A. Microstructural characterization

Figure 1 shows the collection of XRD patterns for the samples calcined at different temperatures and indicates the synthesized zinc aluminate powders contain a single phase with the cubic spinel crystal structure (PDF # 74–1136). Figure 1 also reveals sharpening of the peaks with increasing calcination temperature, suggesting grain enlargement due to coarsening. The grain size calculated from the XRD is 25.0 ± 0.9 nm for the powder calcined at 750 °C and 40.6 ± 1.9 nm for the one treated at 900 °C, as shown in Table I. Diffraction peaks for the powder calcined at 1200 °C were too narrow to allow reliable grain size calculation and was estimated to be >100 nm. Figure 2 shows TEM images for the ZnAl₂O₄ nanoparticles calcined at the different temperatures, confirming generally round-shaped particles, and sizes consistent with those calculated from XRD patterns. Solid–solid interfaces between particles can also be seen from these micrographs, suggesting partial sintering caused by calcination. To check ordering of the spinel phase, MAS NMR was performed and showed only a small percentage of Al³⁺ ions occupying tetrahedral positions (2.7, 1.9, and 1.1% for the powders calcined at 750, 800, and 900 °C, respectively). This is consistent with the expectation that Zn²⁺ ions have strong preference for tetrahedral sites,^{Reference Navrotsky and Kleppa16} and therefore, ZnAl₂O₄ is expected to be a normal spinel with a minimal degree of inversion (if any).

FIG. 1.

X-ray diffraction patterns of ZnAl₂O₄ calcined for 10 h at different temperatures (a) 750 °C, (b) 800 °C, (c) 900 °C, and (d) 1200 °C.

TABLE I.

Sample characterization, water adsorption, and drop solution data.

FIG. 2.

Transmission electron micrographs of ZnAl₂O₄ powders calcined at different temperatures for 10 h: (a) 750 °C, (b) 800 °C, (c) 900 °C, and (d) 1200 °C. Arrows show an example of solid–solid interface between particles.

Table I summarizes surface area of different powders based on the crystallite sizes from XRD and measured by nitrogen adsorption (BET). It is no surprise that powders treated at higher temperature have lower surface areas, but the BET results are generally smaller than the areas calculated based on the crystallite size. This is consistent with the fact that BET is a direct assessment of the available surface area, while the results from crystallite sizes represent the total interface area surrounding a particle, implying that both solid–vapor (surface) and solid–solid (grain boundary) areas are included. A direct comparison between the areas from BET and those from crystallite sizes allow the estimation of the grain boundary area using the following equation:

(1)$${\rm{GBA}} = {{{\rm{XSA}} - {\rm{BSA}}} \over 2}\quad ,$$

where XSA and BSA are surface areas determined by grain size and BET, respectively. Since it takes two surfaces to form one solid–solid interface, the factor ½ is needed in Eq. (1).

Note that although the surface area (BET) decreases with increasing calcination temperature, the grain boundary area remains almost constant (∼10–11 m²/g) from 750 to 900 °C. This indicates that as the temperature is increasing, coarsening is taking place but without significant densification. This suggests that the main mechanism for coarsening is Ostwald ripening, as grain growth would lead to grain boundary area decrease, and sintering would lead to neck formation at this stage of the process.^{Reference Castro, Castro and van Benthem17,Reference Castro18}

B. Water adsorption

Investigation of the water adsorption behavior on the surface of the as-synthesized zinc aluminate was performed by recording the adsorbed amount and the enthalpy of adsorption during the exposure of the powder to controlled microdosages of vapor. Figure 3 shows the results for differential enthalpy of adsorption as a function of surface coverage (H₂O per nm²) for two experiments conducted under the same conditions. The results for both tests are consistent, confirming the reliability of the method. The curves show a typical behavior with highly exothermic adsorption behavior at low coverages, and a systematic decrease followed by leveling. At the lowest measured coverage (0.49 H₂O per nm²), the enthalpy of adsorption is −155.46 kJ/mol, indicating a strong chemisorption of water on the anhydrous surface, which is consistent with a dissociative behavior. The measured heat of adsorption decreases (in magnitude) quickly as more water molecules adsorbed and reaches a first plateau around −35 kJ/mol at ∼7 H₂O per nm². This behavior is consistent with recent literature data on the water adsorption on γ-Al₂O₃, which has a similar spinel structure.^{Reference Castro and Quach13} At first, it may sound counterintuitive that the enthalpy of water adsorption is less exothermic than the enthalpy of liquefaction of water at 24 °C (∼−44 kJ/mol). However, though the enthalpy of adsorption is less negative than −44 kJ/mol, the process is still exothermic and simply suggests the existence of hydrophobic sites on ZnAl₂O₄ surfaces, similar to what Bolis et al.^{Reference Bolis, Fubini, Marchese, Martra and Costa19} reported for SiO₂ surfaces. As the coverage increases above 8 H₂O per nm², the enthalpy of water adsorption slowly converges to −44 kJ/mol.

FIG. 3.

Repeated experiments for water adsorption of ZnAl₂O₄ at 24 °C. Differential enthalpy of adsorption converges to −44 kJ/mol (enthalpy of liquefaction of water) at high coverages.

The water adsorption behavior on the surface of ZnAl₂O₄ is indeed similar to that of γ-Al₂O₃ reported elsewhere.^{Reference Castro and Quach13} Although at the lowest dosage measured for ZnAl₂O₄ (0.49 H₂O per nm²), the enthalpy of adsorption is −155.46 kJ/mol for γ-Al₂O₃ and the value is −158.83 kJ/mol at 0.40 H₂O per nm². However, the heat of water adsorption on ZnAl₂O₄ converges to the enthalpy of liquefaction at a lower coverage (∼13 H₂O per nm²) as compared with γ-Al₂O₃ (∼16 H₂O per nm²). Assuming that the size of bound water is 0.25 nm² per H₂O,^{Reference Brunauer, Kantro and Weise20} this coverage is calculated to be roughly three monolayers for ZnAl₂O₄ nanoparticles, whereas for γ-Al₂O₃, −44 kJ/mol is only achieved after four monolayers.^{Reference Castro and Quach13} This contrast between ZnAl₂O₄ and γ-Al₂O₃ suggests that the latter surface holds water molecules more tightly.

The observed adsorption profile is very consistent with the proposed model of water adsorption in two stages reported elsewhere.^{Reference Castro and Quach13} The model describes an initial one-interface adsorption stage where water adsorbs in a dissociative manner. With increasing coverage, a transitional stage takes place before a two-interfaces profile is formed. Under this condition, the outer most water molecules suffer little influence from the surface of the particle and are in a liquid-like configuration. The achievement of this final stage is confirmed by the adsorption isotherm profile and by the enthalpy of adsorption being that of liquefaction of water. Once the liquid-like configuration is achieved by the adsorbed water molecules, it has been suggested that the surface energy of liquid water at 24 °C (0.072 J/m²) can be assigned to the surface itself.^{Reference Castro and Quach13} Therefore, the surface energy of previous hydration states can be calculated using the following equation^{Reference Castro and Quach13,Reference Arrouvel, Digne, Breysse, Toulhoat and Raybaud21} :

(2)$${\rm{\gamma }}_i^{\rm{S}} = {\rm{\gamma }}_{i - 1}^{\rm{S}} + {{\rm{\theta }}_i} \times {\Delta H_{{\rm{ads}}}}\quad ,$$

where ${\rm{\gamma }}_i^{\rm{S}}$ and ${\rm{\gamma }}_{i - 1}^{\rm{S}}$ are surface energies of states i and i − 1, respectively, θ_i is the increase in surface coverage from state i − 1 to state i, and ∆H _ads is enthalpy of adsorption during that process. Using this equation, the surface energy of ZnAl₂O₄ at different water coverages was calculated and is plotted in Fig. 4. From these water adsorption results, the anhydrous surface energy of ZnAl₂O₄ is determined to be 1.36 ± 0.08 J/m². [The error ±0.08 J/m² is a result of incorporating uncertainty in surface area measurements (BET) into calorimetric results during water adsorption.] This value is slightly smaller than for γ-Al₂O₃ (1.57 ± 0.05 J/m² from water adsorption experiments^{Reference Castro and Quach13} and 1.53 ± 0.40 J/m² from high-temperature drop solution calorimetry^{Reference Castro, Ushakov, Gengembre, Gouvea and Navrotsky22}), which is consistent with the water adsorption trend discussed above.

FIG. 4.

Surface energy of ZnAl₂O₄ at different coverages as calculated from water adsorption. Two data sets correspond to two water adsorption experiments performed under same controlled conditions.

Note from Fig. 4 that the surface energy decreases with increasing coverage at two distinct slopes, a steeper one from 0 to about 3 H₂O per nm² and another from that coverage on. This behavior is obviously linked to the differential heat of water adsorption in Fig. 3, which shows a much more exothermic effect on the adsorption of the first water layers. If one plots the surface energy decrease as a function of the relative pressure of water (or relative humidity) during the experiment, one observes that the massive surface energy change (and water coverage) occurs when the pressure increases from 0 to 0.01 (1% relative humidity). This suggests that the surface energy of the anhydrous state will rarely be observed under real processing conditions, where the relative humidity far exceeds 1%. Moreover, the humidity level will define the “effective” surface energy during processing.

For instance, having in mind, the application of the zinc aluminate nanoparticles for fabrication of ceramic parts through sintering in which surface energy is the main driving force,^{Reference Kellett and Lange12} considering the presented results, the humidity level of the processing environment becomes a very important parameter. That is, after compaction in the desired shape, nanoparticles are commonly exposed to relative humidity of at least 30% before densification. This will decrease the surface energy from the anhydrous 1.36 J/m² value to about 0.25 J/m² (effective surface energy) as it corresponds to roughly 10 H₂O per nm². This signifies a major decrease in the driving force for densification. Surely, sintering processes take place at high temperatures (typically 2/3 of the melting point), and one should expect the humidity levels to be very low under that condition. However, humidity in the powder will remain at the initial stages of sintering, leading to lower densification rates and prejudiced microstructural control. Controlling the exposure of the powder to water or establishing a drying schedule may help result in higher densification in ceramic materials. In fact, the role of water vapor atmosphere on sintering behavior has been addressed in the literature. It was shown that highly anhydrous MgO can achieve high densities under field-assisted sintering (>99%) at low temperature, allowing the formation of transparent ceramics.^{Reference Tran, Hayun, Navrotsky and Castro23} Moreover, tin dioxide densification decreases with increasing water relative pressure in the sintering chamber^{Reference Perazolli, Varela, Leite, Longo, Salgado and Filho24} and similar effect is seen for MgO^{Reference Anderson and Morgan25} (though neither effect has been related to a decrease in driving forces due to surface energy change and only kinetic arguments were discussed).

C. High temperature oxide melt solution calorimetry

An independent method to assess anhydrous surface energy of zinc aluminate was used to cross-check the data from water adsorption. High-temperature oxide melt solution calorimetry is a technique capable of providing both surface and grain boundary energies of oxides via measurements of enthalpy of dissolution. The method has been described elsewhere in more details,^{Reference Castro18,Reference Costa, Ushakov, Castro, Navrotsky and Muccillo26} but basically the calorimeter records the enthalpy of dissolution of samples prepared with distinct interface areas in a high temperature solvent and dropped from room temperature. If the bulk structure is kept the same for all specimens, the difference in the enthalpy of solution can be attributed to the energetic contribution of the interfaces, and with adequate microstructural analysis, interface energetics can be singly derived.^{Reference Costa, Ushakov, Castro, Navrotsky and Muccillo26} As the samples to be dissolved are very refractory, oxide melts are used as the solvents at high temperatures as described in the experimental section.

This sort of experiment requires nanoparticles to allow for a large amount of interface areas. That is, interface energies are typically small quantities that can be easily overseen due to other heats evolving during a dissolution process. Nanoparticles enable a multiplication of the interface contribution, allowing measurable and reliable signals. However, due to the inherent reactivity of nanoparticles, if exposed to water vapor (even local humidity), large amount of adsorbed water are expected. During dissolution in the calorimeter, water desorption, evaporation, and heating will take place.

To account for water-related energetic contributions, a thermochemical cycle is required. To control the amount of adsorbed water, the samples were rested for equilibration with the environment in a testing room with controlled humidity and temperature as described in the experimental section. From TG experiments, the adsorbed water contents on different ZnAl₂O₄ powders were determined and are listed in Table I. Combining TG and water adsorption data, one can correct the enthalpy of dissolution data for ZnAl₂O₄ with different surface and interface areas using the following thermodynamic cycle:

$$\eqalign{ {\rm{ZnAl}}_{\rm{2}} {\rm{O}}_{\rm{4}} \cdot x{\rm{H}}_{\rm{2}} {\rm{O}}_{{\rm{(surface,}}\,{\rm{24}}\,{\rm{^\circ C)}}} = {\rm{ ZnAl}}_{\rm{2}} {\rm{O}}_{\rm{4}} _{{\rm{(soln}}{\rm{.,}}\,{\rm{702}}\,{\rm{^\circ C)}}} \,+ x{\rm{H}}_{\rm{2}} {\rm{O}}_{{\rm{(g,}}\,{\rm{702}}\,{\rm{^\circ C)}}} \quad \;\,\,\,\,\,\,\,\hskip-1pt\Delta H_{\rm{1}} = \Delta H_{{\rm{ds}}} \cr x{\rm{H}}_{\rm{2}} {\rm{O}}_{{\rm{(g,}}\,{\rm{702}}\,{\rm{^\circ C)}}} = x{\rm{H}}_{\rm{2}} {\rm{O}}_{{\rm{(g,}}\,{\rm{24}}\,{\rm{^\circ C)}}} \;\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\!\!\Delta H_{\rm{2}} = x{\rm{(}} - {\rm{25}}{\rm{.0 \pm 0}}{\rm{.1) kJ/mol}} \cr x{\rm{H}}_{\rm{2}} {\rm{O}}_{{\rm{(g,}}\,{\rm{24}}\,{\rm{^\circ C)}}} = x{\rm{H}}_{\rm{2}} {\rm{O}}_{{\rm{(surface,}}\,{\rm{24}}\,{\rm{^\circ C)}}} \quad \quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\!\!\Delta H_{\rm{3}} = x\Delta H_{{\rm{ads}}} \cr {\rm{ZnAl}}_{\rm{2}} {\rm{O}}_{\rm{4}} _{{\rm{(nano,}}\,{\rm{24}}\,{\rm{^\circ C)}}} = {\rm{ZnAl}}_{\rm{2}} {\rm{O}}_{\rm{4}} _{{\rm{(soln}}{\rm{.,}}\,{\rm{702}}\,{\rm{^\circ C)}}} \,\,\quad\quad\quad\quad\quad\quad\quad\quad\quad\quad\!\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\Delta H_{\rm{4}} = \Delta H_{{\rm{ds\_corr}}} \cr \Delta H_{\rm{4}} = \Delta H_{\rm{1}} + \Delta H_{\rm{2}} + \Delta H_{\rm{3}} \cr} $$

The enthalpy of drop solution of microsized ZnAl₂O₄ (sample calcined at 1200 °C for 10 h with negligible surface and interface areas) measured in this work was 201.24 ± 1.10 kJ/mol. Navrotsky reported the solution enthalpy of ZnAl₂O₄ in the same solvent (lead borate) at 700 °C to be 92.19 ± 1.02 kJ/mol.^{Reference Navrotsky27} The heat content of ZnAl₂O₄ from 24 to 702 °C is calculated to be 107.54 kJ/mol.^{Reference Robie and Hemingway28} Combining these two literature data give a calculated enthalpy of drop solution of 199.73 ± 1.02 kJ/mol, which is very close to our experimental result (within experimental uncertainties). As one goes from bulk ZnAl₂O₄ to nanopowders, surfaces and grain boundaries introduce excess energies that are released during high-temperature drop solution calorimetry. Therefore, the drop solution enthalpy (after water correction) for these nanopowders should be less endothermic (i.e., more exothermic) as represented in the following equation:

(3)$$\Delta H_{{\rm{ds\_corr,nano}}} = \Delta H_{{\rm{ds, \,bulk}}} - {\rm{\gamma }}_{\rm{s}} {\rm{BSA}} - {\rm{\gamma }}_{\rm{i}} {\rm{IA}}\quad {\rm{,}}$$

where γ_s and γ_i are anhydrous surface energy and interface energy, respectively.

Though one could consider that a solution for the linear system composed of three equations built from the data from Table I (grain boundary and surface areas) into Eq. (3) as a manner to estimate the interface energies because all three nanopowders have very similar grain boundary areas (∼10–11 m²/g), this approach is mathematically impractical. However, the plot of ΔH _{ds_corr, nano} versus BSA should be linear with slope equals −γ_s (anhydrous surface energy) and y-intercept equals (ΔH _{ds, bulk} – γ_i IA) (Fig. 5). By using a linear fitting to the data plotted in Fig. 5, the anhydrous surface energy of ZnAl₂O₄ was determined to be 1.29 ± 0.33 J/m². This value is very consistent with that obtained from water adsorption on ZnAl₂O₄ nanoparticles presented earlier in this paper (1.36 ± 0.08 J/m²). To the best of the authors’ knowledge, no surface energy value for ZnAl₂O₄ has been reported previously. A list of surface energies of various oxides with different crystal structures are given in Table II. The anhydrous surface energy of oxide spinels listed tends to be from 1 to 2 J/m², which is generally smaller than surface energies for metals and rocksalt-structured oxides. The data from the present work fall well within the spinel value range, reinforcing the evidence for this universal trend.

FIG. 5.

Enthalpy of drop solution of ZnAl₂O₄ after correction for water adsorption on surfaces. Note that these three ZnAl₂O₄ powders have similar interface areas (∼2000 m²/mol).

TABLE II.

Surface energies of oxides with different crystal structures.

From the y-intercept in Fig. 5, (ΔH _{ds, bulk} – γ_i IA) is determined to be 202.40 ± 1.35 kJ/mol. Given that the drop solution enthalpy of bulk ZnAl₂O₄ is 201.24 ± 1.10 kJ/mol, the product γ_i IA = −1.16 ± 1.74 kJ/mol. From this, we may conclude that the contribution of grain boundary areas (1985 m²/mol in average) to the energetics of the ZnAl₂O₄ nanopowders used in this study is too small and within the uncertainty of the measurements, suggesting small grain boundary energy. As discussed above and pointed out by Navrotsky et al.,^{Reference Navrotsky, Ma, Lilova and Birkner29} the surface energies of spinel structures are expected small as compared with other crystal structures such as rocksalt. Results from our work also suggest that the grain boundary energy of ZnAl₂O₄ may follow similar trend. It is important to note that the powders prepared for this work were calcined at different temperatures, and the resulting grain boundaries are due to coarsening taking place during the 10 h of treatment (details in the Sec. II). This long annealing may benefit the presence of low energy boundaries. More studies are though required to estimate reliable grain boundary energies for this compound.

Compared with other oxides listed in Table II, the surface energy of ZnAl₂O₄ is on the lower end and slightly smaller than that of isostructural γ-Al₂O₃ and MgAl₂O₄. As discussed previously in the water adsorption section, lower anhydrous surface energy indicates weaker binding between water and surface sites of ZnAl₂O₄. This may contribute to good catalyst performance of ZnAl₂O₄ as water on its surface can be more readily displaced and surface sites become more energetically accessible to other species. Birkner and Navrotsky^{Reference Birkner, Nayeri, Pashaei, Najafpour, Casey and Navrotsky33} argued that the significantly smaller surface energy of calcium manganese oxide (as compared with Mn₃O₄, Mn₂O₃, and MnO₂) was one of the major factors contributing to the superior catalyst performance of this compound in comparison with its manganese oxide counterparts. Similarly, Ma et al.^{Reference Ma, Castro, Zhou and Navrotsky34} suggested that one of the reasons behind the better gas-sensing ability of SnO₂ as compared with its isostructural rutile (TiO₂) was the former’s lower surface energy. The low surface energy of ZnAl₂O₄ supports that this spinel is a strong candidate for high performance catalysts as compared with other spinels, such as gamma-alumina and magnesium aluminate.