- Email: [email protected]
Study of the Co-VPO interaction in promoted n-butane oxidation catalysts
Applied Catalysis A: General 217 (2001) 275–286
Study of the Co-VPO interaction in promoted n-butane oxidation catalysts C. Carrara, S. Irusta, E. Lombardo, L. Cornaglia∗ Instituto de Investigaciones en Catálisis y Petroqu´ımica – INCAPE (FIQ, UNL-CONICET), Santiago del Estero 2829, 3000 Santa Fe, Argentina Received 26 December 2000; received in revised form 5 April 2001; accepted 7 April 2001
Abstract Cobalt-impregnated VPO catalysts were prepared using different cobalt salts and impregnation methods. The catalytic tests showed that cobalt impregnation significantly increased the overall activity. The use of cobalt acetyl acetonate led to a more selective high loading catalyst. To investigate the origin of the cobalt effect, the solids were characterized using XRD, Raman spectroscopy, FT-IR, and XPS. The surface acidity was probed by adsorbing acetonitrile. No structural effects were detected through XRD. After 700 h on stream, the only phase detected in all cases was V(IV) vanadyl pyrophosphate. The surface oxidation state of vanadium was V(IV). The Co 2p XP spectrum showed an intense shoulder at 788 eV, indicating that Co(II) species were present. The concentration of very strong Lewis acid sites increased at higher cobalt loading, but its strength was unaffected. © 2001 Elsevier Science B.V. All rights reserved. Keywords: VPO; n-Butane oxidation; Maleic anhydride; Promoters; Cobalt; Lewis acidity
1. Introduction The VPO system exhibits a unique ability to activate and selectively oxidize alkanes. Vanadyl pyrophosphate is commercially used to catalyze the selective oxidation of n-butane to maleic anhydride (MAN). It is generally agreed upon that the best catalyst precursor is VOHPO4 ·0.5H2 O which is converted to (VO)2 P2 O7 [1] during activation. Besides, several VOPO4 phases present in low proportion have been claimed as necessary for the catalytic act [2]. However, these phases have never been detected in the so-called equilibrated catalysts that have been on stream for 700 h [3].
∗ Corresponding author. Tel.: +54-342-4536861; fax: +54-342-4536861. E-mail address: [email protected] (L. Cornaglia).
A variety of cations have been added to vanadium phosphate catalysts to improve activity and selectivity. It has been claimed that the addition of the first row transition metal produces an improvement in the maleic anhydride yield [4,5]. Several papers [4–10] refer to the effect of cobalt used as a promoter. Takita et al. [6] sustained that the additives were incorporated into the crystal lattice of vanadyl pyrophosphate. Volta et al. [7] using in situ Raman spectroscopy found that the incorporation of Co can change the V(V)/V(IV) balance during the activation period and can therefore change the catalytic performance in the steady state. Other authors [5,7,8] studied the influence of cobalt addition on the composition of the (VO)2 P2 O7 catalyst. The presence of cobalt induced surface phosphorus enrichment, which modified the surface acidity. Zazhigalov et al. [8] proposed that cobalt stabilize the catalyst performance by
0926-860X/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 0 1 ) 0 0 6 1 5 - 9
276
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
forming cobalt phosphate, which improves its catalytic properties. According to Zazhigalov et al. [8], moderate surface acidity facilitates desorption of maleic anhydride and prevents its complete oxidation to carbon oxides. They suggested that promoters play an important role in controlling the surface acidity of promoted VPO solids. Selective oxidation of n-butane requires Lewis–Brönsted acid site in combination with a surface vanadium redox site. In a recent paper, we have studied impregnated Co-VPO catalysts obtained using cobalt acetate [10]. We concluded that on equilibrated catalysts, the addition of cobalt increases the activity while slightly decreasing the selectivity. This effect seems to be a consequence of increased exposure of cobalt on the catalyst surface. In this work, we have extended our studies using different salts and impregnation procedures. The rational was that cobalt acetyl acetonate could allow us to change the cobalt dispersion on the surface, while the high-temperature impregnation may help intercalate the Co atoms between the VPO layers. Besides, in search of additional clues to understand the role of Co on the surface of equilibrated solids, we have now probed the surface acidity using acetonitrile. 2. Experimental 2.1. Catalyst preparation The precursor was obtained by reduction of 5 g of V2 O5 with 30 ml of isobutanol and 20 ml of benzyl alcohol under reflux for 3 h. Then orthophosphoric acid (100%) was added in the desired amount and the solution refluxed for another 2 h. After completion of the reaction, the solid phase was recovered by filtration and dried in air at 390 K overnight. 2.1.1. Promoted catalyst The impregnation was carried out using either cobalt acetate or acetyl acetonate and following two procedures. 1. The impregnation method: the salt, acetate (VPCox-Ia) or acetyl acetonate of Co (VPCox-Iaa where x represents wt.% of Co) was dissolved in 30 ml of isobutanol. The previously prepared precursor was added to this solution and the
suspension was heated to 330 K with continuous stirring. The solvent was evaporated and the wet solid dried at 390 K. The final Co content of all of the preparations was determined by atomic absorption. 2. The high-temperature impregnation method: cobalt acetyl acetonate was used in this case (VPCox-HTI). The salt was dissolved in a mixture of 30 ml of isobutanol and 20 ml of benzyl alcohol, the precursor was added and refluxed overnight. The precipitate was then filtered and dried in air at 390 K. 2.2. Catalyst activation and testing The promoted catalyst precursors and unpromoted VPO were activated in a fixed bed microreactor. Typically, about 1 g of catalyst (screened 170–250 m range Tyler 60–80 mesh) was loaded into the reactor and covered with a layer of quartz wool. The quartz reactor was 60 cm long and 1.0 cm i.d. at the catalyst bed portion. It was mounted vertically in a tubular furnace. A programmable temperature controller was used to assure reproducibility of heating strategies. Mass flow controllers were used to feed the reactants in the right proportion. An on-line gas chromatograph equipped with a FID detector was used to analyze both the reactant and product streams. Separation of C4 H10 and maleic anhydride was accomplished with a 2 ft 1/8 in SS Chromosorb WAW AT-1200+1% H3 PO4 80/100 column. The activation of the catalyst was performed by heating from ambient temperature to 603 K, while flowing a mixture of 0.75% n-butane in air, with a gas hourly space velocity (GHSV) of 900 h−1 . The temperature was kept constant at 603 K for 3 h. Those catalysts which only underwent this treatment are here called nonequilibrated catalysts. Then, the n-butane concentration was increased to 1.5% while still maintaining the GHSV at 900 h−1 . The temperature was raised until either 80% conversion or the maximum allowable temperature of 703 K was reached. Then, the GHSV was raised to 2500 h−1 and the temperature was slowly increased with the limitation that conversion never exceeded 80%. The catalytic data reported here were measured over solids that did not show any change in activity and selectivity after at least 500 h on stream. At this point when the reaction temperature
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
was randomly varied up and down, reproducible values of activities and selectivities were obtained. These solids are called equilibrated catalysts. 2.3. Catalyst characterization 2.3.1. X-ray diffraction The measurements were made with a Shimadzu XD-D1 X-ray diffractometer, using nickel-filtered Cu K␣ radiation with a scanning rate of 1◦ min−1 . 2.3.2. Laser Raman spectroscopy (LRS) The Raman spectra were recorded with a Jasco laser Raman spectrometer model TRS-600-SZ-P, equipped with a CCD (charge coupled device) with the detector cooled to about 153 K using liquid N2 . The excitation source was the 514.5 nm line of a Spectra 9000 Photometrics Ar ion laser. The laser power was set at 30 MW. All the spectra were recorded with the samples under ambient conditions. The powdered solid was pressed into a thin wafer about 1 mm thick. 2.3.3. Infrared spectroscopy The IR spectra were obtained using a Shimadzu FT-IR 8101M spectrometer with a spectral resolution of 4 cm−1 . The solid samples were prepared in the form of pressed wafers (ca. 2 wt.% sample in KBr). The IR spectra of precursors were obtained using fluorolube as a diluting agent in order to explore the high wave number region. The samples for the adsorption experiments were prepared by compressing the used catalysts at 9 tonnes cm−2 in order to obtain a self-supporting wafer (50 mg, 10 cm diameter). They were mounted in a transportable infrared cell with CaF2 windows and external oven. The pretreatment was performed in a high-vacuum system. The sample was first outgassed at 723 K for 12 h in a dynamic vacuum of 7×10−4 Pa. After cooling to room temperature, a spectrum of the catalyst wafer was taken. No bands were observed in the 1600 cm−1 region, this is indicative that molecular water was eliminated. After that, 4.8 × 104 Pa of acetonitrile was admitted into the cell and left in contact with the solid for 10 min, and then a spectrum was recorded. No changes were observed in the 1600 cm−1 region, therefore no water was introduced during acetonitrile adsorption. Spectra were
277
also recorded after evacuation of the cell for 10 min at 298, 353, and 423 K. 2.3.4. X-ray photoelectron spectroscopy The XPS measurements were carried out using an ESCA750 Shimadzu electron spectrometer. Nonmonochromatic Al K␣ X-ray radiation was used. The anode was operated at 8 kV and 30 mA and the pressure in the analysis chamber was about 2 × 10−6 Pa. The data were collected using an ESCAPAC 760 computer interfaced to the spectrometer and analyzed with the Googly software developed at the University of Pittsburgh. The binding energies (BE) were referred to the C 1s signal (284.6 eV). Curve fitting was performed using a Levenberg–Marquardt NLLSCF routine. The background contribution was taken into account by assuming an integral type background which was included in the basic shape of each peak. Account has been taken of the presence of K␣3 and K␣4 spectral lines of the large O 1s signal. These satellite peaks are 9.6 and 11.6 eV downshifted from the O 1s main peak and overlap the V 2p1/2 signal. Since the resolution of the V 2p1/2 is poorer than that of the V 2p3/2 level, we preferably used the latter BEs and widths for comparison with literature data. The V 2p and Co 2p doublets were fitted using a Voigt function with 20% Lorentzian character and 2p1/2 /2p3/2 intensity ratio = 0.5. Additional data for the curve fitting of the V 2p doublets include spin-orbit separation = 7.2 eV. The surface P/V and Co/V atomic ratios were calculated using the areas under the Co 2p3/2 , P 2p, and V 2p3/2 peaks, the Scotfield photoionization cross-sections, the mean free paths of the electrons, and the instrumental function which was given by the ESCA manufacturer. 3. Results and discussion 3.1. Catalytic behavior Fig. 1A and B shows the catalytic results obtained with promoted and unpromoted solids. The impregnated solids exhibit a significantly higher conversion than unpromoted VPO in the whole temperature range, with the exception of the VPCo1.5-Ia. For the promoted catalysts, the selectivity was nearly constant up
278
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
(Fig. 2A and B). Note that at T > 663 K, the yield of VPCo4-Iaa is comparable to industrial catalysts. To investigate how cobalt affects the bulk and surface features of VPO solids, a set of techniques has been used. 3.2. Structure of VPO precursors and catalysts
Fig. 1. Catalytic behavior of equilibrated solids. Reaction conditions: 1.5% n-butane/air; GHSV = 900 h−1 .
to ca. 80% conversion. VPCo4-Iaa is the most selective solid assayed. A higher yield to maleic anhydride is reached with Co-promoted catalysts at GHSV = 900 h−1 . The same behavior is observed at high GHSV
All the FT-IR spectra of the precursors (diluted in fluorolube) show the characteristic fingerprints of the vanadyl hydrogen phosphate hemihydrate. There appears a band at 1640 cm−1 corresponding to the coordinated water, and a very marked shoulder at 1600 cm−1 , which could be due to hydrogen-bonded uncoordinated water. The bands at 1509, 1492, and 1450 cm−1 (C=C vibration in the plane of the aromatic ring) correspond to the benzyl alcohol and benzaldhyde retained in the solid [11]. In the solids prepared by impregnation with Co acetate, with a content ≥ 4%, there appear bands at 1564 and 1430 cm−1 corresponding to the salt, according to what has been reported [10]. The same happens with the solids impregnated with acetyl acetonate in which it is possible to observe signals at 1526 and 1430 cm−1 , characteristic of the salt. This indicates that even after drying, the cobalt salts are present in these precursors. In the catalysts prepared under reflux, the signal at 1526 cm−1 is not observed even for 4% Co. In order to investigate if this was due to the thermal treatment, an experiment was performed in which the solution was refluxed but in the absence of the solid. The solution thus treated presents bands at 1526 and 1430 cm−1 , characteristic of acetyl acetonate. The disappearance of the band at 1526 cm−1 in the precursor could be due to the interaction between the solid and the salt. After activating the solids, the salt bands are no longer observed and the presence of cobalt oxides is not detected by XRD. Rozynek and Polansky [12] studied the bulk and surface reduction properties of the silica-supported cobalt catalysts. The TPR profile of the calcined acetate (supported on silica) was not characteristic of Co3 O4 . Besides, no XRD peak characteristic of oxides could be observed for the calcined acetate, in agreement with our results in VPO catalysts. The Raman Co3 O4 pattern presents sharp bands at 691 and 524 cm−1 [13]. As these bands are not observed in the Raman spectra of the promoted
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
Fig. 2. Maleic anhydride yield of equilibrated catalysts: (A) GHSV = 900 h−1 ; (B) GHSV = 2500 h−1 . Reaction conditions: 1.5% n-butane/air.
279
catalyst, this is considered a confirmation of the XRD results. Zazhigalov et al. [8] observed the appearance of Co2 P2 O7 for the catalysts with Co/V > 0.15. In our case (Co/V = 0.13), neither XRD nor Raman spectroscopy indicate the presence of Co-containing phases. In Table 2, the XRD data for nonequilibrated and equilibrated catalysts are shown. In all cases, vanadyl pyrophosphate was the only crystalline phase detected. The decrease in the FWHMs of the (2 0 0) reflection with a long time-on-stream was observed. No structural modifications (Table 1) were observed on impregnated solids through XRD. The FT-IR spectra of both equilibrated and nonequilibrated solids show the fingerprints of vanadyl pyrophosphate between 700 and 1700 cm−1 , with the following assignments: ν(PO3 ) = 1240 cm−1 , ν(V=O) = 968 cm−1 , ν(V–O–V) = 797 cm−1 , and ν(P–O–P) = 742 cm−1 [14]. In the cases where Co has been introduced during the phosphatization step, it was observed that the P–O–P vibration was modified [9]. The ν(P–O–P) at 742 cm−1 has a similar intensity in the unpromoted and impregnated catalysts. This is substantiated by the high values (∼15 ± 1) of the 1245/742 intensity ratio (Table 1). It can be observed that the values obtained for the intensity ratio are lower in the case of the high-temperature Co-impregnated solids. This results from an increase in the band intensity corresponding to the P–O–P vibration. This increase is independent of both the Co content and the catalyst time-on-stream. This addition method of the Co promoter would also be introducing an increase in the number of P–O–P layer linkages. The same effect was observed when cobalt was added during phosphatization (VPCo4-P). The Raman spectra show that none of the equilibrated catalysts contain VOPO4 phases. This is at variance with the findings of Hutchings et al. [7]. They report that Co and Fe dopants can modify the VOPO4 /(VO)2 P2 O7 dispersion. They concluded that the activity for maleic anhydride formation is a function of the nature and concentration of the V(V) phases present. None of the three instrumental techniques allowed us to detect significant structural differences in the catalysts with and without promoter addition, which could have explained their different catalytic behavior. Let us see now if we can find “clues” on the surface to explain this phenomenon. Presence of V(V), Co oxidation state, P surface enrichment, and
280
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
Table 1 XRD and FT-IR data of precursors, nonequilibrated and equilibrated catalysts Solid
VPO VPCo1.5-Ia VPCo4-Ia VPCo4-Iaa VPCo2-HTI VPCo4-HTI VPCo4-Pe
Precursora
Nonequilibrated
FWHMb
FWHMc
(0 0 1)
1.0 1.0 1.14 1.2 1.4 1.4 0.9
Equilibrated
(2 0 0)
I1240 /I742
1.2 2.1 2.2 1.8 1.8 1.7 1.6
d
– – 16.3 14.2 12.8 12.6 9.9
FWHMc (2 0 0)
I1240 /I742
1.0 1.0 1.1 1.0 1.15 1.1 1.38
14.0 – 14.0 16.7 10.8 10.9 10.0
Bulk P/V ratio = 1.26. VOHPO4 ·0.5H2 O (2θ = 15.5◦ ); FWHM in 2θ . c (VO) P O (2θ = 23.2◦ ); FWHM in 2θ . 2 2 7 d Relative intensity of FT-IR bands: ν(PO ) = 1240 cm−1 and ν(P–O–P) = 742 cm−1 . 3 e From [9]; cobalt (as cobalt acetate) was added during phosphatization. a
b
variation of the surface acidity are the main variables to be explored.
Table 3 Surface vanadium oxidation states of the equilibrated catalysts Solids
3.3. Surface of equilibrated catalysts
V 2p3/2 a,b
∆[O 1s − V 2p3/2 ] (eV) Vox c
BE (eV) FWHM VPO VPCo4-Ia VPCo4-Iaa VPCo4-HTI
In our catalysts, cobalt impregnation produces an increase of the precursor surface P/V ratio (Table 2). This ratio increases upon equilibration in the unpromoted VPO, whereas it decreases in the Co-containing catalysts. This effect could be due to a phosphorus diffusion into the bulk with time-on-stream. The phosphorus surface enrichment in Co-promoted catalysts was previously reported by other authors [7–10]. The surface oxidation states of vanadium were calculated from the difference in BE between the O 1s and V 2p3/2 signals. This method avoids the use of a reference as we concluded in a previous study [15]. The results are shown in Table 3. This value is equal
517.9 517.8 517.7 517.9
2.3 2.2 2.3 2.2
14.3 14.3 14.5 14.3
4.09 4.09 3.96 4.09
a
BEs determined by curve fitting. O 1s BE = 532.2 eV was used as reference. c Using the following correlation: V ox = 13.82 − 0.68 (∆[O 1s − V 2p3/2 ]), from Coulston et al. [16]. b
to 14.3 ± 0.1 eV for promoted and unpromoted catalysts. Coulston et al. [16] have reported a correlationship between the ∆ value and the average vanadium oxidation state. These calculated values are around 4 in all the cases studied. The curve fitting leads to
Table 2 XPS surface atomic ratios of precursors, nonequilibrated and equilibrated catalystsa Solids
VPO VPCo1.5-Ia VPCo4-Ia VPCo4-Iaa VPCo2-HTI VPCo4-HTI a
Precursors
Nonequilibrated
Equilibrated
P/V
Co/V
P/V
Co/V
P/V
Co/V
2.15 3.80 3.70 3.40 3.50 3.70
– 0.36 0.60 0.70 0.28 0.69
2.9 3.8 3.0 3.9 – –
– – 0.30 0.85 – –
2.8 2.6 2.6 2.5 2.6 2.5
– 0.31 0.27 0.28 0.26 0.30
Bulk P/V ratio = 1.26.
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
281
Table 4 XPS binding energies for cobalt in equilibrated catalystsa Solids
Co 2p3/2 (eV)
Satellite (eV)
Isat /IMP b
Co 2p1/2 (eV)
Satellite (eV)
Isat /IMP c
∆[Co 2p1/2 − Co 2p3/2 ]
CoO
780.5d 780.1e
786.9 788
0.9 –
796.3 795.3
803 804.6
0.8 –
15.8 15.2
Co3 O4
779.5f 779.6e
788 787.1
– –
794.6 794.8
803 804.1
0.2 –
15.1 15.2
Co(OH)2 VPCo4-Ia VPCo4-Iaa VPCo4-HTI
781.0 782.7 (3.0) 783.2 (3.3) 783.3 (3.3)
786.4 787.4 (7.3) 787.4 (5.4) 788 (5.0)
– 1.4 0.8 0.7
796.9 798.4 (3.1) 798.8 (3.2) 799.0 (3.4)
802.6 802.6 (7.0) 803.5 (5.8) 804.2 (6.0)
– 1.5 1.0 1.1
15.9 15.7 15.6 15.7
a
FWHM values (eV) are given in parenthesis. Intensity ratio Co 2p3/2 satellite/main peak. c Intensity ratio Co 2p 1/2 satellite/main peak. d [17]. e [20]. f [18]. b
a single, well-defined binding energy for V 2p3/2 (517.7 ± 0.1 eV) assigned to V(IV) in agreement with the values reported by López Granados et al. [3] for unpromoted catalysts. Both methods support the overwhelming presence of V(IV) on the surface of the VPO equilibrated catalysts. The addition of cobalt does not change the oxidation states of surface vanadium. The existence of Co(II) in the impregnated catalyst has been confirmed by XPS measurements (Table 4 and Fig. 3). The binding energies of reference compounds and high Co loading equilibrated catalysts are summarized in Table 4. While the 2p1/2 and 2p3/2 binding energies of Co(II) are relatively close to that of Co(III), the two oxidation states of cobalt can be distinguished by the presence of a distinct shake up satellite structure in Co(II) arising from the presence of unpaired electrons in their valence orbital [17,18]. Co(III) is almost always in a low-spin state and therefore there can be no energy transfer to an unpaired electron, meaning that a shake up satellite structure is not observed. The disappearance of the satellite peaks in the XPS profile would then be indicative of oxidation of Co(II) to Co(III). The Co3 O4 has a spinel structure and contains magnetic Co(II) ions (tetrahedral sites) and diamagnetic Co(III) ions (octahedral sites). These two sites are indistinguishable by XPS. Although, all compounds containing high-spin Co(II) ions have strong satellite peaks in the Co 2p spectra, only Co3 O4 has weak
satellites in this region [20]. These satellites occur about 7–9 eV higher in binding energy from the main bands, which can be observed in Table 4. High-spin Co(II) core level spectra (as in CoO) show strong satellite peaks (Table 4) above the main 2p lines. Frost et al. [21] have shown that high-spin Co(II) compounds exhibit intense satellites, while low-spin Co(II) compounds exhibit weak or missing satellites
Fig. 3. XPS spectra of the Co 2p region.
282
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
in their photoelectron spectra. Thus, it is not difficult to differentiate high-spin Co(II) compounds from Co2 O3 . Cobalt hydroxide has a strong satellite peak associated with both the lines of the Co 2p region. For the VPCo4-Iaa and VPCo4-Ia catalysts, the Co 2p3/2 binding energies were different. The value of 783.3 eV indicates that the presence of CoO on the surface is unlikely. The Co 2p3/2 BE is 780.1 eV for CoO. For Co-promoted vanadyl pyrophosphate, the binding energy is much higher. The Co 2p spectrum has an intense shoulder at ∼788 eV. In addition, the 2p1/2 –2p3/2 spin-orbit splitting of 15.7 eV for this catalyst indicates that Co(II) species are present. Fierro et al. [19] have proposed an interpretation for cobalt, similar to that reported for copper. For Co(II)-containing compounds, the main and the satellite peaks can be related, respectively, to 2p5 3d8 L (L means a ligand hole) and 2p5 3d7 electronic configuration in the final state. One of the most important consequences of this theory is that useful information about the nature of the metal–ligand chemical bonding can be simply gained by evaluating the satellite/main peak intensity ratio and their energy separation. They found that the energy separation increases and the satellite/main peak intensity ratio decreases with increasing covalent character of the metal–ligand bonding. On examining the XPS profile of the impregnated catalysts, the Co 2p satellite intensities approach satellite/main peak intensity ratios observed for bulk CoO, while the strong satellite lines are centered about 4.7–4.2 eV above the principal line. These observations indicate that Co(II) is not being oxidized to Co(III). Besides, the covalent character of metal–ligand bonding is not affected neither by the method nor the salt used for impregnation. The high BEs (783.3–782.7 eV) might be related to strong interactions between cobalt and other atoms of the solid matrix and/or to a highly dispersed cobalt. Binding energies of 783 eV were reported [22] for the ion exchanged form of cobalt in zeolites. In these solids, the higher than normal binding energy is suggestive of cobalt in a highly oxidizing environment. Zazhigalov et al. [8] reported a similarly high binding energy of 783.0 eV for Co-promoted VPO catalysts. They compared their results with the value of 782.8 found in cobalt molybdate or 782.2 and 783.0 for CoCl2 and CoF2 , respectively, and concluded that the metal–ligand bonds are polarized in these reference
compounds and a similar structure may be expected for cobalt phosphates. They sustained that the high binding energy values of 782.7–783 eV observed in their catalysts can be considered as an additional indication of the formation of cobalt phosphate (detected by XRD). On the other hand, they suggested the existence of strong interactions between cobalt atoms and phosphate groups with a shift of bonding electrons towards the phosphate anions. In our catalysts, the Co/V nominal ratio is lower than 0.13, and no cobalt phosphates were detected by XRD, FT-IR, and Raman spectroscopy. In agreement with our results, Hutchings and Higgins [5] have recently reported that the XRD patterns do not even hint the formation of solid solutions at low promoter ratios, M/V ≤ 0.12. Hutchings [4] proposes that the promoters at high promoter/vanadium atomic ratios act as phosphorus scavengers by either formation of metal phosphates or by formation of solid solutions. In our case, considering the high P/V surface ratio and the fact that all the catalysts have similar Co/V surface ratios, it is possible that the high BE of Co 2p3/2 be symptomatic of the presence of surface cobalt pyrophosphate. This surface compound does not have long-range, three-dimensional order, because characteristic X-ray diffraction lines have not been observed. However, binding energies are not a sufficient criterion to confirm the presence of a given compound. A difference in BE was observed between the two impregnated catalysts. This could be due to either different surface Co(II) interactions or to a change of the cation distribution at the surface. Two different impregnating salts (cobalt acetate and acetyl acetonate) might produce catalysts with varying strengths in the interaction between cobalt and VPO or different Co dispersion. 3.4. Lewis acidity of equilibrated solids To ascertain the effect of Co upon the Lewis acidity of the VPO formulations, acetonitrile was used as a probe molecule. When this weak base is adsorbed on the VPO solids, the following bands are observed: ν = 2252 cm−1 assigned to H-bonded and physisorbed molecules, ca. ν = 2270 cm−1 corresponding to CN groups interacting with medium strong Lewis acid sites, ν = 2294 cm−1 associated with the Fermi resonance of the CN group (ν3 + ν4 ),
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
and a fourth band at 2322 cm−1 assigned to the interaction of the base with very strong Lewis acid centers. A shift of 28 cm−1 is observed when this base is adsorbed on ␣-Fe2 O3 [23]. Displacements of ca. 75 cm−1 from the liquid phase frequency are observed when acetonitrile is adsorbed on AlCl3 [24]. The band at 2252 cm−1 rapidly decreases with increasing temperature in all the equilibrated catalysts. A band at 2270 cm−1 ( ν = 18 cm−1 ) appears in the VPO spectrum taken in the presence of the gas phase, and remains when the evacuation temperature is increased. This band is also observed in the VPCo4-Iaa and in VPCo4-HTI in the presence of the gas phase. But this band almost disappears upon temperature increase. The analysis of the spectra obtained with the solids that were fast-activated (nonequilibrated), shows less adsorption in general. However, it is also seen that all the spectra show the signal assigned to the medium strong sites (2270 cm−1 ) when there is acetonitrile in the gas phase. But these signals disappear after evacuation at 298 K. It could be concluded that medium strong Lewis acid sites are present in all nonequilibrated catalysts; when the catalysts become equilibrated, these sites remain only in the unpromoted catalyst and in the catalyst with low percentage of Co (1.5%). The band at 2322 cm−1 , attributed to very strong Lewis acid sites, appears practically at the same frequency in all the solids (Fig. 4). It could accordingly be concluded that the addition of Co in different amounts and by different methods does not introduce any modification in the strength of the very strong Lewis acid sites. Table 5 also shows the intensity ratio of the band at 2322 cm−1 with respect to that at 2297 cm−1 . This Table 5 Acetonitrile adsorption on equilibrated catalysts Catalyst
VPO VPCo1.5-Ia VPCo4-Ia VPCo4-Iaa VPCo4-HTI
Lewis site (cm−1 )
I2322 /I2297 a
Very strong
Medium strong
C/Fb
423 Kc
2322 2322 2322 2322 2322
2270 2270 – – 2270
0.56 0.80 0.60 0.69 0.53
0.48 0.53 1.89 1.14 1.75
Relative intensity of FT-IR bands at ν = 2322 and 2297 cm−1 . Acetonitrile in the gas phase. c After evacuation at 423 K. a
b
283
Fig. 4. FT-IR spectra of acetonitrile adsorption on promoted catalysts after evacuation at 423 K.
band is not influenced by the electronic environment [26], and it does not change with the catalyst acidity. Consequently, this ratio could be used to estimate the very strong Lewis acid site concentration. After evacuation at 423 K, the three catalysts with 4% Co contain a higher concentration of very strong Lewis acid sites than those with 1.5% Co and the unpromoted VPO. According to Busca et al. [25], the very strong Lewis acid sites of the nonequilibrated organically prepared VPO are due to a stretching of the V–O–P linkage. This stretching is a consequence of the stacking fold disorder of the (2 0 0) planes. In our previous study [26], we have demonstrated that the ratio of very strong to medium strong Lewis sites significantly increases with time-on-stream, corresponding to a well-ordered structure of the equilibrated catalysts. 3.5. Role of cobalt The Co addition significantly increases the overall activity. This seems to be a consequence of the presence of cobalt on the equilibrated catalyst surface. Using cobalt acetyl acetonate for the precursor
284
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
impregnation produces a solid with higher selectivities and performances comparable to commercial catalysts. In this complex structure, the promoter might be incorporated in the crystal lattice of (VO)2 P2 O7 , as reported by Takita et al. [6]. They studied the incorporation of promoter elements employing metal acetyl acetonates added during precursor synthesis. Incorporation of the promoter elements into the crystal lattice of (VO)2 P2 O7 brought about the shift in the IR absorption bands of the V=O stretching mode. In our catalysts, the V=O vibration was not modified by the presence of cobalt. They found the same behavior in the case in which Zn was located at the surface of vanadyl pyrophosphate. Guliants et al. [27] found a lower Nb surface ratio of promoter, which probably indicates higher solubility of the Nb promoter in the VPO matrix. Therefore, this suggested that promoter elements may partially form a solid solution. Another promoting effect could be that the cobalt atoms affect the layer linkages by becoming intercalated between the layers of the (VO)2 P2 O7 , as was reported for Fe- and Cr-promoted catalysts [28]. However, the IR band at 742 cm−1 , corresponding to the P–O–P vibration, has not significantly decreased in intensity in our Co-promoted catalysts. Neither a volumetric cobalt phosphate phase nor solid solutions were detected in our catalysts. The solids showed, however, high Co/V and P/V surface ratios plus a significant shift of the Co 2p BE. All this is consistent with the presence of a surface cobalt pyrophosphate in our catalysts. Several authors [25–27,29,30] have recalled the importance of the surface acidity for high activity and selectivity of the bulk VPO. The presence of acid sites should have an effect on the rate of activation of butane and on the rate of its oxidation to maleic anhydride. The acidity of the Co-promoted VPO catalyst was measured with NH3 TPD by Zazhigalov et al. [8]. It has been observed that adsorption of ammonia increases with increasing cobalt content. The ratio of the number of strong to weak adsorption sites decreased with increasing cobalt content. They concluded that the introduction of cobalt ions into the VPO catalyst is the main cause for the appearance of weak acid sites, whereas the number of strong acid sites rises to a lesser extent and only on addition of small amounts of cobalt (Co/V < 0.1 or 0.05 depending on the preparation method).
Busca et al. [25] proposed a model of butane activation in which it proceeds as a concerted simultaneous abstraction of two hydrogen atoms on an active center. This is composed of a strong Lewis acid–base V–O pair in which the empty d orbital of vanadium generates the Lewis acidity, while the occupied 2p orbital of oxygen is responsible for the Lewis basicity. The equilibrated catalysts containing 4% Co show a significant increase in the concentration of very strong Lewis acid sites over both the base VPO and the one containing 1.5% Co. A model of Lewis acid sites in vanadyl pyrophosphate was proposed by Trifirò et al. [1]. They indicated that the presence of defects in the structure of vanadyl pyrophosphate is reflected on the surface in an enhancement of Lewis acidity of surface unsaturated vanadium ions. According to this model, the appearance of very strong Lewis acid sites may be related to a different polarization of the V–O–P bonds, in our case induced by the presence of cobalt. The equilibrated VPCo4-Iaa solid is the most selective one. It has, however, the lowest concentration of very strong Lewis sites among the impregnated solids
Fig. 5. Maleic anhydride selectivities as a function of very strong Lewis acid sites concentration (data from Fig. 1B and Table 5).
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
with 4% Co. The correlation between selectivity and concentration of very strong Lewis sites at different conversions and space velocities is shown in Fig. 5. The presence of a maximum in selectivity might indicate that an optimum value exists for the concentration of very strong Lewis sites. An excess of very strong Lewis acid sites could decrease the rate of desorption of MAN, decreasing the maleic anhydride selectivity. Gleaves et al. [31] observed that the rate of desorption of the olefinic intermediates and furan is low compared to MAN, which implies that MAN should be rapidly desorbed to avoid its combustion. This would justify the former assertion and might indicate that the presence of very strong Lewis sites plays an important role regarding selectivity control. IR observations by Koyano et al. [32] of CO adsorbed al low temperature indicated that n-butane preferably interacts at Lewis acid sites, and recalled the importance of these sites to control selectivity. Note, however, that there could be other factors controlling the selectivity such as microstructure, P/V ratio, or a defect-free structure.
4. Conclusions They only apply to equilibrated catalysts. 1. As reported before [10], the addition of cobalt (≥2 wt.%) to VPO formulations significantly increases the catalytic activity (Fig. 1A). 2. When cobalt acetyl acetonate was used, the highest selectivities to maleic anhydride were observed (Fig. 1B). Besides, these Co-VPO solids give the highest yields (Fig. 2A and B). This ranking is also maintained at T > 660 K if all the previously studied cobalt-containing catalysts were included in Fig. 2 [9,10]. 3. The XPS data is consistent with the presence of Co-pyrophosphate in the first few surface layers (Table 4). However, bulk techniques (XRD, LRS, and FT-IR) could not detect the presence of cobalt phosphates. 4. The presence of cobalt does not modify the oxidation state of vanadium at the surface layers. Only V(IV) is detected through XPS (Table 3), in agreement with our previous studies [9,10]. 5. The presence of Co at the 4% level sharply increases the concentration of very strong Lewis
285
acid sites, although it does not affect their strength (Table 5). However, the best catalyst shows the lowest concentration among the catalysts containing 4% cobalt. This might indicate that an optimum concentration of very strong Lewis acid sites is needed to maximize the selective oxidation to maleic anhydride (Fig. 5). More work is needed to elucidate this point.
Acknowledgements The authors wish to acknowledge the financial support received from UNL (CAI+D 96 Program). They are also grateful to the Japan International Cooperation Agency (JICA) for the donation of the major instruments used in this study. They also acknowledge Prof. Elsa Grimaldi for the edition of the English paper. References [1] G. Centi, F. Trifirò, J. Ebner, V. Franchetti, Chem. Rev. 88 (1988) 55. [2] F. Ben Abdelouahab, R. Olier, M. Ziyad, J.C. Volta, J. Catal. 134 (1992) 151. [3] M. López Granados, J.L. Garc´ıa Fierro, F. Cavani, A. Colombo, F. Giuntoli, F. Trifirò, Catal. Today 40 (1998) 251. [4] G. Hutchings, Appl. Catal. 72 (1991) 1. [5] G. Hutchings, R. Higgins, J. Catal. 162 (1996) 153. [6] Y. Takita, K. Tanaka, S. Ichimaru, Y. Mizihara, Y. Abe, Y. Ishihara, Appl. Catal. A 103 (1993) 281. [7] G. Hutchings, C. Kelly, M.T. Sananés-Schulz, A. Burrows, J.C. Volta, Catal. Today 40 (1998) 273. [8] V.A. Zazhigalov, J. Haber, J. Stoch, A. Pyatnitzkaya, G.A. Komashko, V.M. Belousov, Appl. Catal. A 96 (1993) 135. [9] L. Cornaglia, C. Carrara, J. Petunchi, E. Lombardo, Appl. Catal. A 183 (1999) 177. [10] L. Cornaglia, C. Carrara, J. Petunchi, E. Lombardo, Catal. Today 57 (2000) 313. [11] L. Cornaglia, C. Sánchez, E. Lombardo, Appl. Catal. A 95 (1993) 117. [12] M. Rozynek, C. Polansky, Appl. Catal. 73 (1991) 97. [13] H. Ohtuska, T. Tabata, O. Okada, L. Sabatino, G. Bellusi, Catal. Lett. 44 (1997) 265. [14] G. Busca, F. Cavani, G. Centi, F. Trifirò, J. Catal. 99 (1986) 400. [15] L. Cornaglia, E.A. Lombardo, Appl. Catal. A 127 (1995) 125. [16] G.W. Coulston, E.A. Thompson, N. Herron, J. Catal. 163 (1996) 122. [17] A.S.K. Ainha, V. Shankar, Ind. Eng. Chem. Res. 32 (1993) 1061. [18] S.C. Petrosius, R.S. Drago, V. Young, G.C. Grunewald, J. Am. Chem. Soc. 115 (1993) 6131.
286
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
[19] G. Fierro, M. Lo Iacono, M. Inversi, R. Dragono, P. Porta, Topics Catal. 10 (2000) 39. [20] B. Jit San, K. Klabunde, P. Sherwood, J. Am. Chem. Soc. 113 (1991) 855. [21] D.C. Frost, C.A. McDowell, I. Woosley, Chem. Phys. Lett. 17 (1972) 320. [22] J. Stencel, V. Rao, J. Diehl, K. Rhee, A. Dhere, R. DeAngelis, J. Catal. 84 (1983) 109. [23] V. Lorenzelli, G. Busca, N. Sheppard, J. Catal. 66 (1980) 18. [24] K. Pursell, R. Drago, J. Am. Chem. Soc. 88 (1966) 919. [25] G. Busca, G. Centi, F. Trifirò, Appl. Catal. 25 (1986) 265. [26] L. Cornaglia, E. Lombardo, J. Anderson, J.L. Garc´ıa Fierro, Appl. Catal. A 100 (1993) 37.
[27] V. Guliants, J. Benziger, S. Sundaresan, I. Wachs, A. Hirte, Catal. Lett. 62 (1999) 87. [28] R.L. McCormick, G.O. Alptekin, A.M. Herring, T.R. Ohno, F. Dec Steven, J. Catal. 172 (1997) 160. [29] V. Zazhigalov, J. Haber, J. Stoch, I. Bacherikova, G. Komasko, A. Pyatnitskaya, Appl. Catal. 134 (1996) 225. [30] S. Irusta, A. Boix, B. Pierini, C. Caspani, J. Petunchi, J. Catal. 187 (1999) 298. [31] J.T. Gleaves, J. Ebner, T.C. Kuechler, Catal. Rev. Sci. Eng. 30 (1998) 49. [32] G. Koyano, T. Saito, M. Misono, J. Mol. Catal. A 155 (2000) 31.
Study of the Co-VPO interaction in promoted n-butane oxidation catalysts C. Carrara, S. Irusta, E. Lombardo, L. Cornaglia∗ Instituto de Investigaciones en Catálisis y Petroqu´ımica – INCAPE (FIQ, UNL-CONICET), Santiago del Estero 2829, 3000 Santa Fe, Argentina Received 26 December 2000; received in revised form 5 April 2001; accepted 7 April 2001
Abstract Cobalt-impregnated VPO catalysts were prepared using different cobalt salts and impregnation methods. The catalytic tests showed that cobalt impregnation significantly increased the overall activity. The use of cobalt acetyl acetonate led to a more selective high loading catalyst. To investigate the origin of the cobalt effect, the solids were characterized using XRD, Raman spectroscopy, FT-IR, and XPS. The surface acidity was probed by adsorbing acetonitrile. No structural effects were detected through XRD. After 700 h on stream, the only phase detected in all cases was V(IV) vanadyl pyrophosphate. The surface oxidation state of vanadium was V(IV). The Co 2p XP spectrum showed an intense shoulder at 788 eV, indicating that Co(II) species were present. The concentration of very strong Lewis acid sites increased at higher cobalt loading, but its strength was unaffected. © 2001 Elsevier Science B.V. All rights reserved. Keywords: VPO; n-Butane oxidation; Maleic anhydride; Promoters; Cobalt; Lewis acidity
1. Introduction The VPO system exhibits a unique ability to activate and selectively oxidize alkanes. Vanadyl pyrophosphate is commercially used to catalyze the selective oxidation of n-butane to maleic anhydride (MAN). It is generally agreed upon that the best catalyst precursor is VOHPO4 ·0.5H2 O which is converted to (VO)2 P2 O7 [1] during activation. Besides, several VOPO4 phases present in low proportion have been claimed as necessary for the catalytic act [2]. However, these phases have never been detected in the so-called equilibrated catalysts that have been on stream for 700 h [3].
∗ Corresponding author. Tel.: +54-342-4536861; fax: +54-342-4536861. E-mail address: [email protected] (L. Cornaglia).
A variety of cations have been added to vanadium phosphate catalysts to improve activity and selectivity. It has been claimed that the addition of the first row transition metal produces an improvement in the maleic anhydride yield [4,5]. Several papers [4–10] refer to the effect of cobalt used as a promoter. Takita et al. [6] sustained that the additives were incorporated into the crystal lattice of vanadyl pyrophosphate. Volta et al. [7] using in situ Raman spectroscopy found that the incorporation of Co can change the V(V)/V(IV) balance during the activation period and can therefore change the catalytic performance in the steady state. Other authors [5,7,8] studied the influence of cobalt addition on the composition of the (VO)2 P2 O7 catalyst. The presence of cobalt induced surface phosphorus enrichment, which modified the surface acidity. Zazhigalov et al. [8] proposed that cobalt stabilize the catalyst performance by
0926-860X/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 0 1 ) 0 0 6 1 5 - 9
276
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
forming cobalt phosphate, which improves its catalytic properties. According to Zazhigalov et al. [8], moderate surface acidity facilitates desorption of maleic anhydride and prevents its complete oxidation to carbon oxides. They suggested that promoters play an important role in controlling the surface acidity of promoted VPO solids. Selective oxidation of n-butane requires Lewis–Brönsted acid site in combination with a surface vanadium redox site. In a recent paper, we have studied impregnated Co-VPO catalysts obtained using cobalt acetate [10]. We concluded that on equilibrated catalysts, the addition of cobalt increases the activity while slightly decreasing the selectivity. This effect seems to be a consequence of increased exposure of cobalt on the catalyst surface. In this work, we have extended our studies using different salts and impregnation procedures. The rational was that cobalt acetyl acetonate could allow us to change the cobalt dispersion on the surface, while the high-temperature impregnation may help intercalate the Co atoms between the VPO layers. Besides, in search of additional clues to understand the role of Co on the surface of equilibrated solids, we have now probed the surface acidity using acetonitrile. 2. Experimental 2.1. Catalyst preparation The precursor was obtained by reduction of 5 g of V2 O5 with 30 ml of isobutanol and 20 ml of benzyl alcohol under reflux for 3 h. Then orthophosphoric acid (100%) was added in the desired amount and the solution refluxed for another 2 h. After completion of the reaction, the solid phase was recovered by filtration and dried in air at 390 K overnight. 2.1.1. Promoted catalyst The impregnation was carried out using either cobalt acetate or acetyl acetonate and following two procedures. 1. The impregnation method: the salt, acetate (VPCox-Ia) or acetyl acetonate of Co (VPCox-Iaa where x represents wt.% of Co) was dissolved in 30 ml of isobutanol. The previously prepared precursor was added to this solution and the
suspension was heated to 330 K with continuous stirring. The solvent was evaporated and the wet solid dried at 390 K. The final Co content of all of the preparations was determined by atomic absorption. 2. The high-temperature impregnation method: cobalt acetyl acetonate was used in this case (VPCox-HTI). The salt was dissolved in a mixture of 30 ml of isobutanol and 20 ml of benzyl alcohol, the precursor was added and refluxed overnight. The precipitate was then filtered and dried in air at 390 K. 2.2. Catalyst activation and testing The promoted catalyst precursors and unpromoted VPO were activated in a fixed bed microreactor. Typically, about 1 g of catalyst (screened 170–250 m range Tyler 60–80 mesh) was loaded into the reactor and covered with a layer of quartz wool. The quartz reactor was 60 cm long and 1.0 cm i.d. at the catalyst bed portion. It was mounted vertically in a tubular furnace. A programmable temperature controller was used to assure reproducibility of heating strategies. Mass flow controllers were used to feed the reactants in the right proportion. An on-line gas chromatograph equipped with a FID detector was used to analyze both the reactant and product streams. Separation of C4 H10 and maleic anhydride was accomplished with a 2 ft 1/8 in SS Chromosorb WAW AT-1200+1% H3 PO4 80/100 column. The activation of the catalyst was performed by heating from ambient temperature to 603 K, while flowing a mixture of 0.75% n-butane in air, with a gas hourly space velocity (GHSV) of 900 h−1 . The temperature was kept constant at 603 K for 3 h. Those catalysts which only underwent this treatment are here called nonequilibrated catalysts. Then, the n-butane concentration was increased to 1.5% while still maintaining the GHSV at 900 h−1 . The temperature was raised until either 80% conversion or the maximum allowable temperature of 703 K was reached. Then, the GHSV was raised to 2500 h−1 and the temperature was slowly increased with the limitation that conversion never exceeded 80%. The catalytic data reported here were measured over solids that did not show any change in activity and selectivity after at least 500 h on stream. At this point when the reaction temperature
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
was randomly varied up and down, reproducible values of activities and selectivities were obtained. These solids are called equilibrated catalysts. 2.3. Catalyst characterization 2.3.1. X-ray diffraction The measurements were made with a Shimadzu XD-D1 X-ray diffractometer, using nickel-filtered Cu K␣ radiation with a scanning rate of 1◦ min−1 . 2.3.2. Laser Raman spectroscopy (LRS) The Raman spectra were recorded with a Jasco laser Raman spectrometer model TRS-600-SZ-P, equipped with a CCD (charge coupled device) with the detector cooled to about 153 K using liquid N2 . The excitation source was the 514.5 nm line of a Spectra 9000 Photometrics Ar ion laser. The laser power was set at 30 MW. All the spectra were recorded with the samples under ambient conditions. The powdered solid was pressed into a thin wafer about 1 mm thick. 2.3.3. Infrared spectroscopy The IR spectra were obtained using a Shimadzu FT-IR 8101M spectrometer with a spectral resolution of 4 cm−1 . The solid samples were prepared in the form of pressed wafers (ca. 2 wt.% sample in KBr). The IR spectra of precursors were obtained using fluorolube as a diluting agent in order to explore the high wave number region. The samples for the adsorption experiments were prepared by compressing the used catalysts at 9 tonnes cm−2 in order to obtain a self-supporting wafer (50 mg, 10 cm diameter). They were mounted in a transportable infrared cell with CaF2 windows and external oven. The pretreatment was performed in a high-vacuum system. The sample was first outgassed at 723 K for 12 h in a dynamic vacuum of 7×10−4 Pa. After cooling to room temperature, a spectrum of the catalyst wafer was taken. No bands were observed in the 1600 cm−1 region, this is indicative that molecular water was eliminated. After that, 4.8 × 104 Pa of acetonitrile was admitted into the cell and left in contact with the solid for 10 min, and then a spectrum was recorded. No changes were observed in the 1600 cm−1 region, therefore no water was introduced during acetonitrile adsorption. Spectra were
277
also recorded after evacuation of the cell for 10 min at 298, 353, and 423 K. 2.3.4. X-ray photoelectron spectroscopy The XPS measurements were carried out using an ESCA750 Shimadzu electron spectrometer. Nonmonochromatic Al K␣ X-ray radiation was used. The anode was operated at 8 kV and 30 mA and the pressure in the analysis chamber was about 2 × 10−6 Pa. The data were collected using an ESCAPAC 760 computer interfaced to the spectrometer and analyzed with the Googly software developed at the University of Pittsburgh. The binding energies (BE) were referred to the C 1s signal (284.6 eV). Curve fitting was performed using a Levenberg–Marquardt NLLSCF routine. The background contribution was taken into account by assuming an integral type background which was included in the basic shape of each peak. Account has been taken of the presence of K␣3 and K␣4 spectral lines of the large O 1s signal. These satellite peaks are 9.6 and 11.6 eV downshifted from the O 1s main peak and overlap the V 2p1/2 signal. Since the resolution of the V 2p1/2 is poorer than that of the V 2p3/2 level, we preferably used the latter BEs and widths for comparison with literature data. The V 2p and Co 2p doublets were fitted using a Voigt function with 20% Lorentzian character and 2p1/2 /2p3/2 intensity ratio = 0.5. Additional data for the curve fitting of the V 2p doublets include spin-orbit separation = 7.2 eV. The surface P/V and Co/V atomic ratios were calculated using the areas under the Co 2p3/2 , P 2p, and V 2p3/2 peaks, the Scotfield photoionization cross-sections, the mean free paths of the electrons, and the instrumental function which was given by the ESCA manufacturer. 3. Results and discussion 3.1. Catalytic behavior Fig. 1A and B shows the catalytic results obtained with promoted and unpromoted solids. The impregnated solids exhibit a significantly higher conversion than unpromoted VPO in the whole temperature range, with the exception of the VPCo1.5-Ia. For the promoted catalysts, the selectivity was nearly constant up
278
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
(Fig. 2A and B). Note that at T > 663 K, the yield of VPCo4-Iaa is comparable to industrial catalysts. To investigate how cobalt affects the bulk and surface features of VPO solids, a set of techniques has been used. 3.2. Structure of VPO precursors and catalysts
Fig. 1. Catalytic behavior of equilibrated solids. Reaction conditions: 1.5% n-butane/air; GHSV = 900 h−1 .
to ca. 80% conversion. VPCo4-Iaa is the most selective solid assayed. A higher yield to maleic anhydride is reached with Co-promoted catalysts at GHSV = 900 h−1 . The same behavior is observed at high GHSV
All the FT-IR spectra of the precursors (diluted in fluorolube) show the characteristic fingerprints of the vanadyl hydrogen phosphate hemihydrate. There appears a band at 1640 cm−1 corresponding to the coordinated water, and a very marked shoulder at 1600 cm−1 , which could be due to hydrogen-bonded uncoordinated water. The bands at 1509, 1492, and 1450 cm−1 (C=C vibration in the plane of the aromatic ring) correspond to the benzyl alcohol and benzaldhyde retained in the solid [11]. In the solids prepared by impregnation with Co acetate, with a content ≥ 4%, there appear bands at 1564 and 1430 cm−1 corresponding to the salt, according to what has been reported [10]. The same happens with the solids impregnated with acetyl acetonate in which it is possible to observe signals at 1526 and 1430 cm−1 , characteristic of the salt. This indicates that even after drying, the cobalt salts are present in these precursors. In the catalysts prepared under reflux, the signal at 1526 cm−1 is not observed even for 4% Co. In order to investigate if this was due to the thermal treatment, an experiment was performed in which the solution was refluxed but in the absence of the solid. The solution thus treated presents bands at 1526 and 1430 cm−1 , characteristic of acetyl acetonate. The disappearance of the band at 1526 cm−1 in the precursor could be due to the interaction between the solid and the salt. After activating the solids, the salt bands are no longer observed and the presence of cobalt oxides is not detected by XRD. Rozynek and Polansky [12] studied the bulk and surface reduction properties of the silica-supported cobalt catalysts. The TPR profile of the calcined acetate (supported on silica) was not characteristic of Co3 O4 . Besides, no XRD peak characteristic of oxides could be observed for the calcined acetate, in agreement with our results in VPO catalysts. The Raman Co3 O4 pattern presents sharp bands at 691 and 524 cm−1 [13]. As these bands are not observed in the Raman spectra of the promoted
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
Fig. 2. Maleic anhydride yield of equilibrated catalysts: (A) GHSV = 900 h−1 ; (B) GHSV = 2500 h−1 . Reaction conditions: 1.5% n-butane/air.
279
catalyst, this is considered a confirmation of the XRD results. Zazhigalov et al. [8] observed the appearance of Co2 P2 O7 for the catalysts with Co/V > 0.15. In our case (Co/V = 0.13), neither XRD nor Raman spectroscopy indicate the presence of Co-containing phases. In Table 2, the XRD data for nonequilibrated and equilibrated catalysts are shown. In all cases, vanadyl pyrophosphate was the only crystalline phase detected. The decrease in the FWHMs of the (2 0 0) reflection with a long time-on-stream was observed. No structural modifications (Table 1) were observed on impregnated solids through XRD. The FT-IR spectra of both equilibrated and nonequilibrated solids show the fingerprints of vanadyl pyrophosphate between 700 and 1700 cm−1 , with the following assignments: ν(PO3 ) = 1240 cm−1 , ν(V=O) = 968 cm−1 , ν(V–O–V) = 797 cm−1 , and ν(P–O–P) = 742 cm−1 [14]. In the cases where Co has been introduced during the phosphatization step, it was observed that the P–O–P vibration was modified [9]. The ν(P–O–P) at 742 cm−1 has a similar intensity in the unpromoted and impregnated catalysts. This is substantiated by the high values (∼15 ± 1) of the 1245/742 intensity ratio (Table 1). It can be observed that the values obtained for the intensity ratio are lower in the case of the high-temperature Co-impregnated solids. This results from an increase in the band intensity corresponding to the P–O–P vibration. This increase is independent of both the Co content and the catalyst time-on-stream. This addition method of the Co promoter would also be introducing an increase in the number of P–O–P layer linkages. The same effect was observed when cobalt was added during phosphatization (VPCo4-P). The Raman spectra show that none of the equilibrated catalysts contain VOPO4 phases. This is at variance with the findings of Hutchings et al. [7]. They report that Co and Fe dopants can modify the VOPO4 /(VO)2 P2 O7 dispersion. They concluded that the activity for maleic anhydride formation is a function of the nature and concentration of the V(V) phases present. None of the three instrumental techniques allowed us to detect significant structural differences in the catalysts with and without promoter addition, which could have explained their different catalytic behavior. Let us see now if we can find “clues” on the surface to explain this phenomenon. Presence of V(V), Co oxidation state, P surface enrichment, and
280
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
Table 1 XRD and FT-IR data of precursors, nonequilibrated and equilibrated catalysts Solid
VPO VPCo1.5-Ia VPCo4-Ia VPCo4-Iaa VPCo2-HTI VPCo4-HTI VPCo4-Pe
Precursora
Nonequilibrated
FWHMb
FWHMc
(0 0 1)
1.0 1.0 1.14 1.2 1.4 1.4 0.9
Equilibrated
(2 0 0)
I1240 /I742
1.2 2.1 2.2 1.8 1.8 1.7 1.6
d
– – 16.3 14.2 12.8 12.6 9.9
FWHMc (2 0 0)
I1240 /I742
1.0 1.0 1.1 1.0 1.15 1.1 1.38
14.0 – 14.0 16.7 10.8 10.9 10.0
Bulk P/V ratio = 1.26. VOHPO4 ·0.5H2 O (2θ = 15.5◦ ); FWHM in 2θ . c (VO) P O (2θ = 23.2◦ ); FWHM in 2θ . 2 2 7 d Relative intensity of FT-IR bands: ν(PO ) = 1240 cm−1 and ν(P–O–P) = 742 cm−1 . 3 e From [9]; cobalt (as cobalt acetate) was added during phosphatization. a
b
variation of the surface acidity are the main variables to be explored.
Table 3 Surface vanadium oxidation states of the equilibrated catalysts Solids
3.3. Surface of equilibrated catalysts
V 2p3/2 a,b
∆[O 1s − V 2p3/2 ] (eV) Vox c
BE (eV) FWHM VPO VPCo4-Ia VPCo4-Iaa VPCo4-HTI
In our catalysts, cobalt impregnation produces an increase of the precursor surface P/V ratio (Table 2). This ratio increases upon equilibration in the unpromoted VPO, whereas it decreases in the Co-containing catalysts. This effect could be due to a phosphorus diffusion into the bulk with time-on-stream. The phosphorus surface enrichment in Co-promoted catalysts was previously reported by other authors [7–10]. The surface oxidation states of vanadium were calculated from the difference in BE between the O 1s and V 2p3/2 signals. This method avoids the use of a reference as we concluded in a previous study [15]. The results are shown in Table 3. This value is equal
517.9 517.8 517.7 517.9
2.3 2.2 2.3 2.2
14.3 14.3 14.5 14.3
4.09 4.09 3.96 4.09
a
BEs determined by curve fitting. O 1s BE = 532.2 eV was used as reference. c Using the following correlation: V ox = 13.82 − 0.68 (∆[O 1s − V 2p3/2 ]), from Coulston et al. [16]. b
to 14.3 ± 0.1 eV for promoted and unpromoted catalysts. Coulston et al. [16] have reported a correlationship between the ∆ value and the average vanadium oxidation state. These calculated values are around 4 in all the cases studied. The curve fitting leads to
Table 2 XPS surface atomic ratios of precursors, nonequilibrated and equilibrated catalystsa Solids
VPO VPCo1.5-Ia VPCo4-Ia VPCo4-Iaa VPCo2-HTI VPCo4-HTI a
Precursors
Nonequilibrated
Equilibrated
P/V
Co/V
P/V
Co/V
P/V
Co/V
2.15 3.80 3.70 3.40 3.50 3.70
– 0.36 0.60 0.70 0.28 0.69
2.9 3.8 3.0 3.9 – –
– – 0.30 0.85 – –
2.8 2.6 2.6 2.5 2.6 2.5
– 0.31 0.27 0.28 0.26 0.30
Bulk P/V ratio = 1.26.
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
281
Table 4 XPS binding energies for cobalt in equilibrated catalystsa Solids
Co 2p3/2 (eV)
Satellite (eV)
Isat /IMP b
Co 2p1/2 (eV)
Satellite (eV)
Isat /IMP c
∆[Co 2p1/2 − Co 2p3/2 ]
CoO
780.5d 780.1e
786.9 788
0.9 –
796.3 795.3
803 804.6
0.8 –
15.8 15.2
Co3 O4
779.5f 779.6e
788 787.1
– –
794.6 794.8
803 804.1
0.2 –
15.1 15.2
Co(OH)2 VPCo4-Ia VPCo4-Iaa VPCo4-HTI
781.0 782.7 (3.0) 783.2 (3.3) 783.3 (3.3)
786.4 787.4 (7.3) 787.4 (5.4) 788 (5.0)
– 1.4 0.8 0.7
796.9 798.4 (3.1) 798.8 (3.2) 799.0 (3.4)
802.6 802.6 (7.0) 803.5 (5.8) 804.2 (6.0)
– 1.5 1.0 1.1
15.9 15.7 15.6 15.7
a
FWHM values (eV) are given in parenthesis. Intensity ratio Co 2p3/2 satellite/main peak. c Intensity ratio Co 2p 1/2 satellite/main peak. d [17]. e [20]. f [18]. b
a single, well-defined binding energy for V 2p3/2 (517.7 ± 0.1 eV) assigned to V(IV) in agreement with the values reported by López Granados et al. [3] for unpromoted catalysts. Both methods support the overwhelming presence of V(IV) on the surface of the VPO equilibrated catalysts. The addition of cobalt does not change the oxidation states of surface vanadium. The existence of Co(II) in the impregnated catalyst has been confirmed by XPS measurements (Table 4 and Fig. 3). The binding energies of reference compounds and high Co loading equilibrated catalysts are summarized in Table 4. While the 2p1/2 and 2p3/2 binding energies of Co(II) are relatively close to that of Co(III), the two oxidation states of cobalt can be distinguished by the presence of a distinct shake up satellite structure in Co(II) arising from the presence of unpaired electrons in their valence orbital [17,18]. Co(III) is almost always in a low-spin state and therefore there can be no energy transfer to an unpaired electron, meaning that a shake up satellite structure is not observed. The disappearance of the satellite peaks in the XPS profile would then be indicative of oxidation of Co(II) to Co(III). The Co3 O4 has a spinel structure and contains magnetic Co(II) ions (tetrahedral sites) and diamagnetic Co(III) ions (octahedral sites). These two sites are indistinguishable by XPS. Although, all compounds containing high-spin Co(II) ions have strong satellite peaks in the Co 2p spectra, only Co3 O4 has weak
satellites in this region [20]. These satellites occur about 7–9 eV higher in binding energy from the main bands, which can be observed in Table 4. High-spin Co(II) core level spectra (as in CoO) show strong satellite peaks (Table 4) above the main 2p lines. Frost et al. [21] have shown that high-spin Co(II) compounds exhibit intense satellites, while low-spin Co(II) compounds exhibit weak or missing satellites
Fig. 3. XPS spectra of the Co 2p region.
282
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
in their photoelectron spectra. Thus, it is not difficult to differentiate high-spin Co(II) compounds from Co2 O3 . Cobalt hydroxide has a strong satellite peak associated with both the lines of the Co 2p region. For the VPCo4-Iaa and VPCo4-Ia catalysts, the Co 2p3/2 binding energies were different. The value of 783.3 eV indicates that the presence of CoO on the surface is unlikely. The Co 2p3/2 BE is 780.1 eV for CoO. For Co-promoted vanadyl pyrophosphate, the binding energy is much higher. The Co 2p spectrum has an intense shoulder at ∼788 eV. In addition, the 2p1/2 –2p3/2 spin-orbit splitting of 15.7 eV for this catalyst indicates that Co(II) species are present. Fierro et al. [19] have proposed an interpretation for cobalt, similar to that reported for copper. For Co(II)-containing compounds, the main and the satellite peaks can be related, respectively, to 2p5 3d8 L (L means a ligand hole) and 2p5 3d7 electronic configuration in the final state. One of the most important consequences of this theory is that useful information about the nature of the metal–ligand chemical bonding can be simply gained by evaluating the satellite/main peak intensity ratio and their energy separation. They found that the energy separation increases and the satellite/main peak intensity ratio decreases with increasing covalent character of the metal–ligand bonding. On examining the XPS profile of the impregnated catalysts, the Co 2p satellite intensities approach satellite/main peak intensity ratios observed for bulk CoO, while the strong satellite lines are centered about 4.7–4.2 eV above the principal line. These observations indicate that Co(II) is not being oxidized to Co(III). Besides, the covalent character of metal–ligand bonding is not affected neither by the method nor the salt used for impregnation. The high BEs (783.3–782.7 eV) might be related to strong interactions between cobalt and other atoms of the solid matrix and/or to a highly dispersed cobalt. Binding energies of 783 eV were reported [22] for the ion exchanged form of cobalt in zeolites. In these solids, the higher than normal binding energy is suggestive of cobalt in a highly oxidizing environment. Zazhigalov et al. [8] reported a similarly high binding energy of 783.0 eV for Co-promoted VPO catalysts. They compared their results with the value of 782.8 found in cobalt molybdate or 782.2 and 783.0 for CoCl2 and CoF2 , respectively, and concluded that the metal–ligand bonds are polarized in these reference
compounds and a similar structure may be expected for cobalt phosphates. They sustained that the high binding energy values of 782.7–783 eV observed in their catalysts can be considered as an additional indication of the formation of cobalt phosphate (detected by XRD). On the other hand, they suggested the existence of strong interactions between cobalt atoms and phosphate groups with a shift of bonding electrons towards the phosphate anions. In our catalysts, the Co/V nominal ratio is lower than 0.13, and no cobalt phosphates were detected by XRD, FT-IR, and Raman spectroscopy. In agreement with our results, Hutchings and Higgins [5] have recently reported that the XRD patterns do not even hint the formation of solid solutions at low promoter ratios, M/V ≤ 0.12. Hutchings [4] proposes that the promoters at high promoter/vanadium atomic ratios act as phosphorus scavengers by either formation of metal phosphates or by formation of solid solutions. In our case, considering the high P/V surface ratio and the fact that all the catalysts have similar Co/V surface ratios, it is possible that the high BE of Co 2p3/2 be symptomatic of the presence of surface cobalt pyrophosphate. This surface compound does not have long-range, three-dimensional order, because characteristic X-ray diffraction lines have not been observed. However, binding energies are not a sufficient criterion to confirm the presence of a given compound. A difference in BE was observed between the two impregnated catalysts. This could be due to either different surface Co(II) interactions or to a change of the cation distribution at the surface. Two different impregnating salts (cobalt acetate and acetyl acetonate) might produce catalysts with varying strengths in the interaction between cobalt and VPO or different Co dispersion. 3.4. Lewis acidity of equilibrated solids To ascertain the effect of Co upon the Lewis acidity of the VPO formulations, acetonitrile was used as a probe molecule. When this weak base is adsorbed on the VPO solids, the following bands are observed: ν = 2252 cm−1 assigned to H-bonded and physisorbed molecules, ca. ν = 2270 cm−1 corresponding to CN groups interacting with medium strong Lewis acid sites, ν = 2294 cm−1 associated with the Fermi resonance of the CN group (ν3 + ν4 ),
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
and a fourth band at 2322 cm−1 assigned to the interaction of the base with very strong Lewis acid centers. A shift of 28 cm−1 is observed when this base is adsorbed on ␣-Fe2 O3 [23]. Displacements of ca. 75 cm−1 from the liquid phase frequency are observed when acetonitrile is adsorbed on AlCl3 [24]. The band at 2252 cm−1 rapidly decreases with increasing temperature in all the equilibrated catalysts. A band at 2270 cm−1 ( ν = 18 cm−1 ) appears in the VPO spectrum taken in the presence of the gas phase, and remains when the evacuation temperature is increased. This band is also observed in the VPCo4-Iaa and in VPCo4-HTI in the presence of the gas phase. But this band almost disappears upon temperature increase. The analysis of the spectra obtained with the solids that were fast-activated (nonequilibrated), shows less adsorption in general. However, it is also seen that all the spectra show the signal assigned to the medium strong sites (2270 cm−1 ) when there is acetonitrile in the gas phase. But these signals disappear after evacuation at 298 K. It could be concluded that medium strong Lewis acid sites are present in all nonequilibrated catalysts; when the catalysts become equilibrated, these sites remain only in the unpromoted catalyst and in the catalyst with low percentage of Co (1.5%). The band at 2322 cm−1 , attributed to very strong Lewis acid sites, appears practically at the same frequency in all the solids (Fig. 4). It could accordingly be concluded that the addition of Co in different amounts and by different methods does not introduce any modification in the strength of the very strong Lewis acid sites. Table 5 also shows the intensity ratio of the band at 2322 cm−1 with respect to that at 2297 cm−1 . This Table 5 Acetonitrile adsorption on equilibrated catalysts Catalyst
VPO VPCo1.5-Ia VPCo4-Ia VPCo4-Iaa VPCo4-HTI
Lewis site (cm−1 )
I2322 /I2297 a
Very strong
Medium strong
C/Fb
423 Kc
2322 2322 2322 2322 2322
2270 2270 – – 2270
0.56 0.80 0.60 0.69 0.53
0.48 0.53 1.89 1.14 1.75
Relative intensity of FT-IR bands at ν = 2322 and 2297 cm−1 . Acetonitrile in the gas phase. c After evacuation at 423 K. a
b
283
Fig. 4. FT-IR spectra of acetonitrile adsorption on promoted catalysts after evacuation at 423 K.
band is not influenced by the electronic environment [26], and it does not change with the catalyst acidity. Consequently, this ratio could be used to estimate the very strong Lewis acid site concentration. After evacuation at 423 K, the three catalysts with 4% Co contain a higher concentration of very strong Lewis acid sites than those with 1.5% Co and the unpromoted VPO. According to Busca et al. [25], the very strong Lewis acid sites of the nonequilibrated organically prepared VPO are due to a stretching of the V–O–P linkage. This stretching is a consequence of the stacking fold disorder of the (2 0 0) planes. In our previous study [26], we have demonstrated that the ratio of very strong to medium strong Lewis sites significantly increases with time-on-stream, corresponding to a well-ordered structure of the equilibrated catalysts. 3.5. Role of cobalt The Co addition significantly increases the overall activity. This seems to be a consequence of the presence of cobalt on the equilibrated catalyst surface. Using cobalt acetyl acetonate for the precursor
284
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
impregnation produces a solid with higher selectivities and performances comparable to commercial catalysts. In this complex structure, the promoter might be incorporated in the crystal lattice of (VO)2 P2 O7 , as reported by Takita et al. [6]. They studied the incorporation of promoter elements employing metal acetyl acetonates added during precursor synthesis. Incorporation of the promoter elements into the crystal lattice of (VO)2 P2 O7 brought about the shift in the IR absorption bands of the V=O stretching mode. In our catalysts, the V=O vibration was not modified by the presence of cobalt. They found the same behavior in the case in which Zn was located at the surface of vanadyl pyrophosphate. Guliants et al. [27] found a lower Nb surface ratio of promoter, which probably indicates higher solubility of the Nb promoter in the VPO matrix. Therefore, this suggested that promoter elements may partially form a solid solution. Another promoting effect could be that the cobalt atoms affect the layer linkages by becoming intercalated between the layers of the (VO)2 P2 O7 , as was reported for Fe- and Cr-promoted catalysts [28]. However, the IR band at 742 cm−1 , corresponding to the P–O–P vibration, has not significantly decreased in intensity in our Co-promoted catalysts. Neither a volumetric cobalt phosphate phase nor solid solutions were detected in our catalysts. The solids showed, however, high Co/V and P/V surface ratios plus a significant shift of the Co 2p BE. All this is consistent with the presence of a surface cobalt pyrophosphate in our catalysts. Several authors [25–27,29,30] have recalled the importance of the surface acidity for high activity and selectivity of the bulk VPO. The presence of acid sites should have an effect on the rate of activation of butane and on the rate of its oxidation to maleic anhydride. The acidity of the Co-promoted VPO catalyst was measured with NH3 TPD by Zazhigalov et al. [8]. It has been observed that adsorption of ammonia increases with increasing cobalt content. The ratio of the number of strong to weak adsorption sites decreased with increasing cobalt content. They concluded that the introduction of cobalt ions into the VPO catalyst is the main cause for the appearance of weak acid sites, whereas the number of strong acid sites rises to a lesser extent and only on addition of small amounts of cobalt (Co/V < 0.1 or 0.05 depending on the preparation method).
Busca et al. [25] proposed a model of butane activation in which it proceeds as a concerted simultaneous abstraction of two hydrogen atoms on an active center. This is composed of a strong Lewis acid–base V–O pair in which the empty d orbital of vanadium generates the Lewis acidity, while the occupied 2p orbital of oxygen is responsible for the Lewis basicity. The equilibrated catalysts containing 4% Co show a significant increase in the concentration of very strong Lewis acid sites over both the base VPO and the one containing 1.5% Co. A model of Lewis acid sites in vanadyl pyrophosphate was proposed by Trifirò et al. [1]. They indicated that the presence of defects in the structure of vanadyl pyrophosphate is reflected on the surface in an enhancement of Lewis acidity of surface unsaturated vanadium ions. According to this model, the appearance of very strong Lewis acid sites may be related to a different polarization of the V–O–P bonds, in our case induced by the presence of cobalt. The equilibrated VPCo4-Iaa solid is the most selective one. It has, however, the lowest concentration of very strong Lewis sites among the impregnated solids
Fig. 5. Maleic anhydride selectivities as a function of very strong Lewis acid sites concentration (data from Fig. 1B and Table 5).
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
with 4% Co. The correlation between selectivity and concentration of very strong Lewis sites at different conversions and space velocities is shown in Fig. 5. The presence of a maximum in selectivity might indicate that an optimum value exists for the concentration of very strong Lewis sites. An excess of very strong Lewis acid sites could decrease the rate of desorption of MAN, decreasing the maleic anhydride selectivity. Gleaves et al. [31] observed that the rate of desorption of the olefinic intermediates and furan is low compared to MAN, which implies that MAN should be rapidly desorbed to avoid its combustion. This would justify the former assertion and might indicate that the presence of very strong Lewis sites plays an important role regarding selectivity control. IR observations by Koyano et al. [32] of CO adsorbed al low temperature indicated that n-butane preferably interacts at Lewis acid sites, and recalled the importance of these sites to control selectivity. Note, however, that there could be other factors controlling the selectivity such as microstructure, P/V ratio, or a defect-free structure.
4. Conclusions They only apply to equilibrated catalysts. 1. As reported before [10], the addition of cobalt (≥2 wt.%) to VPO formulations significantly increases the catalytic activity (Fig. 1A). 2. When cobalt acetyl acetonate was used, the highest selectivities to maleic anhydride were observed (Fig. 1B). Besides, these Co-VPO solids give the highest yields (Fig. 2A and B). This ranking is also maintained at T > 660 K if all the previously studied cobalt-containing catalysts were included in Fig. 2 [9,10]. 3. The XPS data is consistent with the presence of Co-pyrophosphate in the first few surface layers (Table 4). However, bulk techniques (XRD, LRS, and FT-IR) could not detect the presence of cobalt phosphates. 4. The presence of cobalt does not modify the oxidation state of vanadium at the surface layers. Only V(IV) is detected through XPS (Table 3), in agreement with our previous studies [9,10]. 5. The presence of Co at the 4% level sharply increases the concentration of very strong Lewis
285
acid sites, although it does not affect their strength (Table 5). However, the best catalyst shows the lowest concentration among the catalysts containing 4% cobalt. This might indicate that an optimum concentration of very strong Lewis acid sites is needed to maximize the selective oxidation to maleic anhydride (Fig. 5). More work is needed to elucidate this point.
Acknowledgements The authors wish to acknowledge the financial support received from UNL (CAI+D 96 Program). They are also grateful to the Japan International Cooperation Agency (JICA) for the donation of the major instruments used in this study. They also acknowledge Prof. Elsa Grimaldi for the edition of the English paper. References [1] G. Centi, F. Trifirò, J. Ebner, V. Franchetti, Chem. Rev. 88 (1988) 55. [2] F. Ben Abdelouahab, R. Olier, M. Ziyad, J.C. Volta, J. Catal. 134 (1992) 151. [3] M. López Granados, J.L. Garc´ıa Fierro, F. Cavani, A. Colombo, F. Giuntoli, F. Trifirò, Catal. Today 40 (1998) 251. [4] G. Hutchings, Appl. Catal. 72 (1991) 1. [5] G. Hutchings, R. Higgins, J. Catal. 162 (1996) 153. [6] Y. Takita, K. Tanaka, S. Ichimaru, Y. Mizihara, Y. Abe, Y. Ishihara, Appl. Catal. A 103 (1993) 281. [7] G. Hutchings, C. Kelly, M.T. Sananés-Schulz, A. Burrows, J.C. Volta, Catal. Today 40 (1998) 273. [8] V.A. Zazhigalov, J. Haber, J. Stoch, A. Pyatnitzkaya, G.A. Komashko, V.M. Belousov, Appl. Catal. A 96 (1993) 135. [9] L. Cornaglia, C. Carrara, J. Petunchi, E. Lombardo, Appl. Catal. A 183 (1999) 177. [10] L. Cornaglia, C. Carrara, J. Petunchi, E. Lombardo, Catal. Today 57 (2000) 313. [11] L. Cornaglia, C. Sánchez, E. Lombardo, Appl. Catal. A 95 (1993) 117. [12] M. Rozynek, C. Polansky, Appl. Catal. 73 (1991) 97. [13] H. Ohtuska, T. Tabata, O. Okada, L. Sabatino, G. Bellusi, Catal. Lett. 44 (1997) 265. [14] G. Busca, F. Cavani, G. Centi, F. Trifirò, J. Catal. 99 (1986) 400. [15] L. Cornaglia, E.A. Lombardo, Appl. Catal. A 127 (1995) 125. [16] G.W. Coulston, E.A. Thompson, N. Herron, J. Catal. 163 (1996) 122. [17] A.S.K. Ainha, V. Shankar, Ind. Eng. Chem. Res. 32 (1993) 1061. [18] S.C. Petrosius, R.S. Drago, V. Young, G.C. Grunewald, J. Am. Chem. Soc. 115 (1993) 6131.
286
C. Carrara et al. / Applied Catalysis A: General 217 (2001) 275–286
[19] G. Fierro, M. Lo Iacono, M. Inversi, R. Dragono, P. Porta, Topics Catal. 10 (2000) 39. [20] B. Jit San, K. Klabunde, P. Sherwood, J. Am. Chem. Soc. 113 (1991) 855. [21] D.C. Frost, C.A. McDowell, I. Woosley, Chem. Phys. Lett. 17 (1972) 320. [22] J. Stencel, V. Rao, J. Diehl, K. Rhee, A. Dhere, R. DeAngelis, J. Catal. 84 (1983) 109. [23] V. Lorenzelli, G. Busca, N. Sheppard, J. Catal. 66 (1980) 18. [24] K. Pursell, R. Drago, J. Am. Chem. Soc. 88 (1966) 919. [25] G. Busca, G. Centi, F. Trifirò, Appl. Catal. 25 (1986) 265. [26] L. Cornaglia, E. Lombardo, J. Anderson, J.L. Garc´ıa Fierro, Appl. Catal. A 100 (1993) 37.
[27] V. Guliants, J. Benziger, S. Sundaresan, I. Wachs, A. Hirte, Catal. Lett. 62 (1999) 87. [28] R.L. McCormick, G.O. Alptekin, A.M. Herring, T.R. Ohno, F. Dec Steven, J. Catal. 172 (1997) 160. [29] V. Zazhigalov, J. Haber, J. Stoch, I. Bacherikova, G. Komasko, A. Pyatnitskaya, Appl. Catal. 134 (1996) 225. [30] S. Irusta, A. Boix, B. Pierini, C. Caspani, J. Petunchi, J. Catal. 187 (1999) 298. [31] J.T. Gleaves, J. Ebner, T.C. Kuechler, Catal. Rev. Sci. Eng. 30 (1998) 49. [32] G. Koyano, T. Saito, M. Misono, J. Mol. Catal. A 155 (2000) 31.