Face-selective tungstate ions drive zinc oxide nanowire growth direction and dopant incorporation

Tailoring the elemental doping of inorganic nanowires remains an important challenge due to complex dopant incorporation pathways. Here we report that the face-selectivity of tungstate ions controls growth direction and dopant incorporation of hydrothermal zinc oxide nanowires. The introduction of tungstate ions on nanowire surface during synthesis unexpectedly enhances nucleation at sidewall 101¯0\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left\{ {10\bar 10} \right\}$$\end{document} planes, while dopant incorporation occurs only on (0001) planes. This conflicting face-selective behavior leads to inhomogeneous dopant distribution. Density functional theory calculations reveal that the face-selective behavior can be interpreted in terms of the effect of coordination structure of the tungstate ions on each zinc oxide crystal plane. In addition, we demonstrate a rational strategy to control the morphology and the elemental doping of tungsten-doped zinc oxide nanowires. Controlling the growth processes of nanowires is vital for tailoring their properties. Here, the presence of tungstate ions on specific surface planes of zinc oxide nanowires causes nanowire growth and chemical doping along specific crystal planes.

H ydrothermal synthesis of inorganic nanomaterials 1-4 offers wide range of applications [5][6][7][8][9][10][11] due to the scalability and the compatibility with various substrates and material processes [12][13][14] originating from its low process temperature which is not available by vapor-phase synthesis [15][16][17][18] . The functions of inorganic nanomaterials strongly depend on their geometry. Therefore, most of previous researches have been devoted to establish predictable models and strategies for controlling the morphology of nanomaterials. As for the hydrothermally synthesized zinc oxide (ZnO) nanowires, the strategies based on the electrostatic interaction between metal ions and crystals 19 , the organic ligand based surface capping 20 , the ligandexchange effect 21 , and the difference in critical nucleation concentrations on crystal planes so-called "concentration window" [22][23][24] have been demonstrated and successfully manipulated the nanowire morphology.
On the other hand, elemental doping is a typical approach for aggressively tuning the functions of inorganic nanomaterials. The elemental doping to nanowires has been intensively investigated in vapor-phase synthesis 25,26 . The feasibility of elemental doping for hydrothermal ZnO nanowires has been also demonstrated together with successful modifications of electrical, optical, and surface chemical properties [27][28][29][30][31] . However, there still remains much room for investigating the elemental doping in hydrothermal nanowire synthesis. For example, the doping often accompanies the variations of anisotropic crystal growth of ZnO nanowires [32][33][34][35] and in other case, the metal impurities solely serve as counter ions rather than being incorporated into the nanowires 19,36 . These results plausibly originate from the variety of chemical reactions and electrostatic effects in the hydrothermal system 19,[37][38][39] , which are not seen in vapor-phase synthesis. Furthermore, the dopant distribution in an individual hydrothermal nanowire has not been discussed so far, while its importance has been highlighted in a vapor-liquid-solid (VLS) nanowire 25 . As such, the elemental doping in hydrothermal synthesis is far from comprehensive understanding and controlling over the morphology and the dopant distribution of hydrothermal nanowire is a challenging issue.
In the present research, we report the unusual face-selectivity of tungstate ion on growth direction and dopant incorporation of hydrothermal ZnO nanowire. Systematic experiments and simulations reveal that tungstate ions (WO 4 2− ) face-selectively act as surfactant and dopant on different crystal planes according to their coordination structure. Furthermore, the dopant distribution in an individual hydrothermal ZnO nanowire is unveiled by means of scanning transmission electron microscopy (STEM) observation with energy dispersive x-ray spectroscopy (EDS) and three-dimensional atom probe (3DAP) analysis. Based on above results, a predictable model and a rational strategy to control over the morphology and the dopant distribution of hydrothermal ZnO nanowires are demonstrated.

Results and discussion
Tungsten is an appropriate dopant species for ZnO in terms of the ionic radius (Zn 2+ 60 pm, W 6+ 42 pm) and the coordination number (Zn 2+ (IV) and W 6+ (IV)). In fact, the improvement of molecular sensing property of ZnO thin film was previously demonstrated by W doping 40,41 . The nanowire growth was conducted by a seed-assisted hydrothermal approach in aqueous solution containing Zn(NO 3 ) 2 (Zn precursor), hexamethylenetetramine (HMTA), polyethyleneimine (PEI), and Na 2 WO 4 (W precursor). A 50-nm-thick ZnO seed layer coated on Si/SiO 2 substrate was utilized for the growth. The temperature and the time for the growth were 95°C and 5 h, respectively. For all experiments, the concentrations of HMTA 25 mM and PEI 1.25 mM were kept constant (the role of PEI is shown in Supplementary Figs. 1, 2 and Supplementary Notes 1, 2). Figure 1a shows side view and top view scanning electron microscopy (SEM) images of ZnO nanowires when varying Na 2 WO 4 concentration (C W ) 0-0.25 mM under the constant Zn(NO 3 ) 2 concentration (C Zn ) 25 mM. When increasing C W , the nanowire tends to be a platelet form. The statistic nanowire length and diameter in Fig. 1b show that the diameter exceeds the length above C W 0.25 mM. This indicates that the preferential growth direction of ZnO nanowires changes from [0001] to 10 10 ½ just by adding 1% of impurity in the growth solution. Figure 1c shows the low and high magnification transmission electron microscopy (TEM) images and the electron diffraction patterns for both nanowire (C Zn 25 mM, C W 0 mM) and nanoplatelet (C Zn 25 mM, C W 0.125 mM) structures. Both samples show single crystallinity and hexagonal wurtzite structure, indicating that the crystallographic features of ZnO nanostructure are not varied by the W addition. Figure 1d, e shows the x-ray photoelectron spectroscopy (XPS) spectra (Zn 2p and W 4f) and the estimated W concentration (at%) of ZnO nanostructures synthesized with C Zn 25 mM and various C W . When increasing C W , W 4f peaks tend to be larger while Zn 2p peaks remain almost unchanged, leading to the systematic increase of W concentration. Figure 1f, g shows the x-ray diffraction (XRD) patterns and the calculated d values of ZnO nanostructures synthesized with C Zn 25 mM and various C W . The ZnO (0004) peak shifts to be higher angle and the d value decreases with increasing C W , which is plausibly due to the substitution of Zn 2+ with W 6+ . The results of XPS and XRD show that W is incorporated into ZnO nanostructures. Alternatively, these results represent that controlling morphology and W doping of hydrothermal ZnO nanowires is quite difficult.
For exploring a way to overcome above dilemma between the dopant incorporation and the morphological variation, we first consider why the preferential growth direction of nanowires is changed by W addition. In previous studies, the formation of nanoplatelet structure was interpreted in terms of (i) the growth inhibition by the electrostatic adsorption of counter ions 19 and/or the chemisorption of inorganic and organic ligands 20,36 on ZnO (0001) plane, (ii) the interfacial energy reduction of ZnO (0001) plane via chemical cappings 42 , and (iii) the dislocation-driven growth by lowering the supersaturation degree [43][44][45][46] (For the detailed review, see Supplementary Table 1). In order to specify the possible metal counter ion species, we performed thermodynamic calculation of equilibrium concentration of ionic species under the given growth condition. We found that WO 4 2− ion is the most plausible adsorption species for the growth inhibition (Supplementary Fig. 3 and Supplementary Note 3). To understand the contribution of WO 4 2− ions for the nucleation phenomenon, next we examine the Zn precursor concentration dependence on ZnO nanowire growth with various C W . Previous our study demonstrated that there exist two threshold concentrations corresponding to the critical concentrations for nucleation on (0001) plane and 10 10 ð Þ plane [22][23][24] . By evaluating the shifts of critical concentrations with varying C W , the influence of WO 4 2− ions for the nucleation events on ZnO crystal planes can be identified. Figure 2a, b shows the Zn precursor concentration dependences on length and diameter of ZnO nanowires with C W 0-0.25 mM. The SEM images for the series of C Zn are shown in Supplementary Fig. 4. The triangle marks in the graphs represent the critical concentration for nucleation. These results are summarized in Fig. 2c. The critical concentration on 10 10 ð Þplane (i.e., sidewall) tends to be lower while that on (0001) plane (i.e., nanowire top) is almost unchanged with increasing C W . This indicates that WO 4 2− ions rather enhance the nucleation on 10 10 ð Þ plane than suppress the nucleation on (0001) plane. Note that the growth inhibition on (0001) plane seems to be negligible at lower C W range, but it becomes significant by increasing C W to 0.25 mM. We found that the nucleation on (0001) plane would be enhanced with decreasing C Zn below the    critical concentration on 10 10 ð Þplane although the W/Zn ratio in precursor becomes higher. Thus, above results indicate that the variation of preferential growth direction of ZnO nanowires by W addition cannot be explained by the existing models and rather suggest the selective nucleation enhancement on 10 10 ð Þ plane. In general, dopant incorporation event is accompanied with crystal growth. Therefore we assume that W incorporation mainly occurs through 10 10 ð Þplane. In fact, the results of XPS analysis in Fig. 2d, e support this model, i.e., the W concentration of ZnO nanowires increases with increasing C Zn above the critical concentration for nucleation on 10 10 ð Þ plane. However, the following XRD results show more complicated situation for the W incorporation. Figure 2f, g shows the XRD pattern of ZnO (0002) peak and the d value for the W-doped ZnO nanowires when varying C Zn with constant C W 0.25 mM. We found that there are two components in (0002) peak and among the one is consistent with bulk ZnO, implying the coexistence of non-doped part and W-doped part in an individual ZnO nanowire. Furthermore, there are two inconsistencies between the results of XPS and XRD. First, the XRD peak associated with Wdoped ZnO still remains when completely suppressing the nucleation on 10 10 ð Þ plane. Second, in XRD result, W incorporation seems to be promoted by decreasing C Zn as can be seen in the decreasing trend of d value for W-doped ZnO in Fig. 2g. These two features are contradictive to the results of XPS. Considering the difference of surface-sensitive XPS and bulk-sensitive XRD, such contradictive features can be interpreted by that ZnO nanowires are composed of W-doped core and non-dope shell. Also this suggests that W incorporation unexpectedly occurs on (0001) plane even though W addition enhances the nucleation on 10 10 ð Þ plane. In order to clarify the dopant incorporation pathway, we examine the feasibility of W incorporation via the crystal growth on 10 10 ð Þ-oriented and (0001)-oriented single-crystalline ZnO substrates as shown in Fig. 3a  (0001)-oriented and 10 10 ð Þ-oriented substrates are 1.03 and 0.04 at%, respectively, supporting that W is mainly incorporated through (0001) plane. This is further confirmed via the compositional analysis using two-step grown nanowires (see Supplementary  Fig. 5 and Supplementary Note 4). According to the face-selective W incorporation on (0001) plane, the result in Fig. 2e can be interpreted in terms of an expansion of (0001) plane size followed by the sidewall growth. Therefore, we perform the TEM observation and analyze the correlation between the (0001) plane size and the W concentration of ZnO nanowires. Figure 3c shows the TEM images of ZnO nanowires grown with various C Zn (10-25 mM) and C W (0-0.125 mM). The (0001) plane tends to be larger with increasing C Zn and/or C W , which is associated with the nucleation enhancement on 10 10 ð Þ plane discussed above. The relationship between (0001) plane width and W concentration is summarized in Fig. 3d. There is a strong correlation between them, and the W concentration is irrelevant to W/Zn ratio in precursors, clearly evidencing that W incorporation selectively occurs on (0001) plane. The TEM observation also reveals the presence of 10 11 ð Þ plane at the tip of nanowires. Note that the 10 11 ð Þ plane exists even suppressing the sidewall growth. The W concentration becomes almost zero when the 10 11 ð Þ plane dominates the nanowire tip with decreasing the diameter. This suggests that W is not incorporated on 10 11 ð Þ plane. To confirm the hypothesis as to the face-selective dopant incorporation, we directly identify the dopant distribution in an individual ZnO nanowire. It should be described that the dopant distribution in hydrothermal ZnO nanowire has not been discussed before unlike that in VLS nanowire [16][17][18]25 . This might be due to that the impinging rate of impurity metal ions on each crystal plane of nanowire is equal in hydrothermal synthesis, but those on liquid-solid interface and vapor-solid interface in VLS synthesis are supposed to be different. In high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image of Fig. 4a, the clear contrast with darker core and brighter shell is seen in the nanowire. For this experiment, the nanowires were grown with C Zn 15 mM and C W 0.125 mM (W concentration in precursor: 0.83%) and we chose a nanowire with sidewall growth. We found that the darker area, which is associated with W-doped part, tends to be larger along the nanowire growth direction. This is well consistent with the model of face-selective W incorporation on (0001) plane discussed above. The qualitative analysis of dopant distribution by EDS shows that the darker area has higher W concentration and in this area dopant seems to be homogeneously distributed as shown and discussed in Supplementary Fig. 6 and Supplementary Note 5. In addition, the dopant distribution in an individual ZnO nanowire is quantitatively analyzed by 3DAP measurement as details can be seen in Supplementary Fig. 7. Figure 4b shows 3D atom maps for Zn, O (left), and W (right) taken around the projected area in Fig. 4a. W atoms are more densely distributed in core side than shell side. This is also confirmed by 2D concentration plot shown at the bottom of 3D atom maps and line profile in Fig. 4c. Note that averaged W concentrations in core area are 0.66 at%, indicating that 80% of WO 4 2− ions impinged on (0001) plane are incorporated into nanowires. On the other hand, W concentration in shell area is limited to 0.11 at%, which corresponds to 13% of WO 4 2− ions impinged on 10 10 ð Þ plane. These results highlight the inhomogeneous dopant distribution in an individual hydrothermal ZnO nanowire, resulting from the faceselective dopant incorporation.
To gain the theoretical insight as to the conflicted face-selective behaviors of WO 4 2− ions on nucleation and dopant incorporation, we performed density functional theory (DFT) calculations. The details of calculation are shown in Supplementary Fig. 8. Figure 5a shows the most stable coordination structure of WO 4 2− ion on ZnO 10 10 ð Þplane, (0001) plane, and 10 11 ð Þplane. WO 4 2− ion is coordinated with tripodal bonds on (0001) and 10 11 ð Þ planes, and with unipodal or bipodal bonds on 10 10 ð Þplane. Here we discuss the WO 4 2− ion-induced nucleation enhancement on 10 10 ð Þ plane by examining the various coordination of WO 4 2− ion and Zn species (Zn 2+ , [ZnOH] + , and Zn(OH) 2 ) on 10 10 ð Þ plane in the presence of water solvent (Fig. 5b). The calculated interaction energies between WO 4 2− ion and various Zn species are summarized in Table 1. The results demonstrate that the sufficient energy gain is obtained for all the examined coordination structures. The growth system is further stabilized by incorporating Zn species into ZnO crystal which is shown as a relative energy in Fig. 5c,   ð Þ plane as experimentally shown above. Next we consider the face-selective dopant incorporation on (0001) plane, which must be correlated with the stability of coordinated WO 4 2− ions. We calculated the adsorption energies of WO 4 2− ions on ZnO 10 10 ð Þ plane, (0001) plane, and 10 11 ð Þ plane, as summarized in Table 2. Note that in the simulation, the WO 4 2− ions are not adsorbed on O-terminated (0001) and 10 11 ð Þ planes, therefore, we calculated the adsorption energies on Zn-terminated planes. The adsorption energies of WO 4 2− ions on ZnO crystal planes become higher in the following order 10 11 ð Þ< 10 10 ð Þ< (0001). This indicates that the WO 4 2− ion is preferentially adsorbed on (0001) plane followed by being incorporated into nanowire. Although the stable adsorption of WO 4 2− ions with tripodal bonds is also available on 10 11 ð Þ plane, it is destabilized by the hydration effect. Although the hydration effects on (0001) plane and 10 11 ð Þ plane are similar, the weaker adsorption of WO 4 2− ions on 10 11 ð Þ plane originating from its distorted coordination structure leads to the negative adsorption energy in water. Thus our calculation clearly explain that the coordination structure of WO 4  Step2 Step3 Step4   On the basis of suggested model, finally we control over the morphology and the elemental doping of ZnO nanowires. To the best of our knowledge, the well-defined synthesis of W-doped hydrothermal ZnO nanowire has not been achieved so far because the W addition seriously causes the nanowire morphology 34,35 . In principle, there are two possible approaches for modulating the dopant incorporation, i.e., a control of W precursor concentration and a control of dopant adsorption on nanowire surface. On the other hand, the morphology can be manipulated by controlling C Zn according to the critical concentration for nucleation on ZnO crystal planes, which is so-called as "concentration window" principle [22][23][24] . However, as shown in Fig. 2c, W addition substantially narrows the concentration window for enhancing the anisotropic nanowire growth. Therefore further increase of C W is not practical for enhancing W incorporation. Here, we control the adsorption of WO 4 2− ions on nanowire surface by varying the thermodynamic parameter. The DFT calculation revealed that the resident time of WO 4 2− ions on (0001) plane becomes longer by decreasing the temperature, as shown in Table 3. Supplementary  Fig. 9a, b shows the TEM images and XRD patterns of W-doped ZnO nanowires (C Zn 12 mM and C W 0.25 mM) grown at 95 and 75°C, respectively. In this experiment, the sidewall growth i.e., nucleation on 10 10 ð Þ plane of each sample is perfectly suppressed by controlling C Zn below the critical nucleation concentration. We found that (0002) peak of W-doped ZnO shifts to higher angle (i.e., smaller d value) by decreasing the growth temperature, meaning that more W is incorporated into nanowires. We successfully demonstrated the controlled morphology and elemental doping of W-doped ZnO nanowires for the first time by the predictable model based on coordination structure of impurity metal ions. The controllable nanowire morphology and elemental doping in hydrothermal synthesis enable to design the property of various ZnO nanostructure-based applications such as photovoltaic devices, photocatalysis, and chemical sensing [47][48][49][50] .
In conclusion, we demonstrated the unusual face-selectivity of WO 4  ð Þplane with unipodal or bipodal bonds serve as surfactant for attracting Zn ion species, whereas those coordinated on (0001) plane with tripodal bonds are easily incorporated into nanowire as dopant due to the different stability of their coordination structure. Based on this predictable model, we successfully demonstrated the rational strategy for controlling over the morphology and the elemental doping of W-doped ZnO nanowire via modulating nucleation events and impurity adsorption with "concentration window" principle. This study clearly highlights the essential importance to understand the coordination structure of metal ions for designing the hydrothermal metal oxide nanowire synthesis and the resultant functions of metal oxide nanowirebased device applications.

Methods
W-doped ZnO nanostructures/nanowires were synthesized by a seed layer-assisted hydrothermal method under atmospheric pressure. Prior to the growth, a 50-nmthick ZnO seed layer was deposited onto a 100-nm-thick SiO 2 coated Si (100) substrate with a 1-nm-thick Ti adhesion layer by radio frequency (RF) sputtering (Ar pressure 0.3 Pa, RF power 50 W). The growth solution was prepared at room temperature by mixing zinc nitrate hexahydrate (Zn(NO 3 ) 2 ·6H 2 O, Wako 99.0 %), HMTA ((CH 2 ) 6 N 4 Wako 99.0 %), PEI (number average m.w. 1800, ALDRICH 50 wt% in H 2 O), and sodium tungstate dehydrate (Na 2 WO 4 ·2H 2 O, ALDRICH ≥ 99%) in deionized (DI) water with varying their concentrations. The concentrations of each reagent was varied in the range of 8-25 mM for Zn(NO 3 ) 2 , 0-3 mM for PEI, and 0-0.5 mM for Na 2 WO 4 . For all experiments, the concentration of HMTA 25 mM was constant. Then the hydrothermal reaction was conducted at 95 and 75°C for 5 h by immersing the seed layer-coated substrate into the growth solution in manner of upside down. In two-step growth for growing non-doped core ZnO nanowires and W-doped ZnO shell, 1st step process and 2nd step process were conducted with Zn(NO 3 ) 2 25 mM and Na 2 WO 4 0 mM for 5 h and with Zn(NO 3 ) 2 25 mM and Na 2 WO 4 0.25 mM for 5 h, respectively. After the growth, the samples were rinsed in DI water and dried by N 2 blow at room temperature. The equilibrium concentrations of ionic species in the growth solution was calculated by Visual MINTEQ 3.1. at constant temperature of 95°C. The DFT calculations for the coordinated structures of WO 4 2− ion and Zn ion species on ZnO surface were performed by VASP 5.4.4 using implicit water solvation model. The resident time of WO 4 2− ion on ZnO crystals was estimated by using the Frenkel equation. The morphology, composition, and crystal structure of synthesized nanostructures were characterized by field effect scanning electron microscopy (FESEM, JEOL JSM-7610F) equipped with EDS at accelerating voltage of 15-30 kV, TEM (JEOL JEM-2100F) at accelerating voltage of 200 kV, XPS (Kratos AXIS-ULTRA) with Al Kα radiation source (12 kV, 5 mA, monochromator), and XRD (Philips X'Pert MRD) with Cu Kα radiation source (45 kV, 40 mA, monochromator), respectively. The binding energies were corrected at C 1s peak. HAADF-STEM and EDS analyses in Fig. 5 and Supplementary Fig. 7 were performed by FEI Titan G2 80-200 at accelerating voltage of 200 kV. 3DAP analysis was carried out using CAMECA LEAP5000XS in a laser mode at specimen temperature of 30 K and the laser pulse energy and pulse rate were 30 pJ and 250 kHz, respectively. For samples preparation of HAADF-STEM and 3DAP, Ni protection layer with 200-nm thickness was first deposited to support nanowires. Then, a sliced sample and a needle-shaped sample composed of a single nanowire, were prepared for HAADF-STEM and 3DAP, respectively, by focused ion beam (FIB) technique (FEI Helios G4UX) using the standard lift-out method at accelerating voltage of 30 kV. Low accelerating voltage of 2 kV was used in the final cleaning steps for reducing the damages caused by 30-kV FIB.

Data availability
The obtained data and the analyzed data are contained in the published article and the Supplementary Information, and are available from the corresponding authors on reasonable request.