Atomic Dipole Imaging
NC-SNDM Study on a Hydrogen-Terminated Si Surface
- Fig. 1: Schematic diagram of NC-SNDM . Reprinted with permission adopted from figure 1, J. Appl. Phys. 113, 014307 (2013). Copyright 2013 American Institute of Physics.
- Fig. 2: NC-SNDM images of a partly hydrogen-terminated Si(111)-(7x7) surface . (a) Topographic image (b) ε3-image (dipole image) (c) and (d) Profiles along the lines A to C in (a) and (b), respectively. Reprinted with permission from figure 3, Appl. Phys. Lett. 103, 101601(2013). Copyright 2013 American Institute of Physics.
- Fig. 3: Charge distribution around (a) bare Si adatom (b) hydrogen-terminated adatom . Large yellow and small green balls denote Si and H atoms, respectively. Blue ellipses indicate the electron distribution around the adatom. The tones in the ellipses indicates charge density. The ellipse bounded by a dotted line in (a) is a weak bond to a Si atom just beneath the adatom . Reprinted with permission adopted from figure 5, Appl. Phys. Lett. 103, 101601 (2013). Copyright 2013 American Institute of Physics.
- Equation 1
- Equation 2
- Equation 3
- Equation 4
Scanning nonlinear dielectric microscopy (SNDM) has been established as an imaging tool for spontaneous polarization or permanent electric dipoles on the nanoscale. In particular, non-contact SNDM (NC-SNDM) can image atomic dipoles on cleaned semiconductor surfaces. In this article, we present NC-SNDM images of a partly hydrogen-terminated Si(111)-(7×7) surface. Our result demonstrates that NC-SNDM is a useful tool for the investigation of surface dipoles in the atomic scale.
Scanning nonlinear dielectric microscopy (SNDM)  is strikingly distinct from other scanning probe microscopy (SPM) methods in its ability to directly image spontaneous polarization or permanent electric dipoles on material surfaces [1-3]. This unique capability is brought about by using a gigahertz range LC self-oscillator coupled with a conductive sharp tip for sensing tip-to-sample capacitance (Cts) . Capacitance variation (∆Cts) as tiny as 10-22 F causes a significant shift in its oscillation frequency (∆ƒ). This extremely high sensitivity allows various applications such as high-resolution imaging of ferroelectric domains , stored charges in flash memories , and dopant profiles in semiconductor devices . In addition, SNDM can be operated in a non-contact mode (NC-SNDM) . NC-SNDM can atomically resolve dipoles on surfaces such as Si(111)-(7×7) [3,8] and Si(100)-(2×1) .
Nanoscale imaging of spontaneous polarization is useful for research on electronic materials as well as ferroelectric ones. Spontaneous polarization formed on surfaces and interfaces has attracted increasing attention, because it can affect carrier mobility , barrier height , and workfunction [12,13] of electronic materials. NC-SNDM is selectively sensitive to spontaneous polarization or permanent dipoles. In contrast, Kelvin probe force microscopy  has sensitivity to monopole (screening) charges and contact potential difference as well.
Here we review the capability of NC-SNDM by atomic-resolution imaging of a hydrogen-terminated Si(111)-(7×7) surface .
The previous scanning tunneling microscopy (STM) studies have shown that Si dangling bonds are terminated by the formation of a Si-H bond and thus the surface atoms are chemically passivated [16-18]. In addition to this chemical insight, the NC-SNDM study reveals that surface dipole moments are significantly reduced upon the hydrogen-termination . This suggests that hydrogen-termination also passivates the Si adatoms electrically.
NC-SNDM shares some components, such as a tube scanner and feedback controller, with other SPM methods, as schematically indicated in figure 1 . As a critical component, a gigahertz range LC oscillator is employed for capacitance sensing . This LC oscillator is equipped with a conductive tip interacting with a sample surface and a ring electrode acting as the ground for high-frequency fields. Since Cts is seen as a capacitance inserted in parallel to a built-in LC resonator, the oscillation frequency is given by (see attached equation 1).
This leads to (see attached equation 2). ∆Cts can be obtained through the demodulation of ∆ƒ using an FM demodulator.
The fundamental idea of SNDM is that ∆Cts can include local information on dielectric phenomena, as represented by spontaneous polarization. If we apply an electric field E between the tip and sample, higher-order effects are locally caused by the high field under the tip. For a simple parallel plate capacitor model, these effects can be described by the following Taylor expansion of electric displacement D: (see attached equation 3) ... (1) where ε2 is a linear dielectric constant and εi,i=3,4,..., are called nonlinear dielectric constants. Ps denotes spontaneous polarization. If E(t)=Epcoswpt and wp <<2πƒ, ∆Cts is given by the sum of different higher harmonics as follows : (see attached equation 4) ... (2)
Since ∆ƒ ∝ ∆Cts, εi+2 is proportional to the i-th component of ∆ƒ(t), which can be measured using a lock-in amplifier. The ε3-channel in figure 1 is used for estimating the orientation of Ps under the tip, because ε3 dominates a second order effect in a polarized material. The sign of ε3 is reversed if Ps under the tip is inverted in the z-direction . The sign is determined by the phase of the fundamental harmonic. ε3=0, if the surface under the tip is not polarized. Thus, a two-dimensional map of ε3, or ε3-image, provides polarization distribution of the surface. In addition, since ε4 strongly depends on the tip-sample distance, the non-contact mode is enabled if ε4-channel is connected to a z-feedback controller . An additional channel is prepared for simultaneous tunneling current measurement (see the detail in ref. ) and STM imaging under dc bias.
Results and Discussion
Figure 2 shows NC-SNDM images and profiles of a partly hydrogen-terminated Si(111)-(7×7) surface . The details of the experiment are described in reference . 20% of Si adatoms were here terminated by atomic hydrogen. The topographic image displayed in figure 2(a) atomically resolves a (7×7) structure remaining even after hydrogen-exposure. Almost all of the adatoms in the image have a similar contrast and thus it is difficult to identify reacted adatoms in the topographic image. This contrasts with the previous STM images where the hydrogen-terminated adatoms appeared dark owing to lower local density of states [16-18]. On the other hand, in the ε4-image shown in figure 2(b), we can determine the reacted adatoms. Some of the adatoms have a dipole moment, as indicated by red. For example, on the atoms denoted by A and C in figure 2(a), the ε3-channel output positively shifted up to about 100 Hz, which is seen in the profile in figure 2(d). These are identified as a bare adatom with one dangling bond . As shown in figure 3(a), the bare adatom has polarized, or asymmetric, charge distribution consisting of a positively charged atomic nucleus protruded towards the vacuum side and negatively charged chemical bonds below the nucleus . This polarized charge distribution was identified by NC-SNDM. In sharp contrast to the bare adatoms, one also finds that some of the surface atom have significantly lower ε3 indicated by green in figure 2(b) (e.g. the ‘B‘ atom. See also the profile in fig. 2(d)). These surface atoms are hydrogen-terminated adatoms. This is because hydrogen-terminated adatoms have nearly central symmetric charge distribution. As shown in figure 3(b), a Si-H bond is formed above the adatom and the charge distribution become close to the non-polar sp3 structure formed by Si atoms in the bulk, resulting in smaller ε3 on the hydrogen-terminated adatoms. These results indicate that the hydrogen-terminated adatoms are electrically passivated as well as chemically.
NC-SNDM is a unique experimental technique that can be used for simultaneous imaging of topography and dipole moments of a surface in the atomic resolution. This method has been successfully applied in the dipole imaging of a Si surface modified by hydrogen. The results presented here demonstrate that NC-SNDM will help the atomic-scale investigation on spontaneous polarization or permanent dipoles on surfaces and interfaces.
This work is partly supported by a Grant-in-Aid for Scientific Research (S) (23226008) from the Japan Society for the Promotion of Science.
 Cho Y. et al.: Rev. Sci. Instrum. 67, 2297 (1996)
 Cho Y. et al.: Appl. Phys. Lett. 75, 2833 (1999)
 Cho Y. and Hirose Y., Phys. Rev. Lett. 99, 186101 (2007)
 Cho, Y. et al.: Jpn. J. Appl. Phys. 36, 3152 (1997)
 Honda K. et al.: Nanotechnology 16, S90 (2005)
 Honda K. et al.: J. Phys.: Conf. Ser. 209, 012050 (2010)
 Ohara Y. and Cho Y., Nanotechnology 16, S54 (2005)
 Hirose R. et al.: Nanotechnology 18, 084014 (2007)
 Suzuki M. et al.: Appl. Phys. Lett. 105, 101603 (2014)
 Ota, H.: Electron Devices Meeting 2007. IEEE International, 2007, 65-68 (2007)
 Yajima T. et al.: Nature Materials 10, 198 (2011)
 Ganzorig C. et al.: Appl. Phys. Lett. 79, 272 (2001)
 Susaki T. et al.: Phys. Rev. B 84, 115456 (2011)
 Nonnenmacher M. et al.: Appl. Phys. Lett. 58, 2921 (1991)
 Mizuno D. et al.: Appl. Phys. Lett. 103, 101601 (2013)
 Tokumoto H. et al.: J. Vac. Sci.Technol. A 8, 255 (1990)
 Rogers D. and Tiedje T.: Phys. Rev. B 53, R13227 (1996)
 Razado I.C. et al.: Phys. Rev. B 71,235411 (2005)
 Yamasue K. and Cho Y.: J. Appl. Phys. 113, 014307 (2013)
 Avouris P. and Wolkow R.: Phys. Rev. B 39, 5091 (1989)
Prof. Dr. Kohei Yamasue
Research Institute of Electrical Communication