Study on the influence of channel structure properties in the dehydration of glycerol to acrolein over H-zeolite catalysts

Applied Catalysis A: General 429–430 (2012) 9–16

Contents lists available at SciVerse ScienceDirect

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

Study on the influence of channel structure properties in the dehydration of glycerol to acrolein over H-zeolite catalysts Yunlei Gu a , Naiyun Cui a , Qingjun Yu a , Chunyi Li a,∗ , Qiukai Cui b a b

State Key Laboratory of Heavy Oil Processing, China University of Petroleum, Qingdao, 266555, PR China Dagang Petrochemical Company, PetroChina Corporation, Tianjin, 300280, PR China

a r t i c l e

i n f o

Article history: Received 8 December 2011 Received in revised form 20 March 2012 Accepted 22 March 2012 Available online 4 April 2012 Keywords: Glycerol Acrolein Dehydration Channel structure property HZSM-11

a b s t r a c t Systematic studies have been conducted over several selected H-zeolites, namely HZSM-5, H-Beta, HY, nano HZSM-5, HZSM-11 and nano HZSM-11, aimed to investigate influence of the channel structure on catalytic performance for gas phase dehydration of glycerol to acrolein. Compared to H-Beta and HY, improved catalytic performance was discovered over HZSM-5, which demonstrated that H-zeolites with smaller channels, the ones marginally larger than the molecular diameter of glycerol, were preferential for the reaction. HZSM-11, with lower channel complexity, was more likely to obtain superior catalytic performance due to enhanced diffusion. Nano HZSM-11 (300–500 nm) exhibited excellent catalytic performance with 81.6 mol% glycerol conversion and 74.9 mol% acrolein selectivity at GHSV as high as 873 h−1 (TOS = 8 h). BET and TEM experiment results indicated that coke was initially deposited at channel intersections of H-zeolites, and when the channel blockage came up to a certain extent, there arrived the onset of coke deposition on the external surface. © 2012 Elsevier B.V. All rights reserved.

1. Introduction With the rapidly growing energy demand, more attention has been devoted to biomass and the conversion of biomass related materials into fuels and chemicals, such as biodiesel. The advancing development of biodiesel production had provided a surplus supply of glycerol with reasonably lower prices [1], which currently has become a suitable feedstock for production of high value chemicals through catalytic conversions, including oxidation, reduction, etherification, etc. [2]. One of the most promising ways of glycerol application is through double-dehydration to produce acrolein, which is an important chemical intermediate for production of acrylic acid, acrylic acid esters and dl-methionine [3]. Acrolein is now commercially produced by selective oxidation of petroleumderived propylene over mixed metal oxides based on BiMoOx [4]. The glycerol dehydration process is conducted either in liquid phase or in gas phase, with liquid or solid acid as catalyst [5]. In liquid phase reaction, sub- or supercritical conditions are necessary to enhance the catalytic activity of homogeneous catalysts [6–9]. However, the harsh conditions together with the presence of liquid acids will generate an extremely corrosive medium, resulting in high equipment investments and maintenance costs. As a result, it is unfeasible to produce acrolein commercially from glycerol in

∗ Corresponding author. Tel.: +86 532 86981862; fax: +86 532 86981718. E-mail address: [email protected] (C. Li). 0926-860X/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.apcata.2012.03.030

liquid phase due to certain technical and environmental problems, such as reactor anticorrosion, catalyst/reaction mixture separation and waste management. Recently, more and more researchers have been focusing on gas phase reaction over solid acid catalysts [10–46], including phosphates, zeolites, supported heteropolyacids, metal oxides, sulfates, and among others. Although supported heteropolyacids exhibited very promising results, the water solubility and poor thermal stability of heteropolyacids restrict their application. H-zeolites possess large surface area and sufficient acid sites, which are necessary conditions for glycerol dehydration to acrolein. Yoda and Ootawa observed that acrolein was selectively formed over H-MFI zeolite at a relatively lower temperature (353 K) by FT-IR [22]. Recently, Jia et al. investigated catalytic performances of nanocrystalline HZSM-5 and bulk HZSM-5 with similar Si/Al ratio and proposed that H-zeolites with smaller particle size were more suitable for glycerol dehydration [20]. Nanocrystalline zeolites could provide large external surface area and short diffusion distance in the channels of particles, thereby facilitating access to active sites and reducing deactivation. However, currently it is hard to tell the predominant factor that controls the reaction process, external surface area or zeolite channels. Kim et al. conducted the dehydration reaction of glycerol over several H-zeolites, and observed a good linear correlation between glycerol conversion and external surface area of catalysts [23]. Therefore, they concluded that external surface area was the main factor controlling glycerol conversion. The disappeared micropore

10

Y. Gu et al. / Applied Catalysis A: General 429–430 (2012) 9–16

surface area of spent catalysts after 2 h of reaction implied the complete blockage of zeolite channels at the initial stage of reaction, which indicated that channels of zeolites contributed quite little for the reaction. Thus, it is highly desirable for the acid sites on the external surface of crystals to be accessible for the reaction, this way similar catalytic performance should be obtained over H-zeolites with similar acid properties and external surface areas but different channel patterns. In order to determine the influence of different types of zeolite channels over catalytic performance, several H-zeolites, namely HZSM-5, H-Beta and HY, were chosen for the dehydration of glycerol. The reaction over HZSM-11 was also conducted to confirm the influence of channel structure. HZSM-11 has similar pore size but simpler channel structure compared to HZSM-5. In the meantime, effects of inert carrier gas (N2 ) on catalytic behavior have also been investigated. 2. Experimental

diffractograms were recorded from 4◦ to 60◦ at a speed of 5◦ /min. The morphologies of different zeolites were studied by S-4800 SEM (Hitachi Company, Japan) and JEM-2100UHR TEM (JEOL, Japan). The acid properties were obtained by NH3 -TPD. About 0.1 g of catalyst (20–60 mesh), sandwiched by quartz wool, was loaded in a quartz tube. The sample was heated from room temperature to 650 ◦ C with a rate of 10 ◦ C/min under Helium flow (30 mL/min), which was then maintained at 650 ◦ C for another 30 min to ensure complete removal of impurities. The sample was cooled down to 100 ◦ C and saturated with ammonia. Then it was flushed by Helium for 1 h to remove physically adsorbed ammonia, afterwards the sample was heated to 650 ◦ C at a heating rate of 10 ◦ C/min in Helium flow (30 mL/min) and the thermal conductivity detector (TCD) signals were recorded as a function of temperature. The amount of desorbed ammonia was quantified by means of a reference test, using a known amount of ammonia pulse. 2.3. Catalytic reaction

2.1. Catalyst preparation HZSM-5 (noted as bulk HZSM-5, SiO2 /Al2 O3 = 31), H-Beta (SiO2 /Al2 O3 = 26) and HY (SiO2 /Al2 O3 = 12) zeolites were purchased from The Catalyst Plant of Nankai University. Nanocrystalline HZSM-5 (noted as nano HZSM-5, SiO2 /Al2 O3 = 30) was generously provided by Dalian Institute of Chemical Physics. HZSM-11 (SiO2 /Al2 O3 = 50) was prepared through a cation exchange and calcination process with NaZSM-11 as starting material, which was synthesized in our own lab according to the procedures in the literature [47]. Nanocrystalline NaZSM-11 was also synthesized via adjusting crystallization conditions, which was subsequently made into HZSM-11 (noted as nano HZSM-11, SiO2 /Al2 O3 = 50). The SiO2 /Al2 O3 ratios in the brackets were analyzed by X-ray fluorescence spectrometer (XRF). Before use, all zeolites were calcined at 550 ◦ C in air atmosphere for 2 h to remove impurities. All H-zeolites were made into shaped catalysts before reaction, in addition, inert support was introduced by diluting active components with the aim of reducing coke formation. The preparation procedures of shaped catalysts began with certain amount of commercial silica sol (as binder, silica concentration = 40 wt%) dispersed thoroughly into deionized water. Then kaolin clay (as support) and H-zeolite were added into the solution separately with vigorous stirring, an additional 2 h of stirring was performed afterwards to ensure uniformity of the mixture. Later the solid solution was dried at 120 ◦ C for about 3 h, and calcined at 700 ◦ C for another 2 h in air atmosphere. Finally the catalyst bulks were crushed and sieved to 80–180 mesh for later use, known as H-zeolite. The composition of shaped catalysts was 35 wt% H-zeolite, 5 wt% silica and 60 wt% kaolin clay. 2.2. Catalyst characterization Several characterization techniques, including nitrogen adsorption, powder X-Ray diffraction (XRD), scanning electron microscope (SEM), transmission electron microscope (TEM) and temperature programmed desorption of ammonia (NH3 -TPD), were used to determine structural and acid properties of catalysts, and to understand their catalytic performances. The BET surface area and pore volume were measured by Micromeritics ASAP 2020 micropore analyzer using nitrogen adsorption method. Prior to measurement, all samples were evacuated at 200 ◦ C for 4 h at a pressure of 6.7 × 10−5 kPa to ensure complete removal of adsorbed moisture. The micropore surface area, external surface area and micropore volume were calculated by the t-plot method. XRD was used to identify structures of zeolites using X’pert PRO MPD diffractometer (PANalytical Company, Netherlands) with Cu-K␣ radiation (40 kV, 40 mA). The X-ray

The gas phase dehydration of glycerol was conducted in a vertical fixed bed reactor (8 mm i.d.) under atmospheric pressure using 1 g of catalyst. Prior to reaction, catalysts were pretreated at reaction temperature (320 ◦ C) in flowing dry N2 (40 mL/min) for about 1 h. A preheater was located on top of the reactor to vaporize the feed (vaporization temperature 250 ◦ C). In the meantime the reactor had provided an additional preheating zone (about 10 cm long spiral chute) to ensure complete vaporization. The reaction feed, an aqueous solution with 20 wt% glycerol, was introduced into the system from the top of the preheater by a HPLC pump at a fixed rate of 0.13 mL/min. The reaction was carried out at a gas hourly space velocity (GHSV) of glycerol as high as 873 h−1 , which is much higher than that of previously published experiments. The experiments without carrier gas were conducted under the same reaction conditions as the ones with carrier gas, except for an additional stripping step after reaction. For experiments with carrier gas, catalyst bed was heightened appropriately by mixing it with inert quartz sand (with the same particle size of catalysts) to make sure that contact time is the same as those without carrier gas. In the stripping step, catalyst bed was swept by a N2 flow (40 mL/min) for 10 min. During the reaction and stripping steps, effluent was collected in a receiver located at the exit of reactor, which was kept at 0 ◦ C by means of a mixture of water and ice. Certain amount of ethanol was loaded in the receiver to ensure efficient capture of products. The reaction was carried out for several hours and products were collected every half hour for analysis. The sample obtained during the first half hour was disregarded considering the poor material balance. The products were analyzed on Agilent 6820 gas chromatograph (GC) equipped with HP-INNOWAX capillary column (30 m × 0.32 mm × 0.25 ␮m) and a flame ionization detector (FID). Known amount of internal standard (ethylene glycol) was mixed within the sample prior to analysis for quantitative determination. The amount of coke deposition was determined by a Vario EL III CHNS analyzer (Elementar Corporation, Germany) equipped with a TCD by monitoring the amount of CO2 from flash-burning of spent catalyst. The glycerol conversion, product selectivity and yield are defined as follows:



Glycerol conversion (mol%) =

Product yield (mol%) =

1−

moles of glycerol in the sample moles of glycerol in feed



× 100;

moles of carbon in a defined product in the sample × 100; moles of carbon in glycerol in feed

Product selectivity (mol%) =

Product yield × 100. Glycerol conversion

Y. Gu et al. / Applied Catalysis A: General 429–430 (2012) 9–16

Fig. 1. NH3 -TPD profiles of shaped catalysts based on different H-zeolites.

3. Results and discussion 3.1. Characterization of catalysts 3.1.1. Acid properties of H-zeolite catalysts The crystalline structures of all zeolites were analyzed by XRD. All zeolites shared typical XRD peaks representing their inherent structures (data not shown). NH3 -TPD was employed to determine the acidity of catalyst. It is well acknowledged that the position and area of TPD peaks are closely related to the acid strength and acid amount respectively. The TPD profiles (Fig. 1) show that all catalysts exhibit two distinct peaks at low and high temperature respectively, indicating the existence of two kinds of acid sites on these catalysts. The low temperature desorption peaks, with maxima around 200 ◦ C, are probably related to extra-framework aluminum species. While framework aluminum sites correspond to the high temperature desorption peaks, maximized at 330–380 ◦ C [48]. According to Kim et al. [19], the area under the desorption peak curve in the range of 100–300 ◦ C represents the amount of weak acid, while that above 300 ◦ C represents the amount of strong acid. The calculated acid amounts are listed in Table 1. HY catalyst consists of large amount of weak as well as strong acid sites. Compared to H-Beta catalyst, there are fewer strong acid sites in catalysts based on bulk HZSM-5. The total acid amount increases in the following order: HZSM-11 < nano HZSM-5 ≈ nano HZSM-11 (profile not shown) < bulk HZSM-5 < H-Beta < HY. 3.1.2. Structure properties of different H-zeolites Besides acidity, structure properties of H-zeolites also have a tremendous impact on catalytic performance. The SEM and TEM images could provide data of particle size, particle morphology and aggregation degree of zeolite crystals. Nano HZSM-5 (Fig. 2(a)) has clubbed crystal morphology in the size of 70–100 nm. Table 1 Acid properties of shaped catalysts based on different H-zeolites. Catalysts

HZSM-11 Nano HZSM-5 Bulk HZSM-5 H-Beta HY Nano HZSM-11

Acid amount (mmol/g) Weak acid

Strong acid

Total acid

0.116 0.136 0.172 0.170 0.217 0.112

0.071 0.081 0.100 0.152 0.133 0.106

0.187 0.217 0.272 0.322 0.350 0.218

11

HZSM-11 (Fig. 2(b)) contains near-spherical particles with a diameter of approximately 4 ␮m, which are formed by aggregation of rod-like microcrystals (500 nm × 50 nm). For H-Beta and nano HZSM-11 (Fig. 2(c) and (f)), most particles are irregular spherical and their diameters vary between 300 and 500 nm. Bulk HZSM-5 (Fig. 2(d)) is composed of hexagonal columnar crystals (about 2 ␮m × 2 ␮m × 1 ␮m), while the crystals of HY (Fig. 2(e)) exhibit irregular octahedral morphology with the size ranging from 1.0 ␮m to 1.5 ␮m. Nano H-zeolites show higher aggregation degree and lower crystallinity compared to bulk ones. In general, the particle size decreases in the following order: HZSM-11 > bulk HZSM-5 > HY > H-Beta ≈ nano HZSM-11 > nano HZSM-5. In addition to the differences in particle size and morphology, these H-zeolites also differ greatly in channel pattern and channel diameter. HZSM-5 is composed of straight channels (0.51 nm × 0.55 nm) and intersecting sinusoidal channels (0.53 nm × 0.56 nm), the windows of which are both 10-membered rings. While HZSM-11 contains intersecting straight channels (0.53 nm × 0.54 nm) with the same 10-membered ring elliptical openings. Despite of their similar channel diameters, HZSM-11 has a less complex channel structure than HZSM-5. Both zeolites Beta and Y have a 3-dimensional pore structure with 12-membered ring openings. Zeolite Y exhibits the FAU structure with pores running perpendicular to each other. The pore diameter is as large as 0.74 nm, which leads into a large cavity of diameter around 1.3 nm [49], well known as super cage. It has a void volume fraction of 0.48. The pore system of zeolite Beta is composed of three mutually perpendicular 12-membered ring channels, the pore diameters of which are about 0.67 nm. 3.1.3. BET surface area and pore volume of H-zeolite catalysts Both micropore surface area and micropore volume decrease with the channel diameter (Table 2). HY and H-Beta share a similar external surface area, which is lower than that of bulk HZSM-5. One probable reason for the disagreement between crystal size of zeolite and external surface area of catalyst might be the different aggregation degree of zeolites with different crystal size. The other one is probably contributed by the coverage of nano particles during the preparation of shaped catalysts. Similar micropore surface area and external surface area are detected over bulk HZSM-5 and HZSM-11, although HZSM-11 seems to have larger external surface area because of its morphology (Fig. 2). Nano HZSM-11 possesses similar micropore surface area but larger external surface area than nano HZSM-5. 3.2. Catalytic performance of H-zeolite catalysts with different channel types The gas-phase dehydration of glycerol was conducted over different types of H-zeolite catalysts under the same reaction conditions. The reaction results are shown in Fig. 3(a). The reaction activity decreases in the following order: bulk HZSM-5 > HBeta  HY, which is contrary to the selectivity to coke (Table 3). The deactivation of H-zeolites is caused by coke formation, which blocks the pore and covers the acid sites, evidenced by the decreased BET surface area and acid amount of spent catalysts after reaction (Table 3). Coke is the product of multi-molecular condensation reaction or hydrogen transfer reaction, which is mainly affected by acid density and steric hindrance of catalysts [50]. H-Beta and bulk HZSM-5 share similar acid density while HY has a significantly lower acid density (Table 2). While the steric hindrance is related to channel pattern of H-zeolites, descending as follows: HZSM-5 > H-Beta > HY, which is in accordance with the order of selectivity and yield of coke (Table 3). The results imply that steric hindrance or channel pattern

12

Y. Gu et al. / Applied Catalysis A: General 429–430 (2012) 9–16

Fig. 2. SEM and TEM images of different H-zeolites: (a) nano HZSM-5 (TEM), (b) HZSM-11, (c) H-Beta, (d) bulk HZSM-5, (e) HY, and (f) nano HZSM-11.

Table 2 Texture properties and acid densities of shaped catalysts based on different H-zeolites. SBET (m2 /g)a

Catalysts

HY H-Beta Bulk HZSM-5 HZSM-11 Nano HZSM-5 Nano HZSM-11

Acid density (␮mol/m2 )

Pore volume (mL/g)

SM

SE

ST

VMI

VME

VT

181.5 112.0 73.1 75.1 81.4 81.1

56.9 55.8 74.1 76.2 56.4 70.8

238.4 175.7 147.2 151.3 137.8 151.9

0.088 0.060 0.036 0.037 0.040 0.040

0.118 0.122 0.106 0.129 0.097 0.122

0.206 0.182 0.142 0.166 0.137 0.162

1.47 1.83 1.85 1.24 1.57 1.44

a SM = Micropore surface area, SE = External surface area, ST = Total surface area, VMI = Micropore volume, VME = Mesopore volume, VT = Total pore volume, Acid density = Acid amount/SBET . The abbreviations above represent the same meaning in the whole paper.

Table 3 Catalytic performances of different catalysts as well as the acidity, surface areas and coke loadings of spent catalysts after glycerol dehydration. Catalystsa

Xb (mol%)

SAC (mol%)

SC (mol%)

YC (mol%)

HY H-Beta Bulk HZSM-5 HZSM-11 Nano HZSM-5

46.6 69.5 71.8 89.7 93.4

35.5 38.9 42.2 53.4 52.4

0.101 0.040 0.031 0.025 0.015

0.047 0.028 0.022 0.022 0.014

a b

SBET (m2 /g)

Acid amount (mmol/g)

SM

SE

ST

Weak acid

Strong acid

Total acid

2.7 32.2 8.3 30.7 11.8

29.9 16.0 23.2 25.4 41.7

32.6 48.2 31.5 56.1 53.5

0.124 0.088 0.068 0.061 0.093

0.039 0.057 0.017 0.010 0.027

0.163 0.145 0.085 0.071 0.120

X = Glycerol conversion, SAC = Selectivity to acrolein, SC = Selectivity to coke, YC = Yield of coke. The glycerol conversion and acrolein selectivity in the table are the average value for 3 h.

Y. Gu et al. / Applied Catalysis A: General 429–430 (2012) 9–16

13

Fig. 3. Catalytic performances of catalysts based on H-zeolites at similar reaction conditions: (a) and (b) without carrier gas, (c) HZSM-11, and (d) with carrier gas (N2 ). The carrier gas flow rate is 40 mL/min.

has a stronger influence than acid density over coke formation, in this case also the reaction activity. H-Beta exhibits higher catalytic activity than HY in spite of its lower total acid amount. Therefore, the larger amount of strong acid is probably the key factor, indicating that strong acid sites favor the dehydration of glycerol. Compared to H-Beta, bulk HZSM-5 shows similar weak acid amount but smaller strong acid amount. They exhibit similar catalytic activity, however, at the initial stage of reaction glycerol conversion over H-Beta decreases much faster. It is probably because that coke is more likely to be formed on strong acid sites, evidenced by the obvious decrease of strong acid sites in Table 3. The less steric hindrance of H-Beta is the other reason for its faster deactivation. Despite of the larger amount of strong and total acid sites, HY shows much lower catalytic activity than bulk HZSM-5, which confirms the negative influence of lower steric hindrance caused by the existence of super cage. In order to separate acidity effect, the catalytic rates normalized to acid amount are calculated and listed in supplementary Table S1. By comparisons, it confirms that zeolite channel structure has a significant influence on catalytic performance. It is easy to deduce that H-zeolites with smaller channel diameter have better catalytic stability. Acrolein selectivity also decreases as zeolite channel diameter increases. This is because that the OH group on central C of glycerol is preferential to interact with the bridging OH groups on H-zeolites in comparatively small channels, which results in the dehydration of glycerol to acrolein [22]. The lower selectivity to acrolein on Hzeolites with larger channel diameters, especially on HY with super cage, could also be related to large proportion of secondary reactions leading to coke and other by products due to its large void volume fraction. Based on the reaction results, certain conclusion could be made that H-zeolites with channel diameters, marginally larger than the molecular diameter of glycerol (0.5 nm) [51], are most promising for glycerol dehydration. However, Sato and co-workers and Katryniok et al. reported that silicotungstic acid supported on materials with larger mesopores showed more ideal catalytic behavior since the mass-transfer resistance is smaller [29,52]. The

discrepancy of above conclusions is attributed to differences between the two reaction systems. Over mesoporous catalysts, larger mesopores are expected to reduce diffusion resistance, leading to higher reaction activity. Nevertheless, over microporous H-zeolites, the channel diameters are similar to that of glycerol. Therefore, the formation of coke should have significant effect on the catalytic stability. Large channels (0.74 nm) and super cage (1.3 nm) of HY zeolite could provide sufficient space for the formation of coke, resulting in lower catalytic activity; while in the case of HZSM-5, multi-molecular reactions to coke were inhibited by its smaller channels (0.51 nm) and channel intersections (0.9 nm), which could slow down the deactivation of catalyst. Jia et al. [20] reported that diffusion exerts considerable influence on the gas phase dehydration of glycerol to acrolein. Low diffusion resistance could facilitate access to active sites and fasten escape of products from channels of H-zeolites, thereby improving the catalytic performance of catalysts. Although the diffusion of reactant and products could be improved by increasing channel diameter of H-zeolites, it could also increase the chance of coke formation resulted from lower steric hindrance. In conclusion this approach might not be applicable. 3.3. The effect of channel complexity and particle size Besides channel diameter, diffusion resistance is mainly influenced by channel complexity and particle size. Zhang et al. reported that helical channels have larger diffusion resistance than straight channels [53]. Accordingly, HZSM-11 seems to exhibit a more ideal diffusion behavior than HZSM-5. The other approach to reduce diffusion resistance is to decrease particle size of H-zeolites as described in the literature [20]. Nano HZSM-5 exhibits higher selectivity to acrolein and lower decreasing rate of glycerol conversion with time on stream (TOS) than bulk HZSM-5 (Fig. 3(b)). The selectivity and yield of coke (Table 3) are in accordance with the catalytic performance of these two catalysts. For bulk HZSM-5, the catalytic rate normalized to acid amount is quite lower (see supplementary Table S1). This is

14

Y. Gu et al. / Applied Catalysis A: General 429–430 (2012) 9–16

probably because of its longer diffusion distance in channels, attributed to larger crystal size. It indicates larger diffusion resistance for the diffusion of glycerol into channels and acrolein out of channels of H-zeolites. The results above have confirmed the importance of diffusion and indicated that H-zeolites with smaller particle size are more favorable for glycerol dehydration. Compared with bulk HZSM-5, better catalytic stability and acrolein selectivity are obtained over HZSM-11, which are over 50 mol% acrolein selectivity and 80 mol% glycerol conversion at TOS = 3 h (Fig. 3(b)). In spite of its larger particle size, HZSM-11 still shows quite higher catalytic rate normalized to acid amount than bulk HZSM-5 (as shown in supplementary Table S1). In view of the similar channel diameter and BET surface area of the two catalysts (Table 2), the less complex channels are probably the intrinsic reason for the outstanding catalytic performance of HZSM-11. Active sites in straight channels are more accessible for glycerol molecules and the product molecules could get out of straight channels more easily, leading to higher glycerol conversion and acrolein selectivity. The lower selectivity to coke (Table 3) and acetaldehyde (data not shown) over HZSM-11, confirms the effect of channel complexity on reducing secondary reactions. Another conclusion could be made from the above results that low channel complexity, small channel diameter and particle size are essential factors for the production of acrolein from glycerol dehydration over H-zeolites. 3.4. The effect of carrier gas In gas phase dehydration of glycerol, some researches were conducted with carrier gas [16–18], and some without [33,38,39]. To investigate the influence of carrier gas (N2 ), a comparison of the reaction results of HZSM-11 between experiments conducted with or without carrier gas at similar reaction conditions is performed. The glycerol conversion decreases to 80 mol% at 3 h TOS without carrier gas. However, it maintains above 80 mol% before 5 h TOS after the introduction of carrier gas. Besides the improved catalytic stability, selectivity to acrolein is also at a higher level with the existence of carrier gas (Fig. 3(c)). Since the catalysts used in the two reactions possess the same physical and acid properties, the influence of carrier gas seems quite obvious. It is well known that diffusion is virtually motivated by the concentration difference between internal and external space of catalyst [54]. The introduction of carrier gas could enlarge the concentration difference by supplying glycerol (and removing products) continuously to (and from) outer space of catalyst, which could stimulate the diffusion of glycerol into channels and products out of channels of H-zeolites, resulting in higher glycerol conversion and acrolein selectivity. Another reason for the improved catalytic performance is probably the dilution effect of carrier gas, lowering the partial pressures of reactants and products simultaneously, which could reduce the rates of coke formation and catalyst deactivation. In addition, a lower glycerol partial pressure could also reduce the possibility of intermolecular dehydration, which would affect the catalyst performance kinetically. The selectivity to coke decreases from 0.025 mol% to 0.008 mol% after the introduction of carrier gas, although the latter is the results of continuous reaction for 8 h, which confirms the hypothesis that the existence of carrier gas could inhibit secondary reactions. It suggested that carrier gas could enhance the catalytic performance of H-zeolites for glycerol dehydration. 3.5. Catalytic performance of nano HZSM-11 with carrier gas According to Sections 3.3 and 3.4, it is likely to get superior catalytic performance over small-sized HZSM-11 at the existence of

carrier gas. After several attempts, nano ZSM-11 with crystal size of 300–500 nm was successfully synthesized. Compared to nano HZSM-5, both catalytic stability and acrolein selectivity are enhanced over nano HZSM-11, although its particle size is larger. The glycerol conversion is still larger than 80 mol% after 8 h TOS at GHSV as high as 873 h−1 . Moreover the selectivity to acrolein is always larger than 74.0 mol% (Fig. 3(d)). Considering their similar acidity (as shown in Table 1), it is confirmed that the channel system of HZSM-11 is superior to HZSM-5 for acrolein production from glycerol dehydration. The product distribution of both catalysts is shown in Tables 4 and 5. In comparison with nano HZSM-5, less amount of by products, including acetaldehyde, propenol and acrylic acid, is detected over nano HZSM-11 attributed to the lower fraction of secondary reactions. Acetol is considered to be the dehydration product of the OH group at the ␣ position of glycerol over weak acid sites [23]. The lower selectivity to acetol over nano HZSM-11 is due to its smaller amount of weak acid sites (Table 1). The selectivity to acetaldehyde, propenol and acrolein do not vary much as TOS increases, while that of acetol and acrylic acid increase obviously. The disappearance of strong brønsted acid sites is the main cause for the increasing selectivity to acetol with TOS [31,34]. Acrylic acid is the oxidation product of acrolein [43]. Jia et al. [20] reported that nano HZSM-5 exhibited 83% glycerol conversion and 65 mol% acrolein selectivity at TOS = 9–10 h (GHSV = 719 h−1 ), which is slightly lower than results in our study. It is worth noting that particle size of reported nano HZSM-5 is in the range of 20–60 nm, our nano HZSM-11 varies between 300 and 500 nm (Fig. 2). Besides, the catalyst used in this work contains merely 35 wt% of active component. A simple calcination at 600 ◦ C in air atmosphere for 2 h was found to be sufficient for a full regeneration of the catalyst. The regenerated nano HZSM-11 exhibits similar BET surface area, crystallinity and catalytic performance to fresh catalyst (see supplementary Fig. S1 and Fig. S2), which means the zeolite structure remains intact after reaction at harsh hydrothermal conditions. The emphasis of future work will be the synthesis of ZSM-11 in smaller size by optimizing synthesis conditions. 3.6. The effect of external surface In order to determine the role external surface plays and to investigate the mechanism of deactivation of H-zeolite catalysts, several TEM experiments were carried out over spent nano HZSM5. The clear lattice fringes in Fig. 4(a) indicate that nano HZSM-5 is highly crystallized. The framework of zeolite is destroyed by electron beam bombardment to make the structure of deposited coke visible. On the left side of Fig. 4(b) is graphite background, while the sample is laid on the right side. The framework of ZSM-5 has collapsed completely after electron beam bombardment. At the edge of zeolite crystal lie several pieces of curved layers, which are supposed to be coke deposited on the external surface of H-zeolite. The deposited coke has a layered structure similar to graphite, although its layers are curved and arranged disorderly. It is hard to find any coke deposited in channels of nano HZSM-5 in the TEM image, which is because of the channel diameter is only approximately 0.55 nm, whereas the interplanar distance of graphite is about 0.335 nm [55]. Thus the innercrystal coke is not likely to be graphite-like layered coke, but constituted by oxygen-containing alkylated mono- or bi-aromatics or oligomers trapped at intersections of channels [25,56]. The BET data of spent catalysts (Table 6) show that both micropore surface area and micropore volume of nano HZSM-11 decrease significantly after 2 h of reaction. It indicates that coke is initially deposited internally, blocking part of zeolite channels, which is in

Y. Gu et al. / Applied Catalysis A: General 429–430 (2012) 9–16

15

Table 4 Glycerol conversion and product distribution of nano HZSM-5 with carrier gas. TOS (h)

0.5–1

1–2

2–3

3–4

4–5

5–6

6–7

7–8

Xa (mol%) YAC (mol%)

99.6 49.4

99.0 59.0

95.6 52.6

91.9 47.4

88.5 45.7

77.5 48.5

76.1 45.4

72.6 45.3

Product selectivityc (mol%) Acetaldehyde Propenol Acrolein Acetol Acrylic acid

4.7 3.0 49.6 5.2 –

5.6 6.3 59.6 8.8 –

5.0 6.0 55.0 8.0 1.5

5.6 5.7 51.6 8.2 2.2

5.6 5.6 51.6 8.2 2.8

6.4 6.5 62.6 10.1 4.2

6.1 5.9 59.6 9.3 4.2

7.1 6.0 62.4 10.2 4.7

Coke loadingb (wt%)

3.7

a

X = Glycerol conversion, YAC = Yield of acrolein. After the reaction for 8 h. The other by-products included propanal, acetone, methanol, acetic acid, propionic acid, CO, CO2 , light olefins, and oxygenates with unknown structure, which were present in minor amounts. b c

Table 5 Glycerol conversion and product distribution of nano HZSM-11 with carrier gas. TOS (h)

0.5–1

1–2

2–3

3–4

4–5

5–6

6–7

7–8

X (mol%) YAC (mol%)

100 66.0

100 78.4

99.3 73.8

96.1 65.9

93.1 63.6

88.3 64.6

84.5 61.6

81.6 61.1

4.5 1.6 66.0 4.4 –

5.7 3.6 78.4 6.3 –

5.4 4.3 74.3 7.3 0.6

5.0 4.3 68.5 7.7 1.7

6.4 4.1 68.3 7.5 2.2

5.6 4.8 73.1 8.1 2.9

5.6 4.6 72.9 8.0 3.5

5.4 4.3 74.9 8.1 3.8

Product selectivity (mol%) Acetaldehyde Propenol Acrolein Acetol Acrylic acid Coke loading (wt%)

3.5

Fig. 4. TEM images of spent nano HZSM-5 after glycerol dehydration.

accordance with previous research [56]. It is also worth noting that innercrystal coke could only form at the channel intersections containing pairs of acid sites due to steric hindrance [50]. Therefore, Brønsted acid sites, which are in channels or solely at channel intersections, could always catalyze the dehydration of glycerol without afraid of coking unless accesses to them are totally blocked. The high branching of HZSM-5 zeolite microporous structure explains the reason, in spite of the progressive blockage of internal channels,

active sites remain accessible for glycerol. Hence, these acid sites are authentically effective sites for the reaction. As reaction goes on, the micropore surface area and micropore volume remain almost the same, while the external surface area and mesopore volume decrease obviously. It implies the deposition of coke in intercrystalline void volume after a period of reaction. It seems that external surface of H-zeolite serves a role to supply room for coke during the reaction, evidenced by the TEM image of spent nano HZSM-5

Table 6 The texture properties of spent catalysts after glycerol dehydration for different time. Catalysts

Nano HZSM-11-2a Nano HZSM-11-8 HZSM-11-2 HZSM-11-8 a

SBET (m2 /g)

Pore volume (mL/g)

SM

SE

ST

VMI

VME

VT

31.4 31.3 14.7 13.0

51.6 34.7 39.7 37.9

83.0 66.0 54.4 50.8

0.015 0.015 0.007 0.006

0.109 0.082 0.080 0.080

0.124 0.097 0.087 0.086

H-zeolite-t, in which t is the time on stream.

16

Y. Gu et al. / Applied Catalysis A: General 429–430 (2012) 9–16

(Fig. 4(b)). To sum up, coke is initially deposited at channel intersections of H-zeolites, and when the blockage of channels comes up to a certain extent, there arrives the onset of deposition of coke on the external surface. As to HZSM-11, both micropore surface area and external surface area decrease significantly after 2 h of reaction, which then remain basically the same with further reaction. This is probably because of the advancement of external coke deposition caused by larger crystal size of HZSM-11. 4. Conclusions The importance of channel structure and its role in the dehydration of glycerol are demonstrated in this work. Improved catalytic performance was obtained after the channel diameter of H-zeolites had decreased from 0.74 nm to 0.54 nm, which indicated that appropriate steric hindrance was favorable for reaction stability. H-zeolites with smaller channel diameter, marginally larger than the molecular diameter of glycerol, were more likely to obtain better catalytic performance. Ideal catalytic performance was achieved over nano-sized HZSM-5. HZSM-11, composed of fascinating straight channels, exhibited better catalytic stability and acrolein selectivity than HZSM-5, which confirmed the advantage of lower channel complexity. The introduction of carrier gas could improve diffusion process of catalyst and lower the partial pressures of reactants and products simultaneously, resulting in higher glycerol conversion and acrolein selectivity. Notable catalytic performance was obtained over nano HZSM-11 (300–500 nm) at GHSV as high as 873 h−1 , which was 81.6 mol% glycerol conversion and 74.9 mol% selectivity to acrolein at 8 h on stream. It was deduced that low channel complexity, small channel diameter and particle size were essential factors for the production of acrolein from glycerol dehydration over H-zeolites. There were two kinds of coke formed during the reaction, graphite-like layered coke deposited on external surface of Hzeolites and oxygen-containing alkylated mono- or bi-aromatics or oligomers trapped at intersections of channels. The BET and TEM approaches further indicated that coke was initially deposited at channel intersections of H-zeolites, and as the blockage of channels came up to a certain extent, there arrived the onset of deposition of coke on the external surface. The external surface of H-zeolite appeared to participate in the reaction by supplying room for coke deposition. Brønsted acid sites, which were in channels or solely at channel intersections, were authentically effective sites for the dehydration of glycerol. Acknowledgement This work was supported by the National 973 Program (No. 2012CB215006). 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.2012.03.030. References [1] L.C. Meher, D. Vidya Sagar, S.N. Naik, Renew. Sustain. Energy Rev. 10 (2006) 248–268.

[2] M. Pagliaro, M. Rossi, The Future of Glycerol: New Uses of a Versatile Raw Material, The Royal Society of Chemistry, Cambridge, 2008. [3] M. Pagliaro, R. Ciriminna, H. Kimura, M. Rossi, C.D. Pina, Angew. Chem. Int. Ed. 46 (2007) 4434–4440. [4] B.Y. Jo, E.J. Kim, S.H. Moon, Appl. Catal. A: Gen. 332 (2007) 257–262. [5] B. Katryniok, S. Paul, M. Capron, F. Dumeignil, ChemSusChem 2 (2009) 719–730. [6] M. Watanabe, T. Iida, Y. Aizawa, T.M. Aida, H. Inomata, Bioresour. Technol. 98 (2007) 1285–1290. [7] V. Lehr, M. Sarlea, L. Ott, H. Vogel, Catal. Today 121 (2007) 121–129. [8] W. Bühler, E. Dinjus, H.J. Ederer, A. Kruse, C. Mas, J. Supercrit. Fluids 22 (2002) 37–53. [9] L. Ott, M. Bicker, H. Vogel, Green Chem. 8 (2006) 214–220. [10] A.G. Schering-Kahlbaum, FR Patent 695931 (1930). [11] J.L. Dubois, C. Duquenne, W. Hölderlich, WO Patent 2006087083 (2006). [12] B.Q. Xu, S.H. Chai, T. Takahashi, M. Shima, S. Sato, R. Takahashi, WO Patent 2007058221 (2007). [13] A. Neher, T. Haas, D. Arntz, H. Klenk, W. Girke, US Patent 5387720 (1995). [14] X.M. Li, C.L. Zhang, C.H. Qin, C. Chen, J.M. Shao, CN Patent 101070286 (2007). [15] Q. Liu, Z. Zhang, Y. Du, J. Li, X. Yang, Catal. Lett. 127 (2009) 419–428. [16] F. Wang, J.L. Dubois, W. Ueda, J. Catal. 268 (2009) 260–267. [17] F. Wang, J.L. Dubois, W. Ueda, Appl. Catal. A: Gen. 376 (2010) 25–32. [18] J. Deleplanque, J.L. Dubois, J.F. Devaux, W. Ueda, Catal. Today 157 (2010) 351–358. [19] Y.T. Kim, K.D. Jung, E.D. Park, Microporous Mesoporous Mater. 131 (2009) 28–36. [20] C.J. Jia, Y. Liu, W. Schmidt, A.H. Lu, F. Schüth, J. Catal. 269 (2010) 71–79. [21] A. Corma, G.W. Huber, L. Sauvanaud, P. O’Connor, J. Catal. 257 (2008) 163–171. [22] E. Yoda, A. Ootawa, Appl. Catal. A: Gen. 360 (2009) 66–70. [23] Y.T. Kim, K.D. Jung, E.D. Park, Appl. Catal. A: Gen. 393 (2011) 275–287. [24] A.S. Oliveira, S.J.S. Vasconcelos, J.R. Sousa, F.F. Sousa, J.M. Filho, A.C. Oliveira, Chem. Eng. J. 168 (2011) 765–774. [25] W. Suprun, M. Lutecki, T. Haber, H. Papp, J. Mol. Catal. A: Chem. 309 (2009) 71–78. [26] K. Pathak, K.M. Reddy, N.N. Bakhshi, A.K. Dalai, Appl. Catal. A: Gen. 372 (2010) 224–238. [27] L. Cheng, X.P. Ye, Catal. Lett. 130 (2009) 100–107. [28] W. Yan, G.J. Suppes, Ind. Eng. Chem. Res. 48 (2009) 3279–3283. [29] E. Tsukuda, S. Sato, R. Takahashi, T. Sodesawa, Catal. Commun. 8 (2007) 1349–1353. [30] H. Atia, U. Armbruster, A. Martin, J. Catal. 258 (2008) 71–82. [31] A. Alhanash, E.F. Kozhevnikova, I.V. Kozhevnikov, Appl. Catal. A: Gen. 378 (2010) 11–18. [32] S. Erfle, U. Armbruster, U. Bentrup, A. Martin, A. Brückner, Appl. Catal. A: Gen. 391 (2011) 102–109. [33] L. Shen, Y. Feng, H. Yin, A. Wang, L. Yu, T. Jiang, Y. Shen, Z. Wu, J. Ind. Eng. Chem. 17 (2011) 484–492. [34] H. Atia, U. Armbruster, A. Martin, Appl. Catal. A: Gen. 393 (2011) 331–339. [35] S.H. Chai, H.P. Wang, Y. Liang, B.Q. Xu, Green Chem. 10 (2008) 1087–1093. [36] S.H. Chai, H.P. Wang, Y. Liang, B.Q. Xu, Green Chem. 9 (2007) 1130–1136. [37] S.H. Chai, H.P. Wang, Y. Liang, B.Q. Xu, J. Catal. 250 (2007) 342–349. [38] S.H. Chai, H.P. Wang, Y. Liang, B.Q. Xu, Appl. Catal. A: Gen. 353 (2009) 213–222. [39] L.Z. Tao, S.H. Chai, Y. Zuo, W.T. Zhang, Y. Liang, B.Q. Xu, Catal. Today 158 (2010) 310–316. [40] F. Cavani, S. Guidetti, L. Marinelli, M. Piccinini, E. Ghedini, M. Signoretto, Appl. Catal. B: Environ. 100 (2010) 197–204. [41] N.R. Shiju, D.R. Brown, K. Wilson, G. Rothenberg, Top. Catal. 53 (2010) 1217–1223. [42] A. Ulgen, W. Hoelderich, Catal. Lett. 131 (2009) 122–128. [43] A. Ulgen, W.F. Hoelderich, Appl. Catal. A: Gen. 400 (2011) 34–38. [44] P. Lauriol-Garbay, J.M.M. Millet, S. Loridant, V. Bellière-baca, P. Rey, J. Catal. 280 (2011) 68–76. [45] P. Lauriol-Garbay, G. Postole, S. Loridant, A. Auroux, V. Bellière-baca, P. Rey, J.M.M. Millet, Appl. Catal. B: Environ. 106 (2011) 94–102. [46] W. Suprun, M. Lutecki, R. Gläser, H. Papp, J. Mol. Catal. A: Chem. 342–343 (2011) 91–100. [47] L. Zhang, H. Liu, X. Li, S. Xie, Y. Wang, W. Xin, S. Liu, L. Xu, Fuel Process. Technol. 91 (2010) 449–455. [48] G.L. Woolery, G.H. Kuehl, H.C. Timken, A.W. Chester, J.C. Vartuli, Zeolites 19 (1997) 288–296. [49] Z. Ma, T. Kyotani, A. Tomita, Chem. Commun. 23 (2000) 2365–2366. [50] D.M. Bibby, R.F. Howe, G.D. McLellan, Appl. Catal. A: Gen. 93 (1992) 1–34. [51] T. Krebs, G. Andersson, H. Morgner, Chem. Phys. 340 (2007) 181–186. [52] B. Katryniok, S. Paul, M. Capron, C. Lancelot, V. Belliere-Baca, P. Rey, F. Dumeignil, Green Chem. 12 (2010) 1922–1925. [53] L. Zhang, S.Z. Qiao, Y.G. Jin, L. Cheng, Z.F. Yan, G.Q. Lu, Adv. Funct. Mater. 18 (2008) 3834–3842. [54] H. Chen, T. Sun, D. Sui, J. Dong, Anal. Chim. Acta 698 (2011) 27–35. [55] S. Bandow, S. Asaka, Y. Saito, A.M. Rao, L. Grigorian, E. Richter, P.C. Eklund, Phys. Rev. Lett. 80 (1998) 3779–3782. [56] P.L. Benito, A.G. Gayubo, A.T. Aguayo, M. Olazar, J. Bilbao, Ind. Eng. Chem. Res. 35 (1996) 3991–3998.