Silica-Supported Molybdena Catalysts. Surface Structures, Reduction

Dec 19, 1996 - 1040 °C (i.e., TM5) is observable. A shift of Ton ...... Industrial & Engineering Chemistry Research 2013 52 (37), 13297-13304. Abstra...
0 downloads 0 Views 2MB Size
19994

J. Phys. Chem. 1996, 100, 19994-20005

Silica-Supported Molybdena Catalysts. Surface Structures, Reduction Pattern, and Oxygen Chemisorption Francesco Arena† and Adolfo Parmaliana*,†,‡ Dipartimento di Chimica Industriale, UniVersita` degli Studi di Messina, Salita Sperone c.p. 29, I-98166 S. Agata, Messina, Italy, and Istituto CNR-TAE, Salita S. Lucia 39, I-98126 S. Lucia, Messina, Italy ReceiVed: June 24, 1996; In Final Form: September 25, 1996X

The effects of preparation method, nature of the silica support, oxide loading (0.2-7.0 wt %), and pretreatment conditions on the reduction pattern of MoO3/SiO2 catalysts have been systematically investigated by temperature-programmed reduction (TPR) measurements in the range 200-1200 °C. Modeling of TPR spectra of silica-supported molybdena catalysts highlights the presence of several “surface MoVI species” characterized by a different interaction strength with the underlying support, the formation of which is mostly controlled by pretreatment conditions and oxide loading. The reduction path of the above surface species has been probed by catalytic measurements in propane hydrogenolysis reaction. The oxide dispersion of MoO3/SiO2 catalysts and bulk MoO3 has been comparatively evaluated by high- and low-temperature oxygen chemisorption measurements. Evidence of a strong metal oxide support interaction depressing both the reducibility and oxygen chemisorption of low-loaded (800 °C) spectral region. (iii) The fresh MPS 4 catalyst (Figure 3d) shows a very large and sharp reduction peak with a maximum at 498 °C (TM1) contributing ca. 50% to the total H2 consumption. A broad band of low rate of H2 consumption, with unresolved maxima at 600, 721, and 874 °C, spans the range 600-1150 °C. Heat treatment at 600 °C, besides causing the shift of Ton from 368 to 436 °C (Table 3), causes an overall increase in the rate of H2 consumption at T > 600 °C (Figure 3d′) counterbalanced by the disappearance of the TM1 reduction peak (Figure 3d). It results in a convoluted reduction spectrum characterized by the presence of two broad maxima at 588 and 765 °C, respectively (Table 3). (iv) The MPS 7 catalyst is the system undergoing the minor modifications consequent to the heat treatment at 600 °C. Apart from the shift of Ton from 382 to 435 °C, the heat treatment produces a deeper overlap of TM1-TM3 peaks due to the shift of TM1 from 536 (fresh sample) to 574 °C (TM2), while the position of the TM3 peak remains practically unchanged (Table 3). Further, the presence of a tail of H2 consumption, spanning the range 800-1100 °C, is observable in the spectra of both fresh (Figure 3e) and heat-treated samples (Figure 3e′). Finally, a systematic evaluation of the influence of heat treatments at T ranging between 200 and 700 °C on the TPR pattern of the MPS 4 system is outlined in Figure 4 and Table 4. For the sake of comparison, also the spectrum (Figure 4a) and the relative parameters of the untreated sample are shown. The TPR pattern of the sample heat treated at 200 °C (Figure

Figure 4. TPR of MPS 4 catalyst: (a) fresh and treated at (b) 200, (c) 400, (d) 600, and (e) 700 °C.

TABLE 4: Influence of Pretreatment Temperature on TPR of MPS 4 Catalyst: Onset Temperature of Reduction (Ton) and Temperature of Peak Maxima (TMi)a TPret (°C)

Ton (°C)

TM1 (°C)

TM2 (°C)

TM3 (°C)

TM4 (°C)

no 200 400 600 700

368 365 410 436 430

498 505

600 608 551 588 604

721 726

874 878 844

765 787

TM5 (°C)

H2 consumption (molH2/molMoO3) 2.84 2.86 2.91 2.93 2.88

aT M1 < 550 °C; 550 < TM2 < 650 °C; 650 < TM3 < 800 °C; 800 < TM4 < 900 °C; TM5 > 900 °C.

4b) on the whole looks very similar to that of the untreated system (Figure 4a), even if a sharpening of the TM1 peak is evident. Thermal treatment at T g 400 °C leads to a progressive “smoothing” of the contribution of the spectral zone comprised between 400 and 550 °C accounting for a progressive shift of Ton, TM1, and TM3 to higher values (Table 4). However, it is noteworthy that a rise in the pretreatment temperature from 600 (Figure 4d) to 700 °C (Figure 4e) does not modify the TPR pattern of the MPS 4 sample since only a slight shift of TM2 and TM3 to higher T has been recorded (Table 4). 2.C. Reduction Pattern of M-PS Physical Mixtures. Influence of Loading and Thermal Pretreatment. The TPR patterns of fresh (a,b) and heat-treated at 600 °C (a′,b′) M-PS 4 (a,a′), and M-PS 7 (b,b′) physical mixture samples are shown in Figure 5, while the relative values of Ton and TMi are compared with those of the impregnated MPS 4 and MPS 7 systems in Table 5. The stoichiometric reduction of MoVI to Mo0 was also experienced in mechanically mixed M-PS 4 and M-PS 7 samples (Table 5). The TPR spectra of both M-PS 4 and M-PS 7 samples, either in fresh (Figure 5a,a′) or heat-treated (Figure 5b,b′) form, look quite similar to those of same-loaded impregnated systems (Figures 3d,e and d′,e′). In particular, the TPR spectrum of the untreated M-PS 4 sample (Figure 5a) shows a well-resolved reduction peak with the maximum (TM1) at 530 °C and a lower rate of H2 consumption with unresolved maxima at 709 and 886 °C (Table 5). Thermal treatment at 600 °C causes the shift of Ton from 376 to 442 °C and the disappearance of the TM1 peak counterbalanced by an overall increase in the rate of H2 consumption at T > 550 °C, featuring three “broad” maxima at 584, 744, and 865 °C (Figure 5a′).

Silica-Supported Molybdena Catalysts

J. Phys. Chem., Vol. 100, No. 51, 1996 19999

Figure 5. TPR spectra of M-PS 4 (a,a′) and M-PS 7 (b,b′) samples, fresh (a,b) and treated at 600 °C (a′,b′).

TABLE 5: Influence of Thermal Treatment on TPR of “Impregnated” (MPS) and “Mechanically Mixed” (M-PS) MoO3/SiO2 Catalysts: Onset Temperature of Reduction (Ton) and Temperature of Peak Maxima (TMi)a TPret Ton TM1 TM2 TM3 TM4 TM5 H2 consumption catalyst (°C) (°C) (°C) (°C) (°C) (°C) (°C) (molH2/molMoO3) MPS 4

no 600 M-PS 4 no 600 MPS 7 no 600 M-PS 7 no 600

368 436 376 442 382 435 378 436

498

600 588

530 584 536 574 525 612

721 765 709 744 700 703 696 704

874 886 865

2.84 2.93 3.22 3.21 2.87 2.96 3.18 3.23

TM1 < 550 °C; 550 < TM2 < 650 °C; 650 < TM3 < 800 °C; 800 < TM4 < 900 °C; TM5 > 900 °C. a

Similarly to that observed for the MPS 7 sample, the thermal treatment affects to a lower extent the reduction pattern of the M-PS 7 system. In fact, the only evident changes are the shift of Ton from 378 to 456 °C and a deeper overlapping between the two main reduction peaks consequent to the shift of the first one to higher T (Table 5). A tail of H2 consumption at T > 800 °C is observable in the spectrum of both untreated and heat-treated samples (Figure 5b,b′). 2.D. Reduction Pattern of MoO3/SiO2 Catalysts. Influence of the Silica (PS, FS) Support. The TPR profiles of similarly loaded MPS 2 (a), MFS 2(a′), MPS 4 (b), MFS 4 (b′), MPS 7 (c), and MFS 7 (c′) catalysts, heat treated at 400 °C, are compared in Figure 6, while the relative values of onset temperature of reduction (Ton) and temperature of peak maxima (TMi) are reported in Table 6. The bare FS sample does not indicate any reduction process in the T range investigated. Irrespective of the loading, also for MFS catalysts the stoichiometric reduction of MoO3 to Mo0 has been observed (Table 6). The TPR spectra, outlined in Figure 6, indicate that both the level of loading and the nature of the silica support affect the reduction profiles of MoO3/SiO2 catalysts. In fact, the following can be noted: (1) The reduction profile of the MPS 2 catalyst (Figure 6a) is characterized by an increasing rate of H2 consumption starting

Figure 6. TPR spectra of differently loaded MPS (a,b,c) and MFS (a′,b′,c′) catalysts treated at 400 °C: (a) MPS 2, (b) MPS 4, (c) MPS 7, (a′) MFS 2, (b′) MFS 4, and (c′) MFS 7.

TABLE 6: TPR of “Precipitated” (MPS) and “Fumed” (MFS) Silica-Supported MoO3 Catalysts Heat Treated at 400 °C: Onset Temperature of Reduction (Ton) and Temperature of Peak Maxima (TMi)a catalyst

Ton (°C)

MPS 2 MFS 2 MPS 4 MFS 4 MPS 7 MFS 7

445 432 408 400 445 446

TM1 (°C)

TM2 (°C)

TM3 (°C)

590 587 551 551 571 628

700 698 706 695 735

TM4 (°C)

844 888

TM5 (°C)

H2 consumption (molH2/molMoO3)

927 906

2.85 3.14 2.91 3.11 3.12 2.89

a TM1 < 550 °C; 550 < TM2 < 650 °C; 650 < TM3 < 800 °C; 800 < TM4 < 900 °C; TM5 > 900 °C.

at 445 °C (Ton) and bearing two unresolved maxima at 590 and 700 °C, respectively, which gives rise to a main reduction peak centered at 927 °C (TM5). The spectrum of the MFS 2 catalyst (Figure 6a′) is substantially similar to that of MPS 2 sample even if a higher rate of H2 consumption at T < 800 °C is observable. The reduction process starts at 432 °C (Ton), while two not well-resolved reduction maxima at 587 and 698 °C along with a main reduction maximum at 906 °C feature the TPR spectrum of the MFS 2 system. (2) The TPR pattern of the MPS 4 catalyst (Figure 6b) displays the presence of a resolved maximum at 551 °C and a continuous band of H2 consumption with a broad maximum at 844 °C (Table 6). The Ton (408 °C) is shifted to lower T with respect to that (445 °C) of the MPS 2 catalyst. The reduction profile of the MFS 4 sample (Figure 6b′), looking similar to that of the counterpart MPS 4, displays a more enhanced reduction maximum at 551 °C and two broad maxima at 706 and 888 °C, respectively; (3) The TPR spectrum of the MPS 7 sample (Figure 6c) shows a main resolved reduction peak with a maximum at 695 °C strongly convoluted with a less intense peak centered at 571 °C (Table 6). In addition, a tail of H2 consumption in the region 800-1120 °C can be observed (Figure 6c). The Ton is 445 °C, being then higher than that found for the MPS 4 catalyst. The reduction profile of the MFS 7 (Figure 6c′) presents a large band of H2 consumption between 500 and 800 °C ca. with two maxima at 628 and 735 °C, respectively. A tail of H2

20000 J. Phys. Chem., Vol. 100, No. 51, 1996

Arena and Parmaliana

TABLE 7: Activity Data of MPS Catalysts in Propane Hydrogenolysis at TR ) 350 °C catalyst

WMoO3 (mg)

pretreatment T (°C)

MPS 4

14

no

600

M-PS 4

14

no

600

MPS 7

13

no

600

Trid (°C)

C3H8 conv (%)

350 500 550 750 350 500 550 750 350 520 550 750 350 520 550 750 350 500 550 750 350 500 550 750

0.0 17.6 32.3 45.4 0.0 0.0 2.3 42.5 0.0 14.5 24.2 43.3 0.0 0.0 2.2 44.7 0.0 8.1 14.3 45.8 0.0 0.0 1.7 50.2

CH4/C2H6 (mol/mol)

TABLE 8: Oxygen Uptakes and Dispersion Values of Bulk MoO3 and Various MoO3/SiO2 Catalysts Obtained by HTOC and LTOC Measurements: Influences of Thermal Treatment, MoO3 Loading, Preparation Method, and Silica Support

1.60 1.45 2.00

O2 uptake O2 uptake Tpret catalyst (°C) (µmol gcat-1) O/Mo (%) (µmol gcat-1) O/Mo (%)

HTOC

PS 1.82 2.04

MPS 1

1.64 1.70 1.62

MPS 2

1.61 1.39 1.15 1.64 1.70 1.43 1.58

consumption, lying in the range 800-1120 °C, is observable also for this system. The Ton (446 °C) is shifted to higher T with respect to that (400 °C) of the MFS 4 catalyst. 3. Activity of MoO3/SiO2 Catalysts in Propane Hydrogenolysis. The influence of thermal treatment on the oxidation state of Mo in MoO3/SiO2 catalysts reduced at various temperatures has been probed by catalytic measurements in propane hydrogenolysis at 350 °C. The activity data of MPS 4, M-PS 4, and MPS 7 catalysts, reduced in the T range 350-750 °C, are reported in Table 7 in terms of C3H8 conversion and CH4/ C2H6 molar ratio. Irrespective of the preparation method (i.e., impregnation, mechanical mixing) and loading, heat-treated catalysts reduced at T e 550 °C are ineffective in the model reaction. A significant C3H8 conversion level ranging from ca. 42% (MPS 4) to 50% (MPS 7) is measured only upon their reduction at 750 °C. By contrast, untreated samples reduced at 500 °C exhibit appreciable C3H8 conversion levels going from ca. 8% (MPS 7) to 17.6% (MPS 4). A rise in Tred from 500 to 750 °C enhances progressively the C3H8 conversion on all the catalyst samples to an extent comparable with that of the pretreated systems. The CH4/C2H6 molar ratio, varying between 1.15 and 2.0, signals a rather limited occurrence (e25%) of the double hydrogenolysis (C3H8 f CH4 + C2H6 f 2CH4). 4. Oxygen Chemisorption (HTOC, LTOC) on Bulk MoO3 and MoO3/SiO2 CatalystssInfluence of Thermal Pretreatment. The oxygen uptake values of bulk MoO3 and various MoO3/SiO2 catalysts, fresh and heat treated at T ranging between 400 and 700 °C, along with the relative oxide dispersion (O/ Mo, %) values obtained by both HTOC and LTOC measurements are listed in Table 8. For the sake of completeness, the O2 uptake values of the bare PS and FS carriers have been also included. The above data point to a strong dependence of the oxygen chemisorption on MoO3 loading and pretreatment conditions, while the preparation method as well as the nature of the silica carrier exert a minor influence on the chemisorption properties of MoO3/SiO2 catalysts. In particular, the above results can be rationalized as follows. HTOC. (1)Both PS and FS carriers exhibit a very slight O2 chemisorption, not affected by the calcination treatment, which

MPS 4

MPS 7 M-PS 4 M-PS 7 FS MFS 2 MFS 4 MFS 7 MoO3

no 600 no 600 no 400 600 no 400 500 600 700 no 400 500 600 600 600 600 no 600 no 400 600 no 400 600 no 400 500 600

0.5 0.5 23.3 2.5 n.d. 11.9 6.1 232.2 89.4 70.0 66.4 58.8 150.0 71.9 108.4 111.5 69.0 136.8 0.1 n.d. 10.2 n.d. 86.8 71.2 n.d. 50.0 102.9 453.9 46.0 24.3 20.7

LTOC

67.1 7.2 17.1 8.8 167.0 64.4 50.4 47.8 42.3 61.7 29.6 44.6 45.9 49.7 56.3 13.4 55.6 45.6 20.5 42.3 13.1 1.3 0.7 0.6

n.d. 0.5 33.5 9.0 n.d. n.d. 24.7 n.d. n.d. n.d. 103.1 n.d. n.d. n.d. n.d. 85.4 n.d. n.d. 0.1 n.d. 32.0 n.d. n.d. 130.3 n.d. n.d. 93.0 n.d. n.d. n.d. 36.0

96.5 25.9 35.6

74.2

35.1

41.9 83.4 38.3

1.0

is more than 1 order of magnitude lower than that of MoO3based systems; the oxygen chemisorption of the PS support, however, is significantly larger (0.5 µmol g-1) than that of the FS sample (0.1 µmol g-1). Untreated MoO3 provides a high oxygen uptake, more than 1 order of magnitude larger (453.9 µmol g-1) than that of heattreated samples, which corresponds to a dispersion (O/Mo) value of 13.1%. Upon calcination at 400 °C, the oxygen chemisorption of bulk MoO3 drops to 46.0 µmol g-1 pointing to a dispersion value of 1.3%. A rise in pretreatment temperature from 400 to 600 °C leads to a further decrease in O2 uptake (20.7 µmol g-1) resulting in an oxide dispersion value of 0.6%. HTOC measurements of heat-treated systems were affected by a relevant degree ((20%) of unreproducibility. (2) For MPS catalysts pretreated at 600 °C, the O2 uptake steeply rises with the loading going from a value of 2.5 µmol g-1 (MPS 1) to 111.5 µmol g-1 (MPS 7). Very slight dispersion (O/Mo) values are found for the low-loaded MPS 1 (7.2%) and MPS 2 (8.8%) catalysts. Then, the oxide dispersion suddenly increases to a value of 47.8% (MPS 4), leveling off in the loading range between 4 wt % (MPS 4) and 7 wt % (MPS 7). O2 uptake and oxide dispersion values of differently loaded MFS catalysts and M-PS samples, pretreated at 600 °C, were analogous to those of the counterpart MPS systems. (3) Untreated low-medium-loaded (e4 wt % MoO3) MPS samples exhibit O2 uptakes significantly larger than those of heat-treated ones, the highest value being found for the MPS 4 system (232.2 µmol g-1). A rise in the pretreatment temperature from 400 to 700 °C causes a regular decrease in the O2 chemisorption and dispersion values. Likewise, a negative

Silica-Supported Molybdena Catalysts influence of the pretreatment temperature on the chemisorption of MFS 4 catalyst is found. (4) Also for the fresh MPS 7 catalyst, the O2 uptake is larger than that of heat-treated samples; however, an unusual positive influence of the calcination temperature on the chemisorption capability of both MPS 7 and MFS 7 systems is recorded. LTOC. (5)The LTOC uptakes of the bare PS (0.5 µmol g-1) and FS (0.1 µmol g-1) samples pretreated at 600 °C are equal to the HTOC ones. The oxygen uptakes and dispersion values of differently loaded MPS and MFS catalysts pretreated at 600 °C follow a volcano-shaped trend with the MoO3 loading, attaining the maximum on medium-loaded MPS 4 (103.1 µmol g-1; O/Mo ) 74.2%) and MFS 4 (130.3 µmol g-1; O/Mo ) 83.4%) systems. For bulk MoO3 an oxygen uptake equal to 36 µmol g-1, corresponding to a dispersion value close to 1%, was experienced. Discussion A. Morphology and Reduction Pattern of Bulk MoO3. The reduction pattern of bulk and silica-supported MoO3 catalysts has been generally rationalized in the light of the stepwise process MoVI f MoIV f Mo0.13-15 In order to explain the marked differences (i.e., TM, intensity of peaks, etc.) in the TPR pattern of bulk and supported systems various effects such as the different size and morphology of MoO3 crystals besides the presence of several noncrystalline MoO3 forms have been invoked.13,14 Arnoldy et al.,21 investigating the reduction pattern of MoO3 and MoO2, found that the formation of catalytic nucleation sites depends upon H2O partial pressure, concentration of surface defects, and surface area of the sample. Moreover, the TPR pattern of bulk MoO3 is also significantly affected by the sample size, precalcination temperature, and heating rate.21 Therefore, it is not surprising that the reported TPR spectra of bulk MoO3,13-16,21,22 though generally showing two main convoluted reduction peaks, are quite different in terms of reduction maxima, peak shape, and/or intensity. Our TPR spectra of untreated and heat-treated MoO3 samples (Figure 2) show two major resolved peaks, the relative intensity of which roughly accounts for the stepwise reduction stoichiometry Mo6+ f Mo4+ f Mo0,13,14,21 whereas no systematic evidence for the occurrence of such a reduction path arises from the TPR spectra of MoO3/SiO2 catalysts (Figures 3-6). The small peak (≈3%) with a maximum at 515 °C in the spectrum of the fresh sample (Figure 2a) can be attributed to the reduction of an easily reducible surface “MoVI species” (Table 2) stabilized at ambient conditions likely as a consequence of the adsorption of water from the atmosphere on rough crystal planes.16 Calcination at 400 °C negatively affects the reduction of bulk MoO3 (cf. Figure 2a,b) yielding quite remarkable shifts of both Ton and TMi to higher T (Table 2), even if no corresponding changes in morphology and surface area have been noticed (cf. Figure 1a,b). According to Arnoldy et al.,21 this pattern can be related to the disappearance of surface defect sites, monitored by the TM1 peak in the spectrum of the fresh sample (Figure 1a), the reduction of which provides low-Valent Mo atoms (e.g., Mo2+ or Mo0) very effective in catalyzing the subsequent reduction of MoO3. Calcination at 600 °C gives rise to the sintering of the MoO3 crystals (Figure 1c) which assume a quite regular morphology resulting on the whole to be less reducible. However, the preferential exposure of “more reducible” basal planes (Figure 1c′′) enables a higher rate of lattice oxygen diffusion across these crystal planes16 limiting the extent of the shift of TM2 and TM3 to higher T (Table 2). The asymmetry

J. Phys. Chem., Vol. 100, No. 51, 1996 20001 and irregular shape of the TPR profile of bulk MoO3 as well as the unreproducibility affecting TPR and HTOC measurements could be diagnostic of the high “microheterogeneity” of the system (Figure 1). Indeed, it can be inferred that similar reactions take place with different kinetics in MoO3 crystals characterized by a different particle size and morphology.21 B. Surface Structures and Reduction Pattern of MoO3/ SiO2 Catalysts. B.1. General. The nature of the MoO3-SiO2 interaction and consequently the surface structures and the reducibility of the MoO3/SiO2 system are still open matters. Several authors inferred a “weak-type” interaction accounting for an easy reduction of the MoO3/SiO2 system;13,14,22 however, other authors questioned this issue on the basis of either reducibility data15 or spectroscopic findings.2,4,11 De Boer et al.4 in a recent combined EXAFS-Raman study pointed out a remarkable spreading behaVior of molybdena on silica (microspreading) which leads to the formation of Im species. Liu and Rau,15 in addition, documented that the mechanical mixing method can lead to a better molybdena dispersion and a higher noncrystalline molybdena loading with respect to the impregnation, because of spreading of MoO3 crystals (macrospreading) occurring in MoO3-SiO2 physical mixtures calcined at T e 400 °C.6,15 Recently, Smith et al.,2 claimed the presence of Pm, Mc, and Im species in medium-high-loaded (3.5-9.8 wt %) MoO3/SiO2 catalysts attributing to the low effectiveness of the impregnation with AHM solutions the presence of Pm species in dehydrated samples. On the basis of LR evidences3,4,7,18 Wachs and co-workers stated that (a) the preparation method influences the relative “distribution” but not the nature of the “MoO3 surface species” in calcined catalysts3 and (b) the extent of formation of such species is a function of the loading and state of hydration of the samples. Seyedmonir and Howe,12 addressing the redox chemistry of MoO3/SiO2 catalysts, described the structure of molybdenasilica catalysts in terms of three different species, namely, “surface polymolybdates”, a “MoO3 phase”, and “isolated tetrahedral molybdenum forms”, characterized by a progressively lower reducibility. Namely, they found that polymolybdate phase reduces at 400 °C, while terminal oxide vacancies are formed in both polymolybdate and MoO3 phases at 500 °C; however, they did not observe any reduction of the isolated tetrahedral species in that T range.12 The uncertainties still existing on the nature and structure of the “noncrystalline” MoO3 species forming on MoO3/SiO2 catalysts, besides being a consequence of controversial assignments of UV,10 LR,3-5,23-25 and TPR2,13-15,22 spectral features, however, lie in a lack of careful control of the environmental conditions before and during data acquisition.10 B.2. TPR Pattern of MoO3/SiO2 Systems. The difficulty in developing a general “model” describing the complex TPR pattern of the MoO3/SiO2 system mostly arises from the occurrence of numerous overlapping peaks generated by the various MoVI forms having different reducibility and/or by stepwise changes in Mo oxidation state.2,13-15,22 However, also the pretreatment conditions greatly affect the TPR of MoO3/ SiO2 leading to ambiguous assignments of the reduction peaks.2,13-15,22 Then, in the following we infer a first rationalization of the TPR spectra of the MoO3/SiO2 system taking into account previous findings on the influence of the loading and heat pretreatment on the “surface structures” of MoO3/SiO2 catalysts.3-15,22-25 In particular, a systematic inspection of the changes induced by such factors on the TPR spectra (Figures 3-6) allows that the following be ascertained

20002 J. Phys. Chem., Vol. 100, No. 51, 1996 (a) The TM1 peak featuring the TPR spectra of all fresh MPS (Figure 3) and M-PS samples (Figure 5) is associated with the reduction of a “polymolybdate-like” phase, stabilized at ambient conditions by water adsorption and henceforth referred to as Pm species. The easy reduction of this species is enabled by the prompt formation of “low-valent” Mo ions, resulting in the development of an intense dark-blue color characteristic of defective MoOx (2 < x < 3) species, which displaces Ton to T < 400 °C;10-12 the Pm species disappear upon further thermal treatment at T g 400 °C (Figure 4),3,4,10-12,18,23-25 being transformed into less reducible species. (b) The reduction of MoO3 crystallites, Mc, occurs in the T range 500-800 °C2,13-15 as evidenced by the TPR patterns of the highly loaded MPS 7, MFS 7, and M-PS 7 systems. These spectra still do not provide systematic evidence of the occurrence of the stepwise reduction MoVI f MoIV f Mo0. (c) “Isolated strongly-interacting molybdates”, Im, resulting the least reducible species,2,11,12,14,15,22 are reduced in a single step at T > 800 °C14,15,22 according to the TPR features of lowloaded MPS 02, MPS 1, and MPS 2 catalysts (Figure 3). However, while the reduction path of Mc and Im species has been established, no definitive evidence has been provided until now about the TPR pattern of Pm species.10,12 Therefore, in order to highlight the reduction mechanism of Pm and Mc species, probing in particular the occurrence of a single or stepwise reduction path, a series of catalytic measurements in propane hydrogenolysis26,27 on MoO3/SiO2 catalysts reduced at various temperatures has been performed. It has been ascertained that metallic Mo is effective in such a reaction;26-28 thus, the activity data of untreated catalysts (Table 7) essentially indicate that the reduction of Pm species to Mo0 occurs at T < 550 °C in a single step without formation of intermediate oxidation states.21 On the contrary, catalyst samples heat treated at 600 °C become active only upon further to reduction at T > 550 °C. Evidently, the concomitant absence of Pm species and the occurrence of the stepwise reduction of Mc to Mo0 13-15,21,22 allow the stabilization of Mo0 only at T > 550 °C. B.3. Modeling of the TPR Pattern of MoO3/SiO2 System. Although the above findings indicate the formation of the same surface species in all MoO3/SiO2 catalysts, nevertheless the marked differences in the TPR profiles (Figures 3-6) and the irregular variations in the Ton and TMi values (see Tables 3-6) induced by MoO3 loading, pretreatment conditions, preparation method, and silica carrier do not allow an immediate rationalization of the TPR spectra. Thus, using a least-square-fitting program, it was found that it is possible to resolve such spectra into the contribution of similar discrete Gaussian-shaped peaks, each of them being attributable to the reduction of a defined “surface MoVI species”. In particular, the TPR spectra of fresh samples have been resolved into four components corresponding to the reduction of Pm (MoVI f Mo0, one component), Mc (MoVI f MoIV f Mo0, two components), and Im species (MoVI f Mo0, one component), while those of heat-treated catalysts result from the contribution of only three components, Mc and Im species, the Pm species being absent.3-7,10,11,23-25 In order to satisfy the reduction stoichiometry of Mc according to the steps MoVI f MoIV f Mo0,13-15,22 the ratio of the areas of the relative peaks was kept equal to 0.5 ((10%), while no constraints in terms of peak position and full width at halfmaximum were imposed. As an example of such modeling, the deconvoluted TPR spectra of the MPS 4 catalyst, untreated (a) and heat treated at 600 °C (b), are shown in Figure 7. Although a very good fitting (r2 > 0.98) of experimental profiles was attained (Figure 7), the results of such a mathematical procedure assume a qualitative character and must be considered

Arena and Parmaliana

Figure 7. Curve-fitted TPR spectra of the MPS 4 catalyst, fresh (top) and heat treated at 600 °C (bottom).

as a first approximation. The fitting parameters for all MPS catalysts, expressed as peak maximum position (Mi), full width at half-maximum (fwhmi), and percentage peak area (Ai, %), are reported in Table 9. A systematic inspection of these data allows the following remarks to be drawn: (i) With the exception of the highly loaded (MPS 7) system, for which a deep overlapping between first two peaks was observed, in all the other cases the reduction peak of Pm species looks as a discrete component disappearing further to the calcination pretreatment (Figure 3). The analogous peak position (498 e M1 e 529 °C) and fwhm (92-99 °C) in all catalyst samples (Table 9) confirm that it arises from the same surface species. (ii) The stepwise reduction of Mc is monitored by two peaks in the intensity ratio 1:2 centered at 579-641 °C (M2) and 703745 °C (M3), respectively. The intensity of such peaks keeps constant in low-loaded MPS 02 and MPS 1 catalysts being also unaffected by pretreatment at 600 °C. At loadings higher than 2 wt %, the intensity of M2 and M3 peaks steeply rises with MoO3 content, becoming the predominant components in the spectra of the MPS 7 catalyst (Table 9). (iii) The reduction of Im species is monitored by the highest temperature peak (860 e M4 e 1045 °C); such a species is the main component (>80%) in the spectra of MPS 02 and MPS 1 systems. At MoO3 loadings higher than 2 wt % the contribution of the M4 peak sharply decreases, while a significant and regular shift of the M4 value toward lower T (Table 9) signals an improved reducibility of Im species likely enabled by the “autocatalytic effect” exerted by the increased MoO3 concentration.21,22 (iv) The disappearance of the M1 peak in heat-treated samples determines a growth of the M2, M3, and M4 peaks controlled by the MoO3 loading (Table 9).

Silica-Supported Molybdena Catalysts

J. Phys. Chem., Vol. 100, No. 51, 1996 20003

TABLE 9: Fitting Parameters of TPR Spectra of Differently Loaded MPS Catalysts Untreated and Heat Treated at 600 °C Pm Mc Im Tpret catalyst (°C) M1 (°C) fwhm1 (°C) A1 (%) M2 (°C) fwhm2 (°C) A2 (%) M3 (°C) fwhm3 (°C) A3 (%%) M4 (°C) fwhm4 (°C) A4 (%) MPS 02 no 600 MPS 1 no 600 MPS 2 no 600 MPS 4 no 600 MPS 7 no 600

520

94

11.8

525

94

13.8

529

99

31.0

498

97

39.6

526

92

14.7

625 641 614 619 615 623 604 585 599 579

108 100 104 115 100 105 94 129 117 95

Figure 8. Relative abundance of surface species in differently loaded MPS catalysts, fresh (left bar) and treated at 600 °C (right bar). Legend: Im (black cluster), Mc (grey cluster), and Pm (white cluster).

On the basis of these results, the surface composition of differently loaded fresh and heat treated MPS catalysts is shown in Figure 8 in terms of clustered bars corresponding to the relative percentage of Pm, Mc, and Im species. Such data evidentiate the following: 1. Irrespective of the loading (0.2-7.0 wt %), Pm, Mc, and Im species coexist only in fresh samples. 2. The concentration of Pm species in fresh samples attains the maximum development (≈40%) in MPS 4 catalyst, thereafter decreasing in the highly loaded MPS 7 sample (≈15%). 3. The concentration of Im species decreases monotonically with the loading, while the amount of Mc follows a specular increasing trend. In addition, the above data show that only in low-medium (e4 wt %) loaded MPS catalysts a fraction (e25%) of strongly interacting Im species undergoes a reversible transformation into Pm10,11 according to the following reaction scheme: H2O(rt)

Im sssssf rsss Pm

(1)

T > 200 °C

Upon the MoO3 loading rises, because of the inaccessibility of Im species due to the superimposed growth of MoO3 crystals and the analogous acidic IEP value (≈2) of SiO2 and MoO3 surfaces, the above process mainly concerns coordinatively unsaturated sites (CUS) of Mc and to progressively lower extents Im species, the latter resulting negligible in the highly loaded MPS 7 sample (Table 9). In other words, water adsorption besides reaction 1 also enables the partial hydrolysis of defective sites of Mc H2O(rt)

Mc(CUS) sssssf rsss Pm

(2)

T > 200 °C

stabilizing easily reducible Oh Pm-like species.10 The occurrence of reaction 2) is confirmed by the small reduction peak with maximum at 515 °C in the TPR spectrum of the fresh bulk MoO3 (Figure 2a), disappearing upon calcination at T g 400

5.4 4.5 4.2 5.2 4.2 7.6 11.7 22.1 25.7 30.8

725 735 743 730 745 736 716 730 702 703

110 116 140 142 145 139 165 182 97 125

11.1 9.0 8.2 10.5 7.6 15.1 23.2 44.9 52.6 62.1

1045 1039 961 960 935 928 886 878 860 855

170 168 174 170 191 179 190 185 191 172

71.8 86.5 73.8 84.3 57.3 77.3 25.5 33.0 7.0 7.1

Figure 9. Relative abundance of surface species in differently loaded MFS and MPS catalysts treated at 400 °C. For the legend refer to the caption to Figure 8.

(Figure 2b,c). Obviously, the “driving force” of reaction 2 stems from the size and morphology of Mc. In fact, the largest occurrence of the reaction 2 in the MPS 4 catalyst (Table 9) monitors the highest dispersion of Mc (Vide infra). Furthermore, it should be stressed that the amount of Im species levels off at loadings g4 wt %, when Mo surface coverages higher than ca. 0.9 Mo/nm2 (Table 1) are attained.7,23,24 However, taking into account the relative percentage of Im species in differently loaded MPS catalysts (Table 9), an upper Mo surface coverage of 0.26 ( 0.01 Mo atoms/ nm2, well below the value of ca. 1 Mo atoms/nm2 indicated by Roark et al. as limit coverage of silanol groups,24 is obtained. Namely, the low efficiency of the adopted preparation method in achieving a high dispersion of the precursor across the silica surface2-6,14 as well as the calcination treatment at T > 500 °C can concur to determine an agglomeration of the precursor, limiting the formation of Im species. Furthermore, the strong decay in SA occurring in medium-high loaded (g4 wt %) MoO3/ SiO2 systems (Table 1), as a consequence of hydrothermal structural changes in the SiO2 support13,14 and blocking of micropores by Mc,13 could depress the availability of silanol groups further hindering the formation of Im species.7,24 Therefore, as the surface texture of the silica may also play a role in determining the surface distribution of the MoO3 phase and the relative abundance of MoVI species,3,7,23-25 it can be argued that the lower amount of Im species in low-medium loaded (2-4.5 wt %) MFS catalysts with respect to that of the counterpart MPS ones, counterbalanced by a higher abundance of Mc (see Figure 9), arises from both the lower SA of such systems and lower density of silanol groups in fumed silica.29,30 On the other hand, the relative percentages of Pm, Mc, and Im species in fresh (a) and heat-treated at 600 °C (b) mechanically mixed (M-PS 4, M-PS 7) or impregnated (MPS 4, MPS 7) MoO3/SiO2 samples, compared in Figure 10, confirm the effectiveness of MoO3 macrospreading onto the silica surface as to achieve the same “surface distribution” of the impregnation method.6,15 The relative abundance of Pm, Mc and Im in the

20004 J. Phys. Chem., Vol. 100, No. 51, 1996

Arena and Parmaliana TABLE 10: O/Mc Ratio Values for Various MoO3/SiO2 Catalysts Pretreated at 600 °C Obtained by HTOC and LTOC Measurements

Figure 10. Relative abundance of surface species in fresh (left bar) and treated at 600 °C (right bar) M-PS 4, MPS 4, M-PS 7, and MPS 7 samples. For the legend refer to the caption to Figure 8.

Figure 11. Influence of the calcination temperature (°C) on relative abundance of surface species in MPS 4 catalyst. For the legend refer to the caption to Figure 8.

MPS 4 catalyst calcined at T ranging between 200 and 700 °C is outlined in Figure 11. These results clearly indicate that thermal treatment at T g 400 °C does not further affect the surface composition of the supported MoO3/SiO2 system. Thus, it can be concluded that the distribution of the surface Mo species in heat-treated samples tends to an equilibrium state determined by the thermodynamics of MoO3-SiO2 interaction.31 C. Surface Features and Oxygen Chemisorption of Bulk and SiO2 Supported MoO3. Among the various methods proposed to evaluate the dispersity of supported MoO3 systems, those involving oxygen chemisorption on prereduced catalysts have received major attention.5,32,33 In particular, on the basis of the assumption that the reduction of MoO3 is not greatly affected by the interaction with the SiO2 support,5 two techniques were claimed to be equally effective for testing the dispersion in both bulk MoO3 and MoO3/SiO2 catalysts: (1) a first one, referred to as LTOC, involves the reduction at 500 °C with subsequent O2 chemisorption at low temperature (-195 °C/-78 °C)32,33 and (2) a second one, referred to as HTOC, involves a “softer” reduction at 350-360 °C, to avoid “bulk reduction” and sintering phenomena, followed by O2 pulsing at the same temperature to titrate the “surface” reduced Mo sites.5 Our TPR findings prove the occurrence of a strong metal oxide support interaction in the MoO3/SiO2 system which depresses mostly the reducibility of low (e2 wt % MoO3) loaded catalysts. Such a negative influence of the support on the reducibility of the promoter, resulting in the stabilization of Im species (see Figures 8 and 10), accounts for the small oxygen uptakes and very low dispersion values in diluted (e2 wt % MoO3) MoO3/SiO2 catalysts. At MoO3 loadings higher than 2 wt %, both the chemisorption capability and dispersion values (Table 8) rise as a consequence of the increased reducibility linked with the extensive formation of Mc (Figures 8 and 10). Although the LTOC uptakes and relative dispersion values of

catalyst

O(HTOC)/Mc (%)

O(LTOC)/Mc (%)

MPS 1 MPS 2 MPS 4 MPS 7 M-PS 4 M-PS 7 MFS 2 MFS 4 MFS 7

46 40 67 50 74 61 34 61 46

166 157 103 38 102 112 41

low-medium (e4 wt % MoO3) loaded systems pretreated at 600 °C are higher than those obtained by HTOC measurements (Table 8), it is evident that the estimate of the MoO3 dispersion is negatively affected by the unreducibility of such systems. Thus, taking into account the concentration of Mc (Table 9), the “normalized” HTOC and LTOC dispersion values (O(HTOC)/ Mc and O(LTOC)/Mc, respectively) have been calculated (Table 10). The O/Mc values more closely reflect the surface features of MoO3/SiO2 catalysts since the O(HTOC)/Mc ratio in low-loaded catalysts rises considerably with respect to the O/Mo dispersion values (Table 8), now resulting comparable (40-46%) with those of medium-high (g4 wt %) loaded systems (45-65%). However, a rather featureless trend of the O(HTOC)/Mc value with the oxide loading5 is still obtained (Table 10). The O(LTOC)/Mc ratio for MPS and MFS catalysts follows a reasonable decreasing trend with the loading, though the values relative to low-loaded systems are higher than 100%. The origin of such an anomaly could lye in either the underestimation of the concentration of Mc or more reasonably in a sort of systematic error probably arising from the assumed chemisorption stoichiometry O2:Mo ) 2. Indeed, a stoichiometry “O2: Mo” equal to 1 (molecular adsorption) instead of 2 (atomic adsorption)5 would give reasonable dispersion O(LTOC)/Mc values sharply decreasing from 83% to 19% upon the MoO3 loading in MPS catalysts increases from 1 to 7 wt %. Likewise, for MFS catalysts the O(LTOC)/Mc value would decrease from ca. 50% (MFS 2 and MFS 4) to 20% (MFS 7). Thus, in spite of the above limitations, the data reported in Tables 8 and 10 outline very high dispersion values (50-100%) for low-medium (e4 wt %) MoO3/SiO2 catalysts. This, along with the absence of the sharp features at 570-630 and 700750 °C in the related TPR spectra (Figures 3-6), characteristic of the reduction of large MoO3 crystals,22 indicate the presence of a “well-dispersed” heretofore uncharacterized polymerized molybdenum oxide species such as small clusters7,23,24 rather than crystalline MoO3 particles. This species, lying in close interaction with Im species, should be the precursor of MoO3 crystallites.6,7,23,24 The broad profile of TPR spectra of the above systems in the region 500-800 °C is thus connected with the noncrystalline character of such species, the reduction of which is still hindered by the interaction with the SiO2 support. The low reducibility of these species negatively affects the oxygen uptake, resulting still in an underestimation of the O(HTOC)/Mc value in low (e2 wt %) loaded catalysts (see Table 10). In fact, in situ Raman spectra of the MPS 4 sample treated in O2 at 540 °C34 did not provide evidence of any absorbance at 819 cm-1 confirming the absence of crystalline MoO3, while two broad bands at 970-975 and 996 cm-1, diagnostic of the presence of the aforesaid “polymerized molybdenum oxide” and Im species respectively,7,23-24 have been found. Therefore, the collapse of Pm species upon calcination at 400 °C besides the Im species gives rise to “highly disordered” and defective bidimensional MoO3 structures characterized by a higher

Silica-Supported Molybdena Catalysts

J. Phys. Chem., Vol. 100, No. 51, 1996 20005 (3) Polymolybdates are reduced to Mo0 in a single step at T < 500 °C. (4) Modeling of TPR spectra allows evaluation of the influences of loading and calcination treatments on the related distribution of the above surface species in MoO3/SiO2 catalysts. (5) The reliability of HTOC and LTOC methods for evaluating the MoO3 dispersion is limited by the complex reduction pattern of MoO3/SiO2 catalysts. (6) TPR is a fundamental and powerful tool to feature the surface structures of MoO3/SiO2 catalysts. Acknowledgment. Dr. A. Arico` is gratefully acknowledged for SEM analysis on MoO3 samples. The financial support of this work by MURST and CNR (Rome) is also acknowledged. References and Notes

Figure 12. Relationship between HTOC uptakes and concentration of Pm species in fresh MoO3 and various MoO3/SiO2 systems.

reducibility (cf. Tables 3, 4, and 6). The rise in the Tpret to 600 °C strengthens the interaction of such agglomerates with the support, lowering their reducibility (shift of both Ton and TM2 to higher T) and O2 uptake (Table 8). The singular increase in the HTOC uptake and dispersion values with calcination temperature (Table 8) observed for highly loaded MPS 7 and MFS 7 catalysts reflects a change in the morphological properties of such systems also confirmed by the peculiarity of their TPR profile (cf. Figures 3e′ and 6c). Namely, the sharper profile of the TM2 peak and the negative shift of Ton (see Tables 3 and 6) along with the increase of HTOC uptake and dispersion upon Tpret rises from 400 to 600 °C (Table 8) can be attributed to the redispersion of MoO3 crystallites via volatilization7 under the promoting effect of water produced by silica dehydroxylation. Obviously, such hypothesis does not apply for bulk MoO3, and then its very low dispersion degree (Table 8) decreases with Tpret according to SEM and TPR results. The highest HTOC uptakes of untreated MoO3 and MoO3/ SiO2 catalysts (Table 8) parallel the concentration of Pm as shown in Figure 12. The slope of such straight-line relationship (solid line) is 2.17, being larger than the theoretical one (1.5) accounting for the stoichiometric reduction of Pm species to Mo0 (dotted line). This, besides being the definitive proof of the “one-step” reduction pattern of Pm species to Mo0, also indicates the catalytic effect of the Pm species on the reduction of less reducible surface MoO3 species (Mc and Im). Thus, reducibility, morphology, and oxygen chemisorption stoichiometry effects concur with the high undetermination of dispersion data of MoO3/SiO2 catalysts. Conclusions A strong metal oxide support interaction greatly affects reducibility as well as oxygen chemisorption of MoO3/SiO2 catalysts. In particular, the main results of this work can be summarized as follows: (1) Irrespective of the preparation method and silica carrier, molybdenum(VI) oxide stabilizes three defined surface structures on the silica support having a progressively lower reducibility polymolybdates > MoO3 crystallites > isolated molybdates (2) The formation of polymolybdates occurs reversibly at ambient conditions upon the adsorption of water from the atmosphere disappearing upon calcination at T g 400 °C.

(1) Spencer, N. D. J. Catal. 1988, 109, 187. (2) Smith, M. R.; Zhang, L.; Driscoll, S. A.; Ozkan, U. S. Catal. Lett. 1993, 19, 1. (3) Williams, C. C.; Ekerdt, J. G.; Jehng, J.-M.; Hardcastle, F. D.; Turek, A. M.; Wachs, I. E. J. Phys. Chem. 1991, 95, 8781. (4) De Boer, M.; van Dillen, A. J.; Koninsberger, D. C.; Geus, J. W.; Vuurman, M. A.; Wachs, I. E. Catal. Lett. 1991, 11, 227. (5) Desikan, L.; Huang, L.; Oyama, S. T. J. Phys. Chem. 1991, 95, 10050. (6) Liu, T.-C.; Forissier, M.; Coudurier, G.; Vedrine, J. G. J. Chem. Soc., Faraday Trans. 1 1989, 85, 1607. (7) Ban˜ares, M. A.; Hu, H.; Wachs, I. E. J. Catal. 1994, 150, 407. (8) Barbaux, Y.; Elamrani, A. R.; Payen, E.; Gengembre, L.; Bonnelle, J. P.; Grzybowska, B. Appl. Catal. 1988, 44, 117. (9) Ono, T.; Kamiusuki, H.; Hisashi, H.; Miyata, H. J. Catal. 1988, 116, 303. (10) Marcinkowska, K.; Rodrigo, L.; Kaliaguine, S.; Roberge, P. C. J. Mol. Catal. 1985, 33, 189. (11) Rodrigo, L.; Adnot, A.; Roberge, P. C.; Kaliaguine, S. J. Catal. 1987, 105, 175. (12) Seyedmonir, S. R.; Howe, R. F. J. Catal. 1988, 110, 216 and references therein. (13) Cordero, R. L.; Gil Lambias, F. J.; Lo´pez Agudo, A. Appl. Catal. 1991, 74, 125. (14) Ismail, H. M.; Zaki, M. I.; Bond, G. C.; Shukri, R. Appl. Catal. 1991, 72, L1. (15) Liu, T. C.; Rau, J. J. J. Chin. Inst. Chem. Eng. 1989, 20, 269. (16) Smith, M. R.; Ozkan, U. S. J. Catal. 1993, 141, 124. (17) Ban˜ares, M. A; Rodriguez-Ramos, I.; Guerreor-Ruiz, A.; Fierro, J. L. G. In Proceedings of the 10th International Congress on Catalysis, Budapest 1992; Guczi, L., Solymosi, F., Tetenyi, P., Eds.; Akadeˆmiai Kiado´: Budapest, Hungary,1993; Vol. B, p 1131. (18) Kim, D. S.; Wachs, I. E.; Segawa, K. J. Catal. 1994, 146, 268. (19) Malet, P.; Caballero, A. J. Chem. Soc., Faraday Trans 1 1988, 84, 1891. (20) Miceli, D.; Arena, F.; Parmaliana, A.; Scurrell, M. S.; Sokolovskii, V. Catal. Lett. 1993, 84, 283. (21) Arnoldy, P.; de Jonge, J. C. M.; Moulijin, J. A. J. Phys. Chem. 1985, 89, 4517. (22) Regalbuto, J. R.; Ha, J.-W. Catal. Lett. 1994, 29, 189. (23) Roark, R. D.; Kohler, S. D.; Ekerdt, J. G. Catal. Lett. 1992, 16, 71. (24) Roark, R. D.; Kohler, S. D.; Ekerdt, J. G.; Kim, D. S.; Wachs, I. E. Catal. Lett. 1992, 16, 77. (25) Ban˜ares, M. A.; Hu, H.; Wachs, I. E. J. Catal. 1995, 155, 249. (26) Nakamura, R.; Pioch, D.; Bowman, R. G.; Burwell, R. L., Jr. J. Catal. 1985, 93, 388. (27) Chung, J.-S.; Burwell, R. L., Jr. J. Catal. 1989, 116, 519. (28) Touvelle, M. S.; Stair, P. C. J. Catal. 1993, 139, 93. (29) Lygin, V. I. Kinet. Catal. 1994, 35, 480. (30) Zhuravlev, L. T. Langmuir 1987, 3, 316. (31) Machej, T.; Haber, J.; Turek, A. M.; Wachs, I. E. Appl. Catal. 1991, 70, 115. (32) Fierro, J. L. G.; Garcia de la Banda, J. F. Catal. ReV.-Sci. Eng. 1986, 28, 265. (33) Liu, H. C.; Yuan, L.; Weller, S. W. J. Catal. 1980, 61, 282. (34) Arena, F.; Frusteri, F.; Miceli, D.; Parmaliana, A.; Pen˜a, M.; Faraldos, M.; Ban˜ares, M. A.; Lopez-Granados, M.; Fierro, J. L. G.; Baranay, P.; Bota´r, L.; Ku´ti, Zs.; Ne´meth, A.; Vido´czy, T. In Fifth Periodic Report; ECC Joule Programme, Contract N. JOU2-0040, Messina, Italy, August 1995.

JP9618587