Facet-Dependent Electrical Conductivity Properties of Cu2O Crystals

Feb 23, 2015 - ABSTRACT: It is interesting to examine facet-dependent electrical ... When electrical connection was made on two different facets of a...
0 downloads 0 Views 321KB Size
Download PDF
Letter pubs.acs.org/NanoLett

Facet-Dependent Electrical Conductivity Properties of Cu2O Crystals Chih-Shan Tan,‡ Shih-Chen Hsu,† Wei-Hong Ke,† Lih-Juann Chen,*,‡ and Michael H. Huang*,† †

Department of Chemistry and ‡Department of Materials Science and Engineering, National Tsing Hua University, Hsinchu 30013, Taiwan S Supporting Information *

ABSTRACT: It is interesting to examine facet-dependent electrical properties of single Cu2O crystals, because such study greatly advances our understanding of various facet effects exhibited by semiconductors. We show a Cu2O octahedron is highly conductive, a cube is moderately conductive, and a rhombic dodecahedron is nonconductive. The conductivity differences are ascribed to the presence of a thin surface layer having different degrees of band bending. When electrical connection was made on two different facets of a rhombicuboctahedron, a diode-like response was obtained, demonstrating the potential of using single polyhedral nanocrystals as functional electronic components. Density of state (DOS) plots for three layers of Cu2O (111), (100), and (110) planes show respective metallic, semimetal, and semiconducting band structures. By examining DOS plots for varying number of planes, the surface layer thicknesses responsible for the facet-dependent electrical properties of Cu2O crystals have been determined to be below 1.5 nm for these facets. KEYWORDS: Band bending, cuprous oxide, facet-dependent properties, electrical conductivity, nanocrystals

F

measurements on all three low-index facets of Cu2O crystals (that is, the {111}, {100}, and {110} facets) over a fairly large applied voltage range. In addition, it is highly interesting to see if a rectifying I−V curve can be obtained by simultaneously contacting a highly conductive facet and a relatively nonconductive facet of a single Cu2O particle with two measuring probes. Here we show that the three low-index facets of Cu2O crystals exhibit dramatically different electrical conductivity properties. A band diagram is constructed to illustrate possible band structures of Cu2O with consideration of different exposed facets. Electron localization function and electron density distribution diagrams for these facets are also presented. Electrical measurements on two different facets of a Cu2O cuboctahedron and a rhombicuboctahedron show rectifying I− V behavior resembling that of a p−n junction, suggesting the use of single Cu 2 O particles as functional electronic components. Remarkably, density functional theory (DFT) calculations on a few layers of surface planes of Cu2O yielded DOS plots matching the measured facet-dependent electrical conductivity properties of Cu2O crystals. By examining DOS plots with increasing number of surface planes for each facet, approximate thicknesses of surface planes responsible for the observed facet-dependent electrical properties have been determined for the first time. Experimental Details. Cu2O crystals used for the electrical conductivity measurements were synthesized following our previously reported procedures.4,5 The particles were formed from an aqueous mixture of CuCl2, sodium dodecyl sulfate (SDS), NaOH, and NH2OH·HCl reductant. SDS surfactant has

ormation of polyhedral inorganic nanocrystals possessing sharp faces and a series of shapes is essential for the examination of their facet-dependent physical and chemical properties.1−3 In addition, the particle surfaces should be clean to minimize the effects of capping molecules on the measured materials properties. Ionic solids such as metal oxides generally exhibit much greater facet-dependent properties than metals, presumably because the surface charge states for different facets of ionic solids are more different with cationic and anionic components.1 Cu2O, Ag2O, PbS, and PbSe have been synthesized with systematic shape evolution.4−13 They are ideal systems for the facet effect studies. In particular, cuprous oxide crystals have been demonstrated to display strong facetdependent photocatalytic and organocatalytic activities and electrical conductivity behaviors.4,5,14−21 More recently, facetdependent optical properties of Cu2O crystals have been reported by taking UV−vis absorption spectra of Au−Cu2O core−shell nanocrystals with different shell morphologies.22,23 It is believed that the exhibited photocatalytic, electrical, and optical properties of Cu2O crystals are related materials phenomena, because these properties involve charge transfer into and out of the crystal interior, or internal plasmonic field reaching to the crystal surface. Cu2O-catalyzed organic coupling reactions on the other hand are primarily surface-related heterogeneous processes and crystal interior is generally not considered.15−18 Previously electrical conductivity measurements on a single Cu2O cube and octahedron have been performed within a scanning electron microscope (SEM) sample chamber by using two tungsten probes to make contact with the crystal.19 The same approach has recently been employed to obtain I−V curves of a Cu2S−Ag2S superlattice nanowire.24 To better understand the electrical behaviors of different facets of Cu2O, it is necessary to conduct such © XXXX American Chemical Society

Received: January 13, 2015 Revised: February 9, 2015

A

DOI: 10.1021/acs.nanolett.5b00150 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

were not successful. Generally several successful measurements were made for each set of crystal facets to establish their correct electrical conductivity behavior. Figure 1 gives representative

been carefully removed by centrifugation before the electrical conductivity measurements. The large Cu2O rhombicuboctahedra were prepared following a reported procedure with some modification.25 First, 2.9948 g of copper acetate (Cu(OAC)2, 98%, J. T. Baker) and 75 mL of deionized water were added to a beaker and heated to 70 °C. Subsequently 3.6 g of NaOH (98%, Aldrich) dissolved in 5 mL of deionized water was introduced to the beaker and stirred for 5 min. Next, 0.6 g of D(+)-glucose (99%, J. T. Baker) was added with stirring for 40 min until the solution color turned deep red. The solution was centrifuged three times at 5000 rpm for 5 min to collect the Cu2O crystals. A tungsten wire with a diameter of 0.5 mm (99.95%, Alfa Aesar) was dipped into a 1.0 M NaOH solution. A direct current (15 V and 1 A) was applied to sharpen the wire tip. After 1 min of electrolysis, the wire tip was reduced to a diameter of about 100 nm. To remove the tungsten oxide layer, the wire tip was immersed into a 10 M KOH solution for 5 s. Next, a Si(111) substrate was annealed at 900 °C in air atmosphere in a three-zone tube furnace for 48 h to form a SiO2 layer thicker than 500 nm on the substrate as an insulating layer to prevent leakage current flowing from the Cu2O crystal to the Si substrate. The treated tungsten probes were installed on a nanomanipulator (Kammrath & Wiess GmbH), which is connected to a Keithley Model 4200-SCS source measurement unit. The nanomanipulator was loaded inside a JEOL 7000F scanning electron microscope. The two tungsten probes were first brought in contact and an electric current was applied until a linear I−V curve was obtained. This I−V curve indicates purely metallic contact and any surface oxide has been removed. A drop of dilute Cu2O crystals was added to the thermally treated Si substrate. After evaporating the droplet, the substrate was loaded inside the electron microscope. First-principles calculations were performed using DFT26 and the Perdew−Burke−Ernzerhof generalized gradient approximation (PBE-GGA)27 with a 440 eV cutoff for the plane-wave basis set. Cambridge serial total energy package (CASTEP) was used to calculate the electron localization functions (ELFs), electron density, and density of state of the Cu2O (111), (100), and (110) planes. The positions of all the atoms in the one-layer structure were fully relaxed during geometry optimization, and we used 21 k-points to compute the electronic band structures. Results and Discussion. The cubic, octahedral, and rhombic dodecahedral Cu2O crystals with dimensions of hundreds of nanometers and sharp faces were synthesized following our reported procedures.4,5 The nanocrystals have been washed several times to remove sodium dodecyl sulfate surfactant, so they have clean surfaces. Previously, we have used Fourier transform infrared spectroscopy and X-ray photoelectron spectroscopy to confirm that our washing procedure yields surfactant-free Cu2O nanocrystals.15,17 These fairly large nanocrystals were used for the electrical conductivity measurements because the fabricated tungsten probes for making electrical contacts to the particles do not have very sharp tips. The particles were loaded on a silicon substrate coated with an insulating SiO2 layer, so that the substrate does not affect the recorded I−V curves. The substrate was loaded into a SEM chamber equipped with a nanomanipulator to bring the surface oxide-free W probes in contact with a single Cu2O particle. It is not always easy to make correct contacts, as the particles can be pushed away. Sometimes crystal vibration under the influence of a large applied voltage can occur, and such measurements

Figure 1. Tungsten probes contacting single Cu2O crystals for their electrical conductivity measurements. (a−c) SEM images showing two tungsten probes contacting opposite faces of single Cu2O cube, rhombic dodecahedron, and octahedron. (d) I−V curves for the {100}, {110}, and {111} faces of Cu2O crystals. Inset gives the expanded I−V curves to clearly show the electrical response for the {110} faces.

SEM images of tungsten probes making good contacts to a single Cu2O cube, rhombic dodecahedron, and octahedron and the measured I−V curves over the range of −5 to 5 V. Additional I−V measurement data are shown in Figure S1 in the Supporting Information. Consistent with previous measurements, a Cu2O octahedron bound exclusively by the {111} facets is highly electrically conductive with current rising rapidly to reach 1750 nanoamperes (nA) at 5 V.19 Previously a very small current was measured for a pristine cube bound entirely by the {100} facets (less than 20 nA at 4 V).19 Possibly due to better electronics used in this study, Cu2O cubes show better electrical conductivity with a current of ∼20 nA at 3 V. At an applied voltage of 5 V, the current can reach 230 nA. Although the Cu2O cubes are practically not conductive below 2 V, their conductivity can increase substantially under much larger applied voltages. They give an I−V curve more typically observed for a semiconductor with a low turn-on voltage. Transistors fabricated from multigrained Cu2O thin films require a much higher turn-on voltage, so it is advantageous to construct such devices using single crystals.28 By contrast, a Cu2O octahedron is so conductive that its linear I−V response under 1 V resembles that of a metal. Surprisingly, Cu2O rhombic dodecahedra bound by the {110} facets are essentially nonconductive even at 5 V, considering that they are much better photocatalysts than Cu2O octahedra and cubes.4 The exact magnitudes of currents at 5 V for the measured Cu2O octahedron, cube, and rhombic dodecahedron are 1745.47, 131.81, and 0.54 nA, meaning that the current flowing through an octahedron is 244 times higher than a cube, and 3232 times higher than a rhombic dodecahedron. These large differences are related to their surface facets, because they have the same B

DOI: 10.1021/acs.nanolett.5b00150 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

essence, the intrinsic conductivity differences imposed by various crystal surfaces means that the band structure of Cu2O cannot be treated the same without regard to the exposed facets but needs to be drawn to show such differences for better understanding of the observed facet-dependent phenomena. To assist in the explanation of the large electrical conductivity differences, calculations were performed to offer electron localization functions (ELF) and electron density diagrams for the Cu2O (111), (100), and (110) planes (see Supporting Information Figure S2). The ELF and electron density were calculated based on DFT and the Perdew− Burke−Ernzerhof generalized gradient approximation (PBEGGA). While differences in the overall ELF values and electron density over a defined area of these surfaces can be distinguished, the results are less insightful for explaining the large electrical conductivity differences measured. However, DOS plots to be presented later can provide good insights into the observed electrical properties. The existence of very large facet-dependent electrical conductivity differences suggests the possibility to obtain an asymmetric I−V curve similar to that for a p−n junction with rectifying effect by simultaneously contacting two different facets in a measurement. Logically the combinations to try should be the {111}/{100} facets and the {111}/{110} facets because of their large conductivity differences. Because Cu2O cuboctahedra can be easily synthesized using our standard procedure, the {111}/{100} combination was tested first.5 Figure 3a,b gives an SEM image of two W probes touching a Cu2O cuboctahedron and a drawing of the probes connected to the {111} and {100} facets of the crystal to complete the electrical circuit. The recorded I−V curves for this particle with different I−V sweep directions are shown in Figure 3c,d. These two curves are not an exact reversal of each other because the contact areas for the two W probes are not the same, and electrical resistance is related to the area of contact. However, similar I−V curves have been obtained, showing that some difference in the probe contact area is acceptable. In fact, the probes cannot be made to be identical. The turn-on voltage is −0.9 V in Figure 3c, and 0.5 V in Figure 3d. The region denoting {111} to {100} in Figure 3c (the negative voltage portion), corresponding to the forward bias domain in the I−V curve of a p−n junction, refers to electrical current flowing into the crystal through the {111} face and exiting the crystal from the {100} face. The region denoting {100} to {111} (the positive voltage portion) refers to the opposite direction of current flow and should correspond to the reverse bias domain in the I−V curve of a p−n junction. The expected diode-like behavior between the {111} and {100} facets is apparent indicating that the two facets of a Cu2O particle can be used as a switching diode. Supporting Information Figure S3 provides SEM image of another Cu2O cuboctahedron and its measured I−V curve. Very similar I−V response has been recorded. Because the {100} facets are moderately conductive at high voltages, a perfect diode behavior with zero current flowing in the direction of the {100} facet to the {111} facet can only be maintained within 2 V. In Figure 3c, the on/off ratio is about 9.8 between 2.5 and −2.5 V. Figure 4 provides another adjusted band diagram of Cu2O to highlight the relative heights of energy barriers encountered with electrical current flowing into the crystal interior through the {111} facet of the crystal or through the {100}/{110} facet. The band diagram shows that a current flowing into the crystal through the {111} facet meets a lower barrier than that through the {100} facet. Therefore, the

crystal structure and the particles are all single-crystalline with no evidence of defects. Such facet-dependent electrical transport properties cannot be made on individual inorganic nanowires, because nanowires either expose only a single side facet, such as hexagonal ZnO and GaN nanowires, or electrical contacts were made on the same facets or mixed facets as in the case of orthorhombic V2O5 nanowires.29−31 Thus, depending on the exposed surface facets of a Cu2O crystal, it can exhibit electrical properties resembling that of a metal, a semiconductor, or an insulator. Our understanding of electrical and possibly other properties of Cu2O and other semiconducting materials needs to be modified. The electrical conductivity results naturally suggest modifications to the typical band diagram of Cu2O. Figure 2

Figure 2. Adjusted band diagram of Cu2O with consideration of relative band edge energies of different crystal surfaces. The known tungsten band energy level was taken into account to construct this modified band diagram for Cu2O. In the diagram, qX is semiconductor electron affinity, qΦs is semiconductor work function, qΦm is metal work function, qΦBp is energy barrier of the W and Cu2O contact, Ec is conduction band energy, Ev is valence band energy, and Ef is Fermi level.

presents a modified band diagram of Cu2O taking into consideration the electrical conductivity responses of different facets of Cu2O.32 Cu2O is a p-type semiconductor, so the conduction and valence band edges near the particle surfaces should bend down when it is in contact with a metal such as tungsten. The Fermi level of Cu2O should equalize with that of tungsten when they are contacted. The known energy barrier between that of Cu2O and W is 0.82 eV, so a voltage is needed to drive current through the crystal.33 The electron conductivity differences are likely due to different barrier heights between tungsten and the three low-index facets of Cu2O. The metal− semiconductor energy barriers are accordingly highest for the {110} facets, medium for the {100} facets, and lowest for the {111} facets. The band diagram is drawn to show the {110} facet being most deviated from the unmodified valence band and conduction band of Cu2O, and the {111} facet is least deviated, so the energy barrier between W and Cu2O is smallest for the {111} facet. Hole transport across the {111} interface is most efficient with its lowest barrier height. Energy level for the {100} facet is midway between the two cases but closer to that of the {111} facet since it is moderately conductive. The exact band edge energies of these facets are not critical, because the measured conductivity depends on the dimensions of the crystals and the contact areas of the two W probes. Both of these factors can vary from one measurement to another. In C

DOI: 10.1021/acs.nanolett.5b00150 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 3. I−V measurements on a single Cu2O crystal with tungsten probes contacting two different facets. (a) SEM image showing two tungsten probes in contact with the {111} and {100} facets of a cuboctahedron. (b) Illustration of the measurement. (c) I−V measurement by sweeping the voltage from −3 to 3 V. The directions of current flow are given. (d) I−V measurement by reversing the sweeping direction from 3 to −3 V. (e) SEM image showing two tungsten probes in contact with the {111} and {110} facets of a rhombicuboctahedron. (f) I−V measurement of the {111} and {110} facets by sweeping the voltage from −4.8 to 4.8 V.

Figure 4. Adjusted band diagram of a Cu2O crystal with its {111} and {100} or {110} facets in contact with tungsten probes. This figure highlights the relative heights of energy barriers encountered when electrical current flows from the {111} side of the crystal or from the {100}/{110} side of the crystal into the crystal interior. The energy barrier is much smaller when current flows into the crystal from the {111} face.

{111} to {100} direction always has a higher current than that for the reversed direction, as shown in Figure 3c,d. Since the {110} facets are nonconductive, it is highly interesting to perform I−V measurements with the probes

contacting the {111} and {110} facets for a more perfect diodelike behavior. This requires the use of Cu2O crystals exposing sufficiently large areas of these facets for accurate probe contacts. We therefore synthesized large Cu2O rhombicuboctaD

DOI: 10.1021/acs.nanolett.5b00150 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters hedra exposing all three low-index facets by following a reported procedure with some modification.25 Figure 3e shows an SEM image of a Cu2O rhombicuboctahedron having a size of 3.6 μm attached to two W probes for the electrical measurements. I−V curves with a much better diode-like behavior have indeed been recorded with the probes contacting the {111} and {110} facets (Figure 3f). For the {110} to {111} side of the I−V curve, zero current can be maintained up to 3.5 V. Figure 4 also shows that current entering the crystal from the {110} facet meets a greater barrier compared to that for the {100} facet, so a larger applied voltage is required to pass charge carriers through this interface. It appears promising to construct operating transistors and switches by making electrical connections to these two facets of Cu2O nanocrystals. By changing the direction of current flow through the crystal while keeping the voltage constant, the unit can be turned on or off. This is a fundamentally different concept for electronic component design. The measured current for the {111} to {110} side of the I−V curve is quite small. We believe this is due to the huge size of the crystal, which can be solved by using much smaller nanocrystals and probes to fabricate functional structures. Supporting Information Figure S4 offers additional I−V curves obtained using different facet contact combinations to this rhombicuboctahedron. The same rectifying I−V response can also be observed using the {111}/{100} and {100}/{110} combinations, suggesting flexibility in the device design to yield functional components. The observed facet-dependent electrical conductivity properties of Cu2O crystals are attributed to a thin surface layer having different degrees of band bending. DFT calculations of density of state were found to be very insightful in showing the existence of these intrinsic differences. Figure 5 gives DOS plots for the (111), (100), and (110) planes of Cu2O consisting of three layers of these planes. Widths of a single layer of these lattice planes are provided, which will be useful for the determination of the thickness of this surface layer responsible for the facet-dependent effects. A DOS profile with an overlapped band structure in the Brillouin zone, typically seen for metals, was obtained with three layers of (111) planes. The DOS plot for three layers of (100) planes shows a semimetal behavior with an overlapped band structure at a special k-point in the Brillouin zone. A band gap appears between 1.61 and 0.96 eV for three layers of (110) planes, so the (110) planes may behave as a semiconductor. The DOS trend is consistent with the experimentally measured electrical conductivity behaviors of different Cu2O facets. Next, DOS plots for different layers of (111), (100), and (110) planes of Cu2O were constructed to determine the thicknesses of the surface layer responsible for the facet-dependent properties (see Supporting Information Figures S5−S7). By varying the number of (111) planes from three to six layers, the initially overlapped band structure begins to show up a band gap with five layers of (111) planes counting from the surface, which corresponds to a thickness of ∼6.2 Å. The gap widens for six layers of (111) planes to approach a band structure expected for bulk Cu2O. For the (100) planes, it takes 5.5 layers or a surface layer thickness of ∼11.7 Å to start showing up a band gap. The (110) planes develop a gap in just three layers, corresponding to a thickness of ∼4.5 Å. Supporting Information Figure S8 summarizes these surface layer thicknesses. For the first time, the surface layer thicknesses of Cu2O crystals producing facet-dependent effects have been determined. Because the widths of these surface layers are all

Figure 5. Density of state plots for the (111), (100), and (110) planes of Cu2O consisting of three layers of these planes. DOS plots of (a) (111), (b) (100), and (c) (110) planes show notable differences, which match their displayed electrical conductivity behaviors. Models for three layers of lattice planes are shown.

below 1.5 nm, facet-dependent properties should be present in Cu2O nanocrystals of all sizes. Conclusion. Novel facet-dependent electrical conductivity behaviors of Cu2O crystals have been measured. The {111} facets are highly conductive, while the {100} facets are only moderately conductive. The {110} facets are nonconductive. Capitalizing on such dramatic conductivity differences, diodelike electrical responses can be achieved on a single Cu2O crystal by making electrical contacts to two different facets. Theoretical calculations yield DOS plots matching the conductivity measurements. Importantly, thicknesses of the surface layers for the {111}, {100}, and {110} facets of Cu2O giving rise to the observed facet-dependent properties have been determined. Functional electronic components fabricated from single polyhedral nanocrystals offer opportunities for new electronics design. Facet-dependent electrical properties should be general and observable in other semiconductors.

■

ASSOCIATED CONTENT

S Supporting Information *

Electron localization function and electron density simulations of the (111), (100), and (110) planes of Cu2O, additional I−V curves, density of state plots for the (111), (100), and (110) planes of Cu2O consisting of different numbers of plane layers, and summary of the thicknesses of the Cu2O surface layers. E

DOI: 10.1021/acs.nanolett.5b00150 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

(27) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865−3868. (28) Yao, Z. Q.; Liu, S. L.; Zhang, L.; He, B.; Kumar, A.; Jiang, X.; Zhang, W. J.; Shao, G. Appl. Phys. Lett. 2012, 101, 042114. (29) Zhai, T.; Liu, H.; Li, H.; Fang, X.; Liao, M.; Li, L.; Zhou, H.; Koide, Y.; Bando, Y.; Golberg, D. Adv. Mater. 2010, 22, 2547−2552. (30) Duan, X.; Huang, Y.; Cui, Y.; Wang, J.; Lieber, C. M. Nature 2001, 409, 66−69. (31) Kind, H.; Yan, H.; Messer, B.; Law, M.; Yang, P. Adv. Mater. 2002, 14, 158−160. (32) Zhang, Z.; Yates, J. T., Jr. Chem. Rev. 2012, 112, 5520−5551. (33) Sze, S. M. Physics of Semiconductor Devices, 2nd ed; Wiley: New York, 1998; p 342.

This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank the Ministry of Science and Technology of Taiwan for the support of this research (101-2628-M-007-006, 1022633-M-007-002, and 101-2221-E-007-078).

■

REFERENCES

(1) Huang, M. H.; Rej, S.; Hsu, S.-C. Chem. Commun. 2014, 50, 1634−1644. (2) Chiu, C.-Y.; Huang, M. H. J. Mater. Chem. A 2013, 1, 8081−8092. (3) Sun, S.; Yang, Z. RSC Adv. 2014, 4, 3804−3822. (4) Huang, W.-C.; Lyu, L.-M.; Yang, Y.-C.; Huang, M. H. J. Am. Chem. Soc. 2012, 134, 1261−1267. (5) Ho, J.-Y.; Huang, M. H. J. Phys. Chem. C 2009, 113, 14159− 14164. (6) Wang, X.; Wu, H.-F.; Kuang, Q.; Huang, R.-B.; Xie, Z.-X.; Zheng, L.-S. Langmuir 2010, 26, 2774−2778. (7) Lyu, L.-M.; Wang, W.-C.; Huang, M. H. Chem.−Eur. J. 2010, 16, 14167−14174. (8) Kim, M.-J.; Cho, Y.-S.; Park, S.-H.; Huh, Y.-D. Cryst. Growth Des. 2012, 12, 4180−4185. (9) Wang, G.; Ma, X.; Huang, B.; Cheng, H.; Wang, Z.; Zhan, J.; Qin, X.; Zhang, X.; Dai, Y. J. Mater. Chem. 2012, 22, 21189−21194. (10) Wu, J.-K.; Lyu, L.-M.; Liao, C.-W.; Wang, Y.-N.; Huang, M. H. Chem.−Eur. J. 2012, 18, 14473−14478. (11) Chen, H.-S.; Wu, S.-C.; Huang, M. H. Dalton Trans. 2015, DOI: 10.1039/C4DT03345K. (12) Wang, Y.; Dai, Q.; Zou, B.; Yu, W. W.; Liu, B.; Zou, G. Langmuir 2010, 26, 19129−19135. (13) Li, C.; Bai, T.; Li, F.; Wang, L.; Wu, X.; Yuan, L.; Shi, Z.; Feng, S. CrystEngComm 2013, 15, 597−603. (14) Wang, W.-C.; Lyu, L.-M.; Huang, M. H. Chem. Mater. 2011, 23, 2677−2684. (15) Chanda, K.; Rej, S.; Huang, M. H. Chem.−Eur. J. 2013, 19, 16036−16043. (16) Li, L.; Nan, C.; Peng, Q.; Li, Y. Chem.−Eur. J. 2012, 18, 10491− 10496. (17) Chanda, K.; Rej, S.; Huang, M. H. Nanoscale 2013, 5, 12494− 12501. (18) Tsai, Y.-H.; Chanda, K.; Chu, Y.-T.; Chiu, C.-Y.; Huang, M. H. Nanoscale 2014, 6, 8704−8709. (19) Kuo, C.-H.; Yang, Y.-C.; Gwo, S.; Huang, M. H. J. Am. Chem. Soc. 2011, 133, 1052−1057. (20) Wang, L.; Ge, J.; Wang, A.; Deng, M.; Wang, X.; Bai, S.; Li, R.; Jiang, J.; Zhang, Q.; Luo, Y.; Xiong, Y. Angew. Chem., Int. Ed. 2014, 53, 5107−5111. (21) Hua, Q.; Cao, T.; Bao, H.; Jiang, Z.; Huang, W. ChemSusChem 2013, 6, 1966−1972. (22) Yang, Y.-C.; Wang, H.-J.; Lin, F.-C.; Huang, J.-S.; Huang, M. H. Nanoscale 2014, 6, 4316−4324. (23) Hsu, S.-C.; Liu, S.-Y.; Wang, H.-J.; Huang, M. H. Small 2015, 11, 195−201. (24) Tan, C.-S.; Hsiao, C.-H.; Wang, S.-C.; Liu, P.-H.; Lu, M.-Y.; Huang, M. H.; Ouyang, H.; Chen, L.-J. ACS Nano 2014, 8, 9422− 9426. (25) Sun, S.; Song, X.; Sun, Y.; Deng, D.; Yang, Z. Catal. Sci. Technol. 2012, 2, 925−930. (26) Hohenberg, P.; Kohn, W. Phys. Rev. 1964, 136, B864. F

DOI: 10.1021/acs.nanolett.5b00150 Nano Lett. XXXX, XXX, XXX−XXX