ARTICLE pubs.acs.org/JPCA
Decamolybdocobaltate Heteropolyanions in Aqueous Solutions: Chemistry Driving Their Formation and Domains of Stability J. Moreau,†,‡ O. Delpoux,‡ E. Devers,‡ M. Digne,‡ and S. Loridant*,† † ‡
IRCELYON, CNRS, University of Lyon 1, 2 Av. Albert Einstein, Villeurbanne, F-69626, France IFP Energies Nouvelles, BP 3, Solaize, F-69360, France
bS Supporting Information ABSTRACT: In the present study, aqueous solutions of decamolybdocobaltate H4Co2Mo10O386 heteropolyanions were prepared from molybdenum oxide, cobalt carbonate precursors and hydrogen peroxide used as oxidizing agent. The preparation was optimized adding a consecutive hydrothermal treatment at 150 °C to obtain pure H4Co2Mo10O386 aqueous solutions for Co/Mo atomic ratio of 0.5. Combining quantitative Raman and UVvisible measurements and chemometric methods, it was demonstrated that a mixture of H4Co2Mo10O386 and octomolybdate Mo8O264 species is obtained for Co/Mo ratios lower than 0.5, and the relative quantities of H4Co2Mo10O386 are determined by the presence of Mo8O264 species and by the quantity of Co2+ countercations available in the solutions to ensure the electroneutrality. As these quantities can be predicted for each Co/Mo ratio, this finding allows rationalization of the preparation of heterogeneous catalysts using impregnation by H4Co2Mo10O386 aqueous solutions. Parameters relevant of the impregnation step such as the pH, the Co/Mo ratio, and the molybdenum concentration were varied to determine the domains of stability of H4Co2Mo10O386 heteropolyanions after formation. Stable from pH 1 to 4.5, this dimeric Anderson species is destabilized above pH 4.5; Co2+, monomolybdate MoO42 ions, and precipitates are then formed. For Co/Mo ratios lower than 0.5, the relative quantity of dimer does not vary with the pH and with a change of the Co/Mo ratio consecutive to the hydrothermal treatment. On the contrary, the coproduced Mo8O264 species can be transformed into other isopolymolybdates varying the pH according to their domains of stability. For all of the ratios, H4Co2Mo10O386 dimers were also shown to be stable in a wide range of molybdenum concentrations.
’ INTRODUCTION Heteropolycompounds (HPCs) are of fundamental and practical interest for numerous applications.1 In the field of heterogeneous and homogeneous catalysis, numerous and versatile uses were found, for example, as acidic or oxidation catalysts.13 For catalytic applications, HPCs are used either as bulk crystalline compounds or as dispersed molecular species in solvent or on the surface of a porous support. In the later case, the control of the stability of HPC is crucial to ensure reproducible and optimized performances. In solvent medium, experimental conditions such as pH value, concentrations, and ionic strength can lead to partial or total HPC decomposition. The situation is even more complex for HPC deposit on carrier surface: for instance, aqueous impregnation can lead to HPC evolutions due to several correlated effects, like acidobasic reactions (between two molecular species or between surface sites and molecular species), adsorption onto the surface, precipitation into the pores, diffusion, and support partial dissolution. If equilibria of isopoly compounds in solution are well-known,4 the stability of HPC in solution is often more complex. The speciation in aqueous molybdenum phosphorus solution as a function of pH value and P/Mo atomic ratio has been accurately determined by Pettersson et al.5 using potentiometric titrations and 31P NMR. In a similar way, the aqueous molybdenumaluminum system has been studied,6 and the results show the formation of a stable Anderson heteropolyanion r 2011 American Chemical Society
(HPA), the 6-molybdoaluminate Al(OH)6Mo6O183: for an aluminum concentration of 103 M, Mo is mainly present in this HPA species for 2 < pH < 5 and 0.1 < Mo/Al < 10. Some HPA with two or three different transition metals have been prepared, and their stability in solution has been studied: for a given V/Mo ratio, an aqueous PMoV solution exhibits a complex mixture of vanadium-substituted phosphomolybdates H(3+x)PVxMo(12x)O40 (x ranging between 0 and 12).7 Another example is the cobaltsubstituted phosphomolybdate H2PCoMo11O405, the stability of which depends on both P/Mo and Co/Mo ratios.8 Among such complex HPA species, the Anderson-like decamolybdocobaltate H4Co2Mo10O386 with 3 Co2+ as countercations has been proposed as a good starting precursor to prepare highly active sulfided CoMo hydrotreating catalyst.9,10 The high Co/ Mo atomic ratio, equal to 0.5, and the close vicinity of cobalt and molybdenum atoms into the HPA structure seem to be responsible of the enhanced activity as compared to CoMo catalyst prepared using ammonium heptamolybdate and cobalt nitrate as impregnation solution of γ-Al2O3. Nevertheless, the catalyst activity should be likely improved by optimizing the quantity of H4Co2Mo10O386 dimers deposited on the carrier surface and Received: September 5, 2011 Revised: November 15, 2011 Published: November 23, 2011 263
dx.doi.org/10.1021/jp208534p | J. Phys. Chem. A 2012, 116, 263–270
The Journal of Physical Chemistry A
ARTICLE
avoiding the formation and adsorption of undesired species. Impregnation on γ-Al2O3 can be achieved at various pH, Co/Mo ratios, or Mo concentrations. During the maturation or drying steps, the diffusion of the solution inside the alumina porosity may lead to strong modifications of the initial parameters. This is why the knowledge of the H4Co2Mo10O386 stability regarding these parameters was a prerequisite to rationalize its deposition in higher quantity during the preparation of catalysts. In the present study, aqueous H4Co2Mo10O386 solutions were prepared from molybdenum oxide MoO3 and cobalt carbonate CoCO3 precursors using an efficient route described in the literature.11,12 This method, called the “peroxo” route, uses only MoO3, CoCO3, and hydrogen peroxide H2O2 as precursors and avoids the introduction of undesirable counterions for catalysis. However, a mixture of H4Co2Mo10O386 and its monomeric Anderson form H6CoMo6O243 can be obtained using oxidants such as H2O2.13 Because one of the goals was to determine the stability of the dimer as sole precursor in the impregnation solution, a first step consisted of optimizing the preparation method to obtain pure (3Co2+; H4Co2Mo10O386) aqueous solutions. Next, the preparation and quantitative analysis of solutions with Co/Mo lower than 0.5 has allowed one to understand the chemistry driving the formation of the dimer from peroxomolybdates and cobalt carbonate. In a second step, the stability of such solutions was determined varying the pH, Co/Mo ratio, and molybdenum concentration [Mo] in broad ranges. The present investigation was principally achieved using Raman spectroscopy, which is a powerful technique to distinguish different molybdates species in aqueous solution. This technique was combined with chemometric methods for quantification. Moreover, as UVvisible (UVvis) spectroscopy affords characterization of Co3+ cations present in H4Co2Mo10O386 HPA and also Co2+ present as countercations, this technique was complementary used.
Figure 1. Raman spectra of an aqueous solution of (3Co2+; H4Co2Mo10O386) after thermal treatments at various temperatures and comparison with spectra of Mo2O3(O2)42 solutions obtained before CoCO3 addition and reference spectra of (3NH4+; H6CoMo6O243) and (6NH4+; H4Co2Mo10O386) solutions. The 8001100 cm1 spectral range is zoomed on the right side.
the decrease in the Co/Mo ratio on the stability of H4Co2Mo10O386 dimers was studied. For that purpose, the Mo concentration was varied from 0.8 to 0.02 M diluting the prepared solutions, the pH was adjusted from 1 to 9 adding several drops of chloridric acid HCl (5 M) or sodium hydroxide NaOH (3 M) solutions, and the Co/Mo ratio was decreased to 0.1 adding various amounts of ammonium heptamolybdate (NH4)6[Mo7O24] 3 4H2O. Reference aqueous solutions of decamolybdocobaltate ammonium salt (NH4)6[Co2Mo10O38H4] and 6-molybdocobaltate ammonium salt (NH4)3[CoMo6O24H6] were also prepared, respectively, by dissolving in oxygen peroxide solution containing active carbon the appropriate amounts of (NH4)6[Mo7O24] 3 4H2O and cobalt acetate Co(CH3COO)2 3 4H2O and by dissolving in hydrogen peroxide solution the appropriate amounts of (NH4)6[Mo7O24] 3 4H2O and cobalt nitrate Co(NO3)2 3 6H2O.10 Quantitative Spectroscopic Measurements. Raman spectra were recorded from 100 to 1100 cm1 with a LabRam ARAMIS (HORIBA Jobin-Yvon) spectrometer using the exciting line at 532 nm of a diode pumped solid-state laser. The laser beam was focused with a wide angle 50 objective, and the spot size was 2 μm. The diffused light was dispersed with a 1800 lines 3 mm1 grating. The laser power used was 3 mW for all of the solutions, and the measurements were achieved under ambient conditions. A few milligrams of sodium nitrate NaNO3 was added to the prepared solutions to use the ν1(NO3) symmetric stretching vibration band at 1050 cm1 as an internal standard and normalize Raman spectra. An algorithm15 was programmed in MatLab 6.5 to remove the Raman spectral background in a similar way for all of the spectra while keeping the analytical signal intact. The spectra were treated in two parts. Below 750 cm1, the two parameters needed for the calculations were p = 102 and λ = 105. Above 750 cm1, the asymmetric parameter p = 104 and the parameter for smoothness λ = 107. Principal component analysis (PCA) and quantification were achieved using the SIMPLISMA algorithm16 implemented in the UV/IR Processor software ACD Laboratories. As each different compound can be distinguished from another thanks to their principal Raman bands, the SIMPLISMA algorithm is able to extract the pure spectrum of each compound from a series of measured spectra whose signals are the result of the contributions from the different compounds. The general equation used by SIMPLISMA is the following:
’ EXPERIMENTAL SECTION Synthesis. According to the synthesis protocol proposed in the literature,11,12 MoO3 (6 g, Climax) was dissolved in an excess of H2O2 (25 mL, 35%, Sigma-Aldrich) and water (25 mL) during 7 h at room temperature. The pH of the yellow solution obtained was lower than 1. Figure 1 provides the spectrum recorded at this step (lower spectrum): it contained bands at 320, 545, 585, and 975 cm1 that were assigned to peroxomolybdates Mo2O3(O2)42,14 the other band at 880 cm1 being due to ν(OO) stretching vibrations of remaining H2O2. The peroxomolybdates were formed according to the following reaction:
2MoO3 þ 4H2 O2 S Mo2 O3 ðO2 Þ4 2 þ 2Hþ þ 3H2 O To this clear solution were then slowly added at room temperature various amounts of CoCO3 (Alfa Aesar). The resulting exothermic reaction led to a brown mixture. The aqueous solutions as prepared will be labeled xCo:Mo in the following, where x is the Co/Mo atomic ratio. First, as a mixture of H4Co2Mo10O386 and H6CoMo6O243 can be obtained using oxidants such as H2O2,13 the influence of the temperature and the addition of active carbon during the second step on the stability of a 0.5Co:Mo solution were investigated. Conditions to obtain stoichiometric (3Co2+; H4Co2Mo10O386) solutions were determined and applied to prepare Co:Mo solutions for x ranging from 0.3 to 0.5. Next, the influence of [Mo], the pH, and
D ¼ CP þ E 264
dx.doi.org/10.1021/jp208534p |J. Phys. Chem. A 2012, 116, 263–270
The Journal of Physical Chemistry A
ARTICLE
D represents the experimental data matrix, C represents the contribution matrix, proportional to the concentration, P is the pure variable matrix containing the pure spectra, and E is the residual error matrix. After a successful treatment of data, when residual error is minimized, the relative contributions of the different species are given as well as their assigned pure spectra. Next, each calculated pure spectrum has to be assigned to a single or a mixture of molecular species: such assignment is often done using literature data (for instance, comparison with Raman spectra of pure crystalline compound). Finally, the speciation diagrams can be deduced assuming that all of the species are identified in the Raman spectra. For UVvis measurements, the solutions were directly injected into a 1 mm path length quartz Suprasil cell without any dilution. The spectra were recorded from 200 to 900 nm using a Varian Cary 1E spectrophotometer with a scan rate of 100 nm min1 and 0.5 nm data point resolution. A calibration to 0 of the absorbance values in the 800900 nm area was simply carried out.
Figure 2. Raman spectra of solutions prepared with various Co/Mo ratios after hydrothermal treatment at 150 °C. The 8001100 cm1 spectral range is zoomed in on the right side.
’ RESULTS Optimization of the Synthesis Protocol. Figure 1 provides the Raman spectrum of aqueous 0.5Co:Mo solution obtained applying the method described (upper spectrum).11,12 It is noticeable that Mo2O3(O2)42 dimer species were completely decomposed after CoCO3 addition as revealed by the absence in the spectrum of its characteristic bands detailed in the experimental part. Its comparison with the spectra of (3NH4+; H6CoMo6O243) and (6NH4+; H4Co2Mo10O386) solutions, which exhibit their most intense bands at 948 and 956 cm1, respectively, showed that a mixture of monomers and dimers was then obtained. The bands around 210220, 350360, 540610, and 900960 cm1 were assigned, respectively, to bending δ(MoO2t), δ(MoOaMo) and stretching ν(MoOCo), ν(MoO2t) vibrations.17,18 As a consequence, the preparation conditions had to be optimized to get stoichiometric (3Co2+; H4Co2Mo10O386) solutions with only dimers. First, a thermal treatment at 60, 80, and 100 °C was applied after CoCO3 addition, and Raman spectra were recorded after cooling at room temperature (Figure 1). The temperature appeared to be a key parameter to obtain only H4Co2Mo10O386 dimers using the peroxo route method. Indeed, the main monomer band at 948 cm1 dropped with the temperature of the thermal treatment as is easily observed in the zoom (right side of Figure 1). Because after a treatment at 100 °C this band still appeared as a weak shoulder, a hydrothermal treatment at 150 °C was achieved in an autoclave. The corresponding spectrum only contained the bands characteristic of H4Co2Mo10O386 as expected from the stoichiometry of the 0.5Co:Mo solution. After being heated at higher temperature, the formation of a precipitate was observed, and therefore a treatment temperature of 150 °C was chosen and systematically applied to prepare other solutions. Finally, it is noticeable that a pure solution of (3Co2+; H4Co2Mo10O386) was obtained after a treatment at 90 °C, adding small amount of active carbon as catalyst10 simultaneously with CoCO3 and then removing it by filtration (Figure S1). A high amount of active carbon powder would favor the formation of H6CoMo6O243 monomers as shown by the observation of its characteristic bands at 948 and 565 cm1 (see Figure 1 for comparison). The band at 980 cm1, the intensity of which increased with the amount of active carbon and that was observed treating this catalyst with H2O2
Figure 3. UVvisible spectra of solutions prepared with various Co/ Mo ratios after hydrothermal treatment at 150 °C.
(Figure S1), could arise from ν(CO) vibrations of oxidized and suspended carbon particles. As the active carbon powder was difficultly filtered or centrifugated after reaction, the hydrothermal method was used in the following study. Mo and Co Speciation in Co:Mo Solutions. Influence of the Co/Mo Ratio. As this preliminary study took place in the ideal stoichiometric proportions (Co/Mo = 0.5), which privilege the formation of the Anderson dimer, this one was also appraised preparing aqueous solutions with Co/Mo ratios ranging from 0.3 to 0.5. Next, the influence of the Co/Mo ratio on the quantity of H4Co2Mo10O386 anions formed using the optimized synthesis protocol was investigated both by quantitative Raman and by UVvis measurements. The evolution of Raman spectra with this parameter is reported in Figure 2. The relative intensity of the main band of H4Co2Mo10O386 anions dropped, whereas two bands appeared at 970 and 845 cm1 when the Co/Mo ratio was lowered from 0.5 to 0.3. The SIMPLISMA treatment revealed that the spectra contained an individual component with bands at 970, 960, 908, and 845 cm1 attributed to a mixture of β and γ octomolybdate Mo8O264 species1921 in addition to the spectrum of H4Co2Mo10O386. The presence of α Mo8O264 cannot be ruled out because its main Raman bands are located at 959 and 918 cm1.19 UVvis spectra displayed in Figure 3 confirm that a lower quantity of dimers was formed, decreasing the Co/Mo ratio. Indeed the spectra exhibited a band at 605 nm attributed to the 1 A1g w 1T1g transition of Co3+ low-spin cations in octahedral 265
dx.doi.org/10.1021/jp208534p |J. Phys. Chem. A 2012, 116, 263–270
The Journal of Physical Chemistry A
ARTICLE
Figure 4. Raman spectra versus the pH of 0.5Co:Mo solution and speciation deduced from the SIMPLISMA treatment (right side).
coordination.10,22 The band centered around 510 nm was assigned to the 4T1g(F) w 4T1g(P) transition of high-spin Co2+ cations (d7 configuration).10 Finally, the main bands observed around 300 nm corresponded to O2 w Mo6+ charge transfers.22 The intensity of the band at 605 nm clearly decreases with the Co/Mo ratio, leading to the conclusion that the quantity of Anderson dimer seems to be linked to the CoCO3 amount added to the peroxomolybdate solution. The chemistry driving the dimer formation will be highlighted in the discussion section from quantitative Raman and UVvis measurements. Influence of the pH. The stability of all of the prepared solutions was investigated with respect to the pH. The Raman spectra gathered in Figure 4 for 0.5Co:Mo solution exhibit at low pH (from 1 to 4.5) only the bands attributed to the dimer, in particular its main band at 956 cm1. However, an additional band typical of monomolybdate MoO42 was observed at 896 cm1 21 above pH 4.5, which proves that the dimer becomes then unstable. This observation was confirmed by the SIMPLISMA treatment, which revealed only these two individual components (Figure S2, right side) and a negligible amount of MoO42 below pH 4.5 (zoom of Figure 4). The formation of a precipitate, which was further identified by XRD as cobalt molybdate NaCo2(MoO4)2OH 3 H2O, was clearly observed (Figure S3) above pH 5.8; Na+ cations being provided by the NaOH addition achieved to increase the pH. Considering the standard reduction potentials E0(Co3+/Co2+) and E0(O2/H2O) equal in aqueous solution at 25 °C to 1.82 and 1.23 V, respectively,23 it was assumed that Co3+ cations contained in H4Co2Mo10O386 anions are not stable in water after decomposition. Therefore, we propose that the following reactions occurred: H4 Co2 Mo10 O38 6 þ 8OH w 2Co3þ þ 10MoO4 2 þ 6H2 O
ð1Þ
2Co3þ þ H2 O w 2Co2þ þ 1=2O2 þ 2Hþ
ð2Þ
Figure 5. Raman spectra versus the pH of 0.3Co:Mo solution.
Similarly, an amorphous precipitate was evidenced using NH4OH to increase the pH. Interestingly, no shift of the dimer bands was evidenced increasing the pH of 0.5Co:Mo solution, suggesting that H4Co2Mo10O386 was not deprotonated above pH 4 contrarily to what was proposed from cyclic voltammograms.24 Indeed, one can expect upon deprotonation a red-shift of the ν(ModO) bands due to a weakening of the ModO bonds as it is observed for heptamolybdates HxMo7O24(6x) and diphosphomolybdates HxP2Mo5O23(5x).21,25 Considering the results obtained for Co/Mo = 0.5, the pH of solutions with lower Co/Mo was not increased above 5.5 before Raman and UVvisible characterization. The evolution of the Raman spectra of a 0.3Co:Mo solution with respect to the pH plotted in Figure 5 shows that a direct interpretation of the Raman spectra is more complex because of the numerous peaks that appear and disappear according to the pH. As it was also the case for all of the solutions prepared with Co/Mo < 0.5, SIMPLISMA treatments were essential to extract the individual components that corresponded to Mo36O1128, a mixture of β and γ Mo8O264, a mixture of Mo7O246 and HMo7O245, MoO4 species, in addition to H4Co2Mo10O386 (Figure S2, left side).1416 The individual component with the main band at 942 cm1 was attributed to a mixture of Mo7O246 and HMo7O245. These two species were not distinguished by the SIMPLISMA treatment because their main bands located, respectively, at 939 and 946 cm1 were too broad and close to each other. As was already said, the presence of α Mo8O264 cannot be ruled out in the mixture of β and γ Mo8O264. Finally, as there was no shift of their bands with the pH, the presence of protonated forms was excluded. The speciations deduced from these treatments (Figure 6) indicate that when Co/Mo was lower than 0.5, Mo36O1128, Mo8O264, HMo7O245, Mo7O246, and MoO4 species were successively stabilized, increasing the pH, while the dimer concentration seems to remain relatively constant for a given Co/Mo ratio. One can notice in Figure 6 a maximum of this concentration at about pH = 3.5 for each Co/Mo ratio. Either it seems to mean that there is a pH value that maximizes the dimer concentration or this local maximum is an artifact due to the limitations of the SIMPLISMA treatment, which cannot separate the H2Mo7O244 component from the dimer’s one because of their too close main bands that appear respectively at 95821 and 956 cm1. To clarify this point, the dimer concentration was also
Equations 1 and 2 can be replaced by eq 3: H4 Co2 Mo10 O38 6 þ 10OH w 2Co2þ þ 10MoO4 2 þ 7H2 O þ 1=2O2
ð3Þ
The decomposition constant Kd corresponding to this reaction determined in the 4.55 pH range was equal to 1070. Finally, the precipitate can be formed according to: Naþ þ OH þ 2Co2þ þ 2MoO4 2 þ H2 O w NaCo2 ðMoO4 Þ2 OH 3 H2 O
ð4Þ 266
dx.doi.org/10.1021/jp208534p |J. Phys. Chem. A 2012, 116, 263–270
The Journal of Physical Chemistry A
ARTICLE
Figure 7. UVvisible spectra versus the pH of 0.3Co:Mo solution.
Figure 8. Raman spectra versus the [Mo] concentration of 0.5Co:Mo solution.
Regarding the speciation of isopolymolybdates, their relative concentrations are governed both by the dimer concentration and by the pH. Indeed, the overall concentration of isopolymolybdates increases when the Co/Mo is lowered by compensation of the decrease in the dimer concentration. Besides, as the dimer concentration is independent of the pH for a given Co/Mo ratio, the speciation of the isopolymolybdates is only governed by the pH, which means at low pH the Mo36O1128 species is predominant and then Mo8O264, H2Mo7O244, HMo7O245, Mo7O246, and finally MoO42 for a pH higher than 4.5. The domains of stability of isopolymolybdates were found to be in agreement with those reported in the literature.5,6,20 Influence of the [Mo] Concentration. To investigate the influence of the [Mo] concentration, several solutions with [Mo] ranging from 0.02 to 0.8 M were prepared according to the optimized protocol described above. The pH was not adjusted during this experiment. The Raman spectra obtained for 0.5Co:Mo solution gathered in Figure 8 show that the dimer is stable in wide range of Mo concentrations (0.80.08 M) because they contain only the bands at 956, 918, 603, and
Figure 6. Speciations deduced from SIMPLISMA treatments of Raman spectra for Co/Mo < 0.5: (b) H4Co2Mo10O386, () Mo7O246 + HMo7O245, (+) Mo8O264, (]) MoO42, (O) Mo36O1128.
determined by UVvis spectroscopy through the electronic transitions of Co3+ cations in octahedral coordination at 605 nm. Figure 7, which represents the evolution versus the pH of the UVvis spectra of a 0.3Co:Mo solution, evidences that the absorbance at this wavelength remained constant. Thus, the concentration of Anderson dimer formed after CoCO3 addition does not depend on the pH, and the local maximum deduced from Raman spectra treated by SIMPLISMA is an artifact due to the contribution of H2Mo7O244. 267
dx.doi.org/10.1021/jp208534p |J. Phys. Chem. A 2012, 116, 263–270
The Journal of Physical Chemistry A
ARTICLE
Figure 9. Evolution of Raman spectra of 0.5Co:Mo solution after addition of HMA to decrease the Co/Mo ratio and speciation deduced from the SIMPLISMA treatment (right side).
Figure 10. Comparison between the calculated concentrations of H4Co2Mo10O386 and experimental values deduced from quantitative Raman and UVvisible measurements.
570 cm1, which are characteristic of this species. Nevertheless, one can notice that the main band of MoO42 species at 896 cm1 was observed with very weak intensity for [Mo] = 0.02 M in addition to those of H4Co2Mo10O386, but it can be explained by the slight increase in the pH that then reached 4.4, a value above which the dimer becomes unstable as was previously shown. Dilutions of solutions prepared with Co/Mo < 0.5 confirmed the high stability of H4Co2Mo10O386 anions. For instance, as shown in Figure S4, a mixture of H4Co2Mo10O386 and Mo8O264 species was still obtained for 0.3Co:Mo solution, lowering the [Mo] concentration until 3 102 M. The proportions between the two species did not change because the intensity ratio of their main bands located respectively at 956 and 970 cm1 was constant. When [Mo] reached 8 103 M, all of the ν(ModO) bands observed above 900 cm1 were redshifted probably because of higher influence of van der Waals interactions with water molecules that weaken ModO bonds. Influence of a Consecutive Decrease in the Co/Mo Ratio. Attempts to destabilize the dimers formed during the peroxo route synthesis were achieved by modifying the Co/Mo ratio after the hydrothermal treatment. As consecutive addition of CoCO3 to a 0.3Co:Mo solution (to increase the Co/Mo ratio) led to its nondissolution or to the formation of a precipitate, various amounts of (NH4)6[Mo7O24] 3 4H2O were added to a 0.5Co:Mo solution until a Co/Mo ratio of 0.1 was reached. The pH was still not adjusted during these experiments. The Raman spectra plotted in Figure 9 contain four bands located at 956, 918, 603, and 355 cm1, which were attributed to the Anderson dimer, and two other ones at 942 and around 900 cm1, which were due to a mixture of Mo7O246 and HMo7O245 species stable at such pH.21 The SIMPLISMA treatment revealed only these two individual components, and the quantification and shown on the right side of Figure 9 proves that the concentration of the dimer remains constant for Co/Mo ranging from 0.1 to 0.5. Nevertheless, we can notice on the spectrum corresponding to the ratio Co/Mo = 0.1 that a thin band at 896 cm1 characteristic of the MoO42 species21 appears with weak intensity probably because of the increase in the pH value that reached 5.0.
addition to H4Co2Mo10O386 anions from Mo2O3(O2)42 peroxomolybdates and CoCO3, the following chemical equation was considered for a given Co/Mo ratio: 35Mo2 O3 ðO2 Þ4 2 þ 70xCoCO3 þ 70Hþ w ð70x 2yÞCo2þ þ yH4 Co2 Mo10 O38 6 þ ð70 10yÞ=8Mo8 O26 4 þ ð5y 140x þ 35ÞHþ þ ð140 yÞ=2O2 þ 70xCO2 þ ð140x 9y þ 35Þ=2H2 O
ð5Þ where x corresponds to the Co/Mo ratio and y is the number of H4Co2Mo10O386 moles formed during the reaction. When x is equal to 0.5, eq 5 then becomes: 35Mo2 O3 ðO2 Þ4 2 þ 35CoCO3 þ 70Hþ w 7H4 Co2 Mo10 O38 6 þ 21Co2þ þ 133=2O2 þ 35CO2 þ 21H2 O
ð6Þ
4
It is noticeable that no Mo8O26 was then formed in agreement with Raman results, and all of the protons necessary for the charge equilibrium of reactants were consumed. More generally, as the pH of the solutions after thermal treatment ranged from 3 to 4, it was assumed that almost all of the protons were consumed during the reaction, leading to: 5y 140x + 35 ≈ 0 and y ≈ 28x 7. The concentration of H4Co2Mo10O386 at natural pH was deduced from the Raman spectra for Co/Mo ranging from 0.3 to 0.5. For UVvis spectra, as the pre-edge of the O2 w Mo6+ charge transfer strongly varies in the series, the quantification of Co3+ cations was achieved from the amplitude of the band at 605 nm to afford a better precision (calibration curve given in Figure S5). It was assumed that all Co3+ cations were contained in H4Co2Mo10O386 anions considering a new time that Co3+ cations are not stable in water. The theoretical H4Co2Mo10O386 concentrations deduced from the y values varying x (Co/Mo ratio) are compared to those obtained from the Raman and UVvis data in Figure 10. An excellent agreement between theoretical and UVvis data was found with a deviation less than 6% excepting data at Co/Mo = 0.3. The agreement with Raman data was also good but with a higher deviation (10%) that can be explained by the artifact observed on the speciation curves versus the pH (Figure 6) arising from the difficulty to distinguish H4Co2Mo10O386 and H2Mo7O244 contributions in the Raman spectra. Finally, theoretical data were also calculated considering equations of the same type as eq 5 but involving HMo8O263, H2Mo7O244, HMo7O245, Mo7O246, or MoO42 instead of Mo8O264 and
’ DISCUSSION Chemistry Driving the H4Co2Mo10O386 Formation. The
quantitative study of solutions prepared with various Co/Mo ratios allowed for the understanding of the chemistry driving the formation of H4Co2Mo10O386 from peroxomolybdates. Indeed, as Raman spectra revealed the formation of Mo8O264 in 268
dx.doi.org/10.1021/jp208534p |J. Phys. Chem. A 2012, 116, 263–270
The Journal of Physical Chemistry A adjusting the stoichiometric coefficients for each species. As clearly shown in Figure S6, the theoretical results obtained considering the formation of a mixture of H4Co2Mo10O386 and each of these species did not lead to satisfactory agreements contrarily to Mo8O264. As the theoretical H4Co2Mo10O386 concentrations well fit with the experimental ones, it can be concluded that the chemical eq 5 really drives the H4Co2Mo10O386 formation. Almost all of the protons are consumed during the reaction, and, in fact, the quantity of dimers formed is determined by the quantity of Co2+ countercations necessary for electroneutrality. As shown in Figure 10, the quantity of dimer formed strongly decreases with the Co/Mo ratio because its [Mo] concentration drops to only 0.14 M for Co/Mo = 0.3 and should reach 0 for Co/Mo = 0.25. As a consequence, H4Co2Mo10O386 cannot be obtained by the peroxo route synthesis using Co/Mo lower than 0.25. Domains of Stability. The preparation of pure H4Co2Mo10O386 aqueous solutions by the peroxo route using CoCO3 and MoO3 precursors followed by a hydrothermal treatment has allowed us to investigate the domains of stability of this dimeric Anderson HPA. The purpose of this study was to simulate physicochemical phenomena involved in the impregnation step of γ-Al2O3 support during synthesis of catalysts using H4Co2Mo10O386 aqueous solutions. For instance, the pH of the dimer solution will increase in contact with γ-Al2O3 because their surface hydroxyl groups tend to modify the solution pH to reach the PZC value,26 the Co/Mo ratio may vary, and concentration gradients may locally appear during the diffusion of the solution into the support especially for shaped supports like extrudates. H4Co2Mo10O386 aqueous solutions are stable in a wide range of pH values (14.5), Co/Mo ratios (0.10.5), and Mo concentrations (0.020.8 M). In particular, the dimer appears quite stable at pH 1 contrarily to H6AlMo6O243 Anderson anions, for instance, which is unstable for pH lower than 2.6 Additionally, its decomposition constant Kd upon increase in the pH was found much lower (log(Kd) = 70) than the value reported for H6AlMo6O243 (log(Kd) = 51),6 which confirms the high stability of the dimer. When solutions are prepared using a Co/Mo ratio lower than 0.5, Mo8O264 is simultaneously formed. A consecutive dilution until a [Mo] concentration of 0.008 M does not lead to a change in the H4Co2Mo10O386 quantity, confirming the high stability of the dimer. Even when the pH is varied, the quantity of dimers remains constant, and only the Mo8O264 anions present at natural pH are replaced at a given pH by the stable isopolymolybdates according to the Mo speciation diagram.5,6,20
’ CONCLUSIONS In the present study, the peroxo route was used to prepare H4Co2Mo10O386 aqueous solutions from MoO3 and CoCO3. As the standard conditions lead to a mixture of monomers and dimers, the synthesis protocol was improved by adding a consecutive hydrothermal treatment at 150 °C. Hence, pure H4Co2Mo10O386 aqueous solutions were obtained for Co/ Mo equal to 0.5 that corresponds to the ideal ratio to obtain only dimers. When solutions were prepared with lower Co/Mo, a mixture of H4Co2Mo10O386 and Mo8O264 species was obtained after the hydrothermal treatment as evidenced by Raman spectra treated with the SIMPLISMA algorithm. From quantitative and theoretical data, it was demonstrated that the quantity of
ARTICLE
dimers formed is determined by the quantity of Co2+ countercations available in the solutions to ensure the electroneutrality and hence by the quantity of CoCO3 added and can be calculated just from the value of the Co/Mo ratio. After formation, H4Co2Mo10O386 heteropolyanions are stable in a wide range of pH and [Mo] concentrations. However, addition of bases such as NaOH and NH4OH destabilizes the dimer above pH 4.5 and leads to the formation of MoO42 anions and precipitates. A change of the Co/Mo ratio after the peroxo route synthesis does not lead to destabilization of H4Co2Mo10O386 anions. Finally, this work will be useful to rationalize the impregnations of γ-Al2O3 by H4Co2Mo10O386, which lead to strong variation of pH, Co/Mo ratios, and Mo concentrations inside the alumina porosity. These parameters can also influence the maturation or drying steps leading to uncontrolled surface species, which might be detrimental to the catalytic activities of hydrotreatment catalysts prepared from H4Co2Mo10O386 precursors. The methodology used in this study that consisted of combining quantitative measurements and principal component analysis can be transposed to other systems.
’ ASSOCIATED CONTENT
bS
Supporting Information. Evolution of Raman spectra of 0.5Co:Mo solution adding active carbon. Individual components extracted by the SIMPLISMA algorithm of Raman spectra of 0.3Co:Mo and 0.5Co/Mo solutions. XRD patterns of the precipitate obtained above pH 5.5 for 0.5Co:Mo solution and of [NaCo2(MoO4)2OH] 3 H2O reference. Influence of the [Mo] concentration on Raman spectra of 0.3Co:Mo solution. Evolution of UVvis spectra obtained at various [Co] concentrations. Comparison between the [Mo] concentrations of H4Co2Mo10O386 calculated considering Mo8O264, HMo8O263, H2Mo7O244, HMo7O245, Mo7O246, or MoO42 are by-produced in eq 5. This material is available free of charge via the Internet at http://pubs. acs.org.
’ AUTHOR INFORMATION Corresponding Author
*Phone: + 33 [0] 472 445 300. Fax: + 33 [0] 472 445 300. E-mail:
[email protected].
’ ACKNOWLEDGMENT K. Marchand is acknowledged for fruitful discussions during the first year of J.M.’s Ph.D. thesis. ’ REFERENCES (1) Katsoulis, D. E. Chem. Rev. 1998, 1, 359–387. (2) Okuhara, T.; Mizuno, N.; Misono, M. Appl. Catal., A 2001, 222, 63–77. (3) Mizuno, N.; Misono, M. Chem. Rev. 1998, 98, 199–217. (4) Baes, C. F.; Mesmer, R. E. The Hydrolysis of Cations; Wiley: New York, 1976. € (5) Pettersson, L.; Andersson, I.; Ohman, L.-O. Inorg. Chem. 1986, 25, 4726–4733. € (6) Ohman, L.-O. Inorg. Chem. 1989, 28, 3629–3632. (7) Odyakov, V. F.; Zhizhina, E. G.; Maksimovskaya, R. I.; Matveev, K. I. Kinet. Catal. 1995, 36, 733–738. 269
dx.doi.org/10.1021/jp208534p |J. Phys. Chem. A 2012, 116, 263–270
The Journal of Physical Chemistry A
ARTICLE
(8) Bergwerff, J. A.; van de Water, L. G. A.; Visser, T.; de Peinder, P.; Leliveld, B. R. G.; de Jong, K. P.; Weckhuysen, B. M. Chem.-Eur. J. 2005, 11, 4591–4601. (9) Lamonier, C.; Martin, C.; Mazurelle, J.; Harle, V.; Guillaume, D.; Payen, E. Appl. Catal., B 2007, 70, 548–556. (10) Martin, C.; Lamonier, C.; Fournier, M.; Mentr, O.; Harle, V.; Guillaume, D.; Payen, E. Chem. Mater. 2005, 17, 4438–4448. (11) Payen, E.; Guillaume, D.; Lamonier, C.; Marchand, K. Method of preparing at least one cobalt and/or nickel salt of at least one Anderson’s heteropolyanion combining molybdenum and cobalt or nickel in its structure. European Patent 1892038 A1, February 27, 2008. (12) Harle, V.; Martin, C.; Payen, E.; Lamonier, C.; Blanchard, P. Catalyst containing metals from Group VI and VIII at least partly in the form of an heteropolyanion in the oxide precursor. European Patent 1393802 A1, March 3, 2004. (13) Nolan, A. L.; Burns, R. C.; Lawrance, G. A. J. Chem. Soc., Dalton Trans. 1996, 13, 2629–2636. (14) Campbell, N. J.; Dengel, A. C.; Edwards, C. J.; Griffith, W. P. J. Chem. Soc., Dalton Trans. 1989, 1203–1208. (15) Eiler, P. H. C.; Boelens, H. F. M. Baseline Correction with Asymmetric Least Squares Smoothing; 2005; http://www.science.uva.nl/ hboelens/publications/draftpub/Eilers_2005.pdf. (16) Windig, W. Chemom. Intell. Lab. Syst. 1997, 36, 3–16. (17) Martin, C.; Lamonier, C.; Fournier, M.; Mentre, O.; Harle, V.; Guillaume, D.; Payen, E. Inorg. Chem. 2004, 43, 4636–4644. (18) Alizadeh, M. H.; Salimi, A. R. J. Mol. Struct. (THEOCHEM) 2007, 809, 1–10. (19) Himeno, S.; Niiya, H.; Ueda, T. Bull. Chem. Soc. Jpn. 1997, 70, 631–637. (20) Tytko, K. H.; Baethe, G.; Cruywagen, J. J. Inorg. Chem. 1985, 24, 3132–3136. (21) Bergwerff, J. A.; Visser, T.; Weckhuysen, B. M. Catal. Today 2008, 130, 117–125. (22) Cabello, C. I.; Cabrerizo, F. M.; Alvarez, A.; Thomas, H. J. J. Mol. Catal. A: Chem. 2002, 186, 89–100. (23) Geue, R. J.; Hanna, J.; H€ohn, A.; Qin, C. J.; Ralph, S. F.; Sargeson, A. M.; Willis, A. C. In Electron Transfer Reactions; Isied, S., Ed.; ACS Symposium Series; American Chemical Society: Washington, DC, 1997; Chapter 8. (24) Nolan, A. L.; Allen, C. C.; Burns, R. C.; Craig, D. C.; Lawrance, G. A. Aust. J. Chem. 1998, 51, 825–834. (25) Bergwerff, J. A.; Visser, T.; Leliveld, B. R. G.; Rossenaar, B. D.; de Jong, K. P.; Weckhuysen, B. M. J. Am. Chem. Soc. 2004, 126, 14548–14556. (26) Moreau, J.; Delpoux, O.; Marchand, K.; Digne, M.; Loridant, S. Stud. Surf. Sci. Catal. 2010, 175, 397–400.
270
dx.doi.org/10.1021/jp208534p |J. Phys. Chem. A 2012, 116, 263–270
