Playing with Dimensions: Rational Design for Heteroepitaxial p–n

Dec 13, 2011 - A design for a heteroepitaxial junction by the way of one-dimensional wurzite on a two-dimensional spinel structure in a low-temperatur...
0 downloads 5 Views 6MB Size
Download PDF
Letter pubs.acs.org/NanoLett

Playing with Dimensions: Rational Design for Heteroepitaxial p−n Junctions Tae Il Lee,† Sang Hoon Lee,† Young-Dong Kim,† Woo Soon Jang,† Jin Young Oh,† Hong, Koo Baik,† Catherine Stampfl,‡ Aloysius Soon,† and Jae Min Myoung*,† †

Department of Materials Science and Engineering, Yonsei University, Seoul, Korea School of Physics, The University of Sydney, New South Wales, Australia

‡

S Supporting Information *

ABSTRACT: A design for a heteroepitaxial junction by the way of onedimensional wurzite on a two-dimensional spinel structure in a lowtemperature solution process was introduced, and it's capability was confirmed by successful fabrication of a diode consisting of p-type cobalt oxide (Co3O4) nanoplate/n-type zinc oxide (ZnO) nanorods, showing reasonable electrical performance. During thermal decomposition, the 30° rotated lattice orientation of Co3O4 nanoplates from the orientation of β-Co(OH)2 nanoplates was directly observed using high-resolution transmission electron microscopy. The epitaxial relations and the surface stress-induced ZnO nanowire growth on Co3O4 were well supported using the first-principles calculations. Over the large area, (0001) preferred oriented ZnO nanorods epitaxially grown on the (111) plane of Co3O4 nanoplates were experimentally obtained. Using this epitaxial p−n junction, a diode was fabricated. The ideality factor, turn-on voltage, and rectifying ratio of the diode were measured to be 2.38, 2.5 V and 104, respectively. KEYWORDS: Epitaxial growth, spinel cobalt (II,III) oxide nanoplate, wurzite ZnO nanorod, heterogeneous p−n junction

A

Geometrically, 2D single crystalline metal oxide nanoplates are easy to make as an array with a preferred orientation on a substrate due to its unilamellar micro- and macrostructure. If one can efficiently form arrays of it on a substrate, it can be a good free-standing template for certain one-dimensional (1D) metal oxide nanorods, having an epitaxial relationship with it. Based upon this proposition, as a strategy to solve the abovementioned issues, we have introduced a design for heteroepitaxial junctions by the way of 1D wurzite on 2D spinel structures in a low-temperature solution process, and it's capability was confirmed by successful fabrication of a diode consisting of p-type cobalt oxide (Co3O4) nanoplate/n-type zinc oxide (ZnO) nanorods showing reasonable electrical performance. Additionally, the mechanism of structural change during preparation of p-type Co3O4 nanoplates from single crystalline hexagonal β-Co(OH)2 nanoplates was suggested. Furthermore, first-principles calculations concerning the epitaxial growth of the preferred (0001) oriented n-type ZnO nanorods on a large area 2D array of (111) preferred oriented p-type Co3O4 nanoplates were conducted. The reason of the choice for Co3O4 and ZnO as test vehicles is that these two materials can be the most significant and

s a future electronic materials alternative to silicon, single crystalline metal oxide semiconductor nanomaterials have been noticed as building blocks in many applications, for example, energy harvesting,1−3 photonics,4,5 and electronics,6,7 because of their superior properties, multifunctionality, high crystallinity, quantum size effect, high transparency, stability in air, and so on. Recently, to synthesize these nanomaterials, various solution processes including hydrolysis and condensation of metal cations have been developed and matured for their mass production with control of their size and shape.8−10 However, any route to fabricate a functional device based upon these free-standing nanomaterials has not yet been established due to the difficulties in forming arrays of them as a twodimensional (2D) monolayer on a substrate and forming electrical junctions between them or a metal electrode. Discovered by Russell Ohl in 1939, the semiconductor p−n junction has occupied the most important position as an elementary component in current silicon electronics. Therefore, to verify the applicability of single crystalline metal oxide semiconductor nanomaterials in electronics, a reliable fabrication route for a p−n junction device composed of them has to be designed, and also the junction should be epitaxially formed in solution processes to ensure high performance and low process cost. Unfortunately, to date, the fabrication strategy to form this epitaxial junction in a low-temperature solution process has rarely been reported. © 2011 American Chemical Society

Received: August 25, 2011 Revised: December 7, 2011 Published: December 13, 2011 68

dx.doi.org/10.1021/nl202963z | Nano Lett. 2012, 12, 68−76

Nano Letters

Letter

Figure 1. Overall process of building ZnO nanorods on a Co3O4 nanoplate, including 2D arrays of the nanoplates, the bottom side electrical addressing, and the formation of an epitaxial p−n junction.

opposite side of the wall from the dropping position. The study of the mechanism of this method, alcohol spreading on water, is beyond the scope of this communication and was demonstrated in our previous report as an easy route to obtain 2D monolayers of various water compatible nanomaterials. After completing the 2D array of the β-Co(OH)2 nanoplates on water, the array was transferred on to a gold/silicon substrate. For the ohmic contact of p-type Co3O4, the predeposition of gold having high work function (5.1 eV) was adopted. Given that it is often thought that the electrical contact between β-Co(OH)2 nanoplates and gold is poor upon the transfer process, we attempt to improve this electrical contact by thermal annealing at 450 °C for 2 h. This improvement comes about by thermally decomposing βCo(OH)2 to Co3O4, while heating the interface defects between gold and Co3O4 by promoting atomic diffusion of gold to the metal−oxide interface. Therefore, to make ohmic contact of p-type metal oxide semiconductor nanoplates, the procedure, putting their hydroxide phase nanoplate precursors on a noble metal and then annealing for the thermal decomposition of the hydroxide, can be thought to be a simple strategy for bottom side metallization of single crystalline p-type metal oxide nanomaterials. Additionally, the reason why the p-type metal oxide nanoplates is placed first, rather than n-type metal oxide nanomaterials in our fabrication process, is that n-type ohmic metal can be easily oxidized during hydrothermal reaction for epitaxial growth, while p-type ohmic metal is inert against alkali or acid water. Finally, our strategy was completed by aqueous solution-based epitaxial growth of n-type ZnO nanorods on the 2D array of Co3O4 nanoplates at 80 °C. Single crystalline hexagonal β-Co(OH)2 nanoplates were synthesized by the precipitation of an aqueous solution of cobalt(II) chloride hexahydrate (Aldrich 255599) through hexamethylenetetramine (HMTA) hydrolysis in 200 mL of

adequate example to verify the capability of our strategy. They show a different semiconducting type as well as have currently been considered as important multifunctional semiconducting materials for various applications, for example, optoelectronic devices,11 photoelectrolysis cells,12 catalysts,13 sensors,14 solar cells,15 Li ion batteries,16 electrochromic devices,17 and so on. Furthermore, the flexibility in selection of the connecting metal electrode is a merit of our strategy. The connecting electrode can be sophisticatedly tuned with other p-type ohmic materials instead of gold to match exactly the work function of Co3O4 nanoplates; in the case of a n-type nanoplate, a low work function metallic electrode material should be selected, such as aluminum, indium tin oxide, aluminum-doped ZnO, and so on. A schematic diagram shown in Figure 1 illustrates our strategy to build nanorods on individual nanoplates including a 2D array of nanoplates, the electrical addressing, and the formation of an epitaxial p−n junction. As a precursor of p-type Co3O4 nanoplate, the solution of single crystalline hexagonal βCo(OH)2 nanoplates dispersed in ethyl-alcohol(EtOH) was suspended on water. During these dropping events, a 2D array of the β-Co(OH)2 nanoplates was spontaneously formed within a few seconds (see Supporting Information 1). Three sequential steps related to the miscible process of dropping EtOH into the water were employed for the formation of the 2D array: spreading, trapping, and packing. When a droplet of EtOH meets the water, some portion of the EtOH goes into the water, and another portion is spread on the surface because there is a kinetic time delay in the miscible process of EtOH and water. The β-Co(OH)2 nanoplates in the entering portion are mixed with the water; whereas, those in the spreading portion, as the EtOH is gradually dispersed into the water, are trapped by the net force of the surface tension between the air and water. On the surface area confined by the Petri dish, due to the surface pressure from the EtOH spreading, the trapped β-Co(OH)2 nanoplates are stacked two dimensionally at the 69

dx.doi.org/10.1021/nl202963z | Nano Lett. 2012, 12, 68−76

Nano Letters

Letter

Figure 2. Co3O4 nanoplate monolayer preparation. (a) OM image of large scale self-assembled monolayer of hexagonal β-Co(OH)2 nanoplates. (b) Top view SEM image of the monolayer of β-Co(OH)2 nanoplates, an inset shows a magnified hexagonal β-Co(OH)2 nanoplate. (c) Top view SEM image of the monolayer of Co3O4 nanoplates, an inset shows a magnified hexagonal Co3O4 nanoplate. (d) The tendencies of diameter and thickness of β-Co(OH)2 nanoplates as growth time increases, inset photo images of the monolayer deposited on a silicon wafer shows different colors representing different thicknesses. (e) The diameter reduction of β-Co(OH)2 nanoplates after thermal decomposition, inset is 45° tilted view SEM images of the monolayer before and after thermal annealing. (f) The change of XRD peak index of the monolayer before and after thermal annealing, for two different growth conditions, 30 and 150 min.

diameter of β-Co(OH)2 nanoplates reduced to about 300 nm, as shown in Figures 2e. Before and after thermal annealing, the diameter changes and surface conditions with respect to growth time are shown in the insets of Figure 2e. As the growth time increases, the surface of β-Co(OH)2 nanoplates was gradually roughened, and after thermal annealing, the surface of Co3O4 nanoplates was observed to be rough in all cases. To investigate the preferred orientation of the monolayer consisting of β-Co(OH)2 and Co3O4 nanoplates on a substrate, X-ray diffractometry (XRD) indexes were measured for each case, and the results are shown in Figure 2f. From the literature [Joint Committee on Powder Diffraction Standards (JCPDS) 30−0443 (β-Co(OH)2) and JCPDS 74−1657 (Co3O4)], the identities of each XRD peak were identified and referenced. As a result, only {0001} planes were determined for the case of βCo(OH)2 nanoplates as shown in Figure 2f. This means that all β-Co(OH)2 nanoplates were placed with (0001) preferred orientation on the substrate. After thermal annealing at 450 °C for 2 h in air, surprisingly, even though β-Co(OH)2 nanoplates were thermally decomposed into Co3O4 nanoplates, the preferred orientation of the monolayer was maintained by {111} planes of Co3O4 instead of {0001},as shown in Figure 2f. From the XRD peak, changes before and after thermal annealing, it is confirmed that (0001) plane oriented βCo(OH)2 nanoplates were converted into (111) plane oriented Co3O4 nanoplates. Theoretically, volume shrinkage of βCo(OH)2 nanoplates occurs because about 33% of oxygen atoms and 100% of hydrogen atoms were eliminated from βCo(OH)2 during the thermal decomposition. Therefore, to release the mechanical stress generated by this shrinkage, many

mixing solvent, water/EtOH (9:1, v/v). At a fixed mole fraction of divalent cobalt and HMTA (1:1), the temperature (90 °C), concentration (5 mM), and reaction time (from 0.5 to 3 h) were experimental variables. After synthesis, the β-Co(OH)2 nanoplates were purified using three-times centrifugation at 4000 rpm and were then dispersed in EtOH. Figure 2a shows an optical microscope (OM) image of a large area 2D monolayer of the synthesized β-Co(OH)2 nanoplates for 1 h, and the difference in contrast of the image represents the deviation of each β-Co(OH)2 nanoplate thickness. The scanning electron microscope (SEM) result of well-arrayed monolayer of β-Co(OH)2 nanoplates is shown in Figure 2b, and from the inset in Figure 2b, the nanoplates were distributed with about a 100−200 nm gap. After thermal annealing at 450 °C for 2 h in air, single crystalline hexagonal β-Co(OH)2 nanoplates were converted into Co3O4 nanoplates, and the 2D array was maintained as shown in Figure 2c. From the inset in Figure 2c, the reduction of the diameter of each hexagonal βCo(OH)2 nanoplate was observed. As growth time increased from 0.5 to 3 h, the diameter and thickness of the β-Co(OH)2 nanoplates increased as shown in Figures 2d. The lateral growth of β-Co(OH)2 nanoplates was rapidly saturated within 1 h, while the vertical growth was linearly proportional to the time. The visible images of the transferred monolayer of the nanoplates on 1 × 1 cm2 bare silicon wafers are exhibited in the insets of Figure 2d. The changes of color mean a difference in thickness with growth time, and a single color through the entire substrate is represented by the uniformity of the β-Co(OH)2 monolayer. Interestingly, after thermal decomposition, the average 70

dx.doi.org/10.1021/nl202963z | Nano Lett. 2012, 12, 68−76

Nano Letters

Letter

Figure 3. A bright field image and six HRTEM images to investigate the atomic structure change after thermal decomposition of β-Co(OH)2 nanoplates. From this analysis, it was revealed that the lattice orientation of Co3O4 nanoplates was 30° rotated with respect to the direction of each side representing the lattice orientation of β-Co(OH)2 nanoplates.

between the side direction and the (111) spinel lattice orientation of Co3O4 nanoplates, the hexagonal lattice orientation on the (111) plane of Co3O4 nanoplates near each side, from A to F, were determined using high-resolution transmission electron microscopy (HRTEM). From the HRTEM results, it was determined that near the six sides of Co3O4 nanoplates, the hexagonal lattice of the (111) spinel plane was presented with an interplanar spacing of 2.86 Å between oxygen ions, as shown in Figure 3. Interestingly, we observed that the orientation of the hexagonal lattice on (111) spinel-like of Co3O4 nanoplates is exactly rotated by 30° with respect to the orientation of the hexagonal lattice on (0001) brucite-like of β-Co(OH)2 nanoplates, as estimated from the direction of the side of a hexagon shape shown in the bright field image of Figure 3. The 30° rotation shown in Figure 3 is thought to be a clue to reveal the mechanism of the kinetic process from brucite (0001) into spinel (111). The formation of Co3O4 (i.e., Co2+Co3+2O4) from Co(OH)2 is achieved by partially oxidizing the Co2+ in Co(OH)2. In this process, 50% of the octahedral sites are oxidized to the +3 state, while oneeighth of the tetrahedral sites remain in the +2 state. Due to this partial oxidation to its so-called mixed valence oxide, a chemical and structural reordering is observed, manifesting in a

nanocracks are formed in Co3O4 nanoplates, as shown in the inset of Figure 2c. Despite the fact that the volume changes of β-Co(OH)2 nanoplates are dominated by the above-mentioned dehydration, the macroscopic diameter changes of β-Co(OH)2 nanoplates, as shown in Figure 2e, are thought to be related to the microstructural change from brucite-like (0001) β-Co(OH)2 nanoplates with minimum oxygen ions (interplanar spacing db(0001) = 2.329 and db(1000) = 3.182 Å) to spinel-like (111) Co 3 O 4 nanoplates with minimum oxygen ions (interplanar spacing ds(111) = 2.333 and ds(220) = 2.858 Å). To critically assess the microstructural changes when βCo(OH)2 is thermally transformed to Co3O4, we measure and compare the hexagonal lattice orientation of our Co3O4 nanoplates (which is topologically derived from β-Co(OH)2) with that of ideal β-Co(OH)2 (See Figure 3). Single crystalline β-Co(OH)2 has a perfect hexagonal plate-shape expanded from hexagonal close-packed system of O2− ions with a Co2+ ion in its octahedral interstitial site (CoO6) and two H+ ions in their tetrahedral interstitial site (HO4), therefore the microscopic inplane hexagonal lattice orientation can be estimated from the macroscopic hexagonal shape. The bright field image, as shown in the center of Figure 3, represents a perfect hexagonal outline originated from the hexagonal lattice orientation of single crystalline β-Co(OH)2. To directly investigate the relationship 71

dx.doi.org/10.1021/nl202963z | Nano Lett. 2012, 12, 68−76

Nano Letters

Letter

Figure 4. (a) A bright field image of the cross section of ZnO nanorod grown on Co3O4 nanoplates. (b) HRTEM images of region A in the bright field image. (c) Selective area electron diffraction pattern of region B in the bright field image. (d) Schematics describing the 10° tilting between ZnO(0001) and Co3O4 (111) plane. (e) HRTEM images of region B in the bright field image, the 10° tilting is clearly observed.

with reported experimental and other theoretical values in the literature.24,25 The interface structures of ZnO(0001)/Co3O4(111) are modeled using a symmetric slab with a vacuum region of 12 Å. To construct the atomic geometry of the interface, we consider two models based on previous experimental and theoretical propositions and adapted it for this work. First, based on a thermodynamic stability study of various Co3O4 surfaces,24 it is shown that due to its complex bulk structure, there exist many possible surface terminations for the Co3O4(111) surface, with the Cot (i.e., Co at the tetrahedral site, referenced to bulk Co3O4) termination being most favorable under an oxygen atmosphere. On the other hand, using the plasma-assisted molecular beam approach, epilayers of ZnO (for both O- and Zn-polar films) have been grown epitaxially on MgAl2O4,26−29 which has a similar bulk spinel structure as Co3O4. Based on HRTEM images, the authors propose the epitaxial relationship: ZnO[011̅ 0 ] ∥ MgAl 2 O 4 [1̅ 1̅ 2 ] and ZnO[21̅ 1̅ 0 ] ∥ MgAl2O4[1̅1̅0]. It also concluded that the surface of the asprepared substrate should be oxygen terminated.26 We adopt the similar epitaxial relationship for ZnO/Co3O4, given that both Co3O4 and MgAl2O4 have the same bulk spinel structure. This epitaxial relationship of ZnO(0001) and Co3O4(111) is further verified by our own experimental results: The microstructure of a prismatic cross section plane of a ZnO nanorod grown on a Co3O4 platelet is determined, as shown in

30°-off axis rotation from the original Co(OH)2 hexagonal template. For the theoretical prediction of epitaxial relations and surface stress-induced ZnO nanowire growth on Co3O4, our first-principles calculations are based on density functional theory (DFT) as implemented in the Vienna ab initio simulation package (VASP 5.2).18,19 We employ the projector augmented-wave method (PAW)20,21 and the generalizedgradient approximation (GGA) due to Perdew, Burke, and Ernzerhof (PBE)22 for the exchange−correlation functional. The wave functions are expanded in a plane-wave basis set with an energy cutoff of 500 eV. For the Brillouin-zone integrations, a Monkhorst−Pack k-point grid of (6 × 6 × 6) is used for bulk Co3O4, and a γ-centered k-point grid of (12 × 12 × 8) for bulk ZnO, while a γ-centered k-point grid of (4 × 4 × 1) is used for the interface structures of ZnO(0001)/Co 3O4(111). A Methfessel−Paxton smearing of 0.1 eV is used to improve the convergence, and the total energy is extrapolated to zero temperature. For all calculations (except for bulk ZnO), a spinpolarization approach is employed. In addition, we include an on-site Coulomb correction within the DFT+U approach23 for both the Co 3d and Zn 3d states, using an effective U of 3.5 and 5.0 eV, respectively.24,25 The calculated bulk lattice parameter of Co3O4 is 8.153 Å, while those of ZnO are 3.236 and 5.214 Å (for the a and c lattice constants). These are in good agreement 72

dx.doi.org/10.1021/nl202963z | Nano Lett. 2012, 12, 68−76

Nano Letters

Letter

Figure 5. Atomic structures of proposed interface models (side view): (a) O−Coo−O−Zn−O and (b) O−Cot/Coo−O−Zn−O. In the ZnO layer, large white spheres are labeled as zinc atoms, while in the Co3O4 layer, large turquoise (lighter) spheres are labeled as tetrahedrally coordinated Co atoms (Cot), and the octahedrally coordinated Co atoms (Coo) are then shown as blue (darker) large spheres. Oxygen atoms are shown as small red (dark) spheres.

stability of different structures. Given that both interface structures are nonstoichiometric to their corresponding bulk counterparts, the formation energies of these structures will have an explicit dependence on the chosen chemical potential value. Here we consider γ for so-called oxygen-rich and oxygenlean (or Co-rich) conditions. Under oxygen-rich conditions, we equate μO = 1/2EO2 (half the total energy of the O2 molecule), bulk and under oxygen-lean conditions, we set μCo = ECo , which is the total energy of a bulk Co atom. At thermodynamic equilibrium, we assume μCo3O4 = 3μCo + 4μO and μZnO + μZn + μO, where μCo3O4 and μZnO are the chemical potentials of bulk Co3O4 and ZnO, respectively. Thus for the O-rich limit, we have μCo = 1/2(ECo3O4 − 2EO2) and μZn = EZnO + 1/2 EO2, where ECo3O4 and EZnO are the total energies of bulk Co3O4 and ZnO. Similarly, at the O-lean limit, we define μO = 1/4(ECo3O4 − 3EO) and μZn = EZnO + 1/4(ECo3O4 − 2EO) . For the zinc chemical potential, we have also tested taking it from a reservoir of Zn− Co alloy but found it to be more favorable when considered from the ZnO reservoir. Under O-rich conditions, the interface formation energy of O−Coo−O−Zn−O is found to be 0.29 eV/Å2, while under Olean conditions, it is 0.46 eV/Å2. The corresponding values for O−Cot/Coo−O−Zn−O are 0.30 and 0.47 eV/Å2, respectively. Our calculated interface formation energies are clearly endothermic, but it is not an insurmountable energy barrier to overcome. It is also clear that both structures have fairly similar formation energies and are more favored under O-rich growth conditions. It has been reported that when ZnO is grown on MgAl2O4, the interface experiences a large lattice mismatch of about 14% which is considerably large.27 It has been suggested that such a large lattice mismatch induces a large compressive stress at the interface, leading to so-called columnar growth of the deposited material. This has been proposed as a possible mechanism for oxide nanowire growth on its native metal substrate (e.g., CuOx/Cu, FeOx/Fe, etc.),31,32 arguing that due to their structural and density differences and associated large volumetric changes upon oxidation induces substantial accumulated stresses at the metal/oxide interface. Once a critical limit is reached, these stresses are often relieved by forming dislocations along appropriate crystal directions (e.g., the maximum shear strength exists along this direction). Atomic migration then occurs, stacking in the corresponding

Figure 4a,b. From the HRTEM image of ZnO nanorod (region A), it is convinced that the single crystalline ZnO nanorod is well grown in [0001] direction, and its quality is reasonable to use in LED applications. Moreover, the microstructure of Co3O4−ZnO platelet−nanorod heterojunction (region B) is resolved using the selective area electron diffraction (SAED) pattern and HRTEM image as shown in Figure 4c−e. From the SAED pattern of a 100 nm diameter of the focused electron beam diffraction near region B, a typical epitaxial relationship with an about 10° tilting between Co3O4(111) and ZnO(0001) is found, as depicted in Figure 4c,d. In view of nanoscale lattice detection, this tilting is also observed from the HRTEM image of Figure 4e. It is thought that this small tilting is adopted to release the stress by the lattice mismatching between Co3O4(111) and ZnO(0001) during their epitaxial growth. The induced stress upon growth will be addressed by our theoretical results later in the text. From the information described above, we construct two interface models, preserving the hexagonal symmetry of the substrate oxide Co3O4(111), as shown in Figure 5. The first interface structure between ZnO and Co3O4 is bridged by a layer of oxygen, sandwiched between a layer of Zn atoms and a layer of Coo (i.e., Co at the octahedral site, referenced to bulk Co3O4). Thus, we label this structure as O−Coo−O−Zn−O (see Figure 5a). The second structure also consists of a layer of oxygen atoms placed in between a layer of Zn atoms and Co atoms. However for the Co atomic layer, it consists of both Cot and Coo with a ratio of 2:1. This is then labeled as O−Cot/ Coo−O−Zn−O (see Figure 5b). To obtain the optimized atomic geometries of these two models, we allow full atomic relaxation for all atoms in the slab, and residual forces are minimized to be less than 0.01 eV/Å. To evaluate the stability of these interface structures, we consider their Gibbs free energy of formation as defined by

γ=

1 [Etot − 2A

∑ Niμi] i

where Etot is the total energy of the interface structure, Ni and μi are the numbers of atomic species i and its corresponding atomic chemical potential, respectively, and A is the basal interface area of the slab.30 Here, DFT total energies are used instead of the Gibbs free energies, assuming that the vibrational entropic contributions due to finite temperature are similar for both structures and will not drastically change the relative 73

dx.doi.org/10.1021/nl202963z | Nano Lett. 2012, 12, 68−76

Nano Letters

Letter

Figure 6. Epitaxial growth of ZnO nanorods on Co3O4 nanoplates. Temporal growing feature of ZnO nanorods: (a) 10, (b) 30, (c) 60, (d) 180, and (e) 300 min, 45° tilted view. (f) Large area vertically aligned ZnO nanorods on the monolayer of Co3O4 nanoplates, 45° tilted view; and (g) 30° rotated orientation (red hexagon) of ZnO nanorods with respect to the orientation of hexagonal frame (yellow hexagon) of Co3O4 nanoplates, 45° tilted view.

Because ZnO nanorods are epitaxially grown based on the (111) plane of Co3O4 nanoplates, the orientation of the lattice in (0001) plane of ZnO nanorods has to be coincident with that in the (111) plane of Co3O4. Therefore, the 30° rotating relationship between the ZnO nanorod’s hexagon and the Co3O4 nanoplate’s hexagonal outline inherited from β-Co(OH)2 is a natural consequence. Finally, to verify the capability of our strategy for the large scale epitaxial p−n junction based up on single crystalline metal oxide semiconductor nanomaterials, a heteroepitaxial junction diode consisted of p-type Co3O4 nanoplate/n-type ZnO nanorods was fabricated, as shown in the inset in Figure 7.

plane to maintain a nanowire shape. From our calculations, we find that the ZnO layers near the interface are highly compressed to match the basal area of Co3O4(111). The compressed values of the a and c lattice parameters of ZnO are found to be about 2.90 and 5.52 Å, resulting in a large lattice compression (∼ −10%) in the basal plane and a considerable lattice dilation (∼6%) in the vertical direction. To estimate the strain energy due to lattice distortion in the compressed ZnO layers, we take the total energy difference between an artificially strained crystal of ZnO (with a = 2.90 and c = 5.52 Å) and that of the fully relaxed one. In this case, the associated strain energy is found to be 0.428 eV per ZnO formula unit. This large compressive stress in the basal plane could then account for the slight tilting (∼10°) seen in our experiments and would also indicate that the tendency of ZnO growing in the (0001) direction on Co3O4(111) substrate template should be favored, as clearly seen in our experiments. We note that this proposed stress-promoted growth mechanism might not be the full story of our experiments as the presence of solvating ligands could also play an influential role in modifying the relative surface stability, hence directly affecting the growth kinetics to promote preferential growth planes. Coincided with the theoretical prediction, (0001) preferred oriented ZnO nanorods were epitaxially grown on (111) plane of Co3O4 nanoplates, as shown in Figures 6a−g. The ZnO nanorods were grown in a 20 mL mixed aqueous solution of zinc nitrate hexahydrate (Aldrich 96482) (70 mM) and HMTA (70 mM) at 80 °C. The temporal growth features of ZnO nanorods on the Co3O4 nanoplates were shown in Figures 6a− e. In the initial state for 10 min, tiny ZnO nanorods were observed at several points on the Co3O4 nanoplates. As growth time increases, the shape of ZnO nanorods became thicker and longer than their initial state. Figure 6f shows vertically grown ZnO on the large area for 300 min. As observing results from the hexagonal cross sections of all ZnO nanorods on the hexagonal shape Co3O4 nanoplates shown in Figure 6f, the 30° rotating relationship between hexagons for ZnO and Co3O4 nanoplates were determined, as shown as red and yellow hexagons in Figure 6g and Supporting Information, 2-1.

Figure 7. I−V characteristics of Al/n-ZnO nanorods/p-Co3O4 nanoplates/Au, the device parameters, ideality factor, turn-on voltage, and rectifying ratio are 2.38, 2.5 V, and 104, respectively.

To complete this device, a toluene solution 10 wt % of poly(methyl methacrylate) (PMMA) was covered over the ZnO nanorods, shown in Figure 6f, using spin coating at 3000 rpm for 40 s. The top side of ZnO nanorods were exposed using 0.13 Torr pure oxygen plasma at 100 W direct current pulse power for 20 s and then 200 nm aluminum as ohmic electrode for ZnO nanorods was deposited on it using thermal 74

dx.doi.org/10.1021/nl202963z | Nano Lett. 2012, 12, 68−76

Nano Letters evaporator. The electrical characteristics of this device are exhibited in Figure 7, and turn-on voltage and rectifying ratio were measured as 2.5 V and 104, respectively. Using the plot of the ln(current) vs voltage, the ideality factor was calculated as 2.38 within the voltage range 0.01 < V < 0.5 V. To investigate the electrical property of each junction of this device, we determined I−V characteristics of Au/Co3O4 platelet/Au, Al/ ITO/ZnO nanorod/Al, and Al/ZnO nanorod/Au. From the linearity of the I−V characteristics for Au/Co3O4 platelet/Au and Al/ITO/ZnO nanorod/Al shown in Supporting Information 2-2a,b, it is confirmed that the junctions of Al/ZnO nanorod and Au/Co3O4 platelet are ohmic contacts. Additionally, the I−V characteristics of Al/ZnO nanorod/Au shown in Supporting Information 2-2c represent a typical electrical junction property of metal/semiconductor Schottky diode, in which the turn-on voltage is lower than 1 V. Therefore, it is convinced that the rectifying characteristics of p−n diode, ZnO nanorod/Co3O4 platelet shown in Figure 7 is different from that of Schottky diode of ZnO nanorod/Au. In summary, a design for a heteroepitaxial junction by the way of 1D wurzite on a 2D spinel structure in a lowtemperature solution process was introduced, and it's capability was confirmed by successful fabrication of a diode consisting of p-type cobalt oxide (Co3O4) nanoplate/n-type zinc oxide (ZnO) nanorods, showing reasonable electrical performance. From the thermal decomposition of the 2D arrayed βCo(OH)2 nanoplates, (111) spinel Co3O4 nanoplates were obtained, and the 30° rotated lattice orientation of Co3O4 nanoplates from the orientation of β-Co(OH)2 nanoplates was directly observed using HRTEM. This nanoscale structural change was revealed by the macroscopic geometrical relation between the hexagonal ZnO nanorods and that of hexagonal Co3O4 nanoplates. The epitaxial relations and surface stressinduced ZnO nanowire growth on Co3O4 were well supported using the first-principles calculations. Coincided with the theoretical prediction, (0001) preferred oriented ZnO nanorods epitaxially grown on the (111) plane of Co3O4 nanoplates were experimentally obtained. Using this epitaxial p−n junction, a diode was fabricated. The ideality factor, turn-on voltage, and rectifying ratio of the diode were measured to be 2.38, 2.5 V, and 104, respectively. We expected that the epitaxial p−n junction of p-type Co3O4 nanoplate/n-type ZnO nanorods may have future application in electronic devices. Furthermore, we believe that our design rule, 1D wurzite on 2D spinel using solution processes, can be adopted as a fabrication strategy for various single crystalline metal oxide nanomaterials based heteroepitaxial junction devices.

■

ACKNOWLEDGMENTS

■

REFERENCES

Letter

This work was supported in part by the World Class University (WCU) program through the National Research Foundation of Korea funded by the Ministry of Education, Science, and Technology (R32-20031). This work is also supported by the POSCO academic−industrial cooperation program (2011Z088). W.S.J., J.Y.O., and H.K.B. were supported by the National Research Foundation grant 2011-0029856 by funds from MEST. Y.D.K., C.S., and A.S. would like to acknowledge support from the National Research Foundation of Korea (grant no. 2011-8-0952) and the Australian Research Council (ARC). Computational resources have been provided by the Australian National Supercomputing Facility (NCI). We are grateful for discussions on the HRTEM analysis with S. W. Sohn and Prof. D. H. Kim in Center for Non-crystalline Materials in Yonsei University.

(1) Wang, X. D.; Song, J. H.; Liu, J.; Wang, Z. L. Science 2007, 316, 102. (2) Wang, Z. L.; Song, J. H. Science 2006, 312, 242. (3) Law, M.; Greene, L. E.; Johnson, J. C.; Saykally, R.; Yang, P. Nat. Mater. 2005, 4, 455. (4) Huang, M. H.; Mao, S.; Feick, H.; Yan, H.; Wu, Y.; Kind, H.; Weber, E.; Russo, R.; Yang, P. Science 2001, 8, 5523. (5) Yan, R.; Gargas, D.; Yang, P. Nat. Photonics 2009, 3, 569. (6) Li, Y.; Qian, F.; Xiang, J.; Lieber, C. M. Mater. Today 2006, 9, 18. (7) Ju, S.; Facchetti, A.; Xuan, Y.; Liu, J.; Ishikawa, F.; Ye, P.; Zhou, C.; Marks, T. J.; Janes, D. B. Nature Nanotechnol. 2007, 2, 378. (8) Park, J.; An, K.; Hwang, Y.; Park, J.; Noh, H.; Kim, J.; Park, J.; Hwang, N.; Hyeon, T. Nat. Mater. 2004, 3, 891. (9) Jana, N. R.; Chen, Y.; Peng, X. Chem. Mater. 2004, 16, 3931. (10) Ding, Y.; Zhang, G.; Wu, H.; Hai, B.; Wang, L.; Qian, Y. Chem. Mater. 2001, 13, 435. (11) (a) Jeong, M. C.; Oh, B. Y.; Ham, M. H.; Lee, S. W.; Myoung, J. M. Small 2007, 3, 568. (b) Bao, J.; Zimmer, M. A.; Capasso, F.; Wang, X.; Ren, Z. F. Nano Lett. 2006, 6, 1719. (12) Walter, M. G.; Warren, E. L.; McKone, J. R.; Boettcher, S. W.; Mi, Q.; Santori, E. A.; Lewis, N. S. Chem. Rev. 2010, 110, 6446. (13) Yan, L.; Zhang, X.; Ren, T.; Zhang, H.; Wang, X.; Suo. J. Chem. Commun. 2002, 8, 860. (14) (a) Wang, X.; Zhou, J.; Song, J.; Liu, J.; Xu, N.; Wang, Z. L. Nano Lett. 2006, 6, 2768. (b) Wan, Q.; Li, Q. H.; Chen, Y. J.; Wang, T. H.; He, X. L.; Li, J. P.; Liu, C. L. Appl. Phys. Lett. 2004, 84, 3654. (c) Ando, M.; Kobayashi, T.; Lijima, S.; Haruta, M. J. Mater. Chem. 1997, 7, 1779. (15) Levi-Clement, C.; Tena-Zaera, R.; Ryan, M. A.; Katty, A.; Hodes, G. Adv. Mater. 2005, 17, 1512. (16) (a) Poizot, P.; Laruelle, S.; Grugeon, S.; Dupont, L.; Tarascon, J. M. Nature 2000, 407, 496. (b) Li, W. Y.; Xu, L.; Chen, J. Adv. Funct. Mater. 2005, 15, 852. (17) Svegl, F.; Orel, B.; Hutchins, M. G.; Kalcher, K. J. Electrochem. Soc. 1996, 143, 1532. (18) Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47, 558; 1994, 49, 14251. (19) Kresse, G.; Furthmüller, J. Comput. Mater. Sci. 1996, 6, 15. (20) Kresse, G.; Furthmüller, J. Phys. Rev. B 1996, 54, 11169. (21) Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758. (22) Blöchl, P. E. Phys. Rev. B 1994, 50, 17953. (23) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865. (24) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.; Sutton, A. P. Phys. Rev. B 1998, 57, 1505. (25) Montoya, A.; Haynes, B. S. Chem. Phys. Lett. 2011, 502, 63. (26) Walsh, A.; Silva, J. D.; Wei, S. H. Phys. Rev. Lett. 2008, 100, 256401.

■

ASSOCIATED CONTENT * Supporting Information Supporting Information 1 (a real time video file related to the spontaneous formation of a 2D array of the β-Co(OH)2 nanoplates on water surface); 2-1 (30° rotating relationship between hexagons for ZnO and Co3O4 nanoplates); and 2-2 (I−V characteristics and schematics of Au/Co3O4 platelet/Au, Al/ITO/ZnO nanorod/Al, and Al/ZnO nanorod/Au). This material is available free of charge via the Internet at http:// pubs.acs.org. S

■

AUTHOR INFORMATION Corresponding Author *E-mail: [email protected]. Telephone: (82-2)21232843. 75

dx.doi.org/10.1021/nl202963z | Nano Lett. 2012, 12, 68−76

Nano Letters

Letter

(27) Chen, Y.; Hong, S.-K.; Ko, H. J.; Nakajima, M.; Yao, T.; Segawa, Y. Appl. Phys. Lett. 2000, 76, 245. (28) Zeng, Z. Q.; Liu, Y. Z.; Yuan, H. T.; Mei, Z. X.; Du, X. L.; Jia, J. F.; Xue, Q. K.; Zhang, Z. Appl. Phys. Lett. 2007, 90, 081911. (29) Zeng, Z. Q.; Liu, Y. Z.; Yuan, H. T.; Mei, Z. X.; Du, X. L.; Jia, J. F.; Xue, Q. K.; Zhang, Z.; Salamo, G. J. J. Phys. D: Appl. Phys. 2010, 43, 085301. (30) Soon, A.; Todorova, M.; Delley, B.; Stampfl, C. Phys. Rev. B 2007, 75, 125420. (31) Yuan, L.; Wang, Y.; Mema, R.; Zhou, G. Acta Mater. 2011, 59, 2491. (32) Rackauskas, S.; Nasibulin, A. G.; Jiang, H.; Tian, Y.; Kleshch, V. I.; Sainio, J.; Obraztsova, E. D.; Bokova, S. N.; Obraztsov, A. N.; Kauppinen, E. I. Nanotechnology 2009, 20, 165603.

76

dx.doi.org/10.1021/nl202963z | Nano Lett. 2012, 12, 68−76