Klinton Davis,
Ryan Yarbrough,
Michael Froeschle,
Jamel White and
Hemali Rathnayake*
Department of Nanoscience, Joint School of Nanoscience & Nanoengineering, University of North Carolina at Greensboro, Greensboro, NC 27401, USA. E-mail: hprathna@uncg.edu; Tel: +1-336-285-2860
First published on 10th May 2019
A reliable sol–gel approach, which combines the formation of ZnO nanocrystals and a solvent driven, shape-controlled, crystal-growth process to form well-organized ZnO nanostructures at low temperature is presented. The sol of ZnO nanocrystals showed shape-controlled crystal growth with respect to the solvent type, resulting in either nanorods, nanoparticles, or nanoslates. The solvothermal process, along with the solvent polarity facilitate the shape-controlled crystal growth process, augmenting the concept of a selective adhesion of solvents onto crystal facets and controlling the final shape of the nanostructures. The XRD traces and XPS spectra support the concept of selective adhesion of solvents onto crystal facets that leads to yield different ZnO morphologies. The shift in optical absorption maxima from 332 nm in initial precursor solution, to 347 nm for ZnO nanocrystals sol, and finally to 375 nm for ZnO nanorods, evidenced the gradual growth and ripening of nanocrystals to dimensional nanostructures. The engineered optical band gaps of ZnO nanostructures are found to be ranged from 3.10 eV to 3.37 eV with respect to the ZnO nanostructures formed in different solvent systems. The theoretical band gaps computed from the experimental XRD spectral traces lie within the range of the optical band gaps obtained from UV-visible spectra of ZnO nanostructures. The spin-casted thin film of ZnO nanorods prepared in DMF exhibits the electrical conductivity of 1.14 × 10−3 S cm−1, which is nearly one order of magnitude higher than the electrical conductivity of ZnO nanoparticles formed in hydroquinone and ZnO sols. The possibility of engineering the band gap and electrical properties of ZnO at nanoscale utilizing an aqueous-based wet chemical synthesis process presented here is simple, versatile, and environmentally friendly, and thus may applicable for making other types of band-gap engineered metal oxide nanostructures with shape-controlled morphologies and optoelectrical properties.
Among dimensional ZnO nanomaterials, one-dimensional (1D) ZnO nanostructures with defect free high crystallinity are a particular interest due to their unique and inherent intrinsic chemical, electrical, physical, and mechanical properties compared to that of bulk and thin film counterpart.10 However, synthesis of defect free 1D ZnO nanostructures, with desired morphology and composition, has been challenging as most growth techniques involve either high-cost fabrication processes or high temperature wet-chemical syntheses performed in highly toxic solvents. Among such fabrication techniques, chemical vapor deposition (CVD),18 pulse-laser deposition (PLD),19 molecular beam epitaxy (MBE),20 and electro-chemical deposition21 have been utilized to grow ZnO nanostructures directly onto the substrate. In addition to these thin film deposition techniques, sol–gel and solvothermal methods are two of the most common chemical solution methods, which show the promise in terms of scalability, energy efficiency, and cost-effectiveness for the preparation of catalyst-free metal oxide nanostructures with better control over the growth conditions and morphology.22–24 Particularly, the sol–gel process has been extensively investigated for making homogenous, highly stoichiometric, and high-quality metal oxide nanostructures such as nanorods,25,26 nanoflakes,27 nanotubes,25,28 and nanofibers.29 In general, the sol–gel process involves formation of sol from homogeneously mixed solutions of a metal precursor and a base. The sol of metal oxide nanocrystals can either be deposited onto a substrate to grow nanostructures or be continued through the polycondensation process to form gels. The gel can be used to form particles, xerogels, aerogels, glass, and ceramics, depending upon the final processing step involved.30
Up to date, the sol–gel method of either aqueous or non-aqueous hydrolysis and condensation process, has been adapted primarily to make sols of ZnO nanoparticles, followed by casting on substrates to grow ZnO nanostructures, upon subjecting to thermal annealing.1,4,5,31,32 However, as per our knowledge, there is no records of a sol–gel based green synthesis method, which combines the formation of sols and a solvent-driven shape-controlled crystal growth process in solution at low temperature to make ZnO nanostructures, with the wurtzite crystal lattice. This method benefits to make ZnO nanomaterials in powder form for solution processable thin films fabrications in large scale. Thus, herein, we demonstrate a reliable sol–gel approach, which combines the formation of a ZnO nanocrystals sol and a solvent driven shape-controlled in situ crystal growth process to form well-organized ZnO nanostructures at low temperature (<80 °C) in solution. This one-pot synthesis process allows us to make ZnO nanostructures with a variety of morphologies, without disturbing their crystalline structures and compositions. The ZnO nanostructures prepared in this manner showed excellent improvement in crystallinity and optoelectrical properties, with engineered optical band gap. The method developed here is a “green synthesis” where we utilized environmentally friendly and benign materials, with energy efficient wet-chemical approach. Thus, our method is rather advantages in terms of scalability, processability, and reliability compared to other wet-chemical and electrochemical synthesis methods, which typically grow nanostructures directly from thin film casted sol in environmentally friendly manner or high temperature solvothermal methods that utilize highly toxic hydrazine-based solvents and additives.
The ZnO nanostructures were prepared by combining the sol–gel chemical process with a solvent-driven nanocrystals growth at low temperature. The typical sol–gel process involved base catalysed hydrolysis of the metal salt to form metal hydroxide followed by condensation to form nanocrystals of the metal oxide sol. Upon subjecting the sol of nanocrystals to an in situ solvothermal process, ZnO nanostructures with different morphologies were obtained. The solvothermal process induces the nanocrystal elongation along the facet of high energy crystal lattice surface and initiates nanocrystals growth through Ostwald ripening and oriented attachment mechanisms.35 Also, the highest packing efficiency of nanocrystals into well-ordered nanostructures is feasible via solvent-driven self-assembly of nanocrystals through van der Waals interactions.41–44 Such non-covalent interactions enable interacting tiny nanocrystals to form larger crystal aggregates and induce the crystal growth mechanisms (Ostwald ripening and “oriented attachment”).34–36 As a result, nanocrystals with sharing crystallographic orientations directly combine together to form larger ones along the most preferential crystal lattice axis. The dimensionality and morphology of these crystalline nanostructures were tailored by changing the solvent mixture. The difference in polarity and surface adhesion of each organic solvent controlled the shape and size of nanocrystal growth. The solvent molecules act as surfactants that adsorb onto surfaces of the growing crystallites.45 The selective adhesion of solvent molecules onto crystal lattice facets is governed by the surface energy difference, which drives solvent binding onto a selected facet. For example, past literature evidences that if a solvent has high binding affinity to the {001} facet of the crystal lattice, the growth rate along the direction of the [001] plane reduces and consequently, results in the formation of nanorods.37,40,45 Similarly, in our studies, we speculate that depending on the polarity, chemical functionality, hydrophilicity and hydrophobicity, solvents selectively adsorb onto different faces of the nanocrystal lattice, reducing the growth rate along a particular lattice facet to form a shape-controlled nanostructure.40
Scheme 1 The formation of ZnO nanostructures via sol–gel process followed by solvothermal self-assembly. |
The sol of ZnO nanocrystals was prepared by mixing 1:5 molar ratio of the metal precursor (ZnCl2) and the base (NaOH) at room temperature with initially stirring for ∼15 minutes. After ∼15 minutes of stirring at room temperature, the sol of nanocrystal formation was monitored at 15 minutes time intervals up to 60 minutes while homogeneous stirring at 80 °C to yield an opaque solution. The reaction time for the formation of ZnO sol was monitored by acquiring UV-visible spectra of the solution at different time intervals. The time-dependent UV-visible spectra were compared with the absorption traces collected for the initial reactants' solution at room temperature. All the spectra were recorded in solution as either a clear or an opaque solution. As depicted in Fig. 1(a), the UV-visible spectrum, obtained for the reactants' solution after 15 minutes of stirring at room temperature, exhibits the absorption maximum at 264 nm with a shoulder peak at 332 nm, characterizing the Zn2+ absorption. As the reaction progressed, a gradual red shift in the shoulder absorption band at 332 nm was observed. The reaction mixture heated for 60 minutes showed a red-shifted absorption maximum at 347 nm, evidencing the formation of ZnO nanocrystals (Fig. 1(a)). The transmission electron microscopy (TEM) images further confirmed that nanocrystals were somewhat spherical in shape with the size ranging from 3–5 nm (Fig. 1(b)).
The absorption peak in the range of 345–450 nm arises due to the surface plasma resonance effect of ZnO nanocrystals and the peak shift from blue to red is characteristic to the quantum size effect.46–48 The stronger exciton effect is characteristic of the quantum confinement in semiconducting nanostructures, in which the electrons, holes, and excitons have limited space to move; and their movement only is possible for definite values of energies. As a result, the continuum of states in conduction and valence bands exhibits discrete states with an energy spacing relative to band edges; which is inversely proportional to the square of the particle size and reduced mass.48 Thus, their energy spectrum is quantized. With the increase in crystallites size, a red shift in the absorption spectra can be observed because of narrowing the band gap, which leads to the effective band gap smaller than its bulk value. In our case, the absorption peak shift from 332 nm in initial precursor solution, to 347 nm for ZnO nanocrystals sol, and finally to 375 nm for ZnO nanorods (Fig. 1(a) and (c)) evidence the gradual growth and ripening of nanocrystals to dimensional nanostructures. As a result, we observed a gradual decrease in the band gap from 3.57 eV to 3.20 eV, confirming the presence of highly confined carriers in ZnO nanocrystals compared to that of in solvothermal grown ZnO nanorods. This is in good agreement with the previously reported results.48
The sol prepared in this manner was subjected to in situ condensation and shape-controlled nanocrystal growth process by transferring the reaction mixture onto a sand bath with no stirring while maintaining the temperature at 80 °C for overnight (24 h). The nanostructures' formation, morphologies, and crystallinities were characterized by acquiring UV-visible spectra, SEM images, and X-ray powder diffraction (XRD) traces combined with selective area electron diffraction (SAED) patterns. The atomic compositions and binding energies of Zn, O, and C present in ZnO nanostructures with respect to different morphologies were analysed using X-ray photoelectron spectroscopy.
Trial # | Solvent system (6:1 v/v) | Nanostructures morphology and dimension description |
---|---|---|
a A molar ratio of the metal precursor (ZnCl2) to NaOH was maintained at 1:5 for each trial. In each case, pH was maintained at the range of 12–13. | ||
1 | DMF:H2O | Hexagonal rods with the average length of 1–5 μm and width of 50–180 nm |
2 | Acetonitrile:H2O | Hexagonal rods with the average length of 1–3 μm and width of 50–180 nm |
3 | DMSO:H2O | Thin slates like structures ranging from 500–2 μm |
4 | Toluene:H2O | A mix of thin hexagonal rods and wide slates. The average length of rods 1–5 μm and width of 100–200 nm |
5 | Hydroquinone:H2O | Globular shaped particle-like structures ranging from 100–500 nm |
6 | m-Xylene:H2O | Very short hexagonal rods with the average length of <300 nm and width of 40–50 nm |
Fig. 2 SEM images of ZnO nanostructures prepared using different solvent systems; scale bars represent 200 nm. |
The formation of ZnO nanorods, nanoparticles, and random slates like structures with respect to the solvent type can be explained by understanding the solvent-driven shape-controlled crystal growth process. The polarity and chemical nature of the organic solvent in each reaction mixture may act as a selective adhesion surfactant to facilitate the shape-controlled crystal growth. As studied in the past literature, the morphology of the final nanostructure controls by the surface energy and selective adhesion of solvent molecules onto crystallites facets that modulate the crystal growth direction in the nanocrystal unit cell.35,36 Since the crystal growth rate is correlated exponentially to the surface energy, surface-energy differences induce much faster growth of the higher surface energy planes and keep the slower growing planes (lower surface energy) as the facets of the product. Comparing the intensities of crystal planes reflections for the crystallites facets of {100}, {001}, and {101} of XRD traces for each solvent system, the preferential crystal growth direction and the morphology of the final nanostructure can be realized. Therefore, we investigated the diffraction patterns for the final products of ZnO nanostructures by acquiring X-ray powder diffraction traces and SAED patterns from the TEM under dark field diffraction mode. As shown in Fig. 3(a), the diffraction patterns of ZnO nanostructures prepared in five of the six solvent systems indexed to hexagonal phase wurtzite crystal structure, and is in good agreement with the previous published literature.10 We observed a clear difference in the intensities of [100], [002], and [101] Bragg peaks as well as the peak resolution of the [002] reflection plane in each case. The diffraction pattern obtained for the ZnO nanorods prepared in DMF/water solvent system shows a poorly resolved second order reflection of [001] plane as a shoulder that merges onto the [101] reflection plane. The XRD traces of ZnO nanorods, prepared in the mixtures of acetonitrile, toluene, and xylene with water show a weak intensity peak for the [002] diffraction plane of the first order {001} facet, while maintaining the intensity of the [100] peak by three-fold higher than the intensity of [002] reflection plane. However, the intensity ratios of [100]:[101] diffraction planes for the nanorods prepared in the presence of DMF, acetonitrile, and xylene were 1:2, which is slightly higher compared to the intensity ratio of the [100]:[101] peak for the nanorods prepared in the toluene and was found to be 1:1.5. The reason for the slight deviation in the intensity ratio could be due to the dimensional difference between thin hexagonal nanorods and wide slates; as observed in Fig. 2. Overall, from these observations, it is evidenced that the presence of poorly resolved second order reflection planes of {001} facets in the XRD patterns of the ZnO nanorods, formed in DMF, acetonitrile, xylene, and toluene could be preferentially bonded to {001} and {002} facets and reduced the crystal growth rate along the second order reflection planes in the [001] direction to yield nanorods. Similarly, in the past literature, it was shown that TiO2 nanorods was formed when lauric acid was used and acted as the surfactant that reduces the growth rate along the [001] direction, binding strongly to the {001} facet.40
Fig. 3 (a) The powder XRD patterns, and (b) the SAED patterns along with the TEM images of ZnO nanostructures prepared in different solvent systems. |
In contrast to the XRD patterns of ZnO nanorods, the XRD pattern obtained for the ZnO nanoparticles exhibits well-resolved first and second order reflection planes for the Bragg peaks of [100], [002], and [101]. The difference in the intensity of these Bragg peaks confirms that the crystal growth rate along the directions of [100], [002], and [101] planes is same. This suggests that there is no preferential adsorption of hydroquinone molecules onto a particular facet over others. In turn, it exhibits equal binding affinity for all facets of crystal lattice. So, the crystals grew into nanoparticles, enabling the crystal growth along all the directions of crystal facets. In contrast, the XRD traces obtained for ZnO nanostructures, prepared in the presence of DMSO show only the well-resolved [101] diffraction plane with a poorly resolved [100] reflection peak. This suggests that DMSO molecules hinder the crystal growth direction along both [100] and [001] faces by selectively binding to both surfaces of {100} and {001} facets.
The SAED pattern of a single nanostructure prepared in each solvent mixture was taken from the TEM under dark field diffraction mode by directing the electron beam at 90° angle to a one of the nanostructure's faces. The SAED patterns collected in this manner are depicted in Fig. 3(b). The SAED patterns viewed along the [001] axis show defect free and single crystalline nanocrystals growth for the nanostructures prepared in all other solvent systems except for the nanostructures prepared in DMSO. The single crystal unit cell diffraction planes are well aligned with the crystal growth directions, revealed from the peak intensities of the respective XRD powder spectra. The SAED patterns for the ZnO nanorods and nanoparticles, show defect free single crystalline pattern with first, second, and third order facets of the nanocrystals' ordering along the [100] diffraction planes. The difference in the well-resolved spacing of higher order reflections at [100] inter-diffraction planes confirms the unit cell packing and unit cell distances of crystalline facets. As observed from the SAED patterns, ZnO nanorods prepared in DMF and toluene exhibit larger nanocrystal unit cell spacing of higher order reflections whereas ZnO nanoparticles, formed in the presence of hydroquinone, show the smallest unit cell spacing in the crystal packing. The crystal growth pattern and nanocrystal packing during the formation of ZnO nanorods in acetonitrile and xylene are alike and exhibit same unit cell spacing among the higher order reflections. The SAED pattern obtained for ZnO nanostructures formed in DMSO (also see Fig. S2†) reveals weak polycrystalline pattern with poorly resolved nanocrystals ordering along [100] plane, confirming truncated crystal facets with polycrystallinity to amorphous.
Fig. 4 XPS wide survey spectra of ZnO sol, nanorods (formed in m-xylene), and nanoparticles (formed in hydroquinone). |
The Table S1† summarizes binding energy (BE) for the core levels Zn 2p, O 1s, C1s and their corresponding full-width at half maximum (FWHM) along with the respective %atomic concentrations for two types of ZnO nanostructures and sol. The C 1s atomic concentration for nanoparticles is twenty times higher than that of ZnO sol and five times higher than that of nanorods. Overall, with the increase in carbon and oxygen atomic concentrations and a gradual decrease in Zn atomic concentration clearly evidence that the presence of adsorbed solvent molecules on crystallites facets that affect in modulating the crystal growth direction in the nanocrystal unit cell. As we discussed in the morphology analysis section, it is clear that m-xylene prefers binding only onto a particular facet, hindering the crystal growth along the respective facet, yielding rod shape nanostructures. Whereas hydroquinone shows equal binding affinity onto all the facets of the crystal lattice and adsorb solvent molecules onto all the facets, yielding spherical shape nanostructures.
The noticeable differences in the binding energy spectra of Zn 2p, O 1s, and C 1s of ZnO sol, nanorods, and nanoparticles reveal the chemical environment interaction between solvent molecules and the surface atoms of the crystal facets. Fig. 5 demonstrates the comparison of orbital binding energy states of high resolution XPS spectra of the Zn 2p region, O 1s core-level, and C 1s for ZnO sol, nanorods, and nanoparticles. The binding energies of the Zn 2p components of sol are slightly lower than the binding energies of the Zn 2p components of nanorods and nanoparticles (Fig. 5(a)). The low binding energy of Zn 2p in sol evidences that the chemical environment interaction of Zn–O bonding surfaces of sol could be different from the chemical environment interactions of nanorods and nanoparticles. This may also be due to the difference in surface morphology and crystal size in nanocrystals compared to nanorods and nanoparticles.50 Surprisingly, the binding energy peaks of the Zn 2p components for ZnO nanorods and nanoparticles are located at the same positions. This suggests that the nature of chemical interactions in both rods and particles are attributed to strong Zn–O bonding interactions. In overall, Zn 2p components binding energy spectral traces indicate that the chemical valence of Zn at the surface of nanostructure morphologies and sol is Zn2+ oxidation state. The binding energy difference between the Zn 2p1/2 and Zn 2p3/2 is 23 eV for ZnO nanostructures and ZnO sol.
The O 1s spectrum for ZnO sol in Fig. 5(b) shows a sharp peak at 529.1 eV and a shoulder peak at 530.5 eV. The lower energy peak is attributed to O2− in Zn–O bonding of the wurtzite structure of ZnO.51 The higher binding energy (530.5 eV) is usually related to OH group absorbed onto the surface of the ZnO nanoparticles.50 However in our case, sol usually contains a mixture of ZnO and Zn(OH)2. It is expected to have OH groups on nanocrystal surface. The O 1s spectra of nanorods and nanoparticles, shown in Fig. 5(b), show only O−2 in Zn–O bonding with a shift in binding energy of 2 eV and 2.5 eV, respectively, compared to the binding energy peak of O−2 in ZnO sol. The peaks positions at 531.1 eV and 531.6 eV in nanorods and nanoparticles are in good agreement with the previous work of ZnO nanorods and nanoparticles.50,51 The C 1s spectra of sols, nanorods, and nanoparticles show three different binding energy peaks, which are uniquely characteristic to each morphology of ZnO nanostructures. The C 1s spectrum of ZnO sol shows two very weak intensity peaks at 283.8 eV and 288.4 eV, reflecting C–C/C–H and C–O bonding. In our case, these weak C–C/C–H and C–O bonding energies are attributed to traces of ethanol molecules adsorbed onto the nanocrystal surface during the washing of centrifuged powder form of ZnO sol. However, the binding energy peak for C–C/C–H in nanorods and nanoparticles are clearly visible and slightly shifted to yield binding energy of 284.4 eV and 284.8 eV respectively. In nanoparticles, the intensity counts for the C–C/C–H bonding binding energy is significantly higher than nanorods and confirms the presence of adsorbed organic molecules onto the nanoparticle surface. The higher binding energy peaks at 289.8 eV (weak peak) and 288.9 eV (strong peak) in nanorods and nanoparticles, respectively, are attributed to carbonyl (CO) bonds. The presence of a strong peak for carbonyl bonds in nanoparticles further confirms that the adsorbed hydroquinone molecules on nanoparticle surface.
Fig. 6 (a) Thin film UV-visible spectra of the ZnO nanostructures with respect to the solvent type; (b) the calculated band structure for ZnO wurtzite hexagonal lattice structure obtained by matching with the simulated XRD spectra and the experimental XRD traces (Fig. S3†). The energy of the valence-band maximum (VBM) was set to zero and used Hubbard U+DFT hybrid function for the calculation; and (c) respective ZnO crystal structure extracted from the crystallography data base to compute the band structure. |
The comparison in the energy absorption maxima and optical band gaps of the ZnO nanostructures formed in different solvent systems are summarized in Table 2. The ZnO nanorods and nanoparticles formed in DMF and hydroquinone solvent systems show maximum absorption energies at 3.42 eV and 3.58 eV respectively. The nanoparticles' optical band gap, calculated from the UV on-set was ranged from 3.33–3.41 eV, which is considerably higher than the optical band gap range of 3.16–3.24 eV for nanorods made in DMF solvent system. The ZnO nanorods prepared in acetonitrile, toluene, and xylene solvent systems exhibit the absorption maxima ranging from 3.30–3.39 eV with the lowest optical band gap of 3.10 eV for nanorods formed in acetonitrile. The optical band gaps of the ZnO nanorods formed in toluene and xylene are 3.22 ± 0.10 and 3.19 ± 0.04 respectively. The changes in the optical band gaps with respect to the ZnO nanostructure morphology further confirms that nanostructure crystallinity, crystal growth facets, and crystal grain size lead to the effective band gap of nanostructured ZnO smaller than its bulk value of 3.37 eV.
Solvent system | Energy (hv) of the absorption maximum (eV) | Optical band gap (eV) |
---|---|---|
DMF:H2O | 3.40 ± 0.02 | 3.20 ± 0.04 |
Acetonitrile:H2O | 3.36 ± 0.03 | 3.15 ± 0.05 |
Toluene:H2O | 3.33 ± 0.03 | 3.22 ± 0.10 |
Hydroquinone:H2O | 3.50 ± 0.08 | 3.33 ± 0.04 |
m-Xylene:H2O | 3.34 ± 0.04 | 3.19 ± 0.04 |
The optical band gap of ZnO nanorods formed in DMF solvent systems was compared with the theoretical band gap computed from the respective ZnO crystal structure, which was extracted from a crystallographic data base, and followed by adjusting the unit cell parameters comparing the simulated XRD pattern of the extracted ZnO crystal structure with the experimental XRD traces of ZnO nanorods. The theoretical band structure was computed using the plane-wave form of Density Functional Theory (DFT) implemented in the open source Quantum ESPRESSO (QE) suite. Ultra-soft pseudopotentials created with low-density approximation (LDA) functions were used and were obtained from QE. The band gap calculated using only DFT was severely underestimated, forcing a change to DFT+U; this scheme uses an additional Hubbard U potential for each element to correct band overlap discrepancies. The potentials for zinc and oxygen to correct the large discrepancy were based on literature, 12 eV and 6.5 eV respectively. The potentials were applied to the 3d orbitals of the Zn atoms and the 2p orbitals of the O atoms.52,53 With this method, we were able to compute the corrected band gap to be 3.29 eV (Fig. 6(b)), which is in good agreement with the optical band gap of ZnO nanorods formed in the DMF solvent system.
Footnote |
† Electronic supplementary information (ESI) available: Additional SEM and TEM images along with XRD powder diffraction traces. See DOI: 10.1039/c9ra02091h |
This journal is © The Royal Society of Chemistry 2019 |