Nanoparticle Characterization Based on Stm and Sts

ab In this review, we describe recent progress made in the study of nanoparticles characterized by scanning tunneling microscopy (STM) and scanning tunneling spectroscopy (STS). Basic principles of STM measurements and single-electron tunneling phenomena through a single NP are summarized. We highlight the results of electrical and photonic properties on NPs studied by STM and STS. Because nanoparticles are single-digit nanometre in diameter, a single-electron transport on individual nanoparticles such as Coulomb blockade and resonant tunneling through discrete energy levels are investigated. Photon-emission from NPs is also introduced based on STM measurements. Novel single-nanoparticle functions such as stochastic blinking and one-write erasing behaviours are presented. This review provides an overview of nanoparticle characterization methods based on STM and STS that include the detailed understanding of the electrical and photonics properties of nanoparticles.


Introduction
Scanning tunneling microscopy (STM) is an indispensable technique in the current research fields of nanoelectronics and surface science. [1][2][3][4][5][6][7][9][10][11] STM was developed by researchers Binnig and Rohrer, at the IBM Zurich Research Laboratory in 1982 to observe the (110) surfaces of CaIrSn 4 and Au. 1 STM has revealed many intriguing surface structures such as the 7 Â 7 reconstruction of a Si(111) surface 6 and highly oriented pyrolytic graphite (HOPG). 7 Gaining atomic resolution in STM is a major motivation in surface science as the technique would allow researchers to obtain images of individual atoms and molecules in real space. Because of the molecular resolution, STM can display local electrical characteristics of nanomaterials by scanning tunneling spectroscopy (STS). The conductivity of a single nanomaterial particle such as carbon

Tsukasa Tada
Tsukasa Tada received the BS, MS, and PhD degrees in quantum chemistry from Kyoto University in 1973, 1975, and 1987, respectively. After joining the Toshiba R & D Center in 1975, he was with the VLSI Cooperative Laboratories from 1976 to 1980. Besides his research and development work in electron resist materials for VLSI, he has carried out quantum chemical studies on electronic materials, Si LSI device processing technology, and molecular devices at Toshiba R & D Center. He has been a research member of Tokyo Institute of Technology since 2014. Dr Tada is a member of the Chemical Society of Japan, the American Chemical Society, and the Japan Society of Applied Physics. nanomaterials 8,[12][13][14][15] and organic molecules [16][17][18][19][20][21][22] has been determined by STS.
Colloidal synthesized nanoparticles (NPs) are interesting nanomaterials for future nanoelectronics and bioelectronics applications because of their unique electronic and optoelectronic properties. [23][24][25][26][27][28][29][30][31][32][33][34][35][36] The electronic properties of NPs are strongly dependent on their size and ligand molecules. The size of NPs is typically in the nanometre scale, and determines the energy band. 37,38 In the synthesis of NPs, the diameter can be precisely controlled to the sub-nanometre scale. [39][40][41][42][43][44][45][46][47][48][49][50] Consequently, the bandgap of semiconductor NPs 36,51,52 and mean energy level spacing 53,54 can be tuned by the synthesis. 24 A variety of chemical properties can be designed by selecting and controlling the ligand molecules surrounding the NPs. This surface functionalization is widely used for enhancing the solubility in a solvent or imparting chemical reactivity to the metallic surface. 32,40,55 The synthesis of NPs has been established and the mass production of size-controlled NPs can be achieved. These advantages lead to practical applications for synthesized NPs in solid-state electronic devices.
Here, we review recent progress in the characterization of NPs based on STM and STS for applications of nanodevices. In Section 2, the basic principles of STM and STS are summarized to understand their advantages for this investigation. In Section 3, several types of synthesized NPs are reviewed. We also discuss established techniques to deposit NPs on a conductive substrate for STM analysis. In Section 4, we highlight the results of electrical and photonic characterization of NPs by STM and STS. Recent literature describing methods for obtaining useful functions of NPs and their ligands is also summarized.

Scanning tunneling microscopy
The atomic-resolution of STM is attributed to the imaging mechanism based on the quantum tunneling phenomena. STM measures the tunneling current I that is generated by a bias voltage V between the atomically sharp STM tip and the material surface. The tunneling current increases by an order of the current for every 1 Å reduction in distance. 3,8 The distance in the xyz-directions is also controlled by a piezoelectric scanner, which provides angstrom-order changes in the distance, and a feedback loop, which controls the z-direction, is installed to keep the tunneling current constant. By observing the tunneling current as the tip scans a surface, the morphology of the surface can be precisely detected. Measurements can be carried out at room temperature in solution as well as at low temperature under ultra-high vacuum (UHV) conditions. 9 We can obtain various electrical properties of an object by analyzing the tunneling current, such as the local density of states (LDOS) from dI/dV and local barrier height from dI/dz. 3 Fig. 1 shows a schematic view of scanning a metallofullerene and alkanethiol self-assembled monolayer (SAM) on an Au(111) surface by STM. Alkanethiol SAMs self-assemble on an Au (111) surface to form a ffiffi ffi 3 p Â ffiffi ffi 3 p R30 structure. 15,[88][89][90][91] Fig. 2 shows the STM image of a Lu endohedral metallofullerene (Lu@C 82 ) on an octanethiol SAM/Au(111) surface. 15 The octanethiol molecules form SAMs on the Au(111) surface at a distance of 0.50 nm. It is noted that the corrugated surface of the metallofullerene reflects the LDOS in the molecule. The reason why the LDOS affects the STM topography images will be explained later.  Here, we describe the basic physical principles of STM measurements. Fig. 3(a) shows a typical STM tunneling junction comprising alkanethiol molecules on an Au (111) surface. To elucidate the basic principles of STM measurements as simply as possible, we have introduced an approximation that the electric potential of an alkanethiol monolayer is nearly identical to that of the metal electrode (Au). 92 By assuming this simple approximation, a schematic energy diagram of the STM junction can be illustrated as shown in Fig. 3(b).
In this simple model, when a bias voltage V is applied between the tip and the metal electrode, the transmission coefficient T 0 of the electrons through the potential barrier is approximated by the WKB approximation as 93 where d is the distance between the tip and the sample, m e is the electron mass, E is the electron energy, f = (f tip + f sample )/2 is the average potential barrier height between the tip and the sample, h is the Dirac constant, and e is the elementary charge of an electron. If a hard wall potential barrier (i.e. infinite potential barrier f = N) exists in the direction of the wave, the electron is completely reflected by the wall (T 0 = 0). However, if a finite potential barrier exists ( f a N), there is some probability that the electron goes through the potential barrier even though the electron does not have enough energy to overcome the barrier (T 0 a 0).
The tunneling current can be obtained by the integration of the electron density r and the transmission coefficient where r sample (E) and r tip (E) are the density of states for the sample and the tip, respectively, at an electron energy E from the Fermi energy of the sample E Fs , V is the bias voltage (defined as the sample bias with respect to the tip), and T 0 (E, eV, d) is the tunneling transmission probability for electrons with energy E and applied voltage V over the distance from the sample to the tip d. This relationship indicates that the tunneling current between two metallic surfaces is determined by the distance and local density of states of the substrate. Therefore, STM images of a single molecule or a substrate itself reflect both their LDOSs and height information. 13,15,22,[94][95][96][97][98] By considering the tunneling-distance dependence, the transconductance G of the material can be represented as follows, where a is the decay constant of the material, z is the tipsample separation, and A is a constant value. This equation leads to the tunneling current changing by an order of magnitude when the value of z changes by 1 Å (Fig. 4). Measuring dI/dz indicates the conductivity of the observed nanomaterial, for example, a = 2.3 Å À1 in a vacuum or a = 1.1 Å À1 in alkanethiol molecules. 8,16,18,20 The conductance of a single nanomaterial particle can be measured using STM by two methods. One is a break junction technique, which measures the conductance between the tip   and the substrate while raising the tip. 19,[99][100][101] The peak conductance in the statistical histogram represents the conductance of a single molecule.
The other method is the STS technique for directly measuring the current-voltage (I-V) characteristics of a single molecule. In the investigation of a single NP, STS is usually adopted because single-electron tunneling phenomena can be simultaneously characterized in the current as a function of voltage. Therefore, we only consider the STS technique in further sections of this review.

Scanning tunneling spectroscopy
STS can achieve electrical characterization of a local point on a nanomaterial by monitoring the I-V characteristics. An STM tip is positioned over a sample keeping the feedback loop off during the measurements; therefore, the STS measurements can obtain the electrical characteristics of a truly single molecule or NP.
If the tunneling current through the sample is represented by eqn (2), the value of dI/dV is given as follows, assuming that d is constant during the measurements. Therefore, the differential conductance dI/dV is proportional to the LDOS of the sample r sample if the applied voltage is low and r tip (0) is constant. 1,4,5,11 For NPs, the LDOS corresponds to the discrete zero-dimensional (0D) energy spectrum or energy bandgap. Spatial mapping of dI/dV in the STM image can be achieved by current-imaging tunneling spectroscopy (CITS). CITS is useful for investigating the spatial distribution of the LDOS over an entire molecule. 15,22,82,102 Electron tunneling is treated as an elastic process in the above discussion; however, a small amount of current is also provided by inelastic tunneling processes of electrons through the tunneling barrier. Vibrational modes of organic molecules can be detected by measuring dI 2 /dV 2 characteristics using a lock-in amplifier, which is called inelastic tunneling spectroscopy (IETS). [103][104][105] In IETS measurements, electrons release an energy of ho through the inelastic tunneling process as phonon vibrations o of the nanomaterials. STM is also useful in nanophotonics to observe locally excited luminescence and plasmonic modes in nanostructures, as well as electrical characterization. [106][107][108] In this case, the energy released from the tunneling electrons is converted into photonic energy, which excites photons from the nanomaterial. The simultaneous observation of topographies and photon-emission maps is realized by the combination of STM and an avalanche photodiode to count the number of photons. 108 However, we can exclude the effects of inelastic tunneling processes in the STS measurements as long as the contribution of the inelastic tunneling process is small. Therefore, we keep two viewpoints in mind to analyze the STS characteristics: direct tunneling and single-electron tunneling.
2.2.1 Direct tunneling. When a single tunneling barrier, such as a vacuum or an organic molecule, insulates the STM tip from the metallic substrate, tunneling current flows between them ( Fig. 3(a)). Simmons summarized three types of tunneling behaviors in the analysis of electrical characteristics: (a) eV C 0, (b) eV o f, and (c) eV 4 f, where the average barrier height of the insulator was f. 93 The three types of tunneling behaviors are shown in Fig. 5. When the temperature T = 0, the tunneling current density J can be written as , h is the Planck constant, Dd is the tunneling distance through an insulator, and b is a correction factor.
Characterization based on this basic analysis has been carried out to understand the electrical properties of organic molecules and nanogap electrodes. 18,[109][110][111] To understand the electrical behaviors, tunneling phenomena at the severalnanometre scale must be considered.
2.2.2 Single-electron tunneling. If double-barrier tunneling junctions (DBTJ) comprising a single conductive island between two metallic electrodes are formed between the STM tip and the substrate, single-electron tunneling appears through the barriers. The single conductive island in a DBTJ is called a Coulomb island. Several nanomaterials on a metallic substrate, such as organic molecules and NPs, can behave as Coulomb islands. Fig. 7(a) shows a schematic diagram of a DBTJ comprising a decanethiol-protected Au NP anchored by a decanedithiol molecule on an Au(111) surface.
Typical experimental I-V characteristics of the DBTJ consisting of an STM tip/vacuum/octanethiol ligand/Au core/decanedithiol anchor molecule/Au (111) substrate are shown in Fig. 6 (solid line), which was measured by STS. A clear Coulomb staircase including Coulomb gap and first and second stairs for positive and negative bias voltages has been observed.
An equivalent circuit diagram of this DBTJ is shown in Fig. 7(b). Tunneling junction 1 (2) is expressed as a parallel of tunneling resistances R 1 (R 2 ) and capacitances C 1 (C 2 ), respectively. Here, we consider that junction 1 consists of a vacuum and ligand molecules, and junction 2 consists of only a decanedithiol.
Single-electron tunneling can be described by the orthodox theory. 81,[112][113][114] A single electron can tunnel into the island when electrostatic conditions are present for the entire circuit; the total energy after changing the number of electrons in the island is smaller than the energy of the initial states. The change in the total energy when an electron tunnels through junction 1 DE AE 1 is given by and for tunneling through junction 2, DE AE 2 is given by where n is the number of electrons in the Coulomb island, C S = C 1 + C 2 is the total capacitance of the DBTJ, and Q 0 is the fractional residual charge surrounding the island. When DE AE 1 and DE AE 2 are larger than zero, single-electron tunneling is suppressed because the electrostatic energy after electron tunneling becomes larger than that of the initial states. This suppression is called a Coulomb blockade (Fig. 7(c)). Electrons can tunnel through the junction one by one only if DE AE for the junction becomes negative ( Fig. 7(d)).
To observe the Coulomb blockade clearly, it is necessary to satisfy the energy-scale condition: where k B is the Boltzmann constant, T is the temperature, R is the resistance of the tunneling junction, and E c = e 2 /2C S is the charging energy. This condition limits the observation temperature of the Coulomb blockade. The tunneling resistance of R 1 can be adjusted by a tunneling current through a DBTJ (set-point current). In contrast, the resistance of R 2 is expected to be constant owing to the ligand molecules. The overall shape of the Coulomb staircase strongly depends on the tunneling resistance ratio R 1 /R 2 as shown in Fig. 8; therefore, we can estimate the values of the tunneling resistance in the junctions. 72,81,84,87 In evaporated Au NPs on dodecanethiol/Au(111) surfaces, it was reported that the total resistances (R 1 + R 2 ) and the total capacitances (C 1 + C 2 ) of DBTJs can be controlled by changing the set-point current in the STM system. 72 The tunneling resistances in the DBTJs can be estimated as a unique value if 1 o R 1 /R 2 o 50 when R 1 is larger than R 2 . 81 Under these conditions, we can estimate the resistance R 2 between a NP and an Au(111) surface, which corresponds to the resistance of the linker molecules, by fitting the I-V curves with the orthodox theory.
One interesting application of a Coulomb blockade is a single-electron transistor, which comprises a DBTJ and gate electrodes capacitively coupled to a Coulomb island. 115 Single-electron transistors can be used as single-electron memory, 116 a single-electron pump, 117 and multiple logic circuits. 63 Room-temperature single-electron transistors have been reported using metallic NPs, 58,59 Si NPs, [118][119][120][121] and carbon nanotubes. 122,123 The charging energy of a single-electron transistor is an important parameter to determine the operating temperature following eqn (8). We can evaluate the charging energy of a single NP by the STS measurement and the orthodox theory. Thus, STM and STS are useful tools to explore candidates for Coulomb islands in single-electron transistors.

Charging energy of nanoparticles
Here, we evaluate the charging energy of NPs to show that synthesized NPs are promising for practical applications of quantum effects. Fig. 9(a) shows the theoretical capacitance between a single conductive sphere and a conducting plate, which is proportional to the capacitance of a sphere in free space. 124,125 The corresponding charging energy is estimated in Fig. 9(b). The charging energy increases as the size of the Coulomb island decreases. The simulated E c /k B T dependence of the Coulomb staircase is shown in Fig. 9(c). The I-V curves are calculated by the orthodox theory and it is clear that this simulation follows eqn (8). Moreover, this simulation indicates that E c is required to be much larger than a thermal fluctuation to observe a clear zero conductance region. The zero conductance region of the Coulomb blockade is critical to maintain the on/off ratio in a single-electron transistor. Therefore, Coulomb islands less than 2 nm in diameter are desired for room-temperature operation, e.g., E c 4 260 meV where E c is 10 times larger than the thermal energy at 300 K. When the diameter of NPs is lower than 2 nm, we need to consider the effect of the discrete 0D energy levels caused by quantum confinement, even at room temperature. Fig. 9(d) shows an example of the discrete energy level spacing DE in Au NPs and Si nanoclusters (NCs). The value of DE in the metallic Au NPs is directly derived from the density of states at the Fermi energy and volume DE E 2p 2 h 2 /(m Au k F V), where m Au is the electron mass in Au (1.1 m e ) and k F is the Fermi wavenumber in Au. 23,126 This assumption is valid when the Fermi level is in a half-filled band (i.e., metallic NPs), which implies that a number of electrons exist in the NPs. Therefore, the 0D levels are randomly dispersed along the energy axis, and the value of DE in metallic NPs describes a mean spacing between the levels.
In contrast, the value of DE in semiconductor NPs is estimated by the potential well of a quantum sphere DE E p 2 h 2 /(2m Si (R/2) 2 ), where m Si is the electron mass in Si (0.40 m e ). 127,128 This corresponds to an energy difference between the ground state and the first excited state of the classical quantum sphere model. Semiconductor NPs have a few electrons because the Fermi level is in its bandgap; thus, the 0D levels tend to follow the electron-shell filling described by the potential-well approximation. This simple 0D-level spectrum is attributed to the observation of a predictable atomic-shell filling in semiconductor NPs (shown later).  Semiconductor NPs generally have larger energy spacings than metallic NPs of the same diameter. 129 The condition to observe the effect of discrete energy levels is DE 4 3.5k B T, which was derived from the thermal broadening of Fermi-Dirac distributions in metallic lead electrodes. 130 To harness these quantum effects practically, it is critical to control the size of the Coulomb islands. Electron configurations of elemental atoms are precisely determined by the number of atomic nuclei. Coulomb islands of a single-donor atom in Si have been reported as single-atom transistors, which show a charging energy or resonant spectrum in accordance with the excited state of the single donor. [131][132][133] These results indicate that precisely size-controlled Coulomb islands show reproducible quantum effects. Synthesized NPs have several advantages: size control in the sub-nanometre scale, ability for mass production, and reproducibility. Thus, synthesized NPs are promising materials to exhibit the above-mentioned quantum effects.

Synthesized nanoparticles
For stable metallic NPs, noble metals such as Au, Ag, and Pt are used because they are not easily oxidized. 31 Synthesized NPs are usually surrounded by ligand molecules to prevent aggregation.
One of the most famous techniques to synthesize metallic NPs is the Brust-Schiffrin method for Au NPs. 40,43 Temperatureand air-stable, size-controlled Au NPs can be mass produced by this method and can be functionalized by ligand molecules. 27,[134][135][136] Teranishi et al. controlled the core diameter of decanethiol-protected Au NPs from 3.4 AE 0.3 nm to 6.8 AE 0.5 nm by revealing the appropriate heat-treatment temperatures during synthesis. 45 The number of Pt atoms (13, 55, 147, 309, etc.) 137,138 and Au atoms (10, 15, 18, 22, 25, etc.) 48 occurs in the well-known magic numbers. As the magic number is determined by the crystalline structure of the NPs, the size of the nanoparticles can be identical. Therefore, if we utilize the magic number of atoms in single-electron devices, the charging energy of the Coulomb islands and the 0D-level spacing can be identical.
Several thiol-derivatized molecules are used as ligands, the same as alkanethiol molecules, such as amino-type molecules, 46 glutathione thiolate, 47,48 porphyrin, 49 and oligo(phenyleneethynelene) (OPE) molecules. 50 The synthesized NPs can be stored in the solid state under ambient conditions and redissolved in common organic solvents. Semiconductor NPs can also be synthesized by established methods to fine-tune the diameter in the sub-nanometre scale. 51,52,55,[139][140][141][142] Because of these advantages, synthesized NPs are intensively used in the field of nanoelectronics. Reported synthesized NPs are summarized in Table 1.

View Article Online
The deposition of synthesized NPs is important in the practical application of NPs; such deposition is classified into two types of absorption: physisorption and chemisorption. Dropcast, spincoat, and Langmuir-Blodgett methods are widely used as physisorption processes to form monolayers of NPs on flat substrates. The interacting force between the NPs and the substrates is mainly caused by only van der Waals interactions. Conversely, chemisorption processes are based on chemical reactions between the NPs and the functionalized substrates with SAMs. The interacting forces in chemical interactions are stronger than van der Waals forces, which make the NPs stable on the substrates.
Controlling the density of NPs by the density of tether molecules on the substrate is a notable advantage in the chemisorption process. Alkanethiol-alkanedithiol-mixed SAMs can be formed by immersing Au(111) substrates into alkanethiol and alkanedithiol solutions successively. [145][146][147] The inserted alkanedithiols in alkanethiol monolayers can be observed by STM or atomic force microscopy (AFM). 83,[145][146][147] In ref. 83, the density of chemisorbed Au nanoparticles was controlled in 1/100 Â 100 nm 2 in 0.5 mM octanedithiol solution and 39/100 Â 100 nm 2 in 50 mM solution. Electrically stable SETs (chemically assembled SETs) have been demonstrated by using the same chemisorption process to deposit a single synthesized Au NP between size-regulated nanogap electrodes formed by electroless-gold plating. 54,60-65 Therefore, chemisorption is more promising than physisorption for stable and ideal electrical transport through the NPs by STM and STS.
In this section, we explained the advantages of the synthesized NPs. Chemical assembly processes are more suitable for stable chemisorption of NPs on a substrate. In the early stages, single-electron transport in metallic NPs was mainly investigated by using STM. The NPs for STM investigation were made by thermal evaporation. Because metals naturally tend to nucleate into particles, a small amount of evaporation (less than 1 nm) can generate hemispherical NPs for small metal droplets on a thin insulator. [148][149][150] In the first report of single-electron transport in NPs, a Coulomb staircase was observed at 4.2 K in a structure comprising a granular layer of evaporated Au NPs in the oxide layer between two Ag or Cu films. 151 Because multiple current paths exist as parallel arrays of junctions when this technique is used, the observed Coulomb staircase strongly depends on the variation in size or the number of NPs. Therefore, it is necessary to observe single-electron transport in a single NP by using STM and STS measurements.
A Coulomb staircase was first observed by STM in a structure comprising a granular Al layer on oxidized Al films. 152 The measurement temperature was 1.2 K and STS was carried out on a single granular Al dot. The observed Coulomb staircase had a periodic structure, which was explained by Monte Carlo simulations based on the DBTJ of a single Coulomb island. This analysis proved that the single NP acted as a Coulomb island in this measurement. After this report, several evaporated NPs were investigated by STM and STS at low temperature, e.g., Al, 153 Au, 114,154,155 Co, 156 In, 157,158 Pt, 159,160 Pd, 161,162 and Si. 79 To increase the observation temperature of the Coulomb blockade, it is necessary to form smaller NPs for Coulomb islands. To observe the Coulomb blockade at room temperature (T = 300 K), the charging energy needs to be much larger than the thermal fluctuation k B T (25.8 meV). We have shown in Fig. 9(c) that a much larger charging energy is required to observe singleelectron transport without any random tunneling events. Therefore, the size of the NPs has to be on the order of 1 nm to observe the Coulomb blockade at room temperature.
Schönenberger et al. reported a Coulomb staircase on Au-evaporated NPs/ZrO 2 (1 nm)/Au film at room temperature in air. 71 The schematic image of the structure and STS characteristics are shown in Fig. 10(a) and (b), respectively. To increase the charging energy, the ideal height and diameter of the NPs were 0.2 nm and 4 AE 0.5 nm, respectively. The charging energy of the DBTJ was 70 meV, which is larger than 26 meV at 300 K; therefore, a room-temperature Coulomb staircase was observed.
Next, organic SAMs were introduced as tunneling barriers. Initially, the tunneling barrier between the NPs and the substrate was the oxidized layer. Because the tunneling current strongly depends on the distance, the thickness of the insulating layer has to be atomically controlled. When a SAM acts as the tunneling barrier, we can control the thickness of the tunneling barrier by the length of the molecules. Alkanethiol molecules are the most famous SAMs, which consist of an alkyl chain with a thiol end group. Alkanethiol works as an insulator because all covalent chemical bonds are localized between the atoms (i.e., s-bond). The length of an alkanethiol molecule increases by 0.13 nm with each addition of a carbon atom in the alkyl chain. 163 Thus, an alkanethiol on the structure of the metal NP/alkanethiol/metallic surface is expected to help control the thickness of the tunneling barrier between the metals on the order of an angstrom.
In 1994, a room-temperature Coulomb staircase was demonstrated by STS for deposited Au NPs/alkanethiol/Au(111) film. 72 The diameter and height of the Au NPs were 5-10 and 2-3 nm, respectively. The charging energy was estimated to be 180 meV by a rough capacitance evaluation using parallel-plate capacitor models. In later reports, double-ended thiol molecules (alkanedithiol) were also used as linkers between the deposited NPs and the Au(111) surfaces for the chemisorption of NPs. To realize more stable Coulomb staircases at room temperature, a,a 0 -xylyldithiol and 4,4 0 -biphenyldithiol SAMs were used to bind NPs deposited on an Au(111) surface by a cluster beam (Fig. 10(c) and (d)). [73][74][75] A multiple expansion cluster source can generate Au NPs on the SAMs with diameters of E2 nm and heights of E1 nm. The resistance between a single Au NP and an Au (111) surface was estimated to be 18 AE 12 MO. This value corresponds to the calculated resistance of 4.5 AE 0.50 MO by the Landauer formula R 0 = (h/2e 2 )T 0 . 164 Deposited CdS, 165,166 Co, 167 and Pt NPs 168,169 on SAMs were also studied. Recently, 0D levels of evaporated Au NPs on a 1,4-benzenedimethanethiol/ Au (111) substrate were confirmed at 78 K. 170 The coverage of Au deposited on octanethiol/Au(111) was investigated to determine the morphology of growing Au NPs. 171 The self-capacitance of a single deposited Au NP on octanedithiol/ Au (111) was strongly dependent on the surface coverage and its diameter. [172][173][174][175][176] Sakotsubo et al. reported that the density of evaporated clusters can be controlled by using thiol-dithiolmixed SAMs. 176 Their results suggest that an isolated Au NP is appropriate for investigating the Coulomb blockade behavior at room temperature because the surrounding Au NPs display a parasitic capacitance and decrease the charging energy.
Synthesized NPs are generally passivated by organic molecules to avoid aggregating and sintering, i.e., they are initially covered by a tunneling barrier. The density of NPs deposited on a substrate can be controlled by changing the concentration of the NPs in solution, which is easier than controlling the density by evaporation methods. Because the NPs were isolated and size-controlled, synthesized NPs covered with ligand molecules were promising materials for Coulomb islands.
Initially, ligand-stabilized Pt NPs (comprising 309 Pt atoms) showed clear Coulomb staircases in STS measurements at 4.2 K. 137,177 Several years later, the Coulomb staircases for a single synthesized Au NP with an atomic mass unit (amu) of 29 k were reported at up to room temperature. 178,179 The Au NPs consisted of 147 AE 10 Au atoms in the core, corresponding to a diameter of 1.68 AE 0.04 nm, which was measured by mass spectrometry. 179 The charging energy of the DBTJ was 75 meV, which was larger than the thermal fluctuations at room temperature ( Fig. 11(a) and (b)).
Electrochemical cyclic voltammetry of Au-NPs solutions was used in an attempt to confirm the values of E c by STS. 41,178,180 Current peaks separated by e/C dot were observed in the voltammograms, which were corresponding to the Coulomb staircases in the STS measurements. Cyclic voltammetry proved that E c of Au NPs (1.64 nm diameters) exceeded ambient thermal fluctuation. 178 Another novel observation in this report was that the 0D levels of the Au NPs affected the I-V characteristics. The observed dI/dV spectrum at 77 K included several peaks that could not be explained by the orthodox theory. These peaks indicated that the tunneling current increased because electrons were resonantly tunneling through the 0D levels. 54 The calculated mean energy level spacing DE was estimated to be 45 meV, which also surpassed the thermal broadening at 77 K (3.5k B T = 23 meV).
Au NPs with various ligand molecules have been investigated by STM: alkanethiols, 77,[81][82][83]178,179,181,182 amino-type molecules, 84,183 dendrimers, 184 and polymers. 185,186 The ligand molecules works as linkers between the NPs and the substrate by physical or chemical interactions. Octanethiol-octanedithiolmixed ligands were used to anchor the 1.5 nm-diameter NPs to a bare Au (111) surface. 181 The driving force in anchoring the NPs is attributed to the strong affinity of free thiols for Au atoms. 187 The force of the Au-S covalent bond is E180 kJ mol À1 , which is stronger than a hydrogen bond or van der Waals interaction. The stability of this bond can be attributed to the observation of a stable Coulomb blockade at room temperature using octanethiol-octanedithiol-mixed ligands. 181 Au NPs surrounded by citric acids are negatively charged; thus, they selectively adsorb onto a positively charged surface functionalized by amino-substituted thiols. 188,189 The Au NPs surrounded by amino-thiol can be anchored on a positively charged surface by carboxyl-substituted thiol as well. 84 Roomtemperature Coulomb blockades have been reported using positively charged 1.8 nm-diameter Au NPs chemisorbed on a thiolactic acid/Au(111) surface (Fig. 11(c) and (d)). 84 Poly-(propylene imine) dendrimers (guest) can be immobilized on SAMs of heptathioether-functionalized b-cyclodextrin (host) through host-guest interactions. 184 The Au NPs protected by the dendrimer showed negative differential resistance peaks, attributed to the resonant tunneling of electrons through the molecular orbitals, in a room-temperature Coulomb staircase. Polyvinylpyrrolidone-protected Au NPs and Pd NPs on HOPG were used to control the charging energy of DBTJs. 185 The charging energy of the NPs was inversely proportional to the diameter of the NPs, which corresponds to the classical expression for the capacitance of a metal sphere in a dielectric medium (4peR, where e is the dielectric constant of the medium and R is the diameter of the NPs).
In our previous studies, we investigated physisorbed and chemisorbed Au NPs on Au(111) surfaces at low temperature. Fig. 12 shows two STS results on physisorbed and chemisorbed octanethiol-protected Au NPs. More clear and predictable Coulomb staircases are observed in the chemisorbed Au NPs anchored by decanedithiol. This result indicates that the chemisorption of NPs can be attributed to the stable electrical characteristics of NPs. 81,83 Electrical characteristics of various synthesized NPs have been reported in addition to Au NPs: Ag, 85,[190][191][192][193] Pt, 137 Co, 156,194 Pd, 168,169,185 Au-Ag alloy, 195 InAs, 85,[196][197][198] CdS, 199 and CdSe. 78,[200][201][202] Tuning the bandgap of InAs NPs linked to an Au surface by hexanedithiol molecules has been demonstrated between 0.9 and 1.9 eV by changing the diameter of the NPs between 1.0 and 3.5 nm (Fig. 13(a)). 196 The tunneling spectrum in the report indicated that an atomic-like Aufbauprinciple filling of the s-and p-shells also occurred. The shell spectrum of CdSe and PbSe was quantitatively evaluated by low-temperature STS at 5 K ( Fig. 13(b)). 201,202 Single-electron tunneling through the 0D levels observed by STS provides significant information about the 0D spectrum of NPs.
Reported single-electron effects of the synthesized NPs by STM are summarized in Table 2. The values of E c are much larger than the thermal fluctuations at room temperature due to the small size of the NPs. It has been demonstrated that controlling the size of the synthesized NPs enables the control of E c in SETs. 62 Therefore, synthesized NPs are good candidates for single Coulomb islands in SETs.
In this section, we described the history of the investigation of single-electron transport in single NPs by STM and STS. As the synthesis of NPs improved, several synthesized NPs have  been characterized for applications in single-electron devices. Single-electron effects and electronic structures of a single NP can be revealed by STS measurements. The charging energy of a NP obtained by STS measurements is summarized in Table 2. STS is a powerful tool to explore candidates for Coulomb islands in single-electron transistors from the viewpoint of the charging energy.

Photon-emission from nanoparticles
Taking inelastic tunneling processes into consideration, it is expected that photons can be generated by inelastic electron tunneling between the STM tip and the substrate. Photonemission in a single Ag NP has been explored with lightcollection optics inside STM. 108,191,193,[203][204][205][206] Ag NPs exhibit a surface plasmon resonance at a wavelength of around 400 nm. 203 The STM images and photon counts of an Ag NP were obtained simultaneously (Fig. 14(a)-(d)). 108,193,205 By using the spectrum analysis of the emitted photon energies, the in-plane and out-of-plane plasmon modes were excited in the SAMs of Ag NPs. 206 Photoluminescence switching was reproducibly observed in a single synthesized dodecanethiolprotected Ag NP on Au(111) at room temperature. 204 Detailed analysis of the photon-emission spectrum enables the observation of single-NP luminescence in several images at the tenth-nm scale. This technique allows us to observe the local excited luminescence of a nanomaterial.

Molecular vibrational modes on nanoparticles
Vibrational modes of molecules on synthesized NPs are important issues when we consider photon-emission from the NPs. The effects of a ligand's vibration have been observed on photon-emission from tunneling barriers, confirmed by STM and other optical spectroscopies. It is well known that Raman signals originating from molecular vibration can be enhanced due to a strong electric field on metallic NPs (Surface-enhanced Table 2 Reported single-electron effects on synthesized NPs by STS measurements. The core diameter (R) of the NPs, the experimental charging energy (E c ) of single NP and the observed temperature (T) of single-electron effects in the papers are summarized. RT: room temperature, N/A: not mentioned in the paper. Raman scattering: SERS). [207][208][209][210] Raman spectroscopy and STM experiments were carried out on a myristate-capped Ag NP and a 2-methyl-1-propanethiolate Ag NP onto dodecanethiol/Au substrates, which shows that the vibration of ligand molecules affects the photon emission from the samples (Fig. 15). 211 High-resolution STM observation and Fourier transform infrared spectroscopy found the relationship between homogeneity of a mixed monolayer (1-octanethiol and its fully deuterated analogues) and CH 2 stretching frequencies. 212 Tunneling current through single NPs can be modulated by inelastic electron tunneling due to molecular vibration on the NPs; 58 therefore, STS measurements are possible to detect the vibration of ligands and absorbed molecules on single NPs. In these two sections, we showed that STM and STS are useful tools in the fields of nanophotonics, the same as nanoelectronics.

Surface morphology of molecules on nanoparticles
STM imaging techniques can obtain fruitful information regarding the ligand molecules on NPs. Structural understanding of NPs has been obtained by transmission electron microscopy and X-ray crystallography. [213][214][215] Transmission electron microscopy is widely used to evaluate the size distribution and crystal structure of NPs. However, it is difficult to detect a thin ligand layer using the highly accelerated electron beam because most of the electrons penetrate the layer. X-ray diffraction gives information about a thin organic monolayer on a metal surface by the diffraction pattern. Conversely, STM can directly create an image of the ligand layer in real space, such as the morphology and electron density distribution. Low-temperature STM has been used to visualize the phenyl rings of ligand molecules at 7 K under UHV conditions, which was correlated with a model ( Fig. 16(a) and (b)). 39,216 High-resolution STM images of a mixed-ligand layer (e.g., octanethiol and mercaptopropionic acid) on Au and Ag NPs have also been reported. 212,[217][218][219][220][221][222] The observed images indicated that each ligand formed domains that were aligned to parallel ripples that spiraled around the NPs (Fig. 16(c) and (d)). The same group revealed that the mechanism for SAM formation on NPs is different from that on flat surfaces; the morphology is dependent on the diameter of NPs and the types of molecules. The height of the NPs in the STM images also affords morphological information about the ligand molecules. The lateral width of the STM images is broader than the true width of the objects because of tip convolution effects. 223,224 In contrast, the height and distance between the NPs are not affected by the convolution of the STM tip and the objects. 223 Therefore, it is more accurate to evaluate the diameter of NPs, including the ligands, from the height of the STM images. Ref. 49 showed that the height of the NPs in the image (2.2 nm) agreed well with the core diameter of the NPs (1.0 nm) and the height of the porphyrin molecules (0.6 nm), and hence proved the coordination pattern of thiol-derivatized porphyrins to Au NPs (e.g., six porphyrin faces surrounding each Au NP in a hexahedron).
By considering the set-point current dependence of a Coulomb staircase, the tunneling resistances of DBTJs can be estimated. 81,84,87 Resistances of alkanethiols, 81 a series of acid and base molecules, 84 and OPE 87    resistance of hexanethiol and octanethiol on Au NPs as 460 MO and 7.6 GO, respectively (Fig. 17). It is noted that these tunneling resistances can be compared with those on reported solidstate single-electron transistors. 62,63 STS is also a powerful tool to explore candidates for Coulomb islands in single-electron transistors from the viewpoint of the tunneling resistance.
In this section, we showed that STM can provide information regarding the ligand molecules, such as the morphology, length, and tunneling resistance. Molecular-resolution STM reveals the geometrical order of ligand molecules on NPs. The length of ligand molecules can be estimated owing to the high sensitivity of the height information given by STM. The setpoint current dependence of a Coulomb staircase provides the tunneling resistance of a ligand molecule on synthesized NPs.

Electrical-switching effects in single nanoparticles
Electrical-switching effects comprising functionalized NPs are interesting in possible applications of NPs. 31,225,226 For example, a photoirradiation-induced switching of nanoparticle monolayers was attributed to the reversible photoisomerization of the diarylethene molecules bridging the NPs. 225 A molecular floating-gate effect obtained by visible-light irradiation was observed in SETs with copper-phthalocyanine-doped Au NPs. 68 In principle, the NP switching is due to the reversible properties of the ligands or linker molecules. STM has demonstrated many single-molecule switching effects in molecules such as OPE, 227,228 endohedral metallofullerenes, 8,15 and porphyrin and phthalocyanine derivatives. 21,95,97,98 Therefore, it is possible to observe the switching phenomena of a single functionalized NP by STM.
Blinking STM images of Au NPs tethered to thiol molecules have been reported. 87,229 These blinking events, which were reversible and stochastic, were observed in alkanedithiol and OPE molecules inserted into alkanethiol SAMs. 228,229 In this measurement, Au NPs were introduced to mark the blinking molecules. The blinking STM images were observed even after the attachment of Au NPs to the linker dithiol molecules; thus, the researchers concluded that the switching was caused by the thermally activated movement of the molecules tethered to the Au surface. The observed reversible switching was not attributed to breaking of the thiol-Au bond or a change in the free thiol group. There is also a report stating that the conductivity of Au NPs can be switched through redox effects of the linker molecule. 86 Because the decay length of an electron changes from 16 nm À1 to 7 nm À1 during redox, the conductivity of the Au NPs through Au(111) changed greatly.
A single Au NP protected by OPE ligands also showed stochastic blinking events ( Fig. 18(a) and (b)). 87 The switching events had been observed several times during 80 min of continuous scanning. Decomposition or removal of the NPs Fig. 17 Set-point current dependence of (a) tunneling resistance R 1 , (b) tunneling resistance R 2 , and (c) tunneling resistance ratio R 1 /R 2 under the structure STM-tip/hexanethiol (octanethiol)-protected Au NP/Au (111). Open circles represent octanethiol protected Au NP in the region of R 1 4 R 2 , open squares represent octanethiol-protected Au NP in the region of R 1 4 R 2 , and closed circles represent hexanethiol-protected Au NP. 81 The dashed and solid lines show the relationship of the set-point current I set and R 1 /R 2 , i.e., I set R 1 /R 2 = À260 pA (V set R 2 = À2 V/7.6 GO) and I set R 1 /R 2 = À4.3 nA (V set R 2 = À2 V/460 MO) for octanethiol and hexanethiol protected Au NPs, respectively. Reprinted with permission from H. Zhang  was not observed because the relative position of the NPs in the images did not change. This result shows the possible functionalization of variously sized NPs by surrounding ligands. In this case, stochastic switching due to single-molecular properties can be observed at scales larger than the molecules.
Sakamoto et al. demonstrated the erasure of Au NPs by STM at room temperature (Fig. 18(c) and (d)). The porphyrinprotected Au NPs were systematically removed by applying a voltage of 3.0 V to the Au NPs for 1 min. The quality of the STM images did not degrade even after erasing the NPs, which implies that the STM tip is kept clean during the erasing process. A possible mechanism for erasure may be the multistep reductive dissociation or heat decomposition of the NPs by inelastic tunneling currents. This erasure can be used as one-time write memory if we define the presence (absence) of NPs as 1 (0).
In this section, we summarized electrical-switching effects on NPs by using STM. The switching effects of NPs are attributed to the reversible physical properties of the surrounding molecules. Stochastic switching and one-time write switching have been reported as intriguing properties of single NPs. If we harness these switching phenomena, there appears a possibility for applying the solid-state memory device based on NPs.

Conclusions
STM and STS have become invaluable tools to characterize electrical and photonic properties of single synthesized NPs owing to the sub-nanometre scale topographies and measurements. Single electron effects of NPs, such as Coulomb blockades, have been demonstrated in detail, when combined with the theoretical calculation. The structural and electrical properties of ligand molecules on NPs have been also investigated. Photonic studies have been carried out on NPs by STM, which detects photons generated from molecular tunneling barriers. Stochastic blinking effects and one-write erasing behaviors of NPs are promising functions for the introduction of solid-state nanoscale memories based on synthesized NPs. Characterization methods for NPs by STM and STS reviewed here cover the significant interest for the NP functions towards the realization of solid-state nanodevices, especially for the diverse applications of single-electron transistors, nanoscale sensors or memories.