Physicochemical and catalytic properties of hybrid catalysts derived from 12-molybdophosphoric acid and montmorillonites

Applied Catalysis A: General 498 (2015) 192–204

Contents lists available at ScienceDirect

Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata

Physicochemical and catalytic properties of hybrid catalysts derived from 12-molybdophosphoric acid and montmorillonites Aleksandra Pacuła a,∗ , Katarzyna Pamin a , Joanna Kry´sciak-Czerwenka a , a ˙ ´ Zbigniew Olejniczak b , Barbara Gil c , Elzbieta Bielanska , Roman Dula a , Ewa M. Serwicka a , a Alicja Drelinkiewicz a

Jerzy Haber Institute of Catalysis and Surface Chemistry, Polish Academy of Sciences, Niezapominajek 8, 30-239 Kraków, Poland Henryk Niewodnicza´ nski Institute of Nuclear Physics, Polish Academy of Sciences, Radzikowskiego 152, 31-342 Kraków, Poland c Jagiellonian University, Faculty of Chemistry, Ingardena 3, 30-060 Kraków, Poland b

a r t i c l e

i n f o

Article history: Received 5 November 2014 Received in revised form 11 March 2015 Accepted 24 March 2015 Available online 31 March 2015 Keywords: Montmorillonite 12-Molybdophosphoric acid Keggin-like structure Catalytic conversion of ethanol

a b s t r a c t The hybrid catalysts derived from 12-molybdophosphoric acid (H3 PMo12 O40 denoted as HPMo) and montmorillonite (Na, H-exchanged montmorillonites derived from Polish bentonite denoted as Na-Mt, HMt and commercially available K10 montmorillonite denoted as Mt-K10) were prepared by conventional and modified impregnation routes. Conventional procedure based on the incipient wetness impregnation method involved stirring of the aqueous dispersion of HPMo and montmorillonite, followed by drying at elevated temperature. Modified preparation method employed ultrasonication as a means of the dispersion homogenization, followed by freeze-drying. Such prepared catalysts were characterized by means of powder X-ray diffraction (XRD), scanning electron microscopy (SEM), nitrogen sorption, electron spin resonance (ESR), Fourier transform infrared (FT-IR), ultraviolet–visible (UV–vis) and 31 P solid-state magic-angle-spinning nuclear magnetic resonance (MAS-NMR) spectroscopy. Acidic/redox properties of HPMo/montmorillonite systems were evaluated by catalytic conversion of ethanol and pyridine sorption. The hybrid catalysts consist of two catalytically active components. In contrast to montmorillonite, especially acid-treated montmorillonite, which has mainly acidic sites, HPMo species in the final hybrid catalysts may offer both acidic and redox catalytic functions. Spreading HPMo on the surface of montmorillonites results in enhancement in the redox reactivity as a consequence of exposing the redox sites (Mo-containing species). The use of Na-Mt leads to the catalysts with dominating redox function as the acidic sites (coming from HPMo) may be obscured by replacement of protons by sodium ions coming from Na-Mt. In addition, the catalyst prepared by modified impregnation shows higher redox reactivity than that prepared by conventional impregnation. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Heteropolyacids with the Keggin-type structure have been of great interest in the field of catalysis owing to their acidic and redox properties [1,2]. They have been studied as catalysts in bulk or supported forms in both homogeneous and heterogeneous reactions. In liquid–solid and gas–solid heterogeneous reactions heteropolyacids are usually examined in a deposited form since they exhibit low specific surface areas. Besides a mechanical role, the support may modify the catalytic properties of heteropolyacids by

∗ Corresponding author. Tel.: +48 12 6395131; fax: +48 12 4251923. E-mail address: [email protected] (A. Pacuła). http://dx.doi.org/10.1016/j.apcata.2015.03.030 0926-860X/© 2015 Elsevier B.V. All rights reserved.

favoring growth of certain structures or by inducing different types of interaction. Many porous materials such as silica, titania, carbon and alumina have been applied for the immobilization of 12molybdophosphoric acid hydrate (H3 PMo12 O40 ) [3–9]. Among them, clay minerals (e.g. montmorillonites) have been also explored as the supports for H3 PMo12 O40 (HPMo). The motivation for choosing montmorillonites as components for the preparation of HPMo-containing catalysts is based on their high thermal stability, high specific surface area, beneficial interaction with HPMo species and their own catalytic properties. According to Ref. [10–12], acid-treated clays themselves can act as catalysts in the reactions such as ethylene polymerization, nitration of bromobenzene, esterification of fatty acids and fatty acid mixtures with methanol.

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

HPMo-containing catalyst derived from acid-treated clays were studied in many reactions catalyzed by acidic sites [13–16]. For instance, HPMo deposited on K10 montmorillonite with 20 wt% loading was tested in etherification of ␤-naphthol with alcohols such as methanol, ethanol, n-propanol, 2-propanol and n-butanol [13]. In this work, the activity of deposited HPMo was compared with other supported heteropolyacids, i.e. dodecatungstophosphoric acid (HPW) and dodecatungstosilicic acid (HSiW). In other work [14], the catalyst prepared by mixing HPMo with K10 montmorillonite was examined in alkylation of benzene with 1-dodecene and its activity was also compared with the systems containing HPW or HSiW. The catalytic performance of heteropolyacids (HPMo, HPW and HSiW) spread on K10 montmorillonite was also evaluated in transesterification of vegetable oils [15]. In this study, various edible/nonedible vegetable oils and alcohols were examined. Moreover, the effect of heteropolyacid loadings (5, 10, 15, 20 and 30 wt%) on clay was also evaluated. The catalytic activity of HPMo (20 wt%) dispersed on acidtreated clay (bentonite from India) was tested in esterification of acetic acid with primary, secondary and tertiary butanol [16]. Its catalytic properties were compared with analogue catalysts containing HPW or HSiW. Based on above-mentioned examples, the catalysts containing 20 wt% of HPMo deposited on acid-treated montmorillonite appeared to be active, selective and stable. According to Refs. [17,18], their catalytic performance may also derive from the synergy between heteropolyacid and montmorillonite. In the present work we describe the hybrid catalysts derived from H3 PMo12 O40 and various montmorillonites (Mt-K10, H-Mt or Na-Mt) prepared via a well-established and commonly applied impregnation procedure and by the method involving ultrasonication and freeze-drying. The conversion of ethanol was selected as a test reaction to evaluate the acidic/redox function of the composites and to demonstrate that not only the form (H- or Na-exchanged) of montmorilonite but also the preparation method affects the catalytic performance of the hybrid catalysts, in particular those derived from Na-Mt. 2. Experimental 2.1. Materials 12-Molybdophosphoric acid, H3 PMo12 O40 ·xH2 O (denoted as HPMo) was purchased from Fluka. The number of structural water molecules (x) in HPMo was determined by thermogravimetric (TG) analysis. Based on TG results the formula of HPMo is the following H3 PMo12 O40 ·14H2 O. Polish bentonite (Milowice in Poland) was used as the raw material to prepare montmorilonite components, i.e. Na- and H-exchanged montmorillonites (denoted as Na-Mt and H-Mt, respectively). The preparation procedure was described in details in our previous publication [19]. Commercially available K10 montmorilonite (denoted as MtK10) was supplied by Fluka. According to [20], Mt-K10 is obtained from the raw material by treatment with mineral acid at high temperature which results in ion exchange and leaching the cations, i.e. aluminum, magnesium and iron. 2.2. Preparation of the hybrid catalysts HPMo was deposited on Na-Mt, H-Mt and Mt-K10 according to either the conventional or the modified impregnation methods described in our previous publication [19]. For instance, the expressions HPMo/Na-Mt and HPMo/Na-Mt/m represent the samples

193

prepared using HPMo and Na-Mt via conventional and modified impregnation methods, respectively. The amount of HPMo used for the deposition (proportional to specific surface area of montmorillonite) was calculated assuming the formation of one HPMo monolayer coverage on both Na-Mt (HPMo content = 9 wt%) and Mt-K10 (HPMo content = 33 wt%), and a 0.7 monolayer on H-Mt (HPMo content = 9 wt%). This “monolayer coverage” model for the HPMo deposition was adopted from Ref. [21–23] in view to potentially achieve a uniform distribution of heteropolyacid species on the surface of montmorillonite.

2.3. Characterization of the physicochemical properties Powder X-ray diffraction (XRD) patterns were obtained with a Philips X’Pert powder diffractometer using Cu-K␣ radiation operating at 35 kV and 30 mA. The patterns were recorded from 2◦ to 72◦ 2 with a scan step time of 2 s and a step size of 0.05◦ . Nitrogen sorption analysis was carried out at −196 ◦ C using an Autosorb-1 QUANTACHROME sorptometer. Prior to analysis, the samples were degassed at 200 ◦ C over night. Adsorption/desorption of nitrogen was performed in the relative pressure range of 0.01 < p/p0 < 1.00, using ca. 0.2 g of sample. Nitrogen sorption measurements were carried out knowing that a similar set of the hybrid catalysts derived from heteropolyacid (HPW) and montmorillonites (Mt-K10, H-Mt and Na-Mt) appeared to be predominantly mesoporous [19]. Specific surface area (SBET ) was calculated using the Brunauer–Emmett–Teller (BET) method based on adsorption data in the partial pressure of 0.1 < p/p0 < 0.35 [24] confirmed by satisfying correlation coefficients (>0.999). Total pore volume (Vtot ) was determined by the amount of nitrogen adsorbed at p/p0 = 0.99. Mesopore volume (Vmeso ) was calculated by using the method based on the non-local density functional theory (NLDFT), whereas micropore volume (Vmicro ) was estimated by the t-method. The pore size distribution (PSD) was calculated by the NLDFT method. Scanning electron micrographs (SEM) were recorded by means of a Philips XL 30 microscope. Ultraviolet–visible (UV–vis) reflectance spectra were recorded in the wavelength range of 200–800 nm using a Shimadzu 160A spectrometer equipped with a reflectance chamber with a 5 mm slit. The samples for analysis were prepared as pellets, and BaSO4 was used as a reference for HPMo, whereas the appropriate montmorillonite was used as a reference for the hybrid catalysts. Fourier transform infrared (FT-IR) spectra were recorded on a Nicolet 800 spectrometer using KBr disk technique. Electron spin resonance (ESR) spectra were recorded at −196 ◦ C with a Technical University of Wroclaw SE/X spectrometer operating in the X-band, at 100 mW microwave power, with a 100 kHz magnetic field modulation. Solid state Magic-Angle-Spinning Nuclear Magnetic Resonance (MAS-NMR) spectra were measured on a Tecmag APOLLO pulse NMR spectrometer in the magnetic field of 7.05 T generated by a Magnex superconducting magnet. A Bruker HP-WB high-speed MAS probe equipped with a 4 mm zirconia rotor and a KEL-F cap was used to record the MAS spectra at the spinning speed of 8 kHz. The solid state 31 P MAS-NMR spectra were measured at 121.28 MHz, using a single 2 ␮s radio-frequency (rf) pulse, corresponding to /6 flipping angle. The acquisition delay used in accumulation was 20 s. The 31 P chemical shifts were referenced using 85% H3 PO4 solution (0 ppm). Thermal stability of the hybrid catalysts was estimated by means of a STA 409PC (Netzsch) thermal analyzer. For comparison, thermal stability of pure HPMo and montmorillonites was also determined. Thermogravimetric (TG) and differential scanning calorimetric (DSC) curves were recorded during heating the

194

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

samples (10 mg) from the ambient temperature up to 1000 ◦ C with a heating rate of 10 ◦ C min−1 under static air condition.

(a)

HPMo = H3PMo12O40·nH2O

2.4. Characterization of the acidic/redox properties Intensity (a.u.)

(001) (004) Q

(020) Q (001)

(020) Q

(001)

(020)

2

12

Q (004) (005) Q

Q

32

(060)

HPMo/Na-mt (060)

Na-mt (060)

Q 42

52

62

2 theta (degree)

(b)

HPMo = H3PMo12O40·nH2O (001)

Q (020) Q

(001)

(001)

HPMo/H-mt/m

(004) (005) Q Q

(020) Q

Q

HPMo/H-mt

(005) Q

Q

(005) Q

Q

(004)

(060)

Q

(060)

Q

Q

(020)

2

12

(004)

22

32

H-mt

42

(060)

52

Q

62

72

2 theta (degree)

(c)

HPMo = H3PMo12O40·nH2O M (001)

Intensity (a.u.)

3. Results and discussion

Q

22

Q

2.4.2. Catalytic conversion of ethanol The conversion of ethanol was conducted in a quartz flow reactor using 0.3 cm3 of catalyst (0.2–0.5 g) with sieve fraction of 0.2–0.4 mm. The reaction was performed according to previously reported procedure [19]. The reactor was heated to the required temperature (100–300 ◦ C) with a heating rate of 2 ◦ C min−1 under a flow of helium (30 cm3 min−1 ) saturated with ethanol vapor (5.7 mol%). The reaction temperature was maintained for 30 min. Prior to catalytic test, the samples were activated in the reactor under helium flow at 300 ◦ C for 2 h. The products were analyzed by means of a Perkin-Elmer 900 gas chromatograph (GC) equipped with a flame ionization detector and a Porapak S column.

HPMo/Na-mt/m

(005) Q

Q (005) (004) Q

Q

Intensity (a.u.)

2.4.1. FT-IR study of pyridine sorption The nature of acidic sites was determined by pyridine sorption followed by FT-IR. FT-IR spectra were recorded on a Bruker Tensor 27 spectrometer equipped with an MCT detector, in the absorption mode, with a spectral resolution of 2 cm−1 . Samples were pressed into self-supported pellets (Mt-K10, HPMo/Mt-K10/m) or deposited on the silicon wafer evaporating on its surface a few drops of ethanol suspension (H-Mt, HPMo/H-Mt). Prior to pyridine adsorption all samples were evacuated at 320 ◦ C under vacuum (10−3 Torr) for 40 min in an IR cell. After activation, samples were cooled down to 170 ◦ C and IR spectra were collected. Subsequently, the excess of pyridine was introduced and allowed to contact with the sample for few minutes, followed by a 20 min evacuation at 170, 250 and 320 ◦ C, respectively. All IR spectra presented here were recorded at 170 ◦ C. The concentration of Brønsted and Lewis acidic sites (in ␮mol g−1 ) was calculated after each desorption step using the following bands and adsorption coefficients: pyridine PyH+ band at 1545 cm−1 , ␧(PyH+ ) = 0.078 cm ␮mol−1 and pyridine PyL band at 1454 cm−1 , ␧(PyL) = 0.165 cm ␮mol−1 [25]. All spectra were recalculated to a ‘normalized’ sample of 10 mg. The acidic strength was determined based on the I170 /I170 , I250 /I170 , I320 /I170 ratio, where I170 , I250 , I320 is the intensity of the pyridine bands after desorption at 170, 250 and 320 ◦ C, respectively.

H3PMo12O40·14H2O

(020) M Q M (020) M Q

(001)

Q

HPMo/mt-K10/m

(004) (005) Q

Q

(004) (005) Q

Q

(060)

Q

HPMo/mt-K10 (060)

Q M (001)

3.1. Powder X-ray diffraction

(020) Q

M

Fig. 1(a–c) shows the XRD patterns of the hybrid catalysts and their starting components: H3 PMo12 O40 ·14H2 O and montmorillonites (Na-Mt, H-Mt and Mt-K10). The reflections characteristic of crystalline HPMo are distinctly seen in the XRD patterns of the catalysts derived from Mt-K10. It implies that HPMo is not well-dispersed on the surface of Mt-K10 and HPMo occurs in the form of crystallites. Similarly, 12tungstophosphoric acid (H3 PW12 O40 denoted as HPW) deposited on Mt-K10 with the amount corresponding to one HPW monolayer loading, does not cover completely the surface of Mt-K10 [26]. The traces of high-intensity reflections of crystalline HPMo can be also observed in the XRD patterns of HPMo/Na-Mt, HPMo/H-Mt and HPMo/H-Mt/m. Different position of these peaks implies that dispersed HPMo contains the crystallites with various numbers of water molecules, i.e. H3 PMo12 O40 and H3 PMo12 O40 ·13H2 O [27,28]. The XRD results clearly demonstrate that HPMo is more evenly dispersed onto Na-Mt or H-Mt than on commercially available Mt-K10. The XRD patterns of the montmorillonites exhibit a set of characteristic reflections ascribed to the (0 0 1), (0 2 0), (0 0 4),

2

12

22

(004) 32

mt-K10

(005) Q

Q 42

52

(060)

Q

62

2 theta (degree) Fig. 1. The XRD patterns of the hybrid catalysts derived from HPMo and various montmorillonites (a) Na-Mt (b) H-Mt and (c) Mt-K10. The XRD pattern of H3 PMo12 O40 ·14H2 O is also displayed.

(0 0 5), (0 6 0) planes. All montmorillonites contain quartz impurities (denoted as Q). In addition, Mt-K10 contains also mica impurities (denoted as M). The XRD pattern of Mt-K10 (Fig. 1c) is similar to that reported elsewhere [29]. The structure of Na-Mt and H-Mt derived from Polish bentonite significantly differs from that of commercially available K10 montmorillonite. According to Ref. [17,20,30], the preparation of K10 montmorillonite involves treatment of raw montmorillonite with strong acid at high temperature which results in a partial distortion of its structure and reduces its crystallinity.

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

HPMo/Na-Mt HPMo/H-Mt HPMo/Mt- K10

0.7 1.3 0.8

Catalysts prepared via modified impregnation

I0 0 1 /I0 2 0

HPMo/Na-Mt/m HPMo/H-Mt/m HPMo/Mt-K10/m

3.1 1.9 2.0

Deposition of HPMo on the clay minerals results in some changes in the montmorillonite structure, similarly as previously observed in HPW deposited on analogous clay minerals [16,17,19]. A decrease in montmorillonite crystallinity is especially pronounced in the case of Na-Mt and H-Mt. The relative intensity of the (0 0 1) and (0 2 0) reflections varies in different way depending on preparation method, i.e. the conventional or the modified impregnation (Table 1). An increase in the ratio of the integrated (0 0 1) and (0 2 0) reflection intensities I0 0 1 /I0 2 0 for the samples obtained according to modified impregnation procedure may be explained by the influence of the ultrasound treatment on the montmorillonite grain organisation along the direction parallel to the (0 0 1) basal plane, which causes a partial ordering in polycrystalline montmorillonite sample (Table 1) [19,31]. Moreover, it has been reported that during impregnation process the chemical composition of montmorillonite may also be changed [17,20,21]. As a consequence, both the interlayer (sodium) and the layer ions (magnesium, aluminium and iron) in the montmorillonite framework may be partially removed. There is no evidence in the XRD patterns (Fig. 1a) that sodium ions in Na-Mt are completely replaced with protons. According to XRD analysis, d-spacing (which corresponds to the sum of one layer thickness and interlayer distance) estimated from the position of the reflection due to the (0 0 1) plane is ca. 12.4 A˚ for Na-Mt. If all sodium ions in Na-Mt are replaced by protons the d-spacing ˚ characteristic of H-exchanged montmoshould be close to 14.8 A, rillonite (like observed for H-Mt or Mt-K10), being a result of a bimolecular water layer associated with proton [32]. After mixing HPMo with Na-Mt, the d-spacing moderately decreases to ca. ˚ which is rather a consequence of lower water content and 11.7 A, therefore rearrangement of a monomolecular layer of water associated with sodium cation [33]. However, we cannot exclude that some Na+ may be replaced with H+ .

dV(w) (cm3/Å/g)

I0 0 1 /I0 2 0

Na-Mt

0.0035

HPMo/Na-Mt

0.0030

HPMo/Na-Mt/m

0.0025 0.0020 0.0015 0.0010 0.0005 0.0000 20

30

40

50

60

70

80

60

70

80

60

70

80

Pore width (Å)

(b) 0.0040

dV(w) (cm3/Å/g)

Catalysts prepared via conventional impregnation

(a) 0.0040

H-Mt

0.0035

HPMo/H-Mt

0.0030

HPMo/H-Mt/m

0.0025 0.0020 0.0015 0.0010 0.0005 0.0000 20

30

40

50

Pore width (Å)

(c) 0.0120

Mt-K10 HPMo/Mt-K10

0.0100

dV(w) (cm3/Å/g)

Table 1 Ratio of integrated (0 0 1) and (0 2 0) reflections intensities I0 0 1 /I0 2 0 from Fig. 1.

195

HPMo/Mt-K10/m 0.0080 0.0060 0.0040 0.0020

3.2. Nitrogen sorption 0.0000

The nitrogen sorption isotherms (Fig. S1 in Supplementary material) of montmorillonites (Na-Mt, H-Mt and Mt-K10) can be classified as type IIb with type H3 hysteresis loop [24]. Such isotherms are observed for mesoporous materials containing assemblages of platy particles as reported previously for Naexchanged montmorillonite [24]. The knee of the isotherm at low relative pressures (p/p0 ) especially pronounced in the case of MtK10 reveals that montmorillonite contains also some micropores. Fig. 2(a–c) illustrates that mesopores are dominating in all montmorillonites with the highest contribution in Mt-K10. Based on NLDFT method, the mean diameter of mesopores is roughly 5.1, 5.7 and 5.3 nm for Mt-K10, H-Mt and Na-Mt, respectively. Mt-K10 has higher specific surface area (211 m2 g−1 ) than Na-Mt and HMt derived from Polish bentonite (42 and 63 m2 g−1 , respectively) (Table 2). Deposition of HPMo onto montmorillonites reduces their specific surface area and porosity (Table 2) as previously observed for similar montmorillonites modified with HPW [19,26,34]. The extent of changes in the textural properties depends on the type of montmorillonite and the impregnation method. Specific surface

20

30

40

50

Pore width (Å) Fig. 2. The pore size distribution (mesoporosity) calculated using nitrogen sorption data and NLDFT method. Table 2 Textural properties of the catalysts prepared by mixing H3 PMo12 O40 ·14H2 O with NaMt, H-Mt, Mt-K10 using conventional and modified impregnations. Specific surface area (SBET ), total pore volume at p/p0 = 0.99 (Vtot ), mesopore and micropore volumes (Vmeso , Vmicro ) were calculated from nitrogen adsorption isotherm. Sample

SBET (m2 g−1 )

Vtot (cm3 g−1 )

Vmeso (cm3 g−1 )

Vmicro (cm3 g−1 )

Na-Mt HPMo/Na-Mt HPMo/Na-Mt/m H-Mt HPMo/H-Mt HPMo/H-Mt/m Mt-K10 HPMo/Mt-K10 HPMo/Mt-K10/m

42 21 4 63 42 4 211 109 111

0.108 0.104 0.030 0.106 0.072 0.037 0.321 0.181 0.185

0.089 0.034 0.006 0.059 0.041 0.011 0.191 0.098 0.091

0.003 0.002 0 0.010 0.009 0 0.006 0.014 0.020

196

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

area of Mt-K10 decreases by ca. 50% after deposition of HPMo (33 wt%), irrespective of the impregnation mode. Spreading HPMo (9 wt%) onto H-Mt and Na-Mt via conventional impregnation has a similar (a ca. 50-% drop) or smaller (a ca. 30-% drop) impact. Modified impregnation dramatically diminishes specific surface area (a ca. 90-% drop) of H-Mt and Na-Mt. It is worth mentioning that there are no micropores in both samples obtained by modified method, i.e. HPMo/H-Mt/m and HPMo/Na-Mt/m (Table 2). Comparison of pore size distributions (PSDs) (Fig. 2(a–c)) illustrates that deposition of HPMo on the surface of Na-exchanged montmorillonite, in particular HPMo/Na-Mt, leads to a partial clogging the pores resulting in the formation of the pores with smaller diameters. The mean mesopore width decreases from 5.3 nm for Na-Mt down to 4.7 nm for HPMo/Na-Mt. PSD for HPMo/Na-mt becomes less uniform than that for Na-Mt. Deposition of HPMo on the surface of H-exchanged montmorillonites (Mt-K10 and H-Mt) also blocks the pores, however, the average diameter of mesopores for the catalysts derived from MtK10 (5.3 nm) or H-Mt (5.7 nm) remains almost unchanged. The extent of pore clogging is reflected in PSD for HPMo/Mt-K10, HPMo/Mt-K10/m and HPMo/H-Mt, which becomes more uniform than in initial montmorillonites (Fig. 2(b, c)). In general, the contribution of mesopores in the HPMocontaining catalysts is lower than in the starting montmorillonites (Table 2). However, the contribution of micropores in both catalysts (HPMo/Mt-K10, HPMo/Mt-K10/m) derived from Mt-K10 is higher than in the starting material – Mt-K10. Micropores may be formed during the impregnation process as a result of a partial destruction of montmorillonite structure in acidic solution, evidenced also by XRD analysis (Fig. 1). In addition, partial clogging the mesopores, in particular by the location of HPMo near or in pore mouth, may lead to virtually narrower pores finally observed as the transformation of mesopores into micropores. It is also likely that the microporosity in HPMo + Mt-K10 is rather developed between HPMo particles (which may make its own pore networks) as previously reported for HPW deposited on activated clay [21]. 3.3. Scanning electron microscopy The grains of montmorillonites (Na-Mt, H-Mt, Mt-K10) consist of irregular agglomerates of plate-like particles as we reported elsewhere [19]. Conventional impregnation does not affect significantly the morphology of the montmorillonite grains as shown in Fig. 3(a, c, e). This means that a decrease in specific surface area of montmorillonites after HPMo deposition is predominately associated with changes in their internal pore structure, which is not reflected in SEM images. A distinct difference in the external image of the grains is only observed for two samples prepared by modified method using Na-Mt and H-Mt (Fig. 3(b, d)). The grains in these samples, i.e. HPMo/Na-mt/m and HPMo/H-mt/m, have a compact arrangement. It seems that HPMo is preferentially deposited on the external surface of montmorillonites, filling the voids between the grains. It looks as though the montmorillonite grains are “glued” together with heteropolyacid, which gives the monolithic appearance of the polycrystalline material. Such observation is consistent with the nitrogen sorption data (Table 2) revealing a very low specific surface area of HPMo/Na-mt/m and HPMo/H-mt/m. It seems that in the case of HPMo/Na-Mt/m and HPMo/H-Mt/m, HPMo was preferably deposited in the external part of montmorillonite particles as a consequence of the use of an excess of impregnating solution volume. In the case of modified method, the amount of impregnating solution was extremely larger than the pore volume of the clays. The volume of aqueous HPMo solution added to 1 g of montmorillonite was 150 cm3 in modified impregnation, whereas it was only 1–2 cm3 in conventional impregnation [19].

Such effect is not observed in the case of HPMo/Mt-K10 and HPMo/Mt-K10/m. No difference in the morphology of these catalysts (Fig. 3(e, f)) is also consistent with their similar specific surface area and porosity (Table 2). It clearly shows that impregnation conditions influence significantly characteristics of the catalysts derived from Na-Mt and H-Mt. 3.4. UV–vis spectroscopy The UV–vis spectrum of pure HPMo exhibits four absorption bands at 250, 295, 320 and 410 nm (Fig. 4). All the bands are due to the ligand-to-metal charge transfer (LMCT) (O2− → Mo6+ ) [21]. These bands are also seen in the spectra of the HPMo-containing catalysts (Fig. 4). The band at ca. 410 nm is shifted into the region of high energy for all the catalysts and it appears at ca. 360–395 nm. Based on Refs. [35,36], this effect may be caused by an increase in the distance between heteropolyanions in the HPMo-containing catalysts in comparison to pure HPMo. This observation is more pronounced for the catalysts derived from Na-Mt and H-Mt than for those derived from Mt-K10. Thus, the UV–vis data are consistent with the XRD data implying that HPMo is less dispersed on commercial montmorillonite. Furthermore, a more precise analysis of the band position, reveals that it is more shifted towards the region of high energy for the catalysts prepared by modified method (HPMo/Na-Mt/m, HPMo/H-Mt/m, HPMo/Mt-K10/m) than for those obtained by conventional procedure. These results demonstrate that the catalysts prepared by modified impregnation show higher heteropolyacid dispersion than those obtained by the conventional method. 3.5. Fourier transform infrared spectroscopy The subtracted FT-IR spectra of the hybrid catalysts are displayed in Fig. 5(a–c). For comparison the spectrum of pure HPMo is also shown. The FT-IR spectrum of heteropolyacid exhibits four bands at 1070, 960, 870 and 780 cm−1 corresponding to as (P–Oa ), as (Mo Od ), as (Mo–Ob –Mo) and as (Mo–Oc –Mo) stretching vibrations, respectively (Fig. 5a), characteristic of (PMo12 O40 )3− anion with the Keggin-like structure [2,9]. The subtracted spectra of HPMo mixed with Mt-K10 are very similar to that of pure HPMo. However, in the spectrum of the sample prepared by modified route a splitting of the bands at 1070 and 870 cm−1 (the shoulders at 1040 and 910 cm−1 , respectively) is observed. The presence of the additional weak band at 1040 cm−1 reveals that symmetry of a fraction of heteropolyanions has been changed and it differs from the Td symmetry characteristic of the hydrated Keggin anions. Several possible reasons of the heteropolyanion symmetry distortion may be considered. They have been described to be a result of (a) a change in interaction between proton and anion being a consequence of lowering the hydration degree, (b) a strong affinity between anions and surface groups of support, (c) a presence of lacunary anions. When a decrease in the symmetry of heteropolyanions is not accompanied by the degradation of the Keggin-like structure the vibrations which can be shifted are mainly those connected with X–O vibrations and vibrations associated with the most external oxygen atoms such as M Od and M–Oc –M in the molecules of Hn (XM12 O40 ). For lacunary anions the symmetry distortion of heteropolyanions is higher so a more complex FT-IR spectrum with a greater number of bands should be expected. For instance, the spectrum of (PMo11 O39 )7− anion in the wavenumber range of 1200–600 cm−1 displays seven bands at 1060, 1010, 930, 900, 860, 790, 740 cm−1 [2]. Thus, the presence of shoulders at 1040 and 910 cm−1 is rather a consequence of the interaction of HPMo with montmorillonite surface groups, which we also observed for HPW deposited on Mt-K10 [19] than a result of heteropolyanion decomposition. This may be further confirmed

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

197

Fig. 3. SEM images of the hybrid catalysts derived from HPMo and various montmorillonites (a, b) Na-Mt (c, d) H-Mt and (e, f) Mt-K10.

by the fact that these shoulders are only observed for the sample obtained by modified impregnation, in which HPMo is more dispersed. HPMo supported on Mt-K10 may be attached to the surface of Mt-K10 by electrostatic attraction, i.e. (SiOH2 )+ (H2 PMo12 O40 )− . The similarity between the spectra of both HPMo-containing samples (HPMo/Mt-K10, HPMo/Mt-K10/m) and the spectrum of pure HPMo implies the presence of HPMo crystallites in both catalysts, confirmed also by the XRD patterns (Fig. 1c). The subtracted FT-IR spectra of H3 PMo12 O40 deposited on HMt (Fig. 5b) reveal that the contribution of Keggin units possessing a disordered symmetry is higher for those catalysts as compared to the samples derived from Mt-K10, and it is especially pronounced for the sample prepared by modified route. The splitting of the band at 790 cm−1 is observed. It suggests that the symmetry of heteropolyanions is reduced as a consequence of their interaction with the support [37]. It implies that HPMo species are more dispersed on H-Mt than on Mt-K10. This observation is consistent with Ref. [4], at low heteropolyacid (HPMo) loading (less than 30%), HPMo is well dispersed on silica, which results in a strong interaction between HPMo and Si–OH on the silica surface. In addition, the interaction between HPMo particles (species) and the montmorillonite surface groups seems to be enhanced in

the sample prepared by modified impregnation and therefore the shoulders at 810 cm−1 in the spectrum of HPMo/H-mt/m are more pronounced. The FT-IR subtracted spectra of HPMo dispersed on Na-Mt significantly differ from that of pure HPMo (Fig. 5c). A decrease in the (Mo Od ) stretching frequency from 870 cm−1 (for pure HPMo) to 860 cm−1 (for HPMo deposited on Na-mt) is observed. According to Ref. [38], this effect is attributed to a weakening of anion-anion (electrostatic) interaction and it may be caused by an increase in counterion size, i.e. replacement of protons from HPMo with sodium ions coming from Na-Mt. Indeed, some interlayer sodium ions in Na-Mt may be replaced with protons derived from HPMo. Thus, sodium cations released from Na-Mt, may compensate the negative charge of heteropolyanions (Na)+ (H2 PMo12 O40 )− by analogy with (SiOH2 )+ (H2 PMo12 O40 )− . The splitting of the bands at 1070 and 940 cm−1 may be attributed to the presence of lacunary anion, i.e. (PMo11 O39 )7− [7], whereas the splitting of the bands at 860 and 790 cm−1 may be attributed to the presence of (P2 Mo18 O62 )6− anion [39], which gives bands at 1080, 1008, 950, 906, 835, 780 cm−1 [40]. The band at 620 cm−1 may indicate that also a small amount of MoO3 is formed during HPMo deposition on Na-mt [9,41].

198

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

resonance line with two maxima at −2.7 and −4.9 ppm is observed. According to Refs. [43,44], the position of the NMR resonance line of H3 PMo12 O40 ·nH2 O may vary with the amount of water of crystallization. For instance, 12-molybdophosphoric acid with n = 30 and n = 8 gives resonances at −3.9 and −3.6 ppm, respectively, whereas the samples with little or no water of crystallization show their NMR signals at −2.0 and 2.9 ppm [45]. However, all the investigated catalysts were pretreated under controlled humidity conditions prior to collecting their NMR spectra and, therefore, the shifts of the NMR signal cannot be explained by the change of their degree of hydration. Another possible reason for the shift of the 31 P NMR signal may be the interaction between heteropolyanions and the surface groups of the support [8]. Similar observation (the NMR signal at −2.5 ppm) was reported for HPMo supported on silica [6]. In the case of HPMo deposited on Na-mt, HPMo may be attached (anchored) to the surface of montmorillonite according to the manner mentioned above, i.e. (SiOH2 )+ (H2 PMo12 O40 )− , (Na)+ (H2 PMo12 O40 )− . After spreading HPMo on Na-Mt, due to alkaline nature of this support, the (PMo12 O40 )3− anions may be partially decomposed as revealed by FT-IR analysis. According to Refs. [1,46–48], in alkaline solution anions with the Kegginlike structure may convert into anions such as (PMo11 O39 )7− and (P2 Mo18 O62 )6− , which give NMR signal at −2 and −2.5 ppm, respectively. In the spectrum of HPMo/Na-Mt/m a broad asymmetric signal, being a superposition of two NMR signals, with maximum at ca. −2.4 ppm is observed. The use of the ultrasound treatment results in more dispersed HPMo in HPMo/Na-mt/m and facilitates the decomposition of the Keggin structure of HPMo leading to the formation of defective anions. Thus, the subsequent downward shift of the 31 P NMR signal points out to a partial degradation of heteropolyanions in the catalysts derived from Na-Mt. Hence, the NMR data confirm the above-mentioned results based on the analysis of the UV–vis and FT-IR spectra (Figs. 4 and 5(a–c)). 3.7. Thermal stability

Fig. 4. UV–vis reflectance spectra of HPMo before and after deposition on various montmorillonites (Mt-K10, H-Mt, Na-Mt) via conventional and modified impregnations.

3.6. Nuclear magnetic angle spinning resonance Fig. 6 shows the 31 P MAS-NMR spectra of the hybrid catalysts prepared via conventional and modified impregnations. In the spectra of HPMo deposited on H-exchanged montmorillonites (H-Mt and Mt-K10) only one broad resonance line at −4.3 ppm is recorded. This signal is characteristic of phosphorous with the symmetry observed for (PMo12 O40 )3− with the Keggin-like structure [42]. This confirms that the structure of heteropolyanions is preserved in the hybrid catalysts derived from both acid-treated montmorillonites. In the 31 P MAS-NMR spectrum of HPMo deposited on Naexchanged montmorillonite by conventional impregnation, a broad

The montmorillonites exhibit four degradation steps associated with the elimination of the water adsorbed on the surface and the interlayer water associated with cations (Na+ , H+ ) (endothermic DSC peak at ca. 100 ◦ C), the dehydroxylation of montmorillonite layers (endothermic DSC peak at ca. 700 ◦ C), the decomposition of dehydrated and defective montmorillonite structure (endothermic DSC peak at ca. 900 ◦ C), the crystallization of new phases from the residue after decomposition of montmorillonite (exothermic DSC at ca. 950 ◦ C). Generally TG and DSC profiles for the hybrid catalysts are similar to those corresponding to appropriate montmorillonites. Thermal effects associated with the decomposition of HPMo are obscured by the degradation steps of montmorillonite. Only for HPMo/Mt-K10 and HPMo/Mt-K10/m the exothermic DCS peak at ca. 390 ◦ C attributed to the decomposition of the compounds with the Keggin-like structure (HPMo) can be observed. Similar DSC peak for pure HPMo is observed at ca. 440 ◦ C. It shows that HPMo (anchored and loosely deposited) dispersed throughout Mt-K10 is less stable than pure HPMo, which is in agreement with the results reported for HPMo deposited on silica [49]. 3.8. Interaction with pyridine Pyridine sorption monitored by FT-IR was employed to characterize the acidity (type of acidic sites and their strength) of the catalysts derived from HPMo and acid-treated montmorillonites (Mt-K10 and H-Mt). For comparison, the acidity of Mt-K10 and H-Mt was also characterized. Among selected catalysts, only two samples namely HPMo/Mt-K10/m and HPMo/H-Mt were able to

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

(a)

(b)

790

199

(c)

790 810

1070 1040

960

1070 1020

HPMo/Mt-K10/m Absorbance (a.u.)

Absorbance (a.u.)

800 965 880

1070

HPMo/Mt-K10 Mo-Oc-Mo 780

Mo=Od 960 P-Oa 1070

790 810 965

870

HPMo/H-Mt 1070

Mo=Od

Mo-Ob-Mo 870

P-Oa

1000

900

800

Wavenumber (cm-1)

790 770 810

1070 1040 1000 940 900

Mo-Oc-Mo

860

P-Oa

620

HPMo/Na-Mt Mo-Oc-Mo

Mo=Od

Mo-Ob-Mo

Mo-Ob-Mo

HPMo

HPMo 1100

HPMo/Na-Mt/m 620

HPMo/H-Mt/m Absorbance (a.u.)

1040

860 940 900

870 910

910

1070

1000

965

870

790 770 810

700

600

1100

1000

900

800

Wavenumber (cm-1)

HPMo 700

600

1100

1000

900

800

Wavenumber (cm-1)

700

600

Fig. 5. The subtracted FT-IR spectra of the hybrid catalysts derived from HPMo and various montmorillonites (a) Na-Mt (b) H-Mt and (c) Mt-K10. The FT-IR spectrum of H3 PMo12 O40 ·14H2 O is also displayed.

prepare for FT-IR analysis of pyridine sorption (Fig. S2 Supplementary material). It is well known from the literature that commercially available Mt-K10 [17,26] has Brønsted and Lewis acidic sites. FT-IR study

- 4.3

HPMo/Mt-K10 - 4.3

HPMo/Mt-K10/m Intensity (a.u.)

- 4.3

HPMo/H-Mt

- 4.3

HPMo/H-Mt/m - 2.7 - 4.9

HPMo/Na-Mt

of pyridine sorption confirms the presence of both types of acidic sites in both acid-treated montmorillonites, Mt-K10 and H-Mt. The concentration of acidic sites is 78 and 51 ␮mol g−1 for Mt-K10 and H-Mt, respectively. The contribution of Brønsted acidic sites in HMt is comparable to that of Lewis ones, whereas the contribution of Lewis acidic sites in Mt-K10 is slightly higher than Brønsted ones (Table 3). The strength of Lewis sites in Mt-K10 and H-Mt is similar, whereas the strength of Brønsted sites of H-Mt is higher than those occurring in Mt-K10 (Fig. S3(a, b) Supplementary material). The addition of HPMo to Mt-K10 increases the number of acidic sites as expected and reported previously in the literature for various heteropolyacids deposited on montmorillonite [26,50,51]. The concentration of acidic sites is 171 ␮mol g−1 in HPMo/Mt-K10/m (Table 3). The increase in the amount of acidic sites is associated with Brønsted sites (from 32 up to 125 ␮mol g−1 ), whereas the concentration of Lewis sites in Mt-K10 (49 ␮mol g−1 ) and HPMo/Mt-K10/m (46 ␮mol g−1 ) are alike. The strength of Brønsted acidic sites of HPMo/Mt-K10 is higher than those occurring in MtK10 (Fig. S3(a, b) Supplementary material) probably because most of them derives from the introduced HPMo. Whereas the strength of Lewis sites of HPMo/Mt-K10/m is lower than those occurring in Mt-K10. Similarly, the addition of HPMo to H-Mt also increases the concentration of acidic sites mainly Brønsted ones. The concentration of acidic sites increases from 51 ␮mol g−1 in H-Mt to 128 ␮mol g−1 in HPMo/H-Mt (Table 3). In contrast to Mt-K10 and HPMo/MtK10/m, the amount of Lewis sites in H-Mt and HPMo/H-Mt is

- 2.4 Table 3 Concentration of acidic sites determined by means of pyridine sorption at 170 ◦ C followed by FT-IR analysis. Activation conditions: temperature (320 ◦ C), vacuum (10−3 Torr), time (40 min).

HPMo/Na-Mt/m 15

10

5

0

-5

-10

-15

Chemical shift (ppm) Fig. 6. The solid-state 31 P MAS NMR spectra of the hybrid catalysts derived from HPMo and various montmorillonites (Na-Mt, H-Mt, Mt-K10) via conventional and modified impregnations.

Sample

Total number of acidic sites (␮mol g−1 )

Number of Brønsted acidic sites (␮mol g−1 )

Number of Lewis acidic sites (␮mol g−1 )

H-Mt HPMo/H-Mt Mt-K10 HPMo/Mt-K10/m

51 128 78 171

24 117 32 125

27 11 46 46

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

3.9. Ethanol conversion The catalytic properties of the hybrid catalysts derived from HPMo and montmorillonites (H-Mt, Mt-K10 and Na-Mt) were evaluated in the conversion of ethanol. Ethene, diethyl ether and acetaldehyde (besides water and hydrogen) are the products of ethanol conversion over HPMo deposited onto H-exchanged montmorillonites (H-Mt, Mt-K10), whereas ethene and acetaldehyde are produced in the presence of HPMo spread on Na-exchanged montmorillonite (Na-Mt). Ethene and diethyl ether are formed on the acidic sites via dehydration of ethanol, whereas acetaldehyde is produced on the redox sites through the dehydrogenation of ethanol. According to Refs. [2,52,53], the catalytic dehydration of ethanol involves protonated ethanol species. Ethene and diethyl ether are formed from a protonated monomer, i.e. (C2 H5 OH)H+ , and a protonated dimer, i.e. (C2 H5 OH)2 H+ , respectively. Alcohol dehydration does not discriminate Brønsted and Lewis acidic sites. Lewis acidic sites may be affected by the water produced in the reaction, turning into Brønsted acidic sites [54]. Acetaldehyde is formed on the redox sites such as transition metal ions, i.e. Mo6+ in HPMo and Fe3+ in montmorillonites [2,20,55]. 3.9.1. Catalytic performance of the systems derived from HPMo and Mt-K10 or H-Mt Fig. 7(a, b) demonstrates that the order of activity (conversion at 300 ◦ C) is the following: HPMo < HPMo + H-Mt < HPMo + MtK10 < H-Mt < Mt-K10. Note that the activity for bulk HPMo is evaluated for HPMo mixed physically with SiO2 using HPMo contents corresponding to HPMo loadings (9 or 33 wt%) in the hybrid catalysts. Deposition of HPMo onto acid-treated montmorillonites (MtK10 or H-Mt) results in the systems, which are more active than bulk heteropolyacid but they are less active than starting montmorillonites. It appears that despite various impregnation methods, the performance of the catalysts in series, i.e. HPMo + Mt-K10 or HPMo + H-Mt, is alike. Fig. 8a illustrates that pure HPMo shows both the acidic and redox reactivity in ethanol conversion. In the presence of bulk HPMo, ethanol converts at 300 ◦ C mainly into ethene (with the selectivity, Se ∼ 89%) and acetaldehyde (with the selectivity, Sac ∼ 11%). In contrast to Ref. [5], we do not notice in the presence of bulk HPMo the transition from redox reactivity (giving

(a) Conversion (%)

different, i.e. 27 and 11 ␮mol g−1 , respectively. The acidic sites (Brønsted and Lewis) of HPMo/H-mt have a comparable strength as those occurring in H-Mt (Fig. S3(a, b) Supplementary material). The acidity estimated via pyridine sorption, i.e. 171 and 128 ␮mol g−1 for HPMo/Mt-K10/m and HPMo/H-Mt, respectively, are markedly lower than the nominal values, i.e. 523 and 175 ␮mol g−1 for HPMo/Mt-K10/m and HPMo/H-Mt, respectively. According to Ref. [25], steric factors may slow down the penetration of large pyridine molecule into the bulk structure of HPMo. The secondary structure of HPMo may also play a role in the process of pyridine sorption. Thus, the sorption process seems to be restricted to some extent and therefore not all acidic sites may be available and some of them may not be detected by pyridine. In addition, in the hybrid catalysts the access of pyridine to acidic sites may be additionally limited by the fact that HPMo species are deeply located inside the cavities. The difference between measured and nominal acidities is higher for HPMo/Mt-K10/m than for HPMo/HMt. It indicates that the penetration of pyridine is more facilitated through HPMo/H-Mt (less porous and with lower HPMo content, 9 wt%) than in the case of HPMo/Mt-K10/m (more porous and with higher HPMo content, 33 wt%).

100

Mt-K10 HPMo/Mt-K10 HPMo/Mt-K10/m HPMo

80 60 40

20 0 100

150

200

250

300

350

300

350

Temperature (oC)

(b) Conversion (%)

200

100

H-Mt HPMo/H-Mt HPMo/H-Mt/m HPMo

80 60 40

20 0 100

150

200

250

Temperature

(oC)

Fig. 7. The conversion of ethanol in the presence of the hybrid catalysts derived from HPMo and H-exchanged montmorillonites (a) Mt-K10 and (b) H-Mt. For comparison, the results for bulk HPMo (mixed physically with SiO2 ) and montmorillonites (MtK10 and H-Mt) are also shown. Two concentrations of bulk HPMo, i.e. 9 and 33 wt%, correspond to the HPMo contents in the hybrid catalysts derived from H-Mt and MtK10, respectively. The reaction was carried out for 30 min at given temperature (in the temperature range of 100–300 ◦ C) under helium flow (30 cm3 min−1 ) saturated with ethanol vapor (5.7 mol%). The heating rate was 2 ◦ C min−1 .

acetaldehyde) into acidic reactivity (giving ethene) observed typically at lower and higher temperatures, respectively. Both products are formed within the whole reaction temperature range used in the tests. However, ethene is a dominant product in the presence of pure HPMo. In addition, as the reaction temperature increases above 250 ◦ C the selectivity to ethene increases at the expense of that to acetaldehyde. Fig. 8b and c show a quite different selectivity pattern for acidtreated montmorillonites, Mt-K10 and H-Mt. Ethene and diethyl ether are the main products in the presence of both supports. Within the whole reaction temperature range, the contribution of acetaldehyde is very low as evidenced by the selectivity, Sac < 10%. As the reaction temperature rises from ca. 200 up to 300 ◦ C, the selectivity to ethene strongly increases reaching ca. 95%, whereas the selectivity to diethyl ether decreases down to ca. 1%. This selectivity pattern agrees well with Ref. [56] indicating that the ethanol-to-diethyl ether reaction is thermodynamically favored at lower temperatures, whereas the ethanol-to-ethene reaction is favored at higher temperatures. It demonstrates that acid-treated montmorillonites have mainly acidic sites active for the conversion of ethanol towards ethene and diethyl ether. The comparison of activities for Mt-K10 and H-Mt (Fig. S4 Supplementary material) shows higher (by 10–15%) conversion of ethanol over the former (Mt-K10). These samples differ in the content of acidic sites (Table 3). The values of TOF calculated for the reaction at 250 ◦ C are 359 and 355 s−1 for Mt-K10 and H-Mt, respectively. Note that these TOF values are calculated using the

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

(a)

100

Selectivity (%)

80

HPMo 60

Ethene Acetaldehyde

40

20 0 100

150

200

250

300

350

300

350

300

350

Temperature (oC)

(b)

100

Selectivity (%)

80

Mt-K10 Ethene

60

Diethyl ether Acetaldehyde

40

20 0 100

150

200

250

Temperature (oC)

(c)

100

Selectivity (%)

80

H-Mt 60

Ethene

40

Diethyl ether Acetaldehyde

20 0 100

150

200

250

Temperature

(oC)

Fig. 8. The selectivity to ethene (Se ), diethyl ether (See ) and acetaldehyde (Sac ) – the products of ethanol conversion over (a) HPMo (b) Mt-K10 and (c) H-Mt. The reaction was carried out for 30 min at given temperature (in the temperature range of 100–300 ◦ C) under helium flow (30 cm3 min−1 ) saturated with ethanol vapor (5.7 mol%). The heating rate was 2 ◦ C min−1 .

number of acidic sites determined by pyridine sorption, which provides underestimated data. In the view of this, the observed slight difference in reactivity of acidic sites coming from Mt-K10 and H-Mt is negligible. Similar acidic reactivity may also indicate that the transport of ethanol through the porous structure of acidtreated montmorillonites (more active) and also HPMo-containing catalysts (less active) do not play an important role. In addition, the conversion of ethanol was performed applying high (constant) value of GHSV (6000 h−1 ) and using small catalyst particle dimensions (0.2–0.4 mm), which eliminate the effects of external and internal mass transfer. Dehydration products (ethene and/or diethyl ether) are dominating over HPMo and acid-treated montmorillonites tested separately as catalysts in ethanol conversion reaction. Thus, we may expect that the combination of heteropolyacid and these

201

montmorillonites should result in the systems with enhanced acidic reactivity. However, the catalytic data clearly reveal that the catalysts derived from HPMo and both montmorillonites (Mt-K10 or H-Mt) have lower acidic function, whereas their redox function in ethanol conversion is preferable (compare Fig. 9 with Fig. 7). Fig. 9(a–c) shows that in the presence of the catalysts derived from HPMo and Mt-K10 ethanol converts at 300 ◦ C mainly into ethene, diethyl ether (with the selectivity, (Se + See ) ∼ 69%) and acetaldehyde (with the selectivity, Sac ∼ 31%). The production of acetaldehyde over these catalysts increases threefold in comparison with that of pure HPMo or Mt-K10. The catalysts derived from HPMo and H-Mt produce even more acetaldehyde than those derived from HPMo and Mt-K10 (compare Fig. 9c and e), especially at temperatures above 200 ◦ C. The selectivity to acetaldehyde at 300 ◦ C is roughly 38% for HPMo/H-Mt and HPMo/H-Mt/m. This improved conversion of ethanol towards acetaldehyde seems to be a result of spreading of HPMo on Mt-K10 or H-Mt, which exposes the Keggin units thus facilitating the access to the redox sites. It is well known that ethanol molecules are able to penetrate the bulk of HPMo. However, it seems that over the hybrid catalysts ethanol prefers to convert rather on the exposed catalytic sites than on those located in the bulk of HPMo. A significant difference in dehydrogenation/dehydration selectivity ratio for bulk HPMo and the catalysts (illustrated in Fig. S5 in Supplementary material) supports this assumption. In addition, the dehydrogenation/dehydration selectivity ratio differs for the catalysts depending on whether they derived from Mt-K10 or H-Mt. It seems that HPMo is less effectively dispersed on the surface of Mt-K10 than on the surface of H-Mt. Thus, the distribution of the catalytic sites on the surface of the catalysts derived from Mt-K10 and H-Mt differs to some extent. Due to a better dispersion of HPMo on H-Mt the redox sites may be more exposed in these catalysts than in those derived from HPMo and Mt-K10. Thus, the selectivity to dehydration products over the hybrid catalysts is less pronounced than expected as the catalysts have also a reasonable amount of the redox sites, which are responsible for the conversion of ethanol towards acetaldehyde. The catalytic data indicate that the selectivity to acetaldehyde increases after spreading of HPMo onto acid-treated montmorillonites. On the other hand, no significant difference in the behavior of the catalysts prepared either via conventional or via modified impregnation (Fig. S5 in Supplementary material) suggests that the access to the redox sites provided by HPMo in both catalysts in series is similar. The conversion of ethanol towards diethyl ether over the hybrid catalysts seems to be rather provided by the acidic sites (in particular, Brønsted ones) of montmorillonite component than by those offered by HPMo. This assumption may be supported by the fact that diethyl ether is produced over the catalysts and acidtreated montmorillonites but it is not detected in the presence of bulk HPMo. In the hybrid catalysts the surface of montmorillonite is mostly covered with HPMo species. The presence of HPMo deposit could make the access to the acidic sites coming from acidtreated montmorillonites more difficult. Therefore, the selectivity to diethyl ether over the hybrid catalysts (Fig. 9(c, f)) differs from that observed for initial montmorillonites. The catalytic data would be consistent with pyridine sorption results if we assume that all acidic sites detected by pyridine are also involved in the conversion of ethanol. For example, the selectivity to dehydration products is 69% and 63% (at 300 ◦ C) (Fig. 9(a, c, d, f)) and the concentration of acidic sites (Brønsted and Lewis) is 171 and 128 ␮mol g−1 for HPMo/Mt-K10/m and HPMo/H-Mt, respectively. It indicates that the acidic reactivity of these catalysts as well as their acidity do not differ so much. Similar trend is observed for acid-treated montmorillonites. The selectivity to dehydration products is 97% and 96% (at 300 ◦ C) (Fig. 9(a, c, d, f)) and the number of acidic sites is 78 and 51 ␮mol g−1 for Mt-K10

202

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

Selectivity to acetaldehyde (%)

(b)

100

Selectivity to ethene (%)

(a)

80 60 40

Mt-K10 HPMo/Mt-K10

20

HPMo/Mt-K10/m HPMo

0 100

150

200

250

300

350

100 Mt-K10 HPMo/Mt-K10

80

HPMo/Mt-K10/m

HPMo

60 40

20 0 150

100

Temperature (oC)

(c)

(d)

100

Selectivity to ethene (%)

Selectivity to ether (%)

HPMo/Mt-K10/m

60 40

20 0 150

200

250

350

300

300

350

HPMo/H-Mt

80

HPMo/H-Mt/m HPMo

60 40

20 0

350

100

150

Temperature (oC)

200

250

Temperature (oC)

(e)

(f)

100

100

H-Mt

H-Mt

HPMo/H-Mt

80

Selectivity to ether (%)

Selectivity to acetaldehyde (%)

300

H-Mt

HPMo/Mt-K10

100

250

100

Mt-K10

80

200

Temperature (oC)

HPMo/H-Mt/m HPMo

60 40

20 0

HPMo/H-Mt

80

HPMo/H-Mt/m

60 40

20 0

100

150

200

250

300

350

Temperature (oC)

100

150

200

250

300

350

Temperature (oC)

Fig. 9. The selectivity to ethene (Se ), diethyl ether (See ) and acetaldehyde (Sac ) – the products of ethanol conversion over the hybrid catalysts derived from HPMo and H-exchanged montmorillonites: (a–c) Mt-K10 and (d–f) H-Mt via conventional and modified impregnations. The reaction was carried out for 30 min at given temperature (in the temperature range of 100–300 ◦ C) under helium flow (30 cm3 min−1 ) saturated with ethanol vapor (5.7 mol%). The heating rate was 2 ◦ C min−1 .

and H-Mt, respectively. The observed variances may be explained by the fact that the sorption of pyridine is not facile as ethanol. We may suppose that more acidic sites could be accessible for ethanol molecules in the catalytic reaction than for pyridine in the process of sorption. 3.9.2. Catalytic performance of the systems derived from HPMo and Na-Mt Na-Mt itself is almost inactive in ethanol conversion reaction. The only conversion recorded at 300 ◦ C is ca. 15% (Fig. 10a). The lack of diethyl ether in the products over Na-Mt suggests that its active sites, in particular the acidic sites differ from those occurring in H-Mt and Mt-K10 and therefore the mechanism of ethanol dehydration may be also different. The selectivity to ethene (Se ∼ 16%) is much lower than the selectivity to acetaldehyde (Sac ∼ 84%), which implies that redox properties are dominating in Na-Mt. The catalysts derived from HPMo and Na-Mt are the least active among the studied hybrid catalysts. In contrast to the catalysts derived from HPMo and acid-treated montmorillonites the impregnation method influences the catalytic performance of HPMo + Na-Mt series. The conversion at 300 ◦ C is ca. 21% and 26% over HPMo/Na-mt and HPMo/Na-mt/m, respectively (Fig. 10a). The catalysts are less active than bulk HPMo but only slightly more active than Na-mt.

Fig. 10(b, c) illustrates that less amounts of ethene is formed in the presence of HPMo/Na-Mt and HPMo/Na-mt/m than over HPMo but more than in the presence of Na-Mt. This observation shows the differences in acidic properties of the samples. The acidic sites (in particular protons) coming from the introduced HPMo seems to be partially blocked by the interaction with Na-Mt. According to Ref. [57], the uptake of alcohols (ethanol) by heteropolyacid (HPW) may be suppressed by the replacement of protons by other cations, for instance Na+ . No diethyl ether is observed in the presence of HPMo deposited on Na-Mt, consequently as diethyl ether is not produced over HPMo and Na-Mt. A lack of ether in the products of methanol conversion was also reported for HPMo supported on silica [5]. Similarly to HPMo mixed with acid-treated montmorillonites, the catalysts derived from Na-Mt have also more pronounced redox function than acidic function. It seems that ethanol molecules convert mainly on the redox sites, which are easily available as spreading HPMo on Na-Mt results in the exposure of heteropolyanions. In addition, Na-Mt provides its own redox sites. The performance of both above-discussed catalysts is different as illustrated in Fig. 10(a–c). HPMo/Na-mt/m is more active than HPMo/Na-mt, probably because of higher concentration of the redox sites provided by defective species formed via degradation of Keggin anions. According to Ref. [49], the degradation

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

(a)

100

Na-Mt HPMo/Na-Mt HPMo/Na-Mt/m HPMo

80

Conversion (%)

203

60 40

20 0 100

150

200

250

300

350

(b)

100

Selectivity to ethene (%)

Temperature (oC)

80

Na-Mt HPMo/Na-Mt HPMo/Na-Mt/m HPMo

60 40

20 0

100

150

200

250

300

350

Temperature (oC)

Selectivity to acetaldehyde (%)

(c)

100 Fig. 11. The EPR spectra of HPMo/Na-mt and HPMo/Na-mt/m (a) after reduction and (b) after reoxidation.

80 Na-Mt

60

HPMo/Na-Mt HPMo/Na-Mt/m

40

HPMo

20 0

100

150

200

250

300

350

Temperature (oC) Fig. 10. The conversion of ethanol (a) and the selectivity to ethene (b) and to acetaldehyde (c) in the presence of the hybrid catalysts derived from HPMo and Na-exchanged montmorillonite (Na-Mt). For comparison, the results for bulk HPMo (mixed physically with SiO2 ) and Na-Mt are also shown. The concentration of bulk HPMo, i.e. 9 wt%, corresponds to the HPMo content in the hybrid catalysts derived from Na-Mt. The reaction was carried out for 30 min at given temperature (in the temperature range of 100–300 ◦ C) under helium flow (30 cm3 min−1 ) saturated with ethanol vapor (5.7 mol%). The heating rate was 2 ◦ C min−1 .

of HPMo leading to the formation of MoO3 facilitates the change from acidic reactivity to redox reactivity. With rising reaction temperature, the selectivity to ethene increases up to ca. 34% and 16% and the selectivity to acetaldehyde decreases down to ca. 66% and 84% over HPMo/Na-mt and HPMo/Na-mt/m, respectively. At 300 ◦ C Sac /Se (dehydrogenation/dehydration) ratio is higher for HPMo/Na-Mt/m (5.25) than HPMo/Na-Mt (2.01). It implies that HPMo/Na-Mt/m has better redox reactivity than HPMo/NaMt. As mentioned above, modified impregnation leads to higher degradation of HPMo deposited on Na-Mt (revealed by NMR) and certainly more efficiently suppresses its acidic function and therefore HPMo/Na-Mt/m has higher redox reactivity than HPMo/Na-Mt. Similar observation was reported for HPW deposited on Na-Mt [19], the selectivity to acetaldehyde was higher for HPW/Na-Mt/m than for HPW/Na-Mt.

ESR spectroscopy was applied to evaluate the influence of the impregnation method on the redox properties of HPMo deposited on Na-Mt. The ESR signal indicates the presence of Mo(V) formed from Mo(VI) via reduction induced by heating the HPMo-containing samples at 300 ◦ C (105 min) in vacuum (8 × 10−4 Torr). According to Refs. [58,59], in Mo-containing heteropolyacid with the Keggin structure Mo(V) centres are formed from Mo(VI) in vacuum by the loss of surface oxygen atoms and charge transfer to neighboring Mo. Fig. 11(a, b) shows that the intensity of ESR signal attributed to Mo(V) is slightly higher for HPMo/Na-Mt/m than for HPMo/NaMt. Thus, the degree of reduction is higher for HPMo/Na-Mt/m. This result suggests that well-dispersed Keggin anions are easily reducible compared to the aggregated HPMo species. The difference in reducibility is relevant to the redox properties of the studied catalysts. The ability of the samples to reoxidize was also checked. After reoxidation, ESR signal is significantly higher for the sample prepared via conventional impregnation than for the sample prepared via modified impregnation (Fig. 11a), which confirms that the reoxidation process is slower for HPMo/Na-Mt than for HPMo/Na-Mt/m. In the case of the sample prepared via modified impregnation, the intensity of its ESR signal decreases after an air exposure. It seems that higher dispersion of heteropolyacid on Na-Mt in HPMo/Na-Mt/m provides better oxygen access to reduced species derived from HPMo. So reoxidation undergoes easier in the case of HPMo/Na-Mt/m than for HPMo/Na-Mt. All the catalysts after being tested in ethanol conversion reaction become black in color probably as a result of being covered with carbonaceous deposit. Judging by appearance, reusing these catalysts is not recommended. The formation of coke on the surface of the catalysts certainly blocks the pores and decreases their specific surface areas. For instance, a 50-% drop in mesopore volume

204

A. Pacuła et al. / Applied Catalysis A: General 498 (2015) 192–204

and an eightfold decrease in specific surface area are observed for HPMo/H-Mt after being tested in the ethanol conversion.The deposition of coke on the surface of the catalysts restricts the access to their catalytic sites and therefore suppresses their catalytic activities. For instance, the conversion of ethanol at 300 ◦ C for fresh and reused Na-Mt drops significantly from ca. 15 to 2%, respectively. The presence of carbonaceous layer on the surface of the catalysts not only may reduce their activity but also may modify their selectivity. For instance, no ethene is formed over used NaMt, while the selectivity to ethene for fresh Na-Mt is ca. 16%. The selectivity to acetaldehyde increases from 84% for fresh Na-Mt up to 100% for used Na-Mt. It implies that redox (basic) sites are dominating on the surface of used Na-Mt, which may result from the presence of carbonaceous deposit offering basic surface sites. Similar performance is observed for HPW (12-tungstophosphoric acid hydrate) deposited on Na-Mt by conventional impregnation [19]. The conversion of ethanol at 300 ◦ C for fresh and reused HPW/NaMt drops significantly from ca. 73 to 4%, respectively. Similarly to Na-Mt, the selectivity to acetaldehyde for fresh and used HPW/NaMt increases from 3 up to 16%, respectively. In contrast to fresh HPW/Na-Mt (See ∼ 26%), no diethyl ether is formed in the presence of used HPW/Na-Mt. The selectivity to ethene for fresh and used HPW/Na-Mt is 71 and 84%, respectively. 4. Conclusions The application of H-exchanged forms of montmorillonite (H-Mt, Mt-K10) helps to preserve the structure of Keggin anions, while Na-exchanged montmorillonite (Na-Mt) (especially after deposition aided by ultrasonication) facilitates the degradation of Keggin anions in the hybrid catalysts derived from 12molybdophosphoric acid and montmorillonites. Spreading HPMo on montmorillonites provides better access to the redox sites associated with heteropolyacid in comparison to bulk HPMo. We cannot exclude that the acidic properties may also be suppressed via the interaction between protons coming from HPMo and the surface (Si–OH) groups of montmorillonite. The acidic sites may be additionally obscured by replacement of protons by sodium ions coming from Na-Mt. Acknowledgments This work was financially supported by the Polish Ministry of Science and Higher Education within the research project 4 T08D 051 24. The IR measurements were carried out with the equipment purchased thanks to the financial support of the European Regional Development Fund in the framework of the Polish Innovation Economy Operational Program (contract no. POIG.02.01.00-12-023/08). Adam Gaweł (AGH-University of Science and Technology) and Zofia Czuła, Małgorzata Ruggiero-Mikołajczyk (Jerzy Haber Institute of Catalysis and Surface Chemistry, Polish Academy of Sciences) are gratefully acknowledged for XRD and nitrogen sorption, respectively. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.apcata. 2015.03.030.

[2] [3] [4] [5] [6] [7] [8] [9]

[10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57]

References [1] M.T. Pope, Heteropoly and Isopoly Oxometalates, Springer-Verlag, Berlin, Heidelberg, 1983.

[58] [59]

T. Okuhara, N. Mizuno, M. Misono, Adv. Catal. 41 (1996) 113–252. Y. Izumi, K. Matsuo, K. Urabe, J. Mol. Catal. 18 (1983) 299–314. S. Kasztelan, E. Payen, J.B. Moffat, J. Catal. 125 (1990) 45–53. M. Fournier, A. Aouissi, C. Rocchiccioli-Deltcheff, J. Chem. Soc. Chem. Commun. (1994) 307–308. T.-H. Chang, J. Chem. Soc. Faraday Trans. 91 (1995) 375–379. S. Damyanova, J.L.G. Fierro, Appl. Catal. A: Gen. 144 (1996) 59–77. P.G. Vázquez, M.N. Blanco, C.V. Cáceres, Catal. Lett. 60 (1999) 205–215. N. Ballarini, F. Candiracci, F. Cavani, H. Degrand, J.-L. Dubois, G. Lucarelli, M. Matgotti, A. Patinet, A. Pigamo, F. Trifirò, Appl. Catal. A.: Gen. 325 (2007) 263–269. C. Camejo-Abreu, V. Tabernero, M.D. Alba, T. Cuenca, P. Terreros, J. Mol. Catal. A: Chem. 393 (2014) 96–104. S.K. Bharadwaj, P.K. Boruah, P.K. Gogoi, Catal. Commun. 57 (2014) 124–128. L. Zatta, E. Jose, M. Paiva, M.L. Corazza, F. Wypych, Energy Fuels 28 (2014) 5834–5840. G.D. Yadav, M.S. Krishnan, Ind. Eng. Chem. Res. 37 (1998) 3358–3365. G.D. Yadav, N.S. Doshi, Org. Process Res. Dev. 6 (2002) 263–272. V.V. Bokade, G.D. Yadav, Ind. Eng. Chem. Res. 48 (2009) 9408–9415. S.K. Bhorodwaj, D.K. Dutta, Appl. Clay Sci. 53 (2011) 347–352. G.D. Yadav, Catal. Surv. Asia 9 (2005) 117–137. C.H. Zhou, Appl. Clay Sci. 53 (2011) 87–96. ˛ A. Pacuła, K. Pamin, A. Zieba, J. Kry´sciak-Czerwenka, Z. Olejniczak, E.M. Serwicka, A. Drelinkiewicz, Appl. Clay Sci. 95 (2014) 220–231. T. Cseri, S. Békássy, F. Figueras, E. Cseke, L.C. Menorval, R. Dutartre, Appl. Catal. A 132 (1995) 141–155. F. Marme, G. Coudurier, J.C. Védrine, Microporous Mesoporous Mater. 22 (1998) 151–163. G.D. Yadav, V.V. Bokade, Appl. Catal. A 147 (1996) 299–323. C. Trolliet, G. Coudurier, J.C. Védrine, Top. Catal. 15 (2001) 73–81. F. Rouquerol, J. Rouquerol, K. Sing, Adsorption by Powders and Porous Solids: Principles, Methodology and Applications, Academic Press, London, 1999. J.G. Highfield, J.B. Moffat, J. Catal. 89 (1984) 185–195. ´ ˙ M. Sliwa, K. Samson, M. Ruggiero-Mikołajczyk, A. Zelazny, R. Grabowski, Catal. Lett. 144 (2014) 1884–1893. File No. 90636 of Inorganic Crystal Structure Database (ICSD). File No. 31128 of Inorganic Crystal Structure Database (ICSD). V.V. Bokade, G.D. Yadav, Appl. Clay Sci. 53 (2011) 263–271. H. Zhao, C.H. Zhou, L.M. Wu, J.Y. Lou, N. Li, H.M. Yang, D.S. Tong, W.H. Yu, Appl. Clay Sci. 74 (2013) 154–162. ´ A. Pacuła, E. Bielanska, A. Gaweł, K. Bahranowski, E.M. Serwicka, Appl. Clay Sci. 32 (2006) 64–72. I. Barshad, Am. Mineral. 35 (1950) 225–238. I. Barshad, Am. Mineral. 34 (1949) 675–684. G.D. Yadav, S.O. Katole, A.K. Dalai, Appl. Catal. A: Gen. 477 (2014) 18–25. M. Fournier, C. Louis, M. Che, P. Chaquin, D. Masure, J. Catal. 119 (1989) 400–414. C. Rocchiccioli-Deltcheff, M. Amirouche, M. Che, J.M. Tatibouët, M. Fournier, J. Catal. 125 (1990) 292–310. ´ ´ ´ A. Bielanski, A. Lubanska, J. Pozniczek, A. Micek-Ilnicka, Appl. Catal. A 238 (2003) 239–250. C. Rocchiccioli-Deltcheff, M. Fournier, R. Franck, R. Thouvenot, Inorg. Chem. 22 (1983) 207–216. J.F. Garvey, M.T. Pope, Inorg. Chem. 17 (1978) 1115–1118. Y. Wu, X. Ye, X. Yang, X. Wang, W. Chu, Y. Hu, Ind. Eng. Chem. Res. 35 (1996) 2546–2560. K. Brückman, M. Che, J. Haber, J.M. Tatibouet, Catal. Lett. 25 (1994) 225–240. J.B. Black, N.J. Clayden, L. Griffiths, J.D. Scott, J. Chem. Soc. Dalton Trans. 12 (1984) 2765–2767. J. Hu, R.C. Burns, J. Catal. 195 (2000) 360–375. E. Silviani, R.C. Burns, J. Mol. Catal. A: Chem. 219 (2004) 327–342. J.B. Black, N.J. Clayden, P.L. Gai, J.D. Scott, E.M. Serwicka, J.B. Goodenough, J. Catal. 106 (1987) 1–15. G.B. McGarvey, J.B. Moffat, J. Mol. Catal. 69 (1991) 137–155. J.A. Rob van Veen, O. Sudmeijer, C.A. Emeis, H. de Wit, J. Chem. Soc. Dalton Trans. (1986) 1825–1831. L. Pettersson, I. Andersson, L.-O. Öhman, Inorg. Chem. 25 (1986) 4726–4733. C. Rocchiccioli-Deltcheff, A. Aouissi, S. Launay, M. Fournier, J. Mol. Catal. A: Chem. 114 (1996) 331–342. G.D. Yadav, N. Kirthivasan, Appl. Catal. A: Gen. 154 (1997) 29–53. G.D. Yadav, N.S. Doshi, J. Mol. Catal. A 194 (2003) 195–209. D. Varisli, T. Dogu, G. Dogu, Chem. Eng. Sci. 62 (2007) 5349–5352. K.Y. Lee, T. Arai, S. Nakata, S. Asaoka, T. Okuhara, M. Misono, J. Am. Chem. Soc. 114 (1992) 2836–2842. A.M. Alsalme, P.V. Wiper, Y.Z. Khimyak, E.F. Kozhevinkova, I.V. Kozhevnikov, J. Catal. 276 (2010) 181–189. J.B. McMonagle, J.B. Moffat, J. Catal. 91 (1985) 132–141. W. Alharbi, E. Brown, E.F. Kozhevnikova, I.V. Kozhevnikov, J. Catal. 319 (2014) 174–181. T. Okuhara, S. Tatematsu, K.Y. Lee, M. Misono, Bull. Chem. Soc. Jpn. 62 (1989) 717–723. P. Rabette, D. Olivier, J. Less Common Met. 36 (1974) 299–304. N. Mizuno, K. Katamura, Y. Yoneda, M. Misono, J. Catal. 83 (1983) 384–392.