Structure–activity relationships for Co- and Fe-promoted vanadium phosphorus oxide catalysts

Abstract

The effect of Co and Fe doping on vanadium phosphate catalysts, prepared by the reaction of V₂O₅ and H₃PO₄ with isobutanol, for the oxidation of n-butane to maleic anhydride is described and discussed. At low levels, both Co and Fe dopants significantly enhance the selectivity and the intrinsic activity to maleic anhydride. A combination of powder X-ray diffraction, ³¹P NMR spin-echo mapping spectroscopy and transmission electron microscopy, together with catalyst test data, is utilised to analyse the origin of the effects of Co doping. Co appears to be essentially insoluble in crystalline (VO)₂P₂O₇ and is preferentially distributed in and stabilises an amorphous VPO material. It is suggested that the origin of the promotional effect of Co is associated with its interaction with the disordered VPO phase. The same techniques have been used to analyse the Fe-doped catalyst, but at present it is not possible to be definitive concerning the specific location of the Fe-dopant within the phases present. Previous studies have indicated that Fe can form a solid solution within (VO)₂P₂O₇ and therefore it is probable that the Fe may act as an electronic promoter for this phase. The role of Co, however, emphasises the importance of amorphous vanadium phosphate phases in the catalyst system.

---

Vanadium phosphates are well established as important industrial catalysts for the conversion of n-butane to maleic anhydride.^1,2 The catalysts are produced by activating a catalyst precursor compound, VOHPO₄·0.5H₂O, termed the hemihydrate, in the n-butane–air reaction mixture in situ in the reactor. The resultant catalyst often consists of a complex mixture of vanadium(IV) and (V) phosphates^1–3 [i.e. (VO)₂P₂O₇; VO(PO₃)₂; α_I-, α_II-, δ-, γ-VOPO₄]. Detailed in situ laser Raman spectroscopy⁴ has been used to study this process and, together with ex situ transmission electron microscopy and ³¹P NMR spectroscopy,⁵ has provided a detailed analysis of the evolution of catalysts in the absence of promoter compounds. Two topotactic transformation processes have been observed,^5,6 namely, the direct transformation of VOHPO₄·0.5H₂O to (VO)₂P₂O₇ and the indirect transformation of VOHPO₄·0.5H₂O to δ-VOPO₄, which subsequently transforms to (VO)₂P₂O₇. However, the ³¹P NMR spectroscopy shows that, in addition to crystalline (VO)₂P₂O₇ , a disordered V⁴⁺–V⁵⁺ phase is present in the activated catalyst. It is generally considered that the oxidation of n-butane to maleic anhydride, a 14-electron process, requires the close proximity of V⁵⁺ and V⁴⁺ cations, but the precise nature of the active site is still a matter of much speculation.^1,7 Recently, Coulston et al.⁸ have shown that the presence of V⁵⁺ is essential for the formation of maleic anhydride, thereby confirming earlier pulse reaction experiments.⁹

Vanadium phosphate catalysts used industrially typically contain additives² (Mo, Co, Fe) to promote the catalytic performance by improving the yield of maleic anhydride and the conversion of n-butane. To date, most structural and mechanistic studies of vanadium phosphate catalysts have been restricted to non-promoted formulations. In this paper, we address the structure of the catalyst precursor and activated catalysts for Co- and Fe-doped vanadium phosphate catalysts using a combination of transmission electron microscopy, powder X-ray diffraction and ³¹P NMR spectroscopy since, as we have shown,⁵ this combination of techniques is very potent in characterising VPO catalysts. We have previously shown¹⁰ that addition of low levels of Co and Fe as dopants can influence the balance of V⁵⁺–V⁴⁺ in the catalyst. In this paper, we further explore the effect of Co and Fe promoters on both catalyst performance and micro-structure, in particular to determine whether these promoters can be associated with a specific VPO phase.

Experimental

Catalyst preparation and testing

The undoped precursor was prepared by adding V₂O₅ (11.8 g) to isobutanol (250 ml). H₃PO₄ (16.49 g, 85%) was then introduced and the mixture was refluxed for 16 h. Doped catalysts (in the range 1 to 5 at% dopant) were prepared by adding the required amount of cobalt or iron acetyl acetonate to the initial reaction mixture. After refluxing, the solid was recovered by filtration and washed with isobutanol (200 ml) and ethanol (150 ml, 100%). The resulting solid was refluxed in water (9 ml g⁻¹ solid), filtered hot and dried in air (110 °C,16 h).

The catalyst precursor was loaded in a fixed bed laboratory microreactor⁵ and activated by heating to 430 °C in the reaction mixture (n-butane–oxygen–helium, 1.6 ∶ 1.8 ∶ 80.4) at a volume hourly space velocity of 1000 ml gas (ml catalyst)⁻¹ h⁻¹ and stabilised prior to testing under a variety of experimental conditions. The catalyst performance was monitored using on-line gas chromatography.

Catalyst characterisation

Powder X-ray diffraction was carried out using a Siemens D 500 diffractometer operating with a Cu-Kα source. BET surface area measurements were carried out using nitrogen adsorption at liquid nitrogen temperatures.

Samples suitable for transmission electron microscopy analysis were prepared by dispersing the catalyst powder onto a lacey carbon film supported on a copper mesh grid. TEM and HREM observations were made in a JEOL 2000EX high-resolution electron microscope operating at 200 kV. This instrument had been fitted with a low-light-level TV camera and a frame-averaging system to allow us to use very low illumination conditions. This latter condition is essential for studying these beam-sensitive vanadium phosphorus oxide compounds. Energy dispersive X-ray (EDX) analysis characterisation was carried out in a JEOL 2000 FR TEM equipped with a LINK system EDS spectrometer.

The ³¹P NMR measurements were performed on a Bruker MSL 300 NMR spectrometer. The ³¹P spin-echo mapping method has been shown by Li et al.¹¹ to be a very powerful technique for evaluating the relative proportions of V⁵⁺ and V⁴⁺ ions surrounding the P atoms in vanadium phosphate compounds. The ³¹P NMR spin-echo spectra were recorded under static conditions, using a 90°x–τ–180°y–τ acquire sequence. The 90° pulse duration was 4.2 μs and τ was 20 μs. For each sample, the irradiation frequency was varied in increments of 100 kHz above and below the ³¹P resonance of H₃PO₄. The number of spectra recorded was dictated by the frequency limits beyond which no spectral intensity was detectable. The ³¹P NMR spin-echo mapping information was then obtained by superposition of all the spectra.

Results and discussion

Characterisation of the doped hemihydrate phase

The powder X-ray diffraction patterns of the fresh hemihydrate precursors are shown in Fig. 1. All of the reflections can be indexed to the hemihydrate phase, VOHPO₄·0.5H₂O for the doped and undoped materials. In the case of the Co-doped samples, the addition of Co leads to a slight variation in the [001]^hemi/[220]^hemi relative intensity ratio (undoped VPO 2.6; VPO Co1, 1.9; VPO Co5, 1.3). The effect is much more pronounced for the Fe-doped precursors with the [001]^hemi/[220]^hemi relative intensity values being 0.84 and 0.42 for the 1 and 5 at% Fe-doped samples, respectively. Furthermore, for the 5 at% Fe-doped sample, there is a significant broadening of the [001]^hemi reflection when compared with the undoped hemihydrate. Indeed, the powder X-ray diffraction pattern of the 5 at% Fe-doped hemihydrate is similar to that for a hemihydrate prepared by reaction of VOPO₄·2H₂O with a primary alcohol.¹² It is possible that the Fe dopant may be playing a similar role to alcohols, which can be occluded within the hemihydrate layer structure. For the 5 at% Co- and 5 at% Fe-doped hemihydrates, an additional reflection at d  = 6.4 and 6.6 Å, respectively, is observed. This cannot be indexed to any known VPO compound and may be due to mixed M–VPO hydrated phases, where M = Co or Fe. For the 5 at% Fe-doped hemihydrate, there is an additional weak reflection at d  = 7.4 Å, corresponding to the [001] reflection of VOPO₄·2H₂O. This indicates that a small amount of oxidation has occurred due to Fe-doping, and this effect was also noted in an earlier study of Fe-doped hemihydrate prepared using aqueous HCl as reducing agent.¹³

Fig. 1 Powder X-ray diffraction patterns of hemihydrates: (a) VPO, (b) VPO Co1, (c) VPO Co5, (d) VPO Fe1, (e) VPO Fe5.

The ³¹P NMR spin-echo mapping spectra of the doped hemihydrates are shown in Fig. 2. All show a signal at 1625 ppm, which is assigned to P atoms coordinated by V⁴⁺ cations in the hemihydrate.¹⁴ A weaker signal at 0 ppm is also observed, which is attributed to P atoms coordinated to V⁵⁺ cations. This also indicates the presence of a VOPO₄ phase in the precursor.

Fig. 2 ³¹P NMR spin-echo mapping spectra of hemihydrates: (a) VPO, (b) VPO Co1, (c) VPO Co5, (d) VPO Fe1, (e) VPO Fe5.

Electron microscopy of the hemihydrate precursors showed that all the doped samples have a similar characteristic rhomboidal plate-like morphology. For the undoped hemihydrate, the dimensions of the rhomboid platelet were typically about 2 μm × 1 μm with a platelet thickness of typically 0.03–0.1 μm. Doping with Co and Fe did not significantly change the morphology (Fig. 3) except that the platelet dimensions increased to typically 3 μm × 2 μm. Selected area diffraction patterns indicated that the major and minor axes of the rhomboid correspond to the [100] and [010] directions of the VOHPO₄·0.5H₂O crystal structure, respectively. Chemical analysis using energy dispersive X-ray analysis indicated that the Fe and Co were homogeneously dispersed throughout the hemihydrate crystal. However, in the case of the 5 at% Co-doped sample, some regions of high Co concentration were noted and these are considered to be due to the formation of cobalt phosphate as a separate phase, which was confirmed from selected area diffraction patterns of selected crystals.

Fig. 3 Bright field transmission electron micrograph and selected area diffraction pattern for VPO Co1 hemihydrates.

Catalytic performance measurements

Catalytic performance data following activation in situ in the reactor in flowing n-butane–oxygen–helium catalysts are shown in Table 1. Addition of Co at the 1 at% and 5 at% levels increased the selectivity to maleic anhydride at high n-butane conversions and also increases the intrinsic activity. A similar effect is noted for the addition of Fe. The enhancement in intrinsic activity, which is observed at both 400 and 430 °C, as well as at two reactant flow rates, can be considered a general feature for the addition of Fe and Co to this vanadium phosphate catalyst. The effects are most pronounced at the 1% doping level and increasing the level of the dopant decreases the promotion effect. This behaviour is similar to that observed in previous studies of promoter concentration on catalyst performance for catalysts prepared using aqueous HCl as a reducing agent.² Of the two promoters investigated, Co gives the greatest enhancement in intrinsic activity. Of particular importance is the observation that the addition of Co and Fe increases the selectivity to maleic anhydride at moderate butane conversions (i.e. S₂₅^MA, Table 1). This effect is significant and contributes to the promotional effect.

Table 1 Catalyst performance data^a

Surface area/m^2 g^−1		Intrinsic activity^e /10^−5 mol MA m^−2 h^−1
Catalyst	Precursor	Activated catalyst	S 0 ^MA ^b (%)	S 25 ^MA ^c (%)	C ^but ^d (%)	430 °C 1000 h^−1	400 °C 1000 h^−1	400 °C 2000 h^−1
a Following activation in C4H10–O2–He = 1.6 ∶ 18 ∶ 80.4 at 1000 ml gas (ml catalyst)^−1 h for ca. 72 h. b Selectivity to maleic anhydride extrapolated to zero n-butane conversion at 400 °C. c Selectivity to maleic anhydride at 25% n-butane conversion at 400 °C. d n-Butane conversion at 400 °C and 2000 ml gas (ml catalyst)^−1 h^−1. e Intrinsic activity is defined as mol maleic anhydride synthesised per unit area of catalyst per unit time. Data are given for three sets of reaction conditions: reaction temperature: 430 or 400 °C; gas hourly space velocity: 1000 or 2000 h^−1.
VPO	11	13	80.0	50.3	16	1.94	1.20	0.98
VPO Col	9	16	74.5	71.4	36	3.52	3.03	1.92
VPO Co5	6	10	85.5	62.4	16	2.51	1.82	1.18
VPO Fe1	7	19	74.5	73.5	41	2.44	2.29	1.82
VPO Fe5	7	14	70.5	72.0	38	1.79	1.58	1.24

---

Characterisation of the doped VPO activated catalysts

Powder X-ray diffraction patterns of the activated catalysts are shown in Fig. 4). The materials can all be classed as poorly crystalline, although the diffraction patterns for the 1% doped samples are a little sharper. A comparison of the diffraction patterns with those obtained from crystalline standards of (VO)₂P₂O₇ and the various VOPO₄ phases¹⁵ does clearly show the presence of (VO)₂P₂O₇ in all the samples as characterised by the 200, 024 and 032 reflections at 23, 28.45 and 29.94°, respectively.

Fig. 4 Powder X-ray diffraction patterns of the activated catalysts: (a) VPO, (b) VPO Co1, (c) VPO Co5, (d) VPO Fe1, (e) VPO Fe5.

Co-doped catalysts. The ³¹P NMR spin–echo mapping spectra of the Co-doped activated catalysts are shown in Fig. 5). The spectra all show three signals, at 0 (P associated with V⁵⁺), 2400 [characteristic of well crystallized (VO)₂P₂O₇] and 200–1500 (characteristic of poorly crystalline/disordered C⁴⁺–V⁵⁺ material) ppm. From a comparison of the undoped and Co-doped activated catalyst, it is apparent that the 1% Co-doped material has a high proportion of well crystallized (VO)₂P₂O₇ but it also has a high proportion of the signal associated with the poorly crystalline/disordered material.

Fig. 5 ³¹P NMR spin-echo mapping spectra of activated catalysts: (a) VPO, (b) VPO Co1, (c) VPO Co5.

Low magnification transition electron micrographs of typical crystallites from VPO and VPO Co1 are shown in Fig. 6. It is clear that the rhomboid platelet morphology of the hemihydrate is retained in these specimens. The undoped VPO catalyst comprises mainly microcrystallites of (VO)₂P₂O₇ , together with a very small amount of disordered material, and the microscopy of this sample has been described in detail previously.⁵ The VPO Co1 platelet is largely disordered material with some (VO)₂P₂O₇ crystallites (30–150 nm in size) that preferentially nucleate at the periphery of the platelet, as shown in Fig. 7. In this image, crossed fringes with an interplane spacing of 3.1 Å and an interplanar angle of 81° (corresponding to [024]^pyro planes) are resolved, and these are characteristic of the [100] projection of (VO)₂P₂O₇. It is clear that doping with 1% Co significantly enhances the proportion of disordered material present in the activated catalysts. Chemical analysis, using energy dispersive X-ray analysis, of the disordered material and an isolated (VO)₂P₂O₇ crystallite are shown in Fig. 8. It is clear that, within the limit of detection of 0.1% Co, the (VO)₂P₂O₇ crystallites do not contain Co, and the Co dopant is preferentially segregated and exists within the disordered material.

Fig. 6 Bright field transmission electron micrographs of activated catalysts: (a) VPO, (b) VPO Co1.

Fig. 7 Bright field transmission electron micrographs of (VO)₂P₂O₇ crystallite that has nucleated at the platelet periphery.

Fig. 8 Energy dispersive X-ray spectra for VPO Co1: (a) (VO)₂P₂O₇ crystallite, (b) disordered platelet interior.

The major morphology present in the VPO Co5 activated catalyst is a very similar platelet morphology to that of VPO Co1. Again, selected area diffraction patterns show reflections due to (VO)₂P₂O₇, together with additional reflections due to non-transformed hemihydrate. It is apparent, at this level of doping, that Co is still inhibiting the transformation of the hemihydrate under the reaction conditions. However, an additional phase is present in the VPO Co5 activated catalyst (Fig. 9) but as a very minor component (<1 vol%). Electron diffraction studies indicate that the latter material is disordered. EDX microanalysis of the material contained significantly more Co than the partially transformed VOHPO₄·0.5 H₂O/(VO)₂P₂O₇ platelets previously described, and it is most likely that the material is cobalt phosphate.

Fig. 9 Bright field transmission electron micrograph of the minor secondary phase observed in VPO Co5.

Fe-doped catalysts. The ³¹P NMR spin-echo mapping spectra for the Fe-doped activated catalysts are shown in Fig. 10. The spectra for VPO Fe1 and VPO Fe5 are very similar and show three signals, at 0, 200–1500 and 2400 ppm, as observed with the Co-doped samples. Qualitative comparison of the spectra indicates that Fe-doping appears to have increased the relative proportions of the signals associated with crystalline (VO)₂P₂O₇ and disordered V⁴⁺–V⁵⁺.

Fig. 10 ³¹P NMR spin-echo mapping spectra of activated catalysts: (a) VPO Fe1, (b) VPO Fe5.

Low magnification transmission electron micrographs of typical crystallites from VPO Fe1 and VPO Fe5 are shown in Fig. 11. It is clear that the rhomboid morphology of the hemihydrate precursor is retained. The platelet interior of the VPO Fe1 material is disordered and it is observed that crystallites of (VO)₂P₂O₇ have formed at the edge of the rhomboid platelet; these (VO)₂P₂O₇ crystals are ca. 40 nm in thickness. In addition to (VO)₂P₂O₇, a small amount (3–4 vol%) of α_II-VOPO₄, as detected by selected area diffraction, was observed to be present. Chemical analysis using energy dispersive X-ray analysis showed that the Fe was uniformly distributed throughout the rhomboidal platelet, but at such low levels the chemical analysis of Fe is inevitably inconclusive, because of the possible fluorescence effect from the microscope pole pieces. In view of this, it is not possible to comment further on the distribution of Fe within this material.

Fig. 11 Bright field transmission electron micrographs of activated catalysts: (a) VPO Fe1, (b) VPO Fe5.

For VPO Fe5, the sample comprises rhomboid platelets in which crystallites have nucleated, preferentially at the platelet edges. High resolution imaging and selected area diffraction patterns confirmed these crystallites to be (VO)₂P₂O₇. In addition, α_II-VOPO₄ is present in this sample (10–20 vol%).

Comments on the promotional effects of Fe and Co dopants

It is clear that addition of low levels of Co and Fe to VOHPO₄·0.5H₂O, prepared by the reaction of V₂O₅ and H₃PO₄ with isobutanol, leads to an improvement in the catalytic performance of the activated material. Both lead to an enhancement in maleic anhydride selectivity and n-butane conversion and, in particular, the intrinsic activity. The detailed structural study of the activated catalysts using trans mission electron spectroscopy allows us to comment on the role of Co as a promoter. The Co dopant is effectively insoluble in the (VO)₂P₂O₇ crystallites. Although Co is homogeneously dissolved within the initial VOHPO₄·0.5H₂O, on activation phase separation takes place and the Co preferentially remains in the disordered V⁴⁺–V⁵⁺ material. In addition, the presence of Co appears to stabilise the disordered material and the VPO Co1 sample contains a significant proportion of the disordered material when compared with the equivalent undoped catalyst.⁵ Since the addition of 1% Co markedly increases the intrinsic activity and maleic anhydride selectivity, we do not consider that this is due to an increased activity of the crystalline (VO)₂P₂O₇ component of the catalyst, since the crystalline (VO)₂P₂O₇ is essentially undoped and must retain the catalytic performance of the undoped material. We propose therefore that the Co dopant is acting as a promoter for the disordered VPO material and this provides an indication that disordered VPO phases may be of significance for the oxidation of n-butane to maleic anhydride. This is in agreement with a number of previous observations concerning the role of disordered material in vanadium phosphate catalysts.¹⁶ As chemical analysis using EDX is inconclusive for the Fe doped material, it is not possible to comment in detail on the origin of the promotional effect observed with Fe. However, previous studies by Otake¹⁷ have shown that Fe can be present as a solid solution in (VO)₂P₂O₇ and a distinct increase in the intrinsic activity for maleic anhydride formation was observed for such (VO)_2−xFe_xP₂O₇ solid solutions. In this case, Fe could act as an electronic promoter specifically for (VO)₂P₂O₇, probably enabling the re-oxidation of the catalyst or aiding oxygen mobility. It is also possible that Fe is associated with the disordered phase and could act in a similar manner with this material. Hence, the promotional effect observed with Fe may originate from a combination of two effects.

Acknowledgements

We thank the EPSRC and EU (European Human Capital and Mobility Program Contract CHRX-CT92-0065) for financial support.

References

1. Forum on Vanadyl Pyrophphate Catalyst, ed. G. Centi, Catal. Today, 1994, 16..
2. G. J. Hutchings, Appl. Catal., 1991, 72, 1.
3. J. W. Johnson, D. C. Johnston, A. J. Jacobson and J. F. Brody, J. Am. Chem. Soc., 1984, 106, 8123.
4. G. J. Hutchings, A. Desmartin-Chomel, R. Olier and J. C. Volta, Nature (London), 1994, 368, 41.
5. C. J. Kiely, A. Burrows, G. J. Hutchings, K. E. Bere, J. C. Volta, A. Tuel and M. Abon, Faraday Discuss., 1996, 105, 103.
6. M. Abon, K. E. Bere, A. Tuel and P. Delichere, J. Catal., 1995, 156, 28.
7. P. L. Gai and K. Kourtakis, Science, 1995, 267, 661.
8. G. W. Coulston, S. R. Bare, H. Kung, K. Birkeland, G. K. Bethke, R. Harlow, N. Herron and P. L. Lee, Science, 1997, 275, 191.
9. J. T. Gleaves, J. R. Ebner and T. C. Kuechler, Catal. Re[italic v]. Sci. Eng., 1988, 30, 49.
10. M. T. Sananés-Schulz, F. B. Abdelouhab, G. J. Hutchings and J. C. Volta, J. Catal., 1996, 163, 346.
11. J. Li, M. E. Lashier, G. L. Schrader and B. C. Gerstein, Appl. Catal., 1988, 38, 83.
12. I. J. Ellison, G. J. Hutchings, M. T. Sananés and J. C. Volta, J. Chem. Soc., Chem. Commun., 1994, 1093.
13. G. J. Hutchings, I. J. Ellison, M. T. Sananés and J. C. Volta, Catal. Lett., 1996, 38, 231.
14. M. T. Sananés, A. Tuel and J. C. Volta, J. Catal., 1994, 145, 251.
15. F. Benabdelouahab, R. Olier, N. Guilhaume, F. Lefevre and J. C. Volta, J. Catal., 1992, 134, 151.
16. A. Bruckner, A. Martin, B. Kubias and B. Lucke, J. Chem. Soc., Faraday Trans., 1998, 94, 2221.
17. M. Otake, US Pat. 1399560, 1975..

---