NANO LETTERS
Imaging of Localized Electronic States at a Nonconducting Surface by Single-Electron Tunneling Force Microscopy
2006 Vol. 6, No. 11 2577-2580
Ezra B. Bussmann, Ning Zheng, and Clayton C. Williams* Physics Department, UniVersity of Utah, 115 S 1400 E RM 201, Salt Lake City, Utah 84112 Received August 25, 2006; Revised Manuscript Received September 19, 2006
ABSTRACT Localized electronic states near a nonconducting SiO2 surface are imaged on a ∼1 nm scale by single-electron tunneling between the states and a scanning probe tip. Each tunneling electron is detected by electrostatic force. The images represent the number of tunneling electrons at each spatial location. The spatial resolution of the single electron tunneling force microscope is determined by quantum mechanical tunneling, providing new atomic-scale access to electronic states in dielectric surfaces and nonconducting nanostructures.
Localized electronic states in dielectric materials are of broad scientific interest and are technologically important. In metaloxide-semiconductor technology, such states in the gate dielectric decrease carrier mobility and increase leakage current.1 Future technologies may utilize single-electronic states as bits for information processing or memory.2,3 Techniques to extract state-specific electronic information with atomic spatial resolution will be important in developing these technologies. In this letter, a scanning probe technique called single-electron tunneling force microscopy (SETFM) is presented. The SETFM captures images of localized states near the surface of completely nonconducting films by counting single electrons tunneling between the probe and each state. Regions where a tunneling event occurs register as a “1”, whereas regions where no tunneling occurs register as a “0” in the images. Near-surface states in a SiO2 film are imaged on a ∼1 nm spatial scale. The spatial resolution is provided by the strong dependence of the tunneling rate on the vacuum gap, as with the scanning tunneling microscope (STM). The SETFM extends the high-resolution imaging capabilities of the STM to localized states in completely nonconducting surfaces. Although the STM has provided unprecedented atomicscale investigation of many different materials, its application to dielectrics is limited.4,5 The STM cannot image surfaces with insufficient conductivity to provide the necessary imaging current (typically ∼0.1 picoampere, i.e., ∼106 e/second).6 Under special conditions, the STM can be applied to dielectrics.7-12 Provided that the dielectric is thin enough, * Corresponding author. E-mail:
[email protected]. 10.1021/nl0620076 CCC: $33.50 Published on Web 10/19/2006
© 2006 American Chemical Society
a current may tunnel through it, either via localized states or directly to the substrate. Localized charge traps in the dielectric, charged by a single-electron, can modify the tunneling current.7,8 This effect has also been measured by techniques such as ballistic electron emission microscopy (BEEM) and tunneling atomic force microscopy (TUNA).9,10 Attempts to use an alternating current (ac) STM (ACSTM) have also been made to image dielectrics at the nanometer scale.11,12 A variety of electrostatic force and potential sensing techniques allow for some electronic materials’ characterization without the need for a tunneling current. Localized single-electron charge may be imaged by techniques such as scanning single-electron transistor microscopy or electrostatic force microscopy. However, these techniques are sensitive only to the charge present at the surface, not to the electronic state itself, so uncharged states cannot be observed. Also, the spatial resolution is not determined by tunneling, and is typically limited by the size of the probe apex (10+ nm).13-25 In order to gain atomic-scale insight into localized electronic states (independent of the initial charge) in nonconducting surfaces, Klein and Williams demonstrated a technique for measuring single-electron tunneling between a scanning probe tip and the surface by force-detection (amplitude detection EFM).18,19 More recently, frequency detection EFM has been employed to measure single-electron tunneling, electron manipulation between probe and state, and electronic spectroscopy.20-22 Similar approaches also perform force-detected spectroscopy and imaging by single-
Figure 1. Schematic of the SETFM apparatus. Single-electron tunneling is measured between the apex of the metallized probe and individual localized states at the nonconducting surface. Tunneling is detected by frequency detection surface potentiometry (SP) measurements.
electron tunneling, but the tunneling occurs between states within the sample.23-25 These techniques, applicable only to surfaces where tunneling between states in the sample is possible, do not take advantage of the high spatial resolution afforded by probe-sample tunneling. In the work presented here, single-electron tunneling between the probe and the sample is used to acquire images of localized electronic states near the surface of a nonconducting film. Single-electron tunneling events are detected between a metallized cantilever probe (Mikromasch NSC15 Ti-Pt) and the surface of silicon dioxide thin films on silicon substrates. Figure 1 shows the experimental apparatus. The samples are 10- and 20-nm-thick thermal silicon dioxide (SiO2) thin films on Si-substrates. Prior to measurements, these samples are heated to 500-600 °C for 1/2 hour to desorb surface contamination. Sample preparation and measurements are performed in a commercial ultrahigh vacuum atomic force microscope (Omicron, Multiprobe S) under 10-8 Torr vacuum at room temperature. In single-electron tunneling force microscopy, the cantilever probe acts as an electrometer to detect electrostatic forces or potentials. Single-electron charge, injected or extracted from the surface, is sensed by the force it exerts on an oscillating cantilever. The cantilever oscillation is driven at a fixed (adjustable) amplitude, a (10-40 nm), at the cantilever’s resonance frequency ( f ≈ 300 kHz) by an external self-oscillating circuit. During the oscillation cycle, the probe approaches the sample surface to a minimum vacuum gap, zm, which is controlled on a sub-angstrom scale. When the probe tip is near the surface, the electrostatic force gradient acting on it causes a shift, ∆f, in the cantilever’s resonance frequency. Any change in the charge on the sample surface under the probe apex changes the electrostatic force gradient, leading to a resonance frequency shift. As shown in previous work, single-electron tunneling events can be detected with a dc voltage, Vdc, applied to the Si-substrate with respect to the (grounded) probe.18-22 Each tunneling electron causes an abrupt step in ∆f.20 In the work presented here, single-electron charge, injected or extracted from the surface by tunneling, is sensed by frequency detection surface potentiometry (SP) measurement. 2578
Figure 2. Single-electron counting by surface potentiometry (SP) measurements. The probe is positioned within tunneling range (zm ≈ 1 nm) of the surface of a 20-nm-thick SiO2 thin film. The SP signal is recorded vs time. During the shaded time intervals, the applied ac voltage is interrupted, and a dc voltage (indicated in the panel) is applied. After the dc voltage is switched off, the SP signal is recorded again. Changes in the value of the SP signal after the application of a dc voltage indicate that charge has tunneled between the probe and sample. Here, an increasing number of individual electrons is extracted from the sample as the magnitude of the voltage is incrementally increased. After application of -3 V and -4 V, electrons are observed to tunnel back into the probe, causing the SP signal to decay back to its initial value.
In SP measurements, an ac voltage, Vac, at a frequency, fac, is applied between the probe and sample. This ac voltage produces a modulation of ∆f at the frequency fac. This frequency modulation is detected with an FM detector and is sent to a lock-in amplifier referenced at fac. The output of the lock-in amplifier (the magnitude of the frequency modulation) is called the surface potentiometry (SP) signal. The SP signal is proportional to the average potential difference between the probe and the sample surface, which depends on the amount of charge below the probe apex. Each single-electron change in the surface charge produces a step in the SP signal. Surface potentiometry accurately counts individual electrons injected into or extracted from the sample surface. Figure 2 a shows a potentiometric measurement of singleelectron tunneling to and from a 10-nm-thick SiO2 film. In this measurement, the probe has been placed at zm ≈ 1 nm. Here a Vac of 2 V amplitude (square wave, 300 Hz) is applied to the sample. Single-electron tunneling from the sample to the probe is induced by application of a dc voltage pulse (indicated in the shaded time periods). The ac voltage is interrupted while the dc voltage is applied. During the time interval between each dc voltage pulse, the SP signal is recorded. The SP signal registers changes in the charge below the probe apex. Each single-electron extracted from the sample by application of a dc voltage is subsequently registered as an incremental shift in the surface potentiometry signal. After dc pulses of -3 V and -4 V, the potentiometry signal decays back toward zero in a step-like fashion, indicating that individual electrons have tunneled back into the surface after several seconds. This is possible because the applied dc voltages that allowed the electrons to tunnel out of the sample have been turned-off. With successively Nano Lett., Vol. 6, No. 11, 2006
Figure 3. The algorithm executed at each point on a 2D grid over the surface to produce an SETFM image. The probe is initially positioned at zm ) 4.0 ( 0.5 nm gap. The SP measurement (ac voltage) is interrupted. Then a dc voltage (Vdc > 0) is applied, and the probe is moved within tunneling range (zm ≈ 1 nm) of the surface, then returned to the starting position. This operation, called a write, allows electrons (q) to tunnel between the probe and sample. The SP signal is then read (read 1). Then the SP measurement is again interrupted, a dc voltage (Vdc < 0) is applied, and the probe is moved within tunneling range of the surface, then returned to the starting position. This operation is meant to extract any electrons injected in the write, so as to avoid net charging of the surface. The SP signal is then read again (read 2). By subtracting the surface potential in read 1 from read 2, the number of tunneling electrons is counted.
larger pulses, an increasing number of electrons are extracted from the sample. In SETFM imaging, an attempt is made to inject and extract electrons at each point in the image (on a regular 2D grid over the surface). The number of tunneling electrons that are counted at each location provides the image contrast. Figure 3 a shows the algorithm used to count tunneling electrons. The probe is initially positioned outside the tunneling range (zm ≈ 3-5 nm). A dc voltage is applied, and the probe is rapidly moved to within zm ≈ 1 nm of the surface, then retracted to the starting position. This measurement, called a write, brings the voltage-biased probe within tunneling range of states in the surface. In previous work, it was shown that this measurement induces tunneling to or from near-surface localized states; provided that there is a state that can be accessed under the given applied voltage. After the probe is retracted to the starting position, the dc voltage is turned off, and the surface potential is read by a SP measurement. Subsequently, a dc voltage, of the opposite polarity to that in the first write step, is applied, and the probe is rapidly moved to within zm ≈ 1 nm of the surface, then retracted to the starting position. This procedure, called an unwrite, is intended to extract the charge deposited in the write step. A combination of the write and unwrite procedure is found to repeatably inject and extract single electrons from states in an SiO2 surface.21 Finally, the surface potential is read again. Any change in the surface charge, due to tunneling induced by the unwrite step, leads to differing values of the potentiometry signal in the first and second reads. The difference between the first and second potentiometry reads is referred to as the surface potentiometry difference signal. The unwrite step is also important to avoid charge buildup on the surface during imaging. Nano Lett., Vol. 6, No. 11, 2006
Figure 4. (a) Repeated (5 × 5 nm2) SETFM images of the same region of 20-nm-thick SiO2. (b) A histogram of the surface potentiometry signal from the two images in panel a. (c) Repeated linetraces made by scanning the probe across a single line in the SiO2 surface. Two electronic states appear in the image at ∼3 and 6 nm lateral position. (d) A linetrace from panel c at the position indicated by the white dotted line. The two regions of tunneling are clearly resolved with spatial resolution of ∼1-2 nm.
Single-electron tunneling force microscopy images are acquired by performing the write-read-unwrite-read sequence in Figure 3a at each point on a regular square grid over the surface. Each pixel in the image is the value of the SP difference signal at that gridpoint. Figure 4 shows typical repeated images of a region of a 20-nm-thick SiO2 surface. The write/unwrite procedures of ∼0.2 s duration have been performed at +6 V/-6 V dc, respectively. The SP read procedures of ∼0.2 s duration are performed at zm ≈ 4.0 ( 0.5 nm. The potentiometry signal is lowpass filtered with a 30 ms time constant. Localized regions of tunneling appear across the image, separated by regions where no tunneling is observed. These localized regions of tunneling are due to localized electronic states near the surface. Although these images have been acquired under identical conditions, the exact pattern of the tunneling varies somewhat between images. This imperfect repeatability suggests that charge may be moving between states near the surface. A histogram of the difference signal from all points in the images supports the claim that the contrast is caused by single-electron tunneling. Figure 4 shows a histogram of the difference signal for both images. The difference signal is quantized. The bulk of the image shows a zero difference 2579
signal (dark blue). The additional peaks correspond to single (green) and double (red) electron charging of states under the probe. The spatial resolution and the number of events observed in these images is a strong function of the vacuum tunneling gap, zm. No tunneling is observed unless the probe is within zm < 2 nm of the surface. Because of the order-of-magnitude per Angstrom change in tunneling probability with probesample gap, sub-Å gap control is necessary for repeatable imaging. These measurements are performed without an automatic feedback circuit to control the probe-sample gap. Instead, measurements are made under conditions where variation of the probe-sample gap due to thermal expansion and drift of the piezoelectric positioner is less than (1 Å/image). The read measurements are performed at a typical vacuum gap, zm ) 4.0 ( 0.5 nm, well outside the maximum experimentally observed tunneling range (zm < 2 nm of the surface). The write brings the probe to zm ≈ 1.0 nm. There is no contact between the probe and sample during the imaging. The gap is set to the same value with sub-angstrom accuracy at the beginning of each image. Figure 4 shows an image acquired by repeatedly scanning the probe across the same region (line) of a SiO2 surface (movement in the vertical direction is disabled). Two regions of tunneling are repeatably observed along this line. These two areas of tunneling are associated with two localized states spatially separated by ∼3 nm. These states are clearly resolved, with a 1.5 nm region between them where a zerosignal is measured, demonstrating ∼1-2 nm spatial resolution. It is likely that with further work an even higher spatial resolution will be achieved (comparable to STM) because the electron transfer occurs by tunneling. In summary, the single electron tunneling force microscope, a technique for imaging localized electronic states at nonconducting surfaces, has been presented. Images of individual electronic states near a SiO2 surface have been acquired with 1-2 nm spatial resolution. Like the STM, the imaging mechanism is electron tunneling, with each electron individually detected by electrostatic force. Future nano-
2580
technologies will require atomic-scale understanding of the electronic properties of individual electronic states in dielectrics. It is expected that this approach will provide unprecedented access to these electronic states with nanometer and possibly subnanometer spatial resolution. References (1) Nicollian, E. H.; Brews, J. R. MOS (Metal Oxide Semiconductor) Physics and Technology; John Wiley & Sons: New York, 1982. (2) M. Xiao; et al. Nature 2004, 430, 435. (3) Guo, L.; Leobandung, E.; Chou, S. Science 1997, 275, 649. (4) Binnig, G.; Rohrer, H.; Gerber, C. H.; Weibel, E. Phys. ReV. Lett. 1982, 49, 57. (5) Chen, C. J. Introduction to Scanning Tunneling Microscopy; Oxford University Press: New York, 1993. (6) Carla`, M.; Lanzi, L.; Pallecchi, E.; Aloisi, G. ReV. Sci. Instrum. 2004, 75, 497. (7) Welland, M. E.; Koch, R. H. Appl. Phys. Lett. 1986, 48, 724. (8) Repp, J.; Meyer, G.; Olsson, F. E.; Persson, M. Science 2004, 305, 493. (9) Kaczer, B.; Pelz, J. P. J. Vac. Sci. Technol., B 1996, 14, 2864. (10) O’Shea, S. J., Atta, R. M.; Murrell, M. P.; Welland, M. E. J. Vac. Sci. Technol., B 1995, 13, 1945. (11) Kochanski, G. P. Phys. ReV. Lett. 1989, 62, 2285. (12) Stranick, S. J.; Weiss, P. S. J. Phys. Chem. 1994, 98, 1762. (13) Yoo, M. J.; Fulton, T. A.; Hess, H. F.; Willett, R. L.; Dunkleberger, L. N.; Chichester, R. J.; Pfeiffer, L. N.; West, K. W. Science 1997, 276, 579. (14) Martin, Y.; Abraham, D. W.; Wickramasinghe, H. K. Appl. Phys. Lett. 1988, 52, 1103. (15) Terris, B. D.; Stern, J. E.; Rugar, D.; Mamin, H. J. Phys. ReV. Lett. 1989, 63, 2669. (16) Scho¨nenberger, C.; Alvarado, S. F. Phys. ReV. Lett. 1990, 65, 3162. (17) Ludeke, R.; Cartier, E. Appl. Phys. Lett. 2001, 78, 3998. (18) Klein, L. J.; Williams, C. C. Appl. Phys. Lett. 2001, 79, 1828. (19) Klein, L. J.; Williams, C. C. Appl. Phys. Lett. 2002, 81, 4589. (20) Bussmann, E.; Kim, D. J.; Williams, C. C. Appl. Phys. Lett. 2004, 85, 2538. (21) Bussmann, E.; Zheng, N.; Williams, C. C. Appl. Phys. Lett. 2005, 86, 163109. (22) Bussmann, E.; Williams, C. C. Appl. Phys. Lett. 2006, 88, 263108. (23) Woodside, M. T.; McEuen, P. L. Science 2002, 296, 1098. (24) Stomp, R.; Miyahara, Y.; Schaer, S.; Sun, Q.; Guo, H.; Grutter, P.; Studenikin, S.; Poole, P.; Sachrajda, A. Phys. ReV. Lett. 2005, 94, 056802. (25) Dana, A.; Yamamoto, Y. Nanotechnology 2005, 16, S125.
NL0620076
Nano Lett., Vol. 6, No. 11, 2006