Enhanced stabilisation of tetragonal (t)-ZrO₂ in the controlled nanotubular geometry

Abstract

Precise control of the structure of nanogranular materials over different polymorphs is directly related to the manifestation of the desired and resultant properties. The room temperature phase stabilisation of nanocrystalline tetragonal (t)-ZrO₂ has been a controversial topic in the literature. Here, we report that the stabilisation of t-ZrO₂ is enhanced with the tubular geometry at the nanoscale and that it can be manipulated by carefully selecting the initial structures as-grown. ZrO₂ nanotubes (ZNTs) produced via template-directed atomic layer deposition (ALD) techniques were in the growth temperature range of 150 through 250 °C, followed by thermal treatment up to 700 °C. The resulting phases of the ZNTs (i.e., tetragonal and/or monoclinic ZrO₂) were strongly affected by the interplay between the original deposition and post-annealing temperatures. We further elucidated that both the initially grown phases and the types of crystalline nuclei in the as-deposited nanotubes determine the final microstructures. This observation gives us an unambiguous clue to understanding the controversial results on the phase stabilisation of nanocrystalline t-ZrO₂. Calculation results also support that the confinement within the thin wall layers is pronounced during nucleation and growth, which results in the enhanced stabilisation of t-ZrO₂. The present results provide us with an insight to the stabilisation of nanocrystalline t-ZrO₂ and with a strategy on how to tailor the structures of ZrO₂ nanostructures in fine-tuning material properties.

---

1. Introduction

Zirconia-based ceramics have many favourable properties, such as high diffusivity for oxygen ions,¹ exceptional resistance against thermal stress and mechanical shock,² electrical insulation³ and high biocompatibility.⁴ These physical aspects have found a wide variety of industrial applications, including solid oxide fuel cells,⁵ catalysts,⁶ thermal barrier coatings,⁷ jet engines,⁸ alternative gate-oxides in microelectronics,⁹ and implantable biomaterials for a hip joint.¹⁰ Bulk ZrO₂ has three well-known polymorphisms at normal atmospheric pressure.¹¹ The monoclinic baddeleyite structure (m-ZrO₂) is thermodynamically stable under ambient conditions, where Zr atoms are in a distorted seven-fold coordination, and O atoms have four- or three-fold coordination. m-ZrO₂ transforms reversibly to the tetragonally distorted fluorite structure (t-ZrO₂) above ∼1175 °C, with Zr in an eight-fold coordination; this phase transformation is known to be accompanied by a substantial volume change (i.e., approximately 5 vol%). Cubic fluorite structured ZrO₂ (c-ZrO₂) is the most stable one among all and is stabilised upon ∼2370 °C. Furthermore, adding aliovalent oxides, such as CaO, Y₂O₃ and Gd₂O₃, stabilises c-ZrO₂, even at room temperatures.¹² Stabilisation of t-ZrO₂ and c-ZrO₂, which is required to be technologically viable over m-ZrO₂, is thus of significant importance in many applications without any structural disintegrations during the phase transition of m-ZrO₂ ↔ t-ZrO₂ and c-ZrO₂ as well.¹³

t-ZrO₂ is known to be stabilised in its nanocrystalline form without any impurity doping, where there exists a critical size. This could be understood simply as an argument of the increased specific surface area by which the excess surface/interface free energies compete with the volumetric counterpart. In practice, the stabilisation is also dependent on the presence of defects, strain energies, and/or many other factors.¹⁴ Significant research efforts over more than two decades on the stabilisation of the pure t-ZrO₂ have been recognised and generally accepted in the nanogranular system. Garvie first reported that the nanocrystalline zirconia possessing the grain size of up to 30 nm could be stabilised as t-ZrO₂, even at room temperature.¹⁵ Since his observation, the phase stabilisation mechanisms of the high-temperature ZrO₂ phases (i.e., t- and c-ZrO₂) have been studied with samples prepared under various experimental conditions and suggested by different research groups.¹⁶ The critical size of approximately 10–30 nm, depending on the processing conditions, is widely accepted because of the above-mentioned cross-over in the excess surface energy. The other explanation is that the surface energy of t-ZrO₂ is lower than that of m-ZrO₂.¹⁷ However, the exact stabilisation mechanism of nanogranular t-ZrO₂ remains controversial because some of the nanocrystalline zirconia with much smaller grains was observed to be not stabilised and had their monoclinic structures preserved.¹⁵

Nanoparticles and other t-ZrO₂ nanomaterials have been reported in the form of nanorods,¹⁸ nanowires,¹⁹ nanobelts,²⁰ and nanotubes.²¹ Among these forms, the nanotubular geometry is of particular interest because the structures should remain intact upon thermal treatments,²² even with a nanometre-scale thickness of the wall layers. ZrO₂ nanotubes (ZNTs) have been fabricated mostly by anodic oxidation of zirconium foils, with the choice of appropriate electrolytes in an electrochemical cell.²³ Many recent reports have successfully demonstrated the fabrication of well-engineered ZNT arrays and the effect of process conditions on the structure and morphology. Fang et al. studied the effect of heat treatment on anodised ZNTs' morphology and crystalline structures.²⁴ They found that F-ions were in the amorphous phase of as-prepared anodised ZNTs, and mixed phases of tetragonal and monoclinic ZrO₂ were observed upon annealing at 400 °C. Subsequently annealing starting from 500 to 900 °C, the anodised ZNTs gradually transformed to wires due to the substantial volume changes by monoclinic-to-tetragonal transformation. It is evident that the fabrication of pure/controlled phases of ZNTs with tailored geometries at nanoscale is of paramount importance and that diverse applications are possible.²⁵ In addition to the anodised Zr foils, only a few reports on ZNTs were found via sol–gel (or solution) processing or even atomic layer deposition (ALD).²⁶

Template-directed ALD coating allows for various hollow nanostructures of high aspect ratio due to the capability of excellent control over the wall thickness because it involves self-limiting surface chemical reactions between gas-phase precursor and solid substrate. Alternating exposure of two precursors, which react sequentially on the surfaces, enables the materials deposited onto the solid surfaces to be formed with atomic precision in a layer-by-layer manner. Although the ALD technique has the unprecedented capability of atomic level control over the thickness and conformality under the most demanding conditions, it is fully dependent on the saturation kinetics and the flow dynamics of gas-phase precursors. If the desirable samples have nanoporous structures, then the significance of the diffusion kinetics of precursor molecules will be increased. Moreover, the controlled formation of thin layers of materials would lead further modification in the free energy and thereby the phase stabilisation/selection.

Here, we report on the formation of ZNTs via template-directed ALD techniques. By controlling the initial phases of nuclei, we were able to achieve the stabilised tetragonal modification, which resulted in pure t-ZrO₂ without any structural disintegration after subsequent high-temperature annealing at 700 °C. Calculation results also support that the confined crystallisation within the thin wall layers is beneficial in the room temperature stabilisation of t-ZrO₂. The conformality of Zr-based oxide materials prepared by ALD onto high-aspect ratio nanostructures was confirmed experimentally and showed a good agreement with the previous model. The strategy on the phase selection proposed by the present report should extend to some other high temperature nanoceramics.²⁷

2. Experimental

Porous alumina membranes (PAMs) used as templates in this study were prepared by so-called ‘two-step’ anodisation of aluminium (5N, Goodfellow, UK).²⁸ Briefly, the first anodisation was performed in 0.3 M oxalic acid solutions at 10 °C under applied DC voltage of 40 V for ∼24 h. The anodised alumina layers were removed wet-chemically in mixture solutions of 6 wt% H₃PO₄ and 1.8 wt% H₃CrO₄ at 45 °C overnight, which left the self-organised, imprinted surfaces on Al. The samples were then anodised under similar conditions to the first procedure, except for the anodising time. Approximately 8 μm-thick PAMs were prepared over ∼2 h, and the pore diameter of 52.6 ± 2.9 nm was obtained by the pore-widening step in solutions of 5% H₃PO₄ at 45 °C for approximately 15 min.

ZrO₂ tubes were formed on the PAMs using a commercial ALD (OZONE-100A, ForAll, South Korea) with a shower-head-type reactor in the substrate temperatures of 150 to 250 °C. The working pressure was approximately 3.0 torr. Argon (5N) was used as the purge/carrier gas. Tetrakis(ethylmethylamino)zirconium [TEMAZr] (UP Chem., Korea) was evaporated from a bubbler-type canister held at 80 °C and then introduced into the reactor with the aid of the carrier gas. Deionised water was used as oxygen source and delivered at room temperature. The flow rate and working pressure were automatically controlled by a mass flow controller and a throttle valve, respectively. To study the conformality of the present system, we varied the exposure duration (t_p) of both TEMAZr and H₂O within the range of 5 to 20 s. The purging time of 20 s was applied after each reactant exposure. The growth rate of ZrO₂ was determined to be approximately 0.09–0.11 nm per cycle (see Fig. S1†), which is in good agreement with the literature values. Such growth rates of ZrO₂ films were achieved in our system when t_p was over 2 s and the Ar purging was performed for at least 8 s.

The physical dimensions of the templates and the resulting ZNTs were characterised by field-emission scanning electron microscopy (FE-SEM, JEOL JSM-7000F, Japan). The structures of as-grown and post-annealed ZNTs were investigated using a X-ray diffractometer [XRD, Bruker D8 DISCOVER, Germany, except for Fig. 3(d) (SmartLab, Rigaku, Japan)] and a high-resolution transmission electron microscopy (HRTEM, JEOL JEM-4010, Japan). The wall thicknesses of ZNTs were measured with TEM micrographs using image analysis software (Matrox Inspector, Matrox Electronic Systems Ltd, Quebec, Canada). The grain sizes were analysed on the basis of both TEM and XRD results for as-deposited and annealed samples, respectively.

3. Results and discussion

3.1 Conformality

The fabrication of ZNTs began with template-directed ALD at three different growing temperatures (i.e., substrate temperatures, T_sub of 150, 200 and 250 °C), and the resulting microstructures were characterised using XRD, TEM and HRTEM with selected area electron diffraction (SAED). The template-directed ALD method has been developed by some of the authors for well over a decade.²⁹ Nanoscale tubular structures of metal oxides could be routinely provided by an ultra-precise control over their physical dimensions and, thus, over their physical properties.³⁰ We carefully measured the wall thicknesses of the ZNTs along the PAM pore deep with series of TEM images to verify conformally coated regions with respect to the exposure duration, t_p, of precursor. Fig. 1(a) and (b) shows a representative mosaic of TEM images and the corresponding schematic illustration of the ZNTs fabricated. Due to the limited diffusion of Zr precursors inside the deep pores, the pores at the close end were not conformally coated with ZrO₂, as depicted in Fig. 1(b). This limited diffusion was clearly visible in Fig. 1(a) upon releasing ZNTs from the templates. Under the same conditions, three different ZNT samples were used for the conformality analysis. The measured wall thicknesses along the pores were displayed in Fig. 1(c–g) as a function of t_p (i.e., t_p = 5, 7.5, 10, 15, and 20 s, respectively). The representative mosaics of the TEM images were also given as insets in the corresponding panels in Fig. 1(c–g). In the conformally coated zone, the wall thicknesses of the ZNTs were within 5% deviation. As t_p increases, the conformal zone and thus the aspect ratio were gradually extended.

Fig. 1 Analysis of conformality in the formation of ZNTs on porous templates via ALD. (a) Representative TEM micrograph of a ZNT and (b) schematic depiction of the diffusion limited thickness gradient of the wall of a ZNT. (c–g) Measured thickness profiles of the ZNT wall layers along the pore bottom with varying t_p (from 5 to 20 s). The insets exhibit the corresponding micrographs used for the measurements. (h) A plot of t_p versus the aspect ratio, where the red curve produced using a kinetic model well fits with our experiments.

Gordon et al. previously reported a kinetic model for the step coverage via ALD in a nanosized hole with high aspect ratio.³¹ They assumed that the flux of molecules into the hole can be estimated from the flow of gas molecules in a vacuum system. The model considered a relationship between the partial pressure of the precursor in the chamber (P), the exposure time of the precursor (t), and the aspect ratio of the conformally coated region of fabricated pipes (a), as shown in eqn (1).

where S is the saturation dose, m is the molecular mass of the precursor, k is Boltzmann's constant, and T is the absolute temperature of the chamber. The aspect ratio is defined as the ratio between the length and the diameter of the tube. Only a few experimental verifications of step coverage models have been reported, including those by Gordon et al. and Perez et al. for HfO₂,^31,32 Kim et al. for TiO₂,³³ and Elam et al. for Al₂O₃ and ZnO.³⁴ The experimental data for step coverage of the limited materials above were in good agreement with the model predictions at different experimental parameters, such as precursor exposure time, t (in our work, t_p) and partial pressure, P. Achieving conformality in complex nanostructures with an extremely high aspect ratio remains a major challenge because of the lack of understanding on detailed diffusion dynamics in ALD, although ALD has proven to be the most desirable method for achieving such a conformal deposition.

In our case, we assumed that the factor, in eqn (1) is a constant (C) because we used the same conditions during ALD processes. The relationship between the aspect ratio (a) of the conformally coated region and the precursor exposure time (t) of the precursor is then expressed as quadratic equation, as given in eqn (2) and plotted as a red curve in Fig. 1(h).

where . The resulting plot for t (or t_p) versus aspect ratio (i.e., conformal zone) was found to be in excellent agreement with our experimental observation (Fig. 1(h)). The present results provide us with a method to estimate a practical t_p for preparing ZNTs with the desired high aspect ratio. To obtain an aspect ratio of 100, for example, one can predict that t_p must be over approximately 50 s per half cycle when using similar ALD reactor types.

3.2 Enhanced phase stabilisation in nanotubular geometry

Previously, Shukla and Seal established the room temperature phase stabilisation of nanocrystalline t-ZrO₂, assuming that crystallisation occurred inside the spherical grains (see Table S1†). Now, we consider the tubular geometry at the nanoscale, in which several-nanometre-thick wall layers provide another confinement system during crystallisation. Indeed, such a confinement phenomenon was observed during crystal growth in similarly grown ALD nanotubes.²² The nanotubular geometry remained intact upon thermal treatments. Therefore, a new model that is dominated by surface energy is required if the wall thickness is much thinner than the grain size. In a nanotube, the disc-like geometry for describing a crystalline grain is appropriate when the tube radius is much larger than the wall thickness (e), as shown in Fig. 2(a). Note that introducing the curvature of the nanotube in the disc makes the solution of the problem complex. That is, we assume that the curvature effect could be neglected when the crystal's diameter (defined as r) is much shorter than the tube radius (Fig. 2(a)). As noted below, such an assumption is reasonable by demonstrating that the resulting r upon crystallisation was observed to be well below the tube radius. The effect of the curvature could be treated as a source of non-hydrostatic strain within the crystal, if needed. On the basis of the above assumptions, Fig. 2(a) summarises the descriptions for the free energy of crystal developed along with the wall layers (the detailed derivations can be found in the ESI†). Note that this model would validate the nucleation of crystal grains during the film growth. The fact that the as-grown zirconia is amorphous makes our calculation consistent, as will be seen.

Fig. 2 (a) Surface energy description for the tubular geometry at nanoscale (G, total free energy; r, radius; e, thickness of disc crystal; ψ, volume free energy; σ, surface free energy of the disc-like crystal). (b) Phase diagram of nanocrystalline ZrO₂ with different free energy modifications (solid line for disc-like geometry, and dotted line for sphere).

We first start our calculation by assuming that there are no strain energies inside the zirconia crystals and by next assuming that hydrostatic and non-hydrostatic strain energies come from lattice defects, curvatures and so on. The internal hydrostatic strains are negligible because approximately 5% volume expansion during the monoclinic–tetragonal transformation could be easily accommodated inside the template pores. Because Young's modulus of alumina is higher than that of zirconia, volume changes in zirconia do not affect the surrounding alumina, and such volume changes occur in the inner pores. We also suspect that non-hydrostatic strain energies are low because the system was not treated mechanically (e.g., grinded or milled), and the amount of lattice defects, such as dislocations or ionic defects, should be low. The slow growth nature by ALD also helps accomplish this low number of defects. Moreover, the higher the deposition temperature is, the fewer defects there will be.

Fig. 2(b) shows the comparison of r_c as a function of temperature on spherical and disc geometries, which is indicative of the phase transition between t- and m-ZrO₂. One can see a strong influence of the geometry on the stabilisation of the tetragonal phase. Unambiguously, the disc-like geometry reduces significantly the maximum temperature over which t-ZrO₂ is stable and increases r_c (compare the solid line with the dotted one). We defined that r varies in accordance with the corresponding geometries (i.e., isotropic versus anisotropic crystal growth). Therefore, the excess surface/interface free energies become more significant in competing with the volumetric free energies in the tubular geometry. Both hydrostatic and non-hydrostatic strain energies further lead to the room-temperature phase stabilisation of nanocrystalline t-ZrO₂, if they were built up in similar quantities for both geometric cases (see the ESI† for more details).

3.3 Significance of the phase of initial nuclei

The initial phases of as-prepared ZNTs were likely determined in accordance with T_sub. In Fig. 3, the XRD results testify that the as-deposited ZNTs at 150 and 200 °C were almost amorphous, with no crystalline zirconia detected. The XRD results at 250 °C exhibited small and broad (but still detectable) diffraction peaks (black line in the right panel), which are indicative of m-ZrO₂ and t-ZrO₂ (Joint Committee on Powder Diffraction Standards, JCPDS no. 37-1484 and no. 50-1089, respectively). Peaks with asterisk marks were also visible from the Al metal used as the PAM substrate. Upon subsequent annealing at 700 °C, ZNTs with pure t-ZrO₂ were observed when using samples as-grown at 150 and 200 °C, and ZNTs with a mixture of m-ZrO₂ + t-ZrO₂ using samples prepared at 250 °C. Although the former followed our prediction, as calculated above, the latter requires further understanding on the transformation mechanisms. This issue might provide one a clue regarding the experimental controversy on the phase stabilisation mechanism of nanocrystalline t-ZrO₂.

Fig. 3 (a–c) Phase selectivity in the formation of ZNTs. XRD patterns of ZNTs grown at different T_sub before (black lines) and after (red lines) annealing. T_sub is given as the heading in each column ((a), 150 °C; (b), 200 °C; (c), 250 °C), and the post-annealing temperature was 700 °C (for 1 h) for all the samples. All of the ZNTs were prepared using 150 ALD cycles. The exposure duration of both TEMAZr and H₂O was commonly 5 s, and the purging time was 20 s. All peaks with asterisk marks are due to the Al foils. T denotes the tetragonal phase, and M denotes the monoclinic phase. In the present XRD results, no detectable peaks were found in the as-deposited ZNTs at T_sub = 150 and 200 °C, whereas distinct diffraction peaks were found and matched with the tetragonal and monoclinic phases of ZrO₂ in the as-grown ZNTs at T_sub = 250 °C. Upon thermal treatment, pure t-ZrO₂ was obtained when starting with amorphous ZNTs (T_sub = 150 and 200 °C). ZNTs consisting of the mixed phases of t-ZrO₂ and m-ZrO₂ (T_sub = 250 °C) did not stabilise into any single phase. (d) XRD patterns of ZNTs upon thermal treatments at different annealing temperatures (300–600 °C). Unambiguous splitting was evident for tetragonal (002) and (110) as well as (013) and (121) planes upon annealing (JCPDS #50-1089).

In addition to the XRD investigation, we complementarily studied the structures of our ZNTs using TEM techniques. The crystallisation behaviours of as-grown ZNTs with varying T_sub by performing the subsequent annealing at different temperatures up to ca. 700 °C were investigated. An insufficient amount of material is present for detection via normal XRD when investigating thin-walled ZNTs, although the PAMs contained ∼10¹⁰ pores per·cm². Fig. 4(a) and (b) shows a representative TEM micrograph of the as-deposited ZNT at T_sub = 150 °C. The tubular structure of a ZNT with distinct darker lines as walls is clearly observed. A few darker grains are assumed to be quite small crystallites with the diameter of far less than 10 nm. Remarkably, diffuse rings appeared in the SAED patterns of the TEM analysis (inset of Fig. 4(a)) for both of the samples (grown at T_sub = 150 and 200 °C), despite the lack of detectable crystalline polymorphs in the XRD observation (Fig. 3). The samples grown at T_sub = 200 °C showed more distinct ring patterns than did those at 150 °C (inset of Fig. 5(a)). In higher magnified TEM images, indeed, nanocrystallites were found with distinct lattice images (yellow dotted circles in Fig. 4(b) and (d)). The lattice spacing of the nanocrystallites matched well with tetragonal rather than monoclinic polymorph. Therefore, we were able to conclude that the as-grown ZNTs (T_sub = 150 and 200 °C) have the structures of nanocrystalline t-ZrO₂ embedded in an amorphous (α) matrix, rather than being completely amorphous. The average sizes of nanocrystalline t-ZrO₂ within the α-matrix were 4.3 ± 1.4 nm (T_sub = 150 °C) and 6.8 ± 2.2 nm (T_sub = 200 °C) in diameter (from the statistics by TEM images). Subsequent post-annealing at 300 and 400 °C results in the grain growth of t-ZrO₂ by detecting darker contrast as well as clear diffracted rings (Fig. S4(a and b) and (c and d),† respectively). Whereas ZNTs prepared at T_sub = 150 °C crystallised fully at 400 °C, ZNTs grown at 200 °C were transformed completely, even at 300 °C. Note that ZNTs with larger crystalline nuclei underwent rapid phase transformation. This behaviour might indicate that the amounts of initial nuclei determine the growth kinetics. We also note that the formation of amorphous zirconia as-grown would be the key in the phase selection processes, as will be discussed below.

Fig. 4 TEM observation of the as-grown ZNTs at T_sub = 150 °C (a and b) and 200 °C (c and d). The resulting hollow structures are observed in the TEM bright field images (panels a and c). Nanocrystallites with the sizes of 4.3 ± 1.4 nm (T_sub = 150 °C) and 6.8 ± 2.2 nm (T_sub = 200 °C) in diameter embedded in amorphous matrix are clearly observed in the higher resolution TEM micrographs (panels b and d). When T_sub was increased, the enhanced electron diffractions by larger nuclei on average are also evident in the insets of panels (a and c), even though no diffracted peaks were found in the XRD results (see Fig. 3).

Fig. 5 TEM analysis of the post-annealed ZNTs that had grown originally at T_sub = 150 °C. (a) Bright field image with (b) SAED patterns and (c) higher resolution micrograph (inset, fast Fourier transform result). Nanotubular, pure t-ZrO₂ was found in the post-annealed ZNTs without any structural disintegration upon high-temperature treatment. T denotes tetragonal.

3.4 Crystallization and phase selectivity during post-annealing

Upon post-annealing at 700 °C, fully crystallised ZNTs consisting of pure t-ZrO₂ were achieved, as shown in Fig. 5 and 6, where the initial phases of nuclei were a mixture of t-ZrO₂ and α-ZrO₂. The stabilised tetragonal phase of ZrO₂ in the nanoscale confinement of ∼10 nm-thick wall ZNTs shows that it is truly stable thermodynamically compared with the monoclinic phases and is not just kinetically stable. Furthermore, the annealed ZNTs exhibited no sign of structural disintegrations. According to Livage et al., the distribution function for the amorphous and tetragonal phase indicates that they have the same first-neighbour distance, which means that these phases have a similar short-range order.³⁵ Thus, less or no structural distortion might occur during the amorphous-to-tetragonal transformation. Although the TEM observation provides local information on the samples, we were able to detect forbidden phases, whose amounts were below the XRD detection limit. Indeed, every TEM investigation result on several tubes at various crystallisation stages was consistent with each other.

Fig. 6 TEM analysis of the post-annealed ZNTs that had grown originally at T_sub = 200 °C. (a) Bright field image with (b) SAED patterns and (c) higher resolution micrograph (inset, fast Fourier transform result), indicative of the formation of pure t-ZrO₂ upon high temperature treatment. T denotes tetragonal.

More remarkably, one can see that ZNTs grown at T_sub = 250 °C contain the mixture phases of m-ZrO₂ + t-ZrO₂, as shown in Fig. 7(a–c). The resulting grain sizes upon annealing were summarized in Table S3.† As in other low T_sub results, the resulting structures have amorphous zirconia (Fig. 7(c)). Under the identical annealing conditions (at 700 °C for 1 h), ZNTs (T_sub = 250 °C) were fully crystallised, following the original phases of nuclei by transforming into a mixed phase of m-ZrO₂ + t-ZrO₂ (Fig. 7(d–f)). Indeed, the resulting polymorphs of ZNTs were determined by the original ones that differ from the above-mentioned expectation by the surface free energy argument. We note that under the same post-annealing conditions (i.e., temperature, ambient air, and time) and systems (i.e., geometry, wall thickness, and interfaces with alumina), different types of stabilisation were monitored. This point presents a key advantage when we analyse the resulting microstructures because we are able to rule out any possible source that different preparation conditions lead to different free-energy modifications. For the possible reasons, qualitatively, we suspect that the physical confinement in nanotubular geometry may hinder both the grain growth when m-ZrO₂ (52.7 nm in Scherrer diameter) and t-ZrO₂ (17.1 nm in Scherrer diameter) compete with each other and the volume change, along with the extremely thin wall layers. It is noted that the confinement effects became more pronounced for the formation of m-ZrO₂ than that of t-ZrO₂ (see Table S3†) although the exact mechanism is not clear at the current stage.

Fig. 7 Structural analysis of ZNTs as-deposited at T_sub = 250 °C before (a–c) and after (d–f) post-annealing at 700 °C for 1 h. The initial mixed phases of nanocrystalline t-ZrO₂ and m-ZrO₂ embedded in an amorphous matrix were converted into fully crystallised t- and m-ZrO₂ NTs.

Note that the presence of impurities within our ZrO₂ might affect to the phase stabilisation observed. The incorporation of two different types of impurities could be considered: one is undesired elements such as N and C coming from incomplete chemical reactions of the present ALD chemistry at given thermodynamic conditions; the other is possible doping via the inter-diffusion between alumina templates and ALD zirconia during the thermal treatments. Regarding the former type, other researchers have reported that the ALD-grown zirconia by using TEMAZr as a precursor result in the contamination of N, C, and OH. The amounts of these impurities can be reduced by lowering the activation energy involving plasma and ozone in the reaction, for example.^36–38 In this study, we used the conventional thermal ALD reactor, and our XPS investigation showed trace amounts of impurities upon reaction, as expected (see Fig. S5†). Nevertheless, the starting ZNTs were prepared using the same chemistry and inherently had the similar amounts of impurities. If we assume that such impurities might result in any contributions upon thermal treatments, then the possible effects of the hydrostatic strains on the stabilisation should be treated equally.

Aluminium that originated from the alumina templates is another possible source of impurities. The interfacial diffusion of Al into ZNTs could also lead to the stabilisation of the other polymorphs of zirconia.³⁹ The possible formation of the cubic phase might not be observed when characterising the samples using laboratory XRD equipment. For example, the position of cubic (200) diffraction is very similar with those of tetragonal (002) and (110) planes. Without resolving distinctive tetragonal peaks, it is challenging to determine whether the cubic structures were created. However, we resolved a clear manifestation of tetragonal (002) and (110) planes via finely scanned XRD results, which is indicative of the presence of t-ZrO₂ (data not shown here). Only for the case of doping Y (approximately 8 wt%) into ZrO₂ was the cubic zirconia stabilised,⁴⁰ as confirmed by observing both cubic (200) and (311) peaks from synchrotron X-ray diffraction analysis results (not shown here). A detailed discussion of these results will be published elsewhere. Therefore, the template effects could be ruled out in our stabilisation study.

The calculation results also support the above observation in a quantitative manner. The fact that crystallization occurs in the ‘amorphous’ matrix fits the present assumptions well. Regarding the samples prepared at 150–200 °C, the resulting phases behave as aggregated grains. That is, stable t-ZrO₂ is expected for a wide range of grain radii, whereas the non-hydrostatic strain is low. In contrast, the samples grown at 250 °C behave in a similar manner and produce small tetragonal grains. However, the larger tetragonal grains formed during the deposition behave as though they are isolated in a nanocrystalline matrix and then turn into monoclinic, more stable larger grains (see the ESI† for more details). In the case of non-hydrostatic strains, as in our curved structures, near room temperature, m-ZrO₂ would be predominant for the isolated grains. t-ZrO₂ is only observed for small grains and nearly disappears if non-hydrostatic strain increases. For the aggregated case, t-ZrO₂ is stable, except for high non-hydrostatic strains. Under hydrostatic strains, m-ZrO₂ is only dominant for the isolated grain and only for very low strains at room temperature; even at small strains, t-ZrO₂ is fully stabilised. That is, the initially mixed phase of m-, t-, and α-ZrO₂ become strained (most probably non-hydrostatically) in tubular geometry at the nanoscale during annealing at elevated temperatures. Once m-ZrO₂ exists within the as-deposited ZNTs that were grown at higher deposition temperature and is thus larger, m-ZrO₂ competes with t-ZrO₂, which results in the fully crystallised ZNTs possessing m-ZrO₂ + t-ZrO₂. Therefore, the present results might give us not only an insight into the room temperature stabilisation of nanocrystalline t-ZrO₂ but also a plausible strategy on how to tailor the polymorphs of ZrO₂ nanostructures in fine-tuning material properties by carefully selecting the types and the amounts of the initial nuclei.

4. Conclusions

We fabricated ZNTs via template-directed ALD techniques with excellent conformality and phase selectivity of either pure t-ZrO₂ or a mixed phase of m-ZrO₂ + t-ZrO₂. We studied the conformality in forming ZNTs by changing the precursor exposure time. The measured results were analysed with Gordon's equation, and a simple relation between the aspect ratio and exposure time was obtained, exhibiting a practical guideline to access a desired aspect ratio of ZNTs. Upon thermal treatments, we obtained the resulting polymorphs of ZNTs to be tetragonal and/or monoclinic ZrO₂, which are strongly dependent on a combined effect between the original deposition and post-annealing temperatures. We further elucidated that the initially grown phases and the types/amounts of crystalline nuclei in the as-deposited nanotubes determine the final microstructures. The present results help understand the controversial results on the phase stabilisation of nanocrystalline t-ZrO₂.

Acknowledgements

We acknowledge the financial support from the National Research Foundation of Korea grant, funded by the Korean Government (MEST) (NRF-2013R1A2A2A01068499, 2012M3A7B4049986, and 2014M3A7B4052201) and funding by the Korean Government Ministry of Knowledge Economy. This work was supported in part by the Agency for Defense Development (ADD) of the Republic of Korea.

Notes and references

1. N. Q. Minh, J. Am. Ceram. Soc., 1993, 76, 563.
2. R. M. McMeeking and A. G. Evans, J. Am. Ceram. Soc., 1982, 65, 242.
3. D. Vanderbilt, X. Zhao and D. Ceresoli, Thin Solid Films, 2005, 486, 125.
4. C. Piconi and G. Maccauro, Biomaterials, 1999, 20, 1.
5. (a) J. Will, A. Mitterdorfer, C. Kleinlogel, D. Perednis and L. J. Gauckler, Solid State Ionics, 2000, 131, 79; (b) B. C. H. Steele and A. Heinzel, Nature, 2001, 414, 345; (c) E. D. Wachsman and K. T. Lee, Science, 2011, 334, 935.
6. T. Yamaguchi, Catal. Today, 1994, 20, 199.
7. D. R. Clarke and C. G. Levi, Annu. Rev. Mater. Res., 2003, 33, 383.
8. G. Johner and K. K. Schweitzer, Thin Solid Films, 1984, 119, 301.
9. W.-J. Qi, R. Nieh, B. H. Lee, L. G. Kang, Y. Jeon and J. C. Lee, Appl. Phys. Lett., 2000, 77, 3269; C. M. Perkins, B. B. Triplett, P. C. McIntyre, K. C. Sarawat, S. Haukka and M. Tuominen, Appl. Phys. Lett., 2001, 78, 2357.
10. M. N. Rahaman, A. Yao, B. S. Bal, J. P. Garino and M. D. Ries, J. Am. Ceram. Soc., 2007, 90, 1965.
11. P. Li, I.-W. Chen and J. E. Penner-Hahn, Phys. Rev. B: Condens. Matter Mater. Phys., 1993, 48, 10063.
12. P. Li, I.-W. Chen and J. E. Penner-Hahn, J. Am. Ceram. Soc., 1994, 77, 118.
13. F. Namavar, G. Wang, C. L. Cheung, R. F. Sabirianov, X. C. Zeng, W. N. Mei, J. Bai, J. R. Brewer, H. Haider and K. L. Garvin, Nanotechnology, 2007, 18, 415702; J. Chevalier, L. Gremillard, A. V. Virkar and D. R. Clarke, J. Am. Ceram. Soc., 2009, 92, 1901.
14. S. Shukla and S. Seal, Int. Mater. Rev., 2005, 50, 45; W. Li, H. Huang, H. Li, W. Zhang and H. Liu, Langmuir, 2008, 24, 8358.
15. R. C. Garvie, J. Phys. Chem., 1965, 69, 1238; R. C. Garvie, J. Phys. Chem., 1978, 82, 218; R. C. Garvie and M. F. Goss, J. Mater. Sci., 1986, 21, 1253.
16. B. Tyagi, K. Sidhpuria, B. Shaik and R. V. Jasra, Ind. Eng. Chem. Res., 2006, 45, 8643; M.-H. Tsai, S.-Y. Chen and P. Shen, J. Chem. Phys., 2005, 122, 204708; M. Slavati-Niasari, M. Dadkhah and F. Davar, Inorg. Chim. Acta, 2009, 362, 3677; M. M. Rachad and H. M. Baioumy, J. Mater. Process. Technol., 2008, 195, 178; B. L. Kirsch and S. H. Tolbert, Adv. Funct. Mater., 2003, 13, 281.
17. A. Christensen and E. A. Carter, Phys. Rev. B: Condens. Matter Mater. Phys., 1998, 58, 8050; N.-L. Wu, T.-F. Wu and I. A. Rusakova, J. Mater. Res., 2001, 16, 666; M. W. Pitcher, S. V. Ushakov, A. Navrotsky, B. F. Woodfield, G. Li, J. Boerio-Goates and B. M. Tissue, J. Am. Ceram. Soc., 2005, 88, 160.
18. Y. Liu, C. Zheng, W. Wang, Y. Zhan and G. H. Wang, J. Am. Ceram. Soc., 2002, 85, 3120.
19. H. Cao, X. Qiu, B. Luo, Y. Liang, Y. Zhang, R. Tan, M. Zhao and Q. Zhu, Adv. Funct. Mater., 2004, 14, 243.
20. C. Jiang, F. Wang, N. Wu and X. Liu, Adv. Mater., 2008, 20, 4826.
21. J.-S. Lee, H. Kim, P.-R. Cha, J. Kim and H. Shin, J. Nanosci. Nanotechnol., 2012, 12, 3177.
22. M. Kim, C. Bae, H. Kim, H. Yoo, J. M. Montero Moreno, H. S. Jung, J. Bachmann, K. Nielsch and H. Shin, J. Mater. Chem. A, 2013, 1, 14080.
23. H. Tsuchiya, J. M. Macak, I. Sieber and P. Schmuki, Small, 2005, 1, 722; H. Tsuchiya, J. M. Macak, L. Taveira and P. Schmuki, Chem. Phys. Lett., 2005, 410, 188; S. Berger, J. Faltenbacher, S. Bauer and P. Schmuki, Phys. Status Solidi RRL, 2008, 2, 102; J. Zhao, R. Xu, X. Wang and Y. Li, Corros. Sci., 2008, 50, 1593; L. Guo, J. Zhao, X. Wang, R. Xu and Y. Li, J. Solid State Electrochem., 2009, 13, 1321; Y. Shin and S. Lee, Nanotechnology, 2009, 20, 105301; D. Fang, S. Liu, Z. Luo, C. Xiong and W. Xu, Appl. Surf. Sci., 2012, 258, 6217.
24. D. Fang, K. Huang, Z. Luo, Y. Wang, S. Liu and Q. Zhang, J. Mater. Chem., 2011, 21, 4989.
25. C. N. R. Rao, B. C. Satishkumar and A. Govindaraj, Chem. Commun., 1997, 1581; J. Bao, D. Xu, Q. Zhou and Z. Xu, Chem. Mater., 2002, 14, 4709; H. Ogihara, M. Sadakane, Y. Nodasaka and W. Ueda, Chem. Mater., 2006, 21, 4981; M. C. Tsai, G. T. Lin, H. T. Chiu and C. Y. Lee, J. Nanopart. Res., 2008, 10, 863.
26. H. Shin, D. K. Jeong, J. Lee, M. M. Sung and J. Kim, Adv. Mater., 2004, 16, 1197; D. Gu, H. Baumgart, G. Namkoong and T. M. Abdel-Fattac, Electrochem. Solid-State Lett., 2009, 12, K25.
27. Y. K. Mishra, S. Kaps, A. Schuchardt, I. Paulowicz, X. Jin, D. Gedamu, S. Freitag, M. Claus, S. Wille, A. Kovalev, S. N. Gorb and R. Adelung, Part. Part. Syst. Charact., 2013, 30, 775; I. Paulowicz, V. Hrkac, S. Kaps, V. Cretu, O. Lupan, T. Braniste, V. Duppel, I. Tiginyanu, L. Kienle, R. Adelung and Y. K. Mishra, Adv. Electron. Mater., 2015 DOI:10.1002/aelm.201500081.
28. M. Kim, H. Kim, C. Bae, J. Lee, H. Yoo, J. Montero Moreno and H. Shin, J. Phys. Chem. C, 2014, 118, 26789.
29. B. Lee, K. J. Choi, A. Hamde, M. J. Kim, R. M. Wallace, J. Kim, Y. Senzaki, D. Shenai, H. Li, M. Rousseau and J. Suydam, Microelectron. Eng., 2009, 86, 1658; D. M. Hausmann, E. Kim, J. Becker and R. G. Gordon, Chem. Mater., 2002, 14, 4350.
30. C. Bae, H. Yoo, S. Kim, K. Lee, J. Lim, M. M. Sung and H. Shin, Chem. Mater., 2008, 20, 756; C. Bae, S. Kim, B. Ahn, J. Kim, M. M. Sung and H. Shin, J. Mater. Chem., 2008, 18, 1362; H. Yoo, C. Bae, Y. Yang, S. Lee, M. Kim, H. Kim, Y. Kim and H. Shin, Nano Lett., 2014, 14, 4413; H. Yoo, M. Kim, C. Bae, S. Lee, H. Kim, T. K. Ahn and H. Shin, J. Phys. Chem. C, 2014, 118, 9726.
31. R. G. Gordon, D. Hausmann, E. Kim and J. Shepard, Chem. Vap. Deposition, 2003, 9, 73.
32. I. Perez, E. Robertson, P. Banerjee, L. Henn-Lecordier, S. J. Son, S. B. Lee and G. W. Rubloff, Small, 2008, 4, 1223.
33. J. Y. Kim, J. H. Kim, J. H. Ahn, P. K. Park and S. W. Kang, J. Electrochem. Soc., 2007, 154, H1008.
34. J. W. Elam, D. Routkevitch, P. P. Mardilovich and S. M. George, Chem. Mater., 2003, 15, 3507–3517.
35. J. Livage, K. Doi and C. Mazieres, J. Am. Ceram. Soc., 1968, 51, 349.
36. D. M. Hausmann, E. Kim, J. Becker and R. G. Gordon, Chem. Mater., 2002, 14, 4350.
37. W. Wenke, T. Tina, P. Patrick, D. Maximilian, R. Stefan, S. Jonas, S. Konrad, S. Mahdi, D. E. Simon, O. Susanne, E. Elke and T. Bernhard, J. Vac. Sci. Technol., A, 2013, 31, 01A123.
38. S. Yun, J. Lim and J. Lee, Electrochem. Solid-State Lett., 2004, 7, F81.
39. S. Spiga, R. Rao, L. Lamagna, C. Wiemer, G. Congedo, A. Lamperti, A. Molle, M. Fanciulli, F. Palma and F. Irrera, J. Appl. Phys., 2012, 112, 014107.
40. J. Shim, C. Chao, H. Huang and F. B. Prinz, Chem. Mater., 2007, 19, 3850.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra14481g

---