the tunneling I-V curve, and the remaining 20 us is dead time to allow capacitive transients to die out. During tunneling I-V measurement, the applied voltage is ramped between two voltages while measuring I. One problem is that a wide range of current is generated due to the exponential dependence of I on V.

3. Multiple Stationary I-V Spectroscopy. Varying tip-sample separation while performing multiple V-I scans. The bias is initially set at a relatively large value (+-2V) Feedback is disabled, and a V-shaped voltage is added to the Z-piezo voltage, moving the tip slowly in and back out. The bias voltage is ramped rapidly up and down as the tip moves in and out, and the I vs V curve measured at various tip distances.

4. Stationary S-I Spectroscopy. The tip is moved in and out and the tip current measured as a function of tip-sample distance.

STM Design

• Overview
• Design Objectives
• Desired Tip-Sample Interactions
• Performance Specifications
  • Scanning Range of Motion
  • Electronics Specifications
• Advantages/Disadvantages of Two-Tube Design
• Estimated Cost Analysis

Overview

Software is a major problem. We would like to wind up with a pseudo standard homebrew STM that runs on shareware or freeware. PC clone or MAC. The hardware interface might be a set of RCA audio connectors.

The electronics (scan waveform generation, etc.) should have good basic performance but not too many bells or whistles.

Design Objectives

1. Scan with atomic resolution a designated area.

2. Should be able to reverse polarity of the tunnelling current, as different images are created with each polarity direction. To get the best image, switching is essential. A simple electric switch should suffice for this.

3. Conduct I-V spectroscopic measurements (optional).

4. Tip reference current (the current that the feedback circuit compares itself to) should be variable to vary the tip height for a scan. The lower the tip current set point, the greater the scanning tip distance to the sample. More current means the tip is closer to the sample. Approximate values are from 1 to 10 nanoamps, not counting values needed for tip-sample interaction, which could be higher.

5. Extensive tip-sample interaction capability. Example: After the scanning was done, software would need to plot each point and (hopefully) make it a continuous 3-D surface-type picture. THEN, what would be really neat, would be to somehow have a representation of the tip position appear in the scanned image corresponding to the actual tip position, in all axes. Then, you could move the tip around with a joystick or mouse, digging channels, adsorbing atoms, (maybe), etc., while watching the image on the screen. Of course, you wouldn't really be watching what's really happening. You'd be seeing an old image and a calculated tip position. But with that visual representation of the data, you could do some neat things. Then, you go back and do a repeat scan to see what you've really accomplished. So, in addition to the scanning circuits, you'd need a direct, all axes tip controller.

Desired Tip-Sample Interaction Capability

1. Etching- Direct physical contact: Use the tip to etch a trench, line or pattern into the sample surface. You would need to know the Z voltage for the scanning height at the digging point, then disable the Z feedback circuit, and increase or decrease the z voltage to lower the tip into the sample. Since scanning height is typically from one to ten angstroms, or 0.1 to 1 nanometer, then z voltage would have to be changed by about 1/40th of a volt for each angstrom of tip movement. Controlled dipping of the tip into the sample is also a way to sharpen the tip to regain atomic resolution (probably because a few atoms cling to the tip as it's retracted, forming an atomically sharp tip again).

2. Etching- Blasting pits with electricity: Blast pits in the substrate with pulses of high tip potential (up to 18 volts with 20 nA of current) lasting one to several microseconds.

3. Pinning molecules to the surface: Example: using the tip pulses mentioned above with durations of 100 ns in a gas of tungsten carbonyl to deposite tungsten carbide spots on the surface, and even build columns of tungsten carbide to arbitrary heights. .The feedback circuit remains operational during a pulse. While pinning molecules to the surface repeatedly to build columns of molecules, the tip retracts itself from the surface, maintaining a constant distance from the top of the column.

4. We need to be able to specify a tip pulse's duration (nanoseconds or microseconds to seconds), voltage, amperage (nanoamps on up) and interval between pulses (if doing more than one). We need to be able to disable the feedback circuit during the pulse, or the tip would leap away from the sample.

5. We need to be able to vary the scanning speed (X and Y driver voltage waveforms) to adapt to the degree of surface roughness. The rougher the surface, the slower the scan to prevent the tip crashing into the surface.

6. Molecular herding, version 1: As the tip moves closer to the sample, some molecules that adhere only lightly to the substrate tend to scurry away. The idea would be to start a circular scan of the tip very close to the substrate. The circle's radius would gradually be reduced, herding any loose molecules into the center of the circle. Physically tapping the molecule laterally could also be done.

7. Molecular herding, version 2: Alternatively, using the method pioneered by Eigler, atoms can be made to attach themselves to the tip. The tip is then dragged to where you want the atom to be, and then the tip is retracted from the surface, leaving the atom where you want it to be. this will probably require UHV, though.

Performance Specifications

X, Y Range for Scanning: 140 nm
X,Y Range for Coarse Positioning: Unlimited
Z Range for Scanning: 140 nm

Electronics

Tunnel Current Range: Nominal 1-10nA. Desired 1pA-100nA continuously variable, with preset values 10 pA, 100 pA, 1 nA, 10 nA.

Bias Voltage Range: Required +2 to -2 Volts, selectable to the 100's of millivolts. Desired +10 to -10 Volts (for interaction capability).

Advantages/Disadvantages of a Two-Tube Design

Advantages

1. Ostensibly minimal vibration damping may be required.
2. High resolution claimed. No pictures of actual scans in original article, though.
3. Should be a compact, stiff design. High resonant freqs. Resonant frequency limits the scanning speed.
4. Relatively high speed scanning.
5. No moving parts. No stepper motors, 'inchworm' mechanisms, etc.
6. Good design for vacuum use, due to compact design.
7. No high voltage amplifiers needed. Max 18 volt circuits.

1. Tom Ekkens tried a two-tube concentric design based on the first reference cited and was not able to get the tip to move. Principle of inertial sliding not proved in this design.
2. Two piezotubes must be purchased, versus one for other designs.
3. Sample surface not visible for coarse tip positioning. Can't locate interesting features with an optical microscope and direct the tip there.
4. Sample surface not accessable for cleaning via electron/ion gun while on STM.
5. Limited range due to low-voltage circuits.
6. Electrochemical/Liquid use may be difficult or impossible. Entire microscope would have to be immersed. Liquid would have to be non-conductive and not interact with the nickle electrodes.
7. Large sample analysis not possible, unless the entire STM head could be inverted and placed ON the sample. In this case, the base would be made as small and light as possible, and the preamp would be connected to the STM head via light, thin wires.
8. Generic to all STMs: Conducting samples only.

Estimated Cost Analysis

• Piezoelectric tubes: $350
• Macor Baseplate: $50
• Printed Circuit Board: $200
• Electronic Circuitry Components: $500
• Miscellaneous: $300

STM Scanner

• Piezo Tubes
• Baseplate
• Tip Wire

Piezo Tubes

PZT 5A type piezoceramic is best. It is also used in most STMs.as it has a relatively temperature stable movement/volt compared to, say, PZT-5H which has more movement/volt combined with radical d_3,1 variation with small temperature changes.

Design

Tube thickness in general should be as thin as possible. This enhances performance by enhancing the piezoelectric effect and minimizing the mass which must be moved.

Electrodes should be removed from the portion of the tube(s) that are recessed into the baseplate, to avoid problems with that part of the tube expanding and contracting in the machined recesses.The disadvantage is that the performance of the tube will be a little less than if full length electrodes were used. The effective tube length is reduced.

Scanning Tube Specifications

It should be a PZT-5A type.

The electrodes should be nickel, and coat the exterior of the tube, sectioned longitudinally into quarters, and the interior of the tube, unsectioned. The lead wire positions for exterior and interior should be slightly up from the tube base to allow the tube to be recessed into the baseplate. The interior electrode should be grounded. Two adjacent outer electrodes should be connected to negative X and Y controlling voltages. The other two outer electrodes should be connected to positive X and Y controlling voltages. Opposite electrodes should have opposite, push-me/pull-you voltages.

Baseplate

Attachment to Preamp Board: The wires from the tubes to the preamp should be as close to immobile as possible. Movement can cause erroneous signals.

Tip Wire

With Pt/Ir, cut the tip at a fairly sharp angle (60 deg.) with any scissors. With W, you'll need special shears since W is very tough. It will ruin ordinary scissors. Approximately half of all tips created in this way will provide atomic resolution if the instrument is capable of it, and the sample is atomically flat, such as HOPG.

To image non-atomically flat surfaces, you will have to go to an electrochemically etched tip. Techniques are found in most of the STM books, and some of the articles mentioned. If you etch a tip in air, an oxide layer is created which then must be cleaned off somehow (ion gun, etc.). Of course, this is not a problem if the etching is done in a vacuum.

Etching Tungsten Tips

The procedure for etching W tips is very simple but it requires some skill. One can learn it by practice and careful observation. You can use a 2N NaOH solution for etching W tips, and +15V -15V DC power supply. Connect the positive supply to tip and negative to a length of nickel wire and dip it in the solution. You will have to cover the top portion of tip to restrict tip etching to the end of the wire. It requires nearly 20 minites to drop the top portion of W and to get required tip shape. There are many references on tip etching:

1) Allan J. Melmed J. Vac. Sci. Tech. B9 (2), Mar/Apr 1991
2) H. Bourque and R. M. Leblance Rev. Sci. Instrum. 66(3), March 1995

Printed Circuit Board

• General Specifications/Capabilities
  • Tunneling Voltage Range
  • Tunneling Current Range
• Component/Functions
• Preamp Board
  • Tip current preamp
    • Connection to Tip and Sample
    • Connection to Main Board
• Main Board
  • Tip altitude control feedback loop (Z)
  • Tip approach and removal circuits (Z)
  • Lateral sample movement circuits (X, Y)
    • Coarse Positioning (Inertial Sliding) Circuits
    • Scanning
    • Connection to Piezo Tubes
  • Power source
• Interaction Specifications
  • Manual Controls
    • Tip Current Polarity
    • Tip Bias Voltage
    • Tip Current Control
  • With AD Circuits
  • With DA Circuits
• Provisions for Future Automation

Digital-to-Analog/Analog-to-Digital (DAC/ADC) Board

• Requirements
• Commercial DA/AD Boards
  • Advantages/Disadvantages
  • Specifications
  • Connections to PCB Board
• PC Sound Boards
  • General
  • Advantages/Disadvantages
  • Specifications
  • Connections to PCB Board
• Signals
  • From Preamp
  • To X,Y Positioning Circuits
  • Connectors
  • Shielding

Sound Board DAQ Considerations

DC -vs- AC response: AC response only means the A/D circuit (actually the amp ahead of the A/D) ignores the steady-state input voltage and can only "see" changes that happen above 20Hz frequency or so. On the D/A side, no continuous output DC voltage only AC - audio frequency signals, changes. DC is a problem for audio circuits - causes thumps when you switch things on and off, saturates inputs, etc., so it is filtered out with capacitors (RC high-pass filters). This is called AC coupling.

The D-A part of the sound board's function would be to take a software command for an x-y position and translate that into control voltages.

Two 16bit DAC outputs (such as those provided by a good sound card) for the X and Y scan signals would be needed, as would a single 16bit ADC input for the Z (also: same sound card), if you want to drive tip position independently of the Z feedback circuit. This would allow us to generate X & Y scan signals in software, as well as any arbitrary positioning in X, Y, and Z. If it is too much load on the software to generate the (1KHz + ) scan waveforms, we can get the DSP chip on the sound card to do this, if it has one. Hand positioning of the tip by mouse or joystick should be possible with the proper software. The only limitation is the maximum input sample rate of the A/D on the sound card: probably 44KHz.

The scan waveforms could be a pair of wave tables that are output in the normal way, perhaps even a simple .wav file playback. Set up the input for record, and presto! scano! It is thus possible that we hardly need to write any low-level code at all, just use the drivers that come with the sound card or the functions in Microsoft's Multimedia Software Development Kit. See: Personal Engineering, April 1995, pp.40, "Sound Card Performance proves suitable for many professional DSP tasks"

The Turtle Beach Tahiti and Monterey cards can do simultaneous record and play.

44KHz sample rate is plenty for a low-cost STM. The 20KHz bandwidth limit is also plenty, in fact, it is a very Good Thing, given the sample rate, to "roll off" the frequency response below the "Nyquist Frequency" (1/2 the sample rate), to prevent "aliasing", which is just the appearance of a false low frequency noise signal by a high frequency noise because of a slow sampling rate. Many expensive data acquisition cards do not have this "input sampling filter" feature, as the sound cards do.

A good choice for a sound card might be a Orchid Soundwave 32. Not a great card in any way, but it has an Analog Devices ADSP-2115 DSP chip on board. This can be programmed to generate scan waveforms, sample the Z, and leave the result in the PC's memory, without much invlovement from the host PC itself. The card would have to be modified for DC frequency response. The card costs about $170. Noise is around 72 dB, so actual performance is comparable to a 12-bit laboratory card. 12-bit accuracy translates into about 0.25 angstrom margin of error for X, Y, and Z. Another possibility is the Turtle Beach Monterey. This card has signal to noise ratio of 90 dB, giving true 16-bit performance. It has built-in DC response so no modification is required. It uses the Motorola 56001 DSP. It has a retail price of $399, street price $320. The Tahiti is the same card without the DSP, which may not be needed, for about $240 street price. Another good bet is the Mediatrix Audiotrix Pro (about $240 also).

Preliminary STM Hardware Driver Interface

Also, a good interface spec is needed between the user interface & control & display code on the one side and the hardware drivers & high speed code on the other. That is, what functions need to be handled by the device-specific code? Here's a first stab at it:

Cheap STM hardware driver interface spec version 1.0, 4-7-95

Proposed Functions:

1. Digital XY scan waveform generation, with adjustable rates & amplitudes, offsets, etc.

2. High speed input sampling, say one or two channels at at least 44 KHz, with the resulting data placed off in a memory block for the high-level code to process. At least two full scanline buffers' worth of ram to allow the scanning to continue while the last is being displayed.

Control and Display Software

• User Interface
  • Controls and settable parameters
  • Graphical Interface (if any): Visual Basic?
• Platforms-Considerations
  • IBM: DOS
  • IBM: Windows
  • Macintosh
  • Unix
• Interface to separate DA/AD Board, or on-board DACs/ADCs: General Specifications
• Control Functions
  • Scan variables
    • axis of scan
    • speed
    • scan area
    • scan pattern
    • Use the Look Up Table (LUT) to generate correct X, Y scan voltages for each combination of scan area and speed.
    • Designate which leg to acquire data on: X+ or X-.
  • Data Acquisition Functions
    • Required
      • Record measured Z driver voltage
    • Optional
      • Record tip voltage as it is varied for I-V spectroscopy
      • Record tip current as tip voltage is varied for I-V spectroscopy
      • Record tip voltage AND Z driver voltage as tip height is varied for S-V spectroscopy
  • Image Display and Analysis
    • Realtime
      • Display Image as it's acquired? At least showing each line as it's being acquired would be useful.
    • Post-Acquisition Image Analysis and Display
      • Manual control
      • Scan variables
      • Sense options
      • Analysis Functions
      • Export options
      • Other Options
  • Sample Interactions
    • Nanolithography
    • Molecule/atom rearrangement

  ---

  STM Software: Hardware Driver

  (A loose C notation is used.)
  • initializeSTM();

    This is a mandatory procedure that has to be executed before any other driver procedure. Its content depends solely on the programmer of the driver. E.g. it can be empty.

  • setTipPosition(xPos, yPos, speed);

    This procedure moves the tip to the new x, y position. However, in order to keep things general, the coordinates have to be in some general units (if they were in say, Angstroms, the piezocrystal parameters would have to be passed to the driver, and what's worse, the look-up tables (LUTs) would have to be communicated to the driver too. I will go into more detail on the LUTs later). I think a 16-bit signed integer for both xPos and yPos would be sufficient. The integers are proportional to the voltages applied to the x- and y-piezos with -16384 and 16383 corresponding to the min and the max voltages. As for the speed, this parameter would tell the procedure at what speed the tip should be moving (in units of integers per sec). The reason for this is that any sudden changes of voltages, etc. can result in crushing the tip.

  • getTipPosition(*xPos, *yPos);

    Returns tip's x and y. This is just for completeness/elegance. Some STM hardware can actually have position sensors, or if the driver was treated as an object in C++... It might be a good practice to use this procedure whenever the STM program needs the tip position info.

  • getTipAltitude(*altitude);

    Returns present value of the z-piezo. Again, altitude is an integer either 2 or 4 bytes long, proportional to the voltage applied to z-piezo. I am not sure about this variable's length. We use 16 bits but somebody may wish to use a longer variable.

  • setTunnelCurrent(tunnelCurrent, ampPerSec);

    Allows to set the tunnel current. ampPerSec tells the driver at what rate to change tunnelCurrent. This proc clearly makes sense only if the the STM electronics is capable of this function. Probably optional.

  • setTunnelVoltage(tunnelVoltage, voltPerSec);

    Same thing for tunnel voltage.

  • setFeedback(status);

    Status can be either On, or Off (1 or 0). For some measurements (I-V curves, constant altitude scans) the sample-tip distance has to be kept constant. The electronics would, of course, have to have this built in. Optional.

  • getTunnelCurrent(*tunnelCurrent);
    getTunnelVoltage(*tunnelVoltage);
    getFeedbackStatus(*status);

    Again, these are just for completeness.

  • coarseSampleStep(dir);

    This procedure allows coarse movement of the sample in one of the four lateral directions (e.g. dir= 0 -> X+, 1 -> Y+, 2 -> X-, 3 -> Y-), or any specified direction.

  • coarseTipStep(upOrDown);

    One step up or down (for systems with a separate approach/retract mechanism). Not required for the primary design here, in which tip approach is automatic.

  • scanLine(nOfSamples, samplesPerSec, *XValues, *YValues, *altitudeSamples);

    This procedure is the workhorse of the whole program. It takes the number of samples to be taken in the linescan, pointers to x and y arrays of values (containing nOfSamples items each) which will be sent to x- and y-piezos, respectively, at the rate of samplesPerSec. The proc returns a pointer to an array containing scanned z-piezo values. The reason for setting up this procedure this way is twofold: to keep it compatible with both the waveform and the simpler DAC cards and to allow employing LUTs. LUTs allow one to deal with non-linearities and time delays of the STM hardware through software (i.e. to improve performance at no additional hardware costs- particularly useful for a homebrew STM).

    The first time the STM program would be ran the LUTs can be empty (i.e. have only linear sawteeth stored in them). When scanning a surface of known geometry later the LUTs can recalculated (calibrated) so the tip actually scans through a RECTANGULAR grid (which it necessarily does not have to do if just rectangular grid points are sent to x- and y-piezos!).

    The only disadvantage of this is that there would have to be different LUTs (and therefore different calibrations) for different scanned area sizes and scanning speeds. But that can be taken care of through software.

  • measureIVCurve(nOfSamples, samplesPerSec, *voltageValues, *CurrentSamples);

    The idea here is similar to the previous proc. voltageValues is a pointer to an array of nOfSamples sample-tip voltages that are applied at the rate of samplesPerSec. A pointer to an array of the measured currents is returned.

  ---

  Some comments:
  1. Obviously a system of error messages and error handling has to be worked out.
  2. The rates can be in time-per-one-sample units-- whatever is more convenient for programmers. We just have to agree on something.
  3. There could be a different partition of the actual drivers functions to software procs (e.g. coarseSampleStep(dir) can be written as four separate procs with no parameters etc.). Again, comment (2) applies here.
  4. If the procedures are agreed on in some form the proper C syntax has to be worked out (Like what's passed as a value and what as a pointer, types of variables etc.)

  ---

  Vacuum Systems

  General Considerations: The smaller the STM head, the better, to conserve valuable vacuum space.

  STM Head Design Considerations

  Outgasing Minimization: Special low-outgas epoxies must be used (see Ted Pella Inc., suppliers section). All wire must be bare or have low-outgas insulators.

  Bakeout Considerations

  Vacuum chamber interior surfaces and samples are often cleaned of adsorbed atoms by baking the chamber at high temperature. All STM material must be bakeout capable if this is to be used.

  Vacuum Sample Handlers

  Design: Need access to the head to change samples by vacuum actuators/mechanical feed-throughs. Some sort of standard sample holder or mounting would be useful.

  Use: Often there is a need to clean samples before imaging, as with an electron gun, etc. So handlers must permit moving the sample from cleaning station to the STM. Better would be access to the sample surface while on the STM for cleaning.

  Tip Replacement Mechanisms

  Design:Need access to the head to change tips by vacuum actuators/mechanical feed-throughs. It would be nice if you could change the tip without moving the sample.

  Vibration Damping A vacuum chamber is bolted to the floor, or at least rests on it without damping, so the floor transmits vibration quite efficiently to the chamber. Space in the chamber is at a premium, so there isn't much room for vibration damping. This makes a small, stiff scanner head very desirable, and the design of vibration damping apparatus for the vacuum chamber challenging.

• Pumps
  • Vibration control
  • Sources
  • Used
  • New
• Chambers
  • Sources
  • Converted?

Cryogenic Systems

• General Considerations: It would be nice if our instrument was submergable into a liquid nitrogen or helium dewar to do quick scans at low temperatures. Alternatively, it could be designed to be installed into a cryo vacuum system. Cryo is pretty advanced stuff, especially in a vacuum, where you don't have atmosphere to conduct heat. There are numerous non-intuitive design considerations. A vacuum-cryo STM is considered the most advanced and hardest to make.

Samples

• Possible Sample Materials
  • Air: Highly oriented Pyrolytic Graphite (HOPG) is the standard one to see if you have atomic resolution. See suppliers list. Numerous other conducting samples. Oxide is always a problem.
  • Vacuum: Anything conducting that you can get a clean surface.
• Ambient Sample handlers: A standard sample mount would be convenient and allow quicker sample changing. Samples could be mounted separately from the STM.
• Sample preparation: Air. The surface of HOPG does become oxidised after a day or so. To get an oxide-free surface, peel off one layer of graphite from the sample by sticking scotch tape to the sample surface, and pulling gently away. This surface will be oxide-free for many hours.

References

1. "A miniaturized scanning tunneling microscope with large operation range," Rev. Sci. Instrum. 64 (3), March 1993, 692-693. Two-tube design.
2. Grafstrom... J. Vac. Sci. Technol. A8(1) (1990) 357.
3. Altman... Rev. Sci. Inst. 64(5) (1993) 1239.
4. Besenbacher... Rev. Sci. Instr. 59(7) (1088) 1035.
5. Park.. Rev. Sci. Inst. 58(11) (1987) 2010.
6. "Student Scanning Tunneling Microscope" by Lewis et al., Am. J. Phys, 59(1), 38-42. (1991)
7. "Some Design Criteria in Scanning Tunneling Microscopy", by Pohl, IBM J. Res. Dev., 30(4), 417-427, (1986)
8. "Scanning Tunneling Microscopy" by Binnig & Rohrer (these guys invented the STM) IBM J. Res. Dev., 30(4), 355-369, (1986)
9. "Single-Tube Three-Dimensional Scanner for Scanning Tunneling Microscopy", by Binnig & Smith, Rev. Sci. Instrum. 57(8), 1688-1689, (1986)
10. "A Scanning Tunneling Microscope for Undergraduate Laboratories", by Sears et al., Computers in Physics Jul/Aug, 427-430, (1990)
11. "Very Sharp Platinum Tips for Scanning Tunneling Microscopy", by Libioulle, Rev. Sci. Instrum. 66(1), January 1995, 97-99. Describes how to make Êvery high-aspect ratio tips for very precise imaging.
12. "Construction of a Dual-Tip STM: a Prototype Nanotechnology Workstation", by Dr. Mark Voelker. Phd Thesis, 218 pages. Copies available for $40 from Mark Voelker, 1045 Hemolele Place, Honolulu, HI 96817. Recommended.

1. "Introduction to Scanning Tunneling Microscopy", by Julian Chen, Oxford University Press, 1993. Recommended.
2. "Scanning Tunneling Microscopy and Related Methods", by Behm, Kluwer Academic Publishers, 1990
3. "STM and SFM in Biology", by Marti, Academic Press Inc. (Harcourt Brace), 1993
4. "Scanning Tunneling Microscopy", by Stroscio, Academic Press Inc., 1993. Highly recommended. Great chapter on designing, constructing, testing and debugging an STM.

1. "Scanning Tunneling Microscope Data Acquisition and Control in Visual Basic", by Porter, Rev. Sci. Instrum. 64 (12), Dec 1993, 3530-3533
2. "Correction for non-linear behavior of piezoelectric tube scanners used in scanning tunneling and atomic force microscopy" by Akila, Rev. Sci. Instrum. 66(3), March, 1995, 2517-2519. Describes a correction method for X, Y driver voltages to make the scan linear, using a formula rather than a look-up table (LUT).
3. "Scanning Tunneling Microscope Having Proper Servo Control Function", U.S. Patent #5,059,793. Lots of info on general STM control and display with D/A and A/D boards.
4. "Scanning Tunneling Microscope", by Park, Rev. Sci. Instrum. 58 (11), November 1987. Lots of detailed circuit diagrams.
5. "Data acquisition and control system for molecule and atom-resolved tunneling spectroscopy", by Altman, Rev. Sci. Instrum. 64 (5), May 1993, 1239-1243. Details high-speed sample-and-hold circuits for spectroscopy measurements.
6. A compact scanning tunneling microscopy control and data acquisition system based on a Macintosh II workstation", by Grafstrom, J. Vac. Sci. Technol. A8 (1), Jan/Feb 1990.

1. "Sound card performance proves suitable for many professional DSP tasks", Personal Engineering, April 1995, 40-44
2. "Do On-the-fly Signal Processing with your Sound Blaster", Computer Applications Journal, December, 1995, pp 30-38. This article shows how to access a soundcard various functions (especially DSP) from C. A DLL could probably be developed from the sample code provided.

1. "Positioning single atoms with a scanning tunneling microscope," Nature, vol. 344, April 5, 1990, 524-526.
2. "Process for Repositioning Atoms on a Surface Using a Scanning Tunnleing Microscope", Don Eigler, U.S. Patents #4,987,312; Jan 22, 1991, and #5,144,148; Sept 1, 1992
3. "Scanning Tunneling Microscope Nanoetching Method" U.S. Patent #4,896,044
4. "Surface characterization of materials at ambient conditions by STM and AFM", by Magonov, Applied Spectroscopy Reviews, 28(1&2), 1993, 1-121.

Supply Sources

Ted Pella, Inc.
4595 Mountain Lakes Blvd.
Redding, CA 96003
916-243-2200
800-237-3526
Very good free 300 page catalog of SPM and SEM supplies. Conducting glues, tweezers, calibration grids, dessicant, etc.

McAllister Technical Services
800-445-3688
208-772-9527
Makes an UHV STM. Does contract vacuum instrument/STM engineering. Sells numerous vacuum components and instruments.

Huntington Laboratories
1040 L'Avenida
Mtn View, CA 94043
(800) 227-8059
(415) 964-3323
(415) 964-6153 fax
Vacuum Components- Chambers, positioners, connectors- High-end.

Advanced Ceramics Corp.
216-529-3900
800-822-4322
Highly Oriented Pyrolytic Graphite (HOPG) sample, ZYH grade (good enough): $43

Molecular Imaging
1208 E. Broadway, #110
Tempe, AZ 85282
602-894-1653
Cleaved mica with gold evaporated on it, plus Digital Instruments-cloned SPMs, specializing in environmental chambers and picoamp imaging.

MWS Wire Industries
Westlake Village, CA
800-423-5097
Tungsten wire, .010" dia., 100 ft., $85

California Fine Wire
Grover Beach, CA 93433
805-489-5144
Pt/Ir wire, 80/20, .010" dia., very expensive: $75/ft. Anyone know of a cheaper supplier?

Wesgo Inc.
477 Harbor Blvd
Belmont, CA 94002
415-592-9440
415-598-3259 (fax)
Pt/Ir wire, 80/20, .010" dia. Minimum order one troy ounce, approx. $1000.

Sigmund Cohn Co.
121 S. Columbus Ave.
Mount Vernon, NY 10553
914-664-5300
Tip wire.

Note: One tip I'd like to offer is that most of these suppliers make one or two extra tubes for each order they get, in case of breakage. If you call them up and tell them that you aren't particular about exact dimensions or material, you may be able to get one of their extras for a nominal price, hopefully without having to adhere to their minimum order. Especially if you explain that you are a poor student (if you happen to be one).

For example, PZT-5H instead of 5A, or 1" long instead of 1.25" long, or 0.025" thick wall instead of 0.018", or "as cast" quality vs. fully machined. I'd suggest assembling a range of parameters within which you could accept a tube, and call all the manufacturers. Probably the general electrode configuration would be the only hard and fast requirement.

An annotated drawing of your tube would be most helpful in getting quotes.

Also, each manufacturer has useful technical material included with their ordering literature. I suggest ordering a catalog from each, and explaining what your application is and requesting any relevant tech/application notes they may have available.

Morgan Matroc, Inc.
232 Forbes Road
Bedford Ohio 44146
216-232-8600 voice
216-232-8731 fax
Piezoelectric ceramic tubes.

Staveley Sensors, Inc.
91 Prestige Park Circle 203-291-2524 voice
203-289-3189 fax
Piezoelectric ceramic tubes. Minimum order five tubes. Expensive, (around $200/tube) but high quality.

Sensor Technology Ltd
BM Hi-Tech Division
PO Box 97, 20 Stewart Road
Collingwood
Ontario, Canada L9Y3Z4
705-444-1440 voice
705-444-6787 fax
Piezoelectric ceramic tubes.
Low prices on tubes with their standard dimension/electrode configurations. Minimum two tubes per design. Ex.- $65 each for .two 5" dia x 1" long, electrodes fired silver, quartered outside, solid inner, BM-527 material (a modified PZT-5A, apparently, with better d_3,1).
Very useful tech notes.

Scanning Probe Microscope Makers

Digital Instruments
800-873-9750

LK Technologies, Inc.
812-332-4449

Omicron Associates
412-831-2262

Park Scientific Instruments
800-776-1602

Quesant Instruments
818-597-0311

Topac Inc.
617-740-8778

Topometrix
408-982-9700

RHK Technology, Inc.
810-656-3116
810-656-8347 fax

McAllister Technical Services
800-445-3688
208-772-9527

Molecular Imaging
602-894-1653

STM Technology-Related Links

Digital Instruments SPM Mailing List
To subscribe, click on this and include the word "help" in the message body.

LinkSTM.html

spm_www.htm

spm_www.htm

Collection of atomic-scale phenomena.

Molecular Manufacturing Enterprises Inc.
Nanotechnology Links