Volatile organic compounds (VOCs), emitted to the atmosphere by industrial and anthropic processes, are chemically associated to products derived from petrochemical processes and fuel combustion (e.g., automobile exhaust gas). They become hazardous for the environment, and particularly, for the human health.
Several processes, such as adsorption, thermal or catalytic oxidation, have been applied for the conversion of VOCs. Catalytic oxidation to carbon dioxide and water is one of the most promising methods for removing VOCs to carbon dioxide and water, since this process occurs at temperatures lower than those adopted by other non-catalytic processes, and can be used for effluent streams containing low VOC concentration [1
Supported noble metal systems (Pt, Pd, Au, Rh) are recognized as the most active catalysts. However, the high cost of these metals and the easy of deactivation by sintering or poisoning limit industrial application [9
Non-noble metal oxides catalysts (mainly Cu, Co, Ce, Mn, Cr, Fe) as simple, mixed or supported systems, are widely investigated for the oxidation of VOCs, due to many advantages, such as their low cost, low temperature activity, long lifetime, and regeneration capability, among others [10
]. Interestingly, mixed oxide systems and supported systems perform better than the single oxide. This behavior is attributed to the high dispersion of the active phase, metal-support interaction, and to the increase of catalysts’ reducibility, which are critical parameters for VOC oxidation activity [16
is a promising support [19
] as well as an active phase [23
] in oxidation catalysis, because of its high thermal stability, specific surface area and acid-base properties. In addition, the phase composition and physicochemical properties of ZrO2
strongly depend on precursors, preparation methods and conditions. As a support, zirconia can be stabilized in the tetragonal form within temperature range 500–900 °C by addition of foreign species, yielding to variations in the availability of active sites for the catalytic reaction, according to their local symmetries [25
]. On the other hand, the ability of the zirconia lattice to incorporate bi- and tri-valent cations, generates defective solid solutions with oxygen vacancies. In this sense, it is known that the number of bulk and sub-surface defects influences the properties and activity of materials, as it facilitates the migration of oxygen species from bulk to surface.
Although catalytic total oxidations of VOCs have been widely investigated, the investigation in specific types of VOCs, such as polycyclic aromatic hydrocarbons (PAHs), is much lower. PAHs are mainly released by combustion processes [8
], and have been identified as carcinogenic pollutants. Naphthalene is, among them, the most volatile one, therefore, efforts have to be made to reduce its ecotoxicological impact. A few studies have tested the catalytic oxidation of naphthalene using both noble metals [27
] and non-noble metal based catalysts, investigating the role of both the supported species and the support. As for non-noble metal active phase systems, including MnOx
, CuO, TiO2
, and CeO2
, the catalytic performances were found to depend not only on redox properties, but also, on the interaction between reactant molecules and catalyst surface [18
The aim of this work is the development of CuZ, CoZ, and NiZ systems, obtained by the impregnation of ZrO2·nH2O (hydrogel hydrous zirconium oxide) with Cu-, Co-, and Ni-transition metal solutions, to be employed in the catalytic naphthalene oxidation reaction.
BET surface area, SEM-EDS, XRD, DSC, DRS-UV–vis, and TPR, among others, gave bulk information, whereas the Raman spectroscopy allowed the knowledge of the molecular behavior of species at surface level, where volatile organic compounds, oxygen, metal oxide and support can be simultaneously interacting. The physicochemical characterization of fresh and thermally treated materials, was related to the activity of each system, focusing on the comparative behavior related to the symmetry, coordination, and reduction of metal species on the support, and on the structural variation of the support caused by thermal treatments, as well as by the presence of added species.
2. Results and Discussion
collects data for both the support and the supported samples corresponding to specific surface area, transition metal (TM) to zirconium atomic ratios obtained by the semiquantitative surface analysis by energy dispersive X-ray spectroscopy and the transition metal/Zr nominal ratio, structural type of metal oxide or stabilized zirconia polymorphic phase, as well as the reduction capability (expressed as H2
consumed), as will be discussed later. For all the catalysts, surface ratios (EDS) are greater than nominal ratios, suggesting that metallic species are mostly present on the surface of the support. The increase in the time of the thermal treatment generates an increase in the surface ratio TM/Zr. The CoZ 5 h catalyst is the sample that show higher TM surface content.
Thermal treatments and the presence of foreign species affected some support properties, including phase transition and sintering. The contraction of the specific surface area can be associated to the thermal treatment and to the interaction with the supported species; however, the heating time seems not to be a decisive factor for specific surface area modifications.
XRD patterns of undoped ZrO2
(Z) and fresh NiZ catalysts heated for 2 h showed the ZrO2
monoclinic phase (reflections located at 28.2°, 31.5°, 34.2°, and 50.2° of 2θ, PDF N° 03-065-1025), whereas the metastable tetragonal phase (reflections located at 30.0°, 35.0°, and 50.0° of 2θ, PDF N° 01-089-7710) was identified for the CoZ and CuZ samples (Figure 1
A). The calcination for 5 h in air did not modify the structural properties of CoZ and CuZ samples. On the contrary, for NiZ system, the prolonged thermal treatment favored the formation of the metastable tetragonal phase.
As well documented in the literature [32
], the interaction between the supported species and a support surface from the hydrous state yields changes in support properties and in supported species. During thermal treatment, adsorbed species interacting with the support surface hinder crystallization and sintering processes favoring the dispersion of active species.
Depending on the interaction strength, supported species may end up as dispersed or separate phase (crystalline or amorphous) or dissolved in the support to give a solid solution [32
The effect of Cu, Co, and Ni on zirconia crystallization process was different, prevailing the metastable tetragonal form in CuZ and CoZ systems, and the monoclinic one in NiZ for short calcination time. The kinetically dependent process observed for NiZ system suggests that the interaction between zirconia and Ni(II) species is not only confined to the surface, but also involves incorporation into the zirconia structure [35
No appreciable changes were detected in the XRD patterns of CoZ and CuZ samples (thermal treatments for 2 h or 5 h) after catalytic tests. On the contrary, for the NiZ system, a strong contribution of the metastable tetragonal phase was already detected in the NiZ 2 h sample after catalytic runs, confirming the kinetic contribution to the interaction process (Figure 1
In addition to zirconia modifications, reflections due to NiO and Co3
spinel structure (typical lines located at 37.3°, 43.3°, 62.9°, 87.0° of 2θ, PDF N° 03-065-6920 [36
] and at 36.8°, 31.2°, 44.7°, 65.1° and 59.3° of 2θ, PDF N° 01-080-1533 [38
] respectively),were detected for NiZ and CoZ samples after thermal treatment for 5 h.
Likewise, no diffraction lines due to CuO (located 38.8°, 35.5°, 39.0° and 48.7° of 2θ, PDF N° 01-080-1917) were observed in CuZ patterns. This finding can be attributed to the presence of CuO species with crystal size smaller than the XRD detection capacity or to copper incorporation in the zirconia lattice, as already observed [41
Raman spectroscopy was used to determine the structure of the added species and of the zirconia support [43
], both in amorphous and crystalline form, as well as the structure adopted by the metal-O species, either supported or interacting. In fact, this spectroscopic technique is highly sensitive to oxygen displacement, owing to its high polarizability and to intermediate range order without long-range periodicity [44
According to group factor analysis, distinct and characteristic bands in the 100–650 cm−1 range are predicted for the monoclinic (18 Raman active modes) and tetragonal (6 Raman active modes) ZrO2 modifications. In particular, over the range 100–300 cm−1, the monoclinic phase has two bands at 179 and 191 cm−1, well separated from the tetragonal bands appearing at 147 and 261 cm−1. This evidence allows distinguishing different structures in mixed tetragonal/monoclinic zirconia phases.
The most representative Raman spectra of investigated systems, together with those of monoclinic and tetragonal ZrO2
, are illustrated in Figure 2
, Figure 3
and Figure 4
. Spectral analysis of CoZ samples before and after the catalytic test, Figure 2
, revealed the features of both the monoclinic and tetragonal phases (curves (a), (b) and (c)) in all samples, with tetragonal components becoming predominant, increasing the duration of the heat treatment (curve (b)). No significant changes were induced by the catalytic runs. Curve (d) identifies the presence of Co3
nanocrystals randomly distributed in all the CoZ materials. For Co3
spinel structure (Co2+
located at the tetrahedral site (Td) and Co3+
at the octahedral one (Oh)), Raman peaks are predicted to appear at 194 cm–1
(F2g), 485 cm–1
(Eg), 523 cm–1
(F2g), 620 cm–1
(F2g, Td site), and at 693 cm–1
(A1g; Oh site) [45
The CoO presence would contribute with a Raman signal that was relatively weak and superimposed to those of Co3
shows the most representative Raman spectra of NiZ system before and after catalytic investigation. The spectrum of thermally treated sample (curve (a)) consisted mainly of bands due to the monoclinic phase, and showed some weak feature of the tetragonal phase (Figure 3
, curve (a)). After catalytic runs, features of both tetragonal and monoclinic modifications (curve (b)) were revealed, and in addition, a band at about 1080 cm−1
, characteristic of nanosized NiO (curve (c)), was also detected.
For single-crystal or polycrystalline NiO, vibrational bands are predicted at about 420 and 560 cm−1
due to one-phonon (1P) (TO and LO modes), at about 730 cm−1
due to two-phonon (2P) (2TO modes), at about 906 cm−1
(TO + LO), and 1090 cm−1
(2LO modes) [47
]. Furthermore, the intensities of the (1P) band (about 560 cm−1
) and that of (2P) band (about 1100 cm−1
) strongly depend on NiO particle sizes and on their defectivity; the smaller the size the higher the intensity and defectivity. Although the band at 560 cm−1
is obscured by ZrO2
bands, the strong feature at about 1080 cm−1
is a clear indication of nanosized NiO presence.
The Raman spectrum collected on the CuZ sample after 2 h heating contains mainly the bands due to monoclinic phase, and small contributions because of tetragonal phase, Figure 4
, curve (a). After treatment at 600 °C 5 h and after catalytic test with C10
, bands due to tetragonal ZrO2
, curve (b) and (c)) were recorded. No evidence of CuO separate phase could be found by Raman analysis. Because three active modes at 298, 345, and 632 cm−1
are predicted for CuO [49
], the CuO presence should be evidenced by a band resulting from the overlapping of its most intense band (298 cm−1
), and the one at about 270 cm−1
of the ZrO2
Comparison of the Raman spectra among catalysts with those of monoclinic and tetragonal ZrO2 support indicates a local order decrease in the supported systems that broaden the Raman bands. Because Raman scattering comes both from the surface and the bulk of samples, the band broadening indicates that ZrO2 phase transition starts at the surface region. Differences between results from Raman spectroscopy and XRD analysis can also be seen for almost all samples. For CoZ and NiZ systems and for CuZ 2 h sample, both tetragonal and monoclinic ZrO2 phases were detected, whereas XRD results show the presence of tetragonal ZrO2. These findings suggest, under our experimental conditions, that catalysts are heterogeneous materials.
Transformations occurring during calcination of hydrous zirconia, ZrO2
O, as well as of catalytic precursors (M(II)/ZrO2
O (M(II) = Co, Cu and Ni), were studied using differential scanning calorimetry (DSC). Figure 5
reveals that the process occurs in two steps. The first one is characterized by a broad endothermic zone between 40 and 300 °C. Signals evolved by the broadband correspond to the removal of weakly bound water and to the decomposition of residual metal nitrate, used as impregnating agents. The second one is characterized by exothermic signals between 300 and 600 °C attributed to zirconia crystallization process. From DSC curves of Figure 5
, it can be clearly observed that metal cations delay the crystallization process. The undoped zirconia presents two exothermic signals, the first one at 380 °C is associated to the oxolation of hydroxyl groups to form embryonic oxide nuclei and the second one at 420 °C is associated to the crystal growth process [50
]. The introduction of metal ions reduces the amplitude of exothermic peaks and shifts them to a higher temperature. It is evident that the dopant presence retards significantly the processes as well documented in the literature [51
]. The crystallization process is complex, and involves an arrangement which is different from the kinetic point of view than that observed in undoped zirconia. There must be a small degree of substitution of the Zr4+
ions for ions of lower charge (Cu2+
], a process that can be favored for similar ionic radii and higher surface areas. In addition, the host metals preferential coordination must be taken into account. Table 2
shows Shannon–Prewitt ionic radii values of studied species [53
]. The formation of the oxygen-defective structure increases the mobility of oxygen atoms [54
]. In fact, as observed by Raman spectroscopy, the tetragonal phase increases at surface level with the calcination time.
It is well known that Zr(IV) in tetragonal phase presents octahedral coordination. This structural form is stabilized by the presence of bi and tri-valent host ions. In this context, Cu(II) ionic radius in octahedral environment is similar to that of Zr(IV) in similar coordination, although according to the electronic configuration, Cu(II) is more susceptible to adopting a distorted octahedral or square planar environment, due to the Jahn–Teller effect. This aspect implies that the tetragonal stabilization must be kinetically less favorable, requiring consequently higher energy.
In fact, comparing the exothermic peaks exhibited by the catalytic precursors, the amorphous–crystalline tetragonal transformation occurs at the highest temperature for CuZ system. Differences with respect to the other metal species are responsible for the higher temperature observed for the copper re-arrangement in the zirconia lattice. Likewise, there are limits in the incorporation of copper in zirconia networks; so, the Cu-O species, which are not incorporated, contain Cu(II) ions with axial distortion. In this context, the broad signal in the 600–800 nm range for the CuZ 2 h UV–vis diffuse reflectance spectrum (Supplementary Materials
) corroborates the presence of this coordination out the zirconia lattice [56
Temperature programmed reduction (TPR) analysis was used to study redox properties of catalysts. Figure 6
shows the evolution of H2
consumption as a function of temperature for catalysts. Furthermore, the reduction peaks were integrated, and the values of H2
consumed (mmol), given in Table 1
, are in acceptable agreement with that expected by the supported phases concentration. In the studied temperature range, zirconia does not show an appreciable consumption of hydrogen, although it is known that it can be partially reduced at high temperature (about 800 °C).
As it is shown in the Figure 6
, depending on the supported transition metal oxide, the catalysts present different reducibility. Moreover, signals for CoO or CuO supported oxides appear in a temperature range slightly shifted with respect to that observed for pure oxides (Supplementary Materials
]. So, the pure Co3
TPR profile is characterized by two reduction steps, assigned to Co(III)–Co(II) and Co(II)–Co° (with maximums at 324 and 397 °C respectively) [19
], while the TPR profiles of CoZ 2 h and CoZ 5 h catalysts show a series of signals in the range 260–360 °C, which indicate the presence of diverse types of cobalt oxide species, or species with different interactions over the support. Crystalline Co3
is not detected in the CoZ 2 h catalyst by DRX, revealing that cobalt oxide is highly dispersed in the zirconia, as was detected by Raman.
It is interesting to observe that the CuO reductions in the CuZ 2 h and CuZ 5 h catalysts occur at about 160 °C below the temperature observed for the pure CuO (around of 360 °C) [58
], revealing a high dispersion of copper species, in agreement with CuO phase absence in the XRD pattern and Raman spectroscopy results.
The small size of the Co3
or CuO particles and/or the interaction of transition metal species with the zirconia surface may explain their reduction at lower temperature [21
]. This suggests that the support favors the surface dispersion of the oxide, increasing the reducibility. In this context, it is also interesting to note that CuZ 2 h and CuZ 5 h catalysts present higher surface areas and lower sintering. Centi et al. [64
] have reported that the reducibility of copper species increases on ZrO2
, especially on tetragonal phase. These authors suggest that during the hydrous zirconia calcination to give anhydrous ZrO2
, the OH groups were responsible for the reduction of Zr4+
ions, whereas the presence of a lower oxidation state of Zr generates an increase of anionic vacancies concentration. Similar behavior is expected by the isomorphous substitution of Zr(IV) by transition metals, revealing that the effect of the vacancies of the zirconia lattice affects the redox properties of supported oxide species, increasing the reducibility, because these supported particles are in a more reducing environment.
Finally, in contrast to that observed with the CoZ and CuZ catalysts, the presence of zirconia does not generate an increase in the reducibility of the supported NiO species. The pure NiO TPR profile shows H2 consumption between 340 and 420 °C, while both NiZ 2 h and NiZ 5 h catalysts’ TPR profiles show H2 consumption in the same temperature range. It is possible to suggest a low interaction NiO-support in agreement with DRX and Raman results.
Likewise, the reduction of catalysts calcinated for 5 h occurs at temperatures slightly higher than those thermally treated during 2 h. This result can be attributed to a stronger metal-support interaction, and increase of diffusion by thermal effect (mobility of oxygen atoms), factors that improve the crystallization. Besides, it can be suggested a small crystal growth of oxide particles. In the case of the CoZ catalyst, these results are consistent with that observed in XRD, where the presence of Co3O4 crystals has been detected just in CoZ 5 h catalyst. Considering all the above, the reducibility of catalysts follows the order: CuZ 2 h > CuZ 5 h > CoZ 2 h > CoZ 5 h > NiZ 2 h > NiZ 5 h.
Catalytic activity results for the naphthalene oxidation reaction, expressed as CO2
yield vs. temperature, are presented for all catalysts in Figure 7
. Table 2
shows T50 and T90 values (temperature where conversions reached are 50% and 90%, respectively) obtained in each experiment. No CO or other by-products were observed in the outlet gas, and then, the selectivity towards CO2
is higher than 99%.
When performing the experiment in catalyst absence, the T50 and T100 (temperature where conversions reached are 50% and 100%) are 430 °C and 485 °C, respectively. For the ZrO2 support, a CO2 yield of 50% is reached at 270–280 °C, obtaining the complete conversion (CO2 yield of 100%) at 400 °C; it is evident that zirconia is active for naphthalene oxidation.
Transition metal oxide-modified catalysts show improved activities; with them, the total conversion of naphthalene is achieved at lower temperatures, in the range 220–300 °C. These results revealed that all catalysts were more active than the bare support. The relative order of catalyst activity can be analyzed by comparing the required temperature to reach 50% of CO2
yield (Table 3
). By comparing the T50 values the activity of catalysts follows the order: CuZ 5 h > CuZ 2 h > CoZ 5 h > CoZ 2 h > NiZ 2 h ≈ NiZ5 h >> Z.
On the other hand, the increase of the calcination time generates an increase in the activity of the catalysts, with this effect being more marked in CoZ and CuZ catalysts. This slight increase in activity may be associated with the higher surface content of the oxide transition metal species (Table 1
). It may also be due to an increase in the concentration of oxygen vacancies. According to Raman spectroscopy, it has been observed that the CuZ 5 h catalyst has higher content of tetragonal phase on the catalytic surface than the CuZ 2 h catalyst.
The structural and redox behavior, as well as electronic properties of the different supported transition metal species, are expected to play a decisive role in the naphthalene oxidation, considering that the mechanism of hydrocarbons (HC) oxidation reaction (e.g., naphthalene) occurs through a redox process. One of the most used mechanisms to explain the catalytic oxidation of hydrocarbons is the Mars—van Krevelen mechanism, where active oxygen comes from catalytic oxide species [14
]. In this context, Solsona et al. [67
] carried out the propane oxidation process in oxygen absence using a CuOx
catalyst, and detected oxygen containing compound (CO2
) as reaction product, suggesting the participation of lattice oxygen.
On the other hand, the Langmuir–Hinshelwood accepted mechanism proposes, as an important step of the process, the adsorption of the reactants in active sites. Labaki et al. [21
] have studied the total oxidation of propene and toluene in the presence of zirconia doped by copper. These authors have proposed that the oxidation mechanism implies the intervention of chemisorbed activated oxygen. On zirconia, the adsorption of O2
can be explained by the intervention of Zr3+
redox cycles, the formation of O2− ads
, and anionic vacancies.
Taking into account the obtained results, both mechanisms may contribute to catalytic activity of catalyst in this work. Zirconia, with its oxygen vacancies, may be an oxygen adsorption site and may be a site of adsorption for the hydrocarbon.
Besides, in this work, catalysts that contain highly interacted transition metal oxides present greater activity than zirconia. It is possible to suggest an oxygen supply from the oxides to the reactants, following a Mars–van Krevelen MVK mechanism and generating a redox cycle. In this context, Co(III) can be reduced to Co(II), or Cu(II) to Cu(I). The presence of zirconia oxygen vacancies favors these processes because the reducibility of species increases, as was observed by TPR results. The high metal–support interaction generates a greater mobility of oxygen in the catalytic system, and this property can increase the activity. Moreover, the greater metal–zirconia interaction favors the formation of vacancies in the bulk, and consequently, facilitates the movement of oxygen in the catalytic system from the bulk to the surface. A similar effect has been reported for the Cu–CeO2
]. Finally, reduced sites should be re-oxidized to be regenerate.
From results, it is evident that the activity may also depend on the metal coordination of supported species. Hence, the less active catalyst (ZNi), containing d8Ni(II) species in octahedral coordination, presents lower redox possibilities with respect to CoZ and CuZ catalysts.
In CoZ catalysts, cobalt oxide adopts the easily reducible spinel structure (Co(III) with low spin d6 configuration and octahedral coordination, and Co(II) with d7 configuration and tetrahedral coordination).
For the CuZ system, the lower copper symmetry, due to the Jahn–Teller effect, can explain the higher accessibility to the active site. CuZ catalysts shows the highest activity among the supported catalysts, because of their special copper–zirconia interaction, their high possibility of generated oxygen vacancies, their higher area, and their higher reducibility. CuO particles are nano-dispersed on catalytic surface, and have not been detected with the characterization techniques used (DRX and Raman spectroscopy).
It is interesting to remark that in all cases, the activity is enhanced with the time length of thermal treatment. This fact facilitates diffusion processes, allowing an adequate rearrangement of metal species, accompanied by the stabilization of zirconia tetragonal phase, increasing the oxygen vacancies, and leaving exposed, a higher number of active sites.
On the other hand, under our experimental conditions, no significant changes in the physicochemical properties of CuZ catalysts were induced by the catalytic runs.
Finally, considering the obtained and reported results (Table 3
), it can be suggested that the CoZ, and, more specifically, the CuZ catalysts show an appreciable activity for the catalytic oxidation of naphthalene.