- Email: [email protected]
Selective synthesis of propylene oxide through liquid-phase epoxidation of propylene with H2O2 over formed Ti-MWW catalyst
Journal of Catalysis 342 (2016) 173–183
Contents lists available at ScienceDirect
Journal of Catalysis journal homepage: www.elsevier.com/locate/jcat
Selective synthesis of propylene oxide through liquid-phase epoxidation of propylene with H2O2 over formed Ti-MWW catalyst Xinqing Lu, Haihong Wu, Jingang Jiang, Mingyuan He, Peng Wu ⇑ Shanghai Key Laboratory of Green Chemistry and Chemical Processes, School of Chemistry and Molecular Engineering, East China Normal University, North Zhongshan Rd. 3663, Shanghai 200062, People’s Republic of China
a r t i c l e
i n f o
Article history: Received 20 June 2016 Revised 25 July 2016 Accepted 26 July 2016
Keywords: HPPO process Ti-MWW titanosilicate Propylene oxide Liquid-phase epoxidation Fixed-bed reactor
a b s t r a c t The liquid-phase epoxidation of propylene to propylene oxide (PO) over formed titanosilicate catalysts was investigated in a fixed-bed reactor. The effects of reaction temperature, n(C=3)/n(H2O2) molar ratio, and weight hourly space velocity (WHSV) of H2O2 or solvent on the catalytic performance of the formed Ti-MWW catalyst have been extensively studied. Adding an appropriate amount of ammonia to the reaction mixture prolonged the catalyst lifetime effectively. The main byproduct of propylene glycol (PG) and other heavy byproducts with high boiling points were deposited inside zeolite micropores, which corresponded to the main reason for the catalyst deactivation. The high-temperature calcination in air recovered readily the reactivity of the deactivated catalyst. Fluorine implantation remarkably enhanced the reactivity and lifetime of the catalyst in the hydrogen peroxide propylene oxide (HPPO) process, exhibiting a PO selectivity of >99%. Ó 2016 Elsevier Inc. All rights reserved.
1. Introduction Propylene oxide (PO) is one of the most important epoxide chemicals, useful for the manufacture of polyether polyols, propylene glycols, and propylene glycol ethers [1]. PO production capacity reached approximately 10 million metric tons in 2012 and continues growing at an annual rate of about 5% [2,3]. The current commercial routes for PO production are the classic chlorohydrin process and variations of the hydroperoxide process, including the propylene oxide/styrene monomer process (PO/SM), the propylene oxide/t-butyl alcohol process (PO/TBA), the cumene hydroperoxide-based process (CHPO), and the titanosilicatebased hydrogen peroxide propylene oxide process (HPPO). The chlorohydrin process was first developed by Wurtz for the synthesis of ethylene oxide (EO) and PO in 1859. This route has played a critical role for many decades and it is even currently widely employed in the production of PO. However, it suffers serious problems of equipment corrosion and environmental pollution. Thus, it has been gradually replaced by more environmentally benign processes using hydroperoxides as oxidants. For co-production processes such as the PO/SM and PO/ TBA techniques, the demand and pricing of the co-products styrene
⇑ Corresponding author. E-mail address: [email protected] (P. Wu). http://dx.doi.org/10.1016/j.jcat.2016.07.020 0021-9517/Ó 2016 Elsevier Inc. All rights reserved.
monomer (SM) or t-butyl alcohol (TBA) may not economically match those of PO well, making the process optimization difficult. Moreover, in the past few years, the fairly stagnant styrene monomer market and the increasing limitation of methyl t-butyl ether (MTBE) in gasoline [4] restricted the further development of such PO processes. In 2006, Sumitomo Chemical Co. Ltd. developed an innovative process free of co-product using cumene hydroperoxide as an oxidant, in which the intermediate cumyl alcohol was reduced to cumene with costly hydrogen for recycling. The CHPO process exhibits an obvious advantage in terms of avoiding a coproduct, but it requires energy-consuming multiple steps including cumene oxidation, propylene epoxidation, and reduction of alcohol intermediate. The HPPO process is a more ecoefficient method for selective production of PO, in which the catalytic epoxidation of propylene hydrogen peroxide is carried out on titanosilicate catalysts under mild reaction conditions (313–323 K and 2–3 MPa), giving water as the main byproduct. The development of the HPPO process dates back to 1983, when ENI patented a titanosilicate (TS-1) as a useful catalyst for this process [5]. Subsequently, there have been many fundamental researches focusing on the synthesis of TS-1 catalysts [6–13] and the investigation of propylene epoxidation with different reaction conditions [14–16], reactors [3,17,18], and additives [19,20], as well as the reaction mechanism and kinetics of this process [1,14,21–23]. Based on deep understanding of the
174
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
HPPO process, Evonik and SKC launched a commercial plant in South Korea in 2008 with a capacity of 100 Kton/yr, and this plant is operated continuously at maximum capacity [24]. The next year BASF and DOW Chemical launched a similar plant based on HPPO technology in Belgium, with a higher capacity of 300 Kton/yr [25]. Later, the HPPO plants were widely used for the production of PO and they have been launched around the world [26,27]. However, all the commercial HPPO processes adopt without exception the catalytic system TS-1/H2O2/methanol. With methanol as the optimized solvent, the separation of PO from the reaction system has high energy consumption due to the zeotropic problem of methanol and PO. Moreover, the solvolysis of PO with the protic methanol solvent would inevitably decrease the PO selectivity. In addition, the manufacturing cost of TS-1 is always an obstacle to its industrial application, because extremely expensive tetrapropylammonium hydroxide (TPAOH) is required as the structuredirecting agent (SDA) for the crystallization of TS-1. Hence, it is desirable to develop a more suitable titanosilicate for the HPPO process. In our earlier work [28], we conducted a HPPO batch reactor process over Ti-MWW, which preferred an aprotic solvent of acetonitrile and showed superior performance to TS-1. Benefiting from the weak basicity of acetonitrile, the solvolysis of PO was effectively suppressed to make the PO selectivity high level 99%. A certainly cheaper SDA of piperidine (PI) or hexamethyleneimine (HMI) is used for the synthesis of Ti-MWW [29]. On the other hand, possessing a special lamellar structure different from that of the conventional three-dimensional zeolites with rigid structures, TiMWW is readily structurally modifiable by swelling [30], full delamination [30], partial delamination [31], pillaring [32], and silylation [33,34]. Moreover, the structural rearrangement [35,36] or fluorination treatment [37,38] was effective to enhance the reactivity of Ti-MWW. In the present study, we investigated the HPPO process in the fixed-bed reactor-formed Ti-MWW catalysts. The effects of reaction temperature, the molar ratio of propylene to H2O2, the concentration of ammonia, and the weight hourly space velocity (WHSV) of H2O2 or solvent on this HPPO process were systematically studied. We have studied the causes of catalyst deactivation as well as the optimized reaction conditions for prolonging the catalyst lifetime. Additionally, the formed Ti-MWW catalyst was chemically modified to enhance its HPPO catalytic performance by structural rearrangement or fluorination.
2. Experimental 2.1. Catalyst preparation 2.1.1. Synthesis of the active component of Ti-MWW zeolite powder The Ti-MWW powder was hydrothermally synthesized using boric acid as a crystallization-supporting agent and PI as a SDA in two steps as follows: the synthetic gels with molar composition 1.0 SiO2:0.05 TiO2:1.4 PI:0.67 B2O3:19 H2O were hydrothermally crystallized at 443 K for 7 days, and then the obtained powder was refluxed in a 2 M HNO3 aqueous solution to remove extraframework Ti species and part of framework boron as well [29]. Finally, the acid-washed product was filtered, washed with deionized water to pH 7, dried at 393 K overnight, and calcined at 823 K for 6 h (denoted as Ti-MWW powder).
added dropwise into a mixture of 80 g Ti-MWW powder and 5 g Sesbania cannabina Pers powder under grinding. Then a certain amount of deionized water was added to the mixture to adjust the hardness of the paste. Finally, the paste was extruded into £ 1.8 mm strips, dried at 393 K overnight, and calcined at 823 K for 6 h (denoted as formed Ti-MWW). 2.1.3. Preparation of chemically modified Ti-MWW catalyst Two different modification methods were employed to enhance the performance of the catalyst. First, the extrudates were subjected to hydrothermal structural rearrangement treatment in an aqueous PI solution according to the procedures adopted previously for the Ti-MWW powder [35,36]. The treatment was carried out under rotation (10 rpm) at 443 K for 1 day at a PI/SiO2 molar ratio of 0.4 and a H2O/SiO2 molar ratio of 10. The treated product was washed with deionized water, dried at 393 K overnight, and calcined at 823 K for 6 h, and was denoted as formed Re-TiMWW. Second, following previously reported fluorination for the titanosilicate powder [39], we conducted the treatment of the extruded catalysts in a solution of methanol and NH4F at 443 K for 1 day at an NH4F/SiO2 molar ratio of 0.05 and a MeOH/SiO2 molar ratio of 20. The product was then dried at 393 K overnight and calcined at 823 K for 6 h (denoted as formed F-Ti-MWW). 2.2. Characterization methods The X-ray diffraction (XRD) pattern was recorded on a Rigaku Ultima IV diffractometer using Ni-filtered CuKa radiation (k = 0.1541 nm) in a scanning range of 5–35° at a scanning rate of 10° min1 to confirm the structure and crystallinity of the catalysts. The voltage and current were 35 kV and 25 mA, respectively. The morphologies and crystal sizes were examined by a Hitachi S4800 scanning electron microscope. The UV–visible spectra were collected on a PerkinElmer UV–vis Lambda 35 spectrophotometer using BaSO4 as a reference. The IR spectra of hydroxyl stretching and pyridine adsorption were collected by a Nicolet Nexus 670 FT-IR spectrometer at a resolution of 2 cm1. In order to avoid the influence of absorbed water, a self-supported wafer that was placed in a quartz IR cell sealed with CaF2 windows was evacuated at 723 K for 2 h; then the spectra in the hydroxyl stretching region were collected at room temperature. For pyridine spectra measurement, the pretreated wafer was exposed to a pyridine vapor at 298 K for 20 min, and then the absorbed pyridine was evacuated at 423 K for 0.5 h. In addition, the IR spectra in the region of organic functional group vibration were measured using a KBr technique. All the spectra were collected at room temperature. 29Si MAS NMR spectra were recorded on a VARIAN VNMRS-400 MB NMR spectrometer. The content of titanium and boron was determined by ICP-AES on an IRIS Intrepid II XSP after the samples were dissolved in HF solution. The textural properties of the catalysts were determined by N2 adsorption at 77 K using a BEL SORP instrument. Prior to the adsorption measurements, the samples were degassed in vacuum at 573 K for 6 h. The microporous volume was derived from the t-plot method [40]. According to the reported method [17], the mechanical strength of the formed catalysts was determined under different weights until the extrudates were crushed. Thermogravimetric (TG) analysis was carried out in a Netzsch Sta 4049 F3 apparatus in air with a heating rate of 10 K min1 to 1073 K. 2.3. Liquid-phase epoxidation of propylene
2.1.2. Preparation of formed Ti-MWW catalyst The catalyst shaping was performed by an extrusion procedure from a homogeneous paste of zeolite active component, silica binder, and pore generator. A solution of 66.7 g silica sol (30 wt.%) was
2.3.1. Continuous reaction As shown in Scheme 1, the continuous epoxidation of propylene was carried out in a fixed-bed reactor (11 cm length, 1 cm internal
175
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
Scheme 1. Schematic diagram of experimental apparatus.
diameter, and with a Teflon inner layer) that was immersed in a water bath to control the reaction temperature. The extrudates were crushed and sieved to be 40–80 mesh size particles, 3.0 g of which were loaded in the reactor. The pressure was maintained at 2 MPa with the assistance of N2, which made the propylene liquefied. After the reaction temperature and pressure both reached the determined values, the propylene liquid and a mixed solution of H2O2/CH3CN/H2O were introduced into the reactor independently by two dual-plunger pumps, each equipped with two 50 mL containers. The solution for the reactor outlet, containing the products, unconverted reactants, and solvent, entered a liquid-storage tank where it was cooled to room temperature. During the reaction, the reaction mixture was sampled at a certain time interval from the tank. 2.3.2. Batchwise propylene epoxidation The batchwise epoxidation of propylene was carried out in a 45 mL autoclave reactor equipped with a Teflon inner layer according to our early works [28,41]. In a typical run, 0.1 g catalyst, 10 g MeCN, and 30 mmol H2O2 (30 wt.%) were fed into the reactor. Propylene was charged into the autoclave to replace the air inside three times, and then the propylene was charged into the autoclave at a constant pressure of 0.4 MPa. After the reaction was carried out under vigorous stirring at 313 K for 2 h, the reactor was cooled with ice water and depressurized slowly before being opened. 2.3.3. Products analysis The reaction products were analyzed on a gas chromatograph (Shimadzu 2014) FID detector that was equipped with an RtxWax capillary column (30 m 0.25 mm 0.25 lm). The quantification was performed using isopropanol as an internal standard substance. The remaining amount of H2O2 after reaction was determined by a conventional titration method with 0.05 M Ce (SO4)2 solution. The products formed were further confirmed by a GC-MS (Agilent 6890 series GC system, 5937 network mass selective detector). The result of the reaction was given using these criteria:
H2 O2 conv: ¼
n0H2 O2 nH2 O2 n0H2 O2
100%;
nPO 100%; n0H2 O2 nPO 100%; PO sel: ¼ nPO þ nPG nPO þ nPG H2 O2 eff: ¼ 0 100%: nH2 O2 nH2 O2 PO yield ¼
H2O2 conv., PO yield, PO sel., and H2O2 eff. stand for the conversion of H2O2, the yield of PO, the selectivity of PO, and the utilization efficiency of H2O2, respectively. nPO and nPG stand for the numbers of moles of PO and PG, respectively. n0H2 O2 and nH2 O2 stand for the initial mole content and the final content of H2O2, respectively. 3. Results and discussion 3.1. A comparison of physicochemical and catalytic propylene epoxidation properties between Ti-MWW powder and formed Ti-MWW Ti-MWW-powder showed the XRD pattern of the corresponding MWW-type zeolite structure with a relatively high crystallinity (Fig. 1A) and exhibited a typical platelet morphology with a uniform thickness of about 50 nm (Fig. 2a). The addition of silica binder during the forming process led to a slight decrease in crystallinity that was in accord with the presence of an amorphous phase on the surface of the crystals (Fig. 2b). Additionally, the forming process made the specific surface area and the micropore volume decrease from 554 m2 g1 and 0.14 cm3 g1 for Ti-MWW powder to 505 m2 g1 and 0.12 cm3 g1 for formed Ti-MWW, respectively (Table 1). However, the mesopore volume increased from 0.28 to 0.41 cm3 g1, possibly due to the stacking of amorphous silica and the removal of the pore generator of Sesbania cannabina Pers powder during the preparation of the formed catalysts. Both Ti-MWW powder and formed Ti-MWW contained only isolated Ti species in the framework, as evidenced by the predominant adsorption at 210 nm in UV–visible spectra, indicating that
176
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
210
d
Intensity (a.u.)
B
Absorbance (a.u.)
A
c b
d c b
a 5
10
15
20
25
30
a 35
200
300
400
500
Wavelength (nm)
2 Theta (deg.)
Fig. 1. XRD patterns (A) and UV–visible spectra (B) of Ti-MWW powder (a), formed Ti-MWW (b), formed Used-Ti-MWW (c), and formed Regenerated-Ti-MWW (d).
a
b
500 nm
500 nm
c
d
500 nm
500 nm
Fig. 2. SEM micrographs of Ti-MWW powder (a), formed Ti-MWW (b), formed Used-Ti-MWW (c), and formed Regenerated-Ti-MWW (d).
Table 1 Physicochemical properties and catalytic performance of various catalysts. No.
1 2 3 4
Catalyst
Ti-MWW powder Formed Ti-MWW Formed used-Ti-MWWe Formed regenerated-Ti-MWWf
Content (mmol g1)a Ti
B
0.39 0.30 0.29 0.30
0.16 0.12 0.02 0.02
SSA (m2 g1)b
554 505 274 498
Pore volume (cm3 g1)
Horizontal strength (MPa)
Microp.d
Mesop.
0.14 0.12 0.05 0.12
0.28 0.41 0.31 0.40
– 1.78 1.79 1.81
PO (%)c
H2O2d (%)
Yield
Sel.
Conv.
Eff.
94.4 77.5 42.0 85.7
99.6 99.4 99.5 99.6
98.7 81.2 43.5 89.2
96.0 96.0 97.0 96.5
a
Determined by ICP analysis. Langmuir specific surface area (SSA) given by N2 adsorption isotherms at 77 K. All formed catalysts were ground into powder form for the batchwise propylene epoxidation. Reaction conditions: catalyst, 0.1 g; H2O2, 30 mmol; MeCN, 10 g; propylene, 0.4 MPa; temperature, 313 K; time, 2 h. d Measured by N2 adsorption using t-plot method. e Formed Ti-MWW after use in continuous propylene epoxidation for 236 h. f Regenerated by calcination in air at 823 K for 6 h. b
c
the coordination environments of the Ti sites were unchanged in the forming process (Fig. 1B). The Ti content decreased from 0.39 to 0.30 mmol g1 as a result of dilution by the silica binder. Rea-
sonably, formed Ti-MWW gave a relatively lower PO yield and H2O2 conversion than pristine Ti-MWW powder in the batchwise epoxidation of propylene (Table 1).
177
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
3.2. Effect of operating conditions on continuous epoxidation of propylene over formed Ti-MWW
3.2.2. Effect of n(C=3)/n(H2O2) molar ratio Generally, the epoxidation of propylene is carried out in an excess of propylene with the purpose of using up H2O2 completely. The amount of propylene thus had a significant influence on the epoxidation performance, in particular the H2O2 conversion [42]. As shown in Fig. 4, the effect of the n(C=3)/n(H2O2) ratio on the performance of propylene epoxidation was investigated in the range from 1 to 4 by changing the WHSV of C=3 under otherwise similar
Conv., yield., sel. & eff. (%)
100
90
80
313 K
308 K
H2O2 conv.
70
60
318 K
323 K
PO yield PO sel. H2O2 eff.
10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40
Time on stream (h) Fig. 3. Effect of reaction temperature on the epoxidation of propylene over formed Ti-MWW. Reaction conditions: WHSV (H2O2), 0.3 h1; WHSV (solvents), 7.2 h1; n(C=3)/n(H2O2) molar ratio, 3.
Solvolysis
Epoxidation [H2O2]
O
H+
Ti4+ Propylene
[H2O]
PO
OH OH PG
Scheme 2. Reaction pathways of Ti-MWW/H2O2-catalyzed propylene epoxidation in the presence of aprotic solvent.
Conv., yield, sel. & eff. (%)
3.2.1. Effect of reaction temperature The effect of the reaction temperature on the performance of propylene epoxidation was investigated in the temperature range 308–323 K (Fig. 3). With the reaction temperature increasing from 308 to 313 K, both H2O2 conversion and PO yield increased and reached a maximum, and they remained almost unchanged from 313 to 318 K. When the reaction temperature was further increased to 323 K, the PO yield and H2O2 utilization efficiency decreased correspondingly, indicating that operating the reaction at high temperatures tended to accelerate nonproductive decomposition of H2O2. MeCN was proved to be the most suitable solvent for alkene epoxidation on Ti-MWW [28]. Benefiting from the weak basicity of this aprotic solvent, the main side reaction, which is the ring-opening of PO to propylene glycol (PG) as a result of acid-sitecatalyzed hydrolysis (Scheme 2), was effectively inhibited. Thus, there was no apparent change in PO selectivity (about 98.0%) with the variation of the temperature.
100
90
=
n(C3 )/n(H2O2)=4
80 =
n(C3 )/n(H2O2)=3 =
n(C3 )/n(H2O2)=2
70 =
n(C3 )/n(H2O2)=1
60
H2O2 conv. PO yield PO sel. H2O2 eff.
10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40
Time on stream (h) n(C=3)/n(H2O2)
Fig. 4. Effect of molar ratio on the epoxidation of propylene over formed Ti-MWW. Reaction conditions: WHSV (H2O2), 0.3 h1; WHSV (solvents), 7.2 h1; temp., 313 K.
reaction conditions. The epoxidation of propylene requires a stoichiometric amount of H2O2. However, when the reaction was performed at an n(C=3)/n(H2O2) molar ratio of 1:1, the H2O2 conversion and PO yield were only 81.9% and 70.9%, respectively. This is because part of H2O2 was decomposed nonproductively. When the n(C=3)/n(H2O2) ratio was increased to 3, the excess of propylene made the H2O2 conversion and PO yield increase significantly to 98.0% and 84.6%, respectively. When the n(C=3)/n(H2O2) ratio increased further to 4, both the H2O2 conversion and the PO yield remained unchanged, as the H2O2 conversion was close to 100%. In consideration of the fact that excess propylene definitely leads to a waste of reactant and energy consumption for the separation and recovery in industrial processes, the n(C=3)/n(H2O2) ratio of 3 is confirmed as the most effective feeding rate for the HPPO process. Nevertheless, the PO selectivity and H2O2 utilization efficiency were maintained at high levels irrespective of the change in the n(C=3)/n(H2O2) ratio, indicating that Ti-MWW is essentially active and selective for developing efficient HPPO processes. 3.2.3. Effect of the WHSV of H2O2 As the residual H2O2 in the reaction mixture cannot be reused and the presence of H2O2 may cause explosions in the purification sections of real processes, the amount of H2O2 charged in has to be exhausted in the epoxidation process. The WHSV of H2O2 determines the production capacity of the HPPO process. Hence, the effect of H2O2 WHSV on the performance of propylene epoxidation was investigated, and the results are shown in Fig. 5. When the H2O2 WHSV was increased from 0.2 to 0.3 h1, the H2O2 conversion and utilization efficiency were nearly invariable; meanwhile the PO yield and selectivity were almost the same. By further increasing the WHSV of H2O2, the H2O2 conversion and utilization efficiency still reached high values of 98.1% and 87.4% at the initial stage of time on stream (TOS, 10 h), respectively. However, the catalyst deactivated rapidly with TOS, resulting in a lowered PO selectivity of 93.7% and a decreased PO yield of 80.4% at TOS = 35 h. Since the aqueous solution of H2O2 is relatively acidic (pH 3.5– 4.5), at a higher WHSV of H2O2, the acidity of the reaction solution would be increased, which would result in aggravated hydrolysis of PO to PG [3]. Moreover, the H2O2 conversion and the PO yield decreased significantly with TOS. The above results showed that the WHSV of H2O2 should be matched well with the capacity of the catalyst in order to maintain high PO selectivity and H2O2 conversion as well as catalyst stability.
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
100
80 -1
0.2h
H2O2 conv.
60
90 80 80 70
70 100
-1
0.3h
70
90
PO yield PO sel. H2O2 eff.
-1
0.4h
90
90 80 80 70
70 0
10 20 30 40 10 20 30 40 10 20 30 40
PO yield (%)
H2O2 conv. (%)
90
PO sel. (%)
Conv., yield, sel. & eff. (%)
100
60
120 180 240 0
60
H2O2 eff. (%)
178
120 180 240
Time on stream (h)
Time on stream (h) Fig. 5. Effect of the WHSV of H2O2 on the epoxidation of propylene over formed TiMWW. Reaction conditions: WHSV (solvents), 7.2 h1; n(C=3)/n(H2O2) molar ratio, 3; temperature, 313 K.
3.2.4. Effect of the WHSV of the solvent It has been reported that MeCN is the best solvent for Ti-MWWcatalyzed HPPO process in terms of PO selectivity and yield [28]. As the H2O2 aqueous solution (30 wt.%) is used as the oxidant, water is definitely brought into or formed as product in the reaction system. The presence of water, in particular its amount relative to that of organic solvent, is presumed to influence the epoxidation activity as well as the PO selectivity. The effect of solvent WHSV was thus investigated, and the results are shown in Fig. 6. A higher PO selectivity and a more stable duration were achieved at higher WHSVs of the solvent (6.0–8.4 h1). However, the low WHSV of the solvent (4.8 h1) lowered the PO selectivity to ca. 95% and made the catalyst deactivate more rapidly with TOS. This means that a low solvent WHSV would lead to a low concentration of MeCN relative to water in the reaction mixture, which is insufficient enough in quantity to inhibit the hydrolysis of PO to PG by water. The formation of a large quantity of heavy byproduct returned to fasten the deactivation of the catalyst. 3.3. Stability and reusability of formed Ti-MWW catalyst The lifetime of a heterogeneous catalyst was the essential factor for actual applications in industrial processes. We conducted a
Conv., yield, sel. & eff. (%)
100
90 -1
8.4 h 80
-1
6.0 h
-1
7.2 h
H2O2 conv.
70 -1
4.8 h 60
PO yield PO sel. H2O2 eff.
10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40
Time on stream (h) Fig. 6. Effect of the WHSV of solvent on the epoxidation of propylene over formed Ti-MWW. Reaction conditions: WHSV (H2O2), 0.3 h1; n(C=3)/n(H2O2) molar ratio, 3; temperature, 313 K.
Fig. 7. Effect of WHSV on the lifetime of the formed Ti-MWW catalyst in the epoxidation of propylene. Reaction conditions: WHSV (solvents), 7.2 h1 (solid) and 3.6 h1 (blank); WHSV (H2O2), 0.3 h1 (solid) and 0.15 h1 (blank); n(C=3)/n(H2O2) molar ratio, 3; temperature, 313 K.
long-term test of propylene epoxidation in this fixed-bed reactor at 313 K while fixing the n(C=3)/n(H2O2) molar ratio at 3 (Fig. 7). When the reaction was carried out at WHSVs of H2O2 and solvent of 0.3 and 7.2 h1, respectively, the H2O2 conversion and the PO yield were maintained stably at >98.0% and 84.0%, respectively, within TOS of 74 h. They then decreased to 86.6% and 74.6%, respectively, when the reaction time was prolonged to TOS of 114 h, indicating a gradual deactivation of the catalyst. Here it is worth noting that the PO selectivity was always maintained at a high level of 99.0%, which had nothing to do with catalytic deactivation. This is different from propylene epoxidation with the TS-1/ MeOH system [18,43]. Stable PO selectivity can be ascribed to the aprotic solvent of MeCN with a weak basicity. Moreover, the lifetime of the catalyst increased from 74 to 201 h when the WHSVs of H2O2 and solvent were decreased from 0.3 and 7.2 h1 to 0.15 and 3.6 h1, respectively. It can be illustrated that the catalyst in the bed is deactivating nonuniformly down the length of the fixed bed and that a small amount of catalyst deactivation at the higher flow rate (lower residence time) is more detrimental to achieving 100% conversion than at the lower flow rate, where only a small portion of the packed bed is needed to achieve 100% conversion. The Used-Ti-MWW formed catalyst was collected after the propylene epoxidation was run under reaction conditions of 313 K, n(C=3)/n(H2O2) ratio of 3, and WHSVs of H2O2 and solvent of 0.15 and 3.6 h1 for 236 h TOS. The spent catalyst was regenerated by calcination in air at 823 K for 6 h, denoted as RegeneratedTi-MWW formed. Propylene epoxidation in a batch reactor was used to evaluate the catalytic performances of formed Ti-MWW, formed Used-Ti-MWW, and formed Regenerated-Ti-MWW. According to the results shown in Table 1, the H2O2 conversion and PO yield decreased from 81.2 and 77.5% for formed Ti-MWW to 43.5 and 42.0% for formed Used-Ti-MWW, respectively, while the Ti content remained unchanged. After regeneration by calcination, formed Regenerated-Ti-MWW showed a totally recovered or even higher reactivity than the fresh formed Ti-MWW-formed catalyst, possibly due to the decrease of the boron species coexisting in the framework (Table 1). Judging from the SEM images shown in Fig. 2b–d, there was no significant difference among the three catalysts, which all showed a platelet-shaped morphology in company with the amorphous silica. As shown in Fig. 1A, the crystallinity of formed Used-Ti-MWW was relatively weaker than that of formed Ti-MWW, due to the deposition of organic byproducts within the pores of zeolite and binder. Formed Regenerated-Ti-MWW showed a crystallinity
179
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
comparable to that of formed Ti-MWW, indicating that the framework remained intact after a long run in the HPPO process. UV–vis spectroscopy was employed as a convenient technique for detecting the coordination state of Ti. There was no significant difference in UV–vis spectra among the three catalysts (Fig. 1B), which were dominated by the isolated Ti-related band around 210 nm. These results verified that the Ti species remained stable in the zeolite framework without change of coordination states during the HPPO process. The FT-IR spectra in the region of the C–H stretching vibration were measured to investigate the organic species (Electronic Supplementary Information, ESI, Fig. S1). Compared with formed TiMWW, two new bands at 2937 and 2978 cm1 were observed in the spectrum of formed Used-Ti-MWW, which are assigned to C– H stretching from CH2 and CH3 groups [44,45]. They should be closely related to the organic species, in particular the heavy byproducts such as PG deposited inside zeolite pores. Calcination in air readily removed these organic species, leading to complete disappearance of these bands (Fig. S1c). The textural properties of the catalysts determined by nitrogen adsorption isotherms are shown in Fig. 8 and Table 1. Compared with those of formed Ti-MWW, the specific surface area and the pore volume (especially the micropore volume) of formed UsedTi-MWW decreased significantly, which can be ascribed to the pore jamming by organic species. This corresponds to the main reason for the catalyst deactivation as well as the lower reactivity of the used catalyst. The specific surface area and pore volume of the catalyst were recovered by reopening the pores after the removal of the organic species by calcination in air. These results indicated that the deactivation of the catalyst could be ascribed to pore blockage by PG or other byproducts with high boiling points, which prevented the H2O2 and propylene molecules from diffusing into the zeolite channels and adsorbing onto the Ti active sites therein. Very similar results have been reported for propylene epoxidation with the TS-1/MeOH catalytic system [22,46,47]. The TG curves of formed Ti-MWW, formed Used-Ti-MWW, and formed Regenerated-Ti-MWW are shown in Fig. S2 in the ESI. The fresh formed Ti-MWW catalyst showed a weight loss of 5% at about 400 K (ESI, Fig. S2a), which can be ascribed to the physically adsorbed water in the zeolite and binder. In the case of formed Used-Ti-MWW, the TG curve showed two stages of weight loss in the temperature range 298–1000 K. The first stage is the removal of the physically adsorbed water, while the second stage
can be ascribed to the decomposition of the occluded organic species. Moreover, the weight loss of 2% at the first stage for formed Used-Ti-MWW was less than that for formed Ti-MWW, indicating that the organic species occluded in the channel of the zeolite lowered the amount of physically adsorbed water. For the formed Regenerated-Ti-MWW, the TG curve turned to be almost the same as that of formed Ti-MWW. It further demonstrated that the occluded organic species could be released from the zeolite channels by calcination, which agreed with the nitrogen adsorption– desorption isotherms. All these results determined that the deactivation of the catalyst was ascribed to the deposition of the byproducts in the micropores. It is possible to regenerate the used Ti-MWW catalyst after removing the byproducts by calcination and to recover its reactivity for propylene epoxidation.
3.4. Effect of ammonia addition on Ti-MWW-catalyzed HPPO process In general, the ring opening of PO closely depends on the concentration of weak acid sites derived from Si–OH, Ti–OH, or Ti– O–O–H in the titanosilicates [48–50]. Moreover, the formed catalyst possessed part of the amorphous silica binder, with a large number of Si–OH groups, which would further increase the acidity amount. Although the solvolysis (or hydrolysis) of PO was inhibited in the aprotic solvent MeCN for the Ti-MWW/CH3CN system [28,41,51], catalyst deactivation still could not be completely avoided due to accumulative production and deposition of PG inside micropores. It has been reported that the presence of ammonia could effectively inhibit the ring opening of PO [49,52]. Hence, the effect of the concentration of ammonia on the performance of propylene epoxidation over formed Ti-MWW was investigated (Fig. 9). The lifetime increased from 74 to 185 h when the concentration of co-fed ammonia was increased from 0 to 10 ppm. The positive effect of the addition of ammonia on the lifetime of the catalyst could be ascribed to the inhibition of the hydrolysis of PO through a neutralization of the acidity of Si–OH, Ti–OH, or Ti– O–O–H with OH. When the concentration of ammonia was further increased to 20 ppm, the H2O2 conversion probably dropped, and it decreased sharply with time on stream, which is due to the poisoning of the active sites by alkalinity [9,52–54]. Hence, the amount of basic additive should be controlled at an appropriate value to achieve a positive effect on PO production. Otherwise, it will have a negative effect on propylene epoxidation.
100
d
600
H2O2 conversion (%)
3
-1
N2 adsorbed (STP) (cm g )
800
+400 400
c
+300
b +100 200
0 0.0
a
0.2
0.4
0.6
0.8
1.0
P/P0 Fig. 8. Nitrogen adsorption desorption isotherms at 77 K of Ti-MWW-powder (a), formed Ti-MWW (b), formed Used-Ti-MWW (c), and formed Regenerated-Ti-MWW (d).
96
92
0 ppm 10 ppm 20 ppm
88
0
40
80
120
160
200
Time on stream (h) Fig. 9. Effect of the concentration of added ammonia on the lifetime of formed TiMWW. Reaction conditions: WHSV (solvents), 7.2 h1; WHSV(H2O2), 0.3 h1; n(C=3)/ n(H2O2) molar ratio, 3; temperature, 313 K.
180
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
pyridine. Fig. 10B showed the FT-IR spectra of formed Ti-MWW and modified catalysts after the physically adsorbed pyridine molecules were removed by evacuation at 423 K. The stretch band at 1599 cm1, related to the hydroxyl groups, was ascribed to hydrogen-bonded adsorbed pyridine [55]. The bands at 1608 and 1550 cm1 were assigned to the vibration of pyridinium ions generated from chemisorbed pyridine on Lewis acid sites (LAS) and Brønsted acid sites (BAS). The presence of BAS could be ascribed to the residual boron in the matrix of silicalite. Additionally, the band at 1490 cm1 associated with both LAS and BAS was also observed in the spectra of all the studied samples. The bands at 1608, 1490, and 1448 cm1 were decreased by fluorination treatment in comparison to the parent sample. However, these bands were significantly increased when the structural rearrangement was performed on formed Ti-MWW, which could be explained by the presence of hexahedral Ti species with stronger acidity than tetrahedral Ti species [56]. The FT-IR spectra in the region of the hydroxyl stretching vibration for Ti-MWW-powder, formed Ti-MWW, formed Re-Ti-MWW, and formed F-Ti-MWW, are shown in Fig. 11. The bands at 3745, 3720, 3676, and 3510 cm1 are attributed to unperturbed external silanols, internal silanols, Ti–OH, and hydrogen-bonded silanol nests, respectively [38]. Compared with Ti-MWW powder, the hydroxyl stretching vibrations of formed Ti-MWW, particularly the 3745 cm1 band, clearly increased in intensity (Fig. 11a and b), due to the presence of amorphous silica in the latter. After fluorination treatment, the 3745 cm1 band decreased in
3.5. Catalyst modification by different methods The formed Ti-MWW catalyst was modified by structural rearrangement and fluorination treatment. When formed Ti-MWW was subjected to hydrothermal treatment in an aqueous solution containing PI at 443 K and then to further calcination at 823 K, it experienced a structural change from 3D MWW to 2D lamellar precursor and then back to 3D MWW, resulting in the formed Re-TiMWW catalyst. This reversible structural change took place as a result of reinsertion of PI molecules into interlayer spaces [35,36]. On the other hand, when formed Ti-MWW was treated with NH4F in methanol to give a fluorinated sample, formed F-TiMWW, the 3D MWW structure was intact but the fluorination of the zeolite framework was expectable [37–39]. As shown in Table 2, the Si/Ti ratio and the specific surface area remained almost the same after different modifications. Moreover, no significant difference was observed in the XRD patterns (ESI, Fig. S3) and the SEM images (ESI, Fig. S4) among formed TiMWW and further modified catalysts. UV–visible spectroscopy was employed to determine the coordination states of the Ti sites in the four catalysts. All the catalysts showed a main absorbance band at approximately 210 nm (Fig. 10A), which was ascribed to the tetrahedral Ti species in the zeolite framework. However, formed Re-Ti-MWW showed a new broad band around 260 nm ascribed to hexahedral Ti species, which was found in our early work [36]. The acidic properties of the catalysts were determined by the FT-IR spectra of adsorbed
Table 2 Effect of post modification on the physicochemical properties and catalytic performance of the formed Ti-MWW catalyst.a No.
Si/Tib
Catalyst
Si/Bb
Ti statec
SSA
Horizontal strength (MPa)
(m2 g1)d 1 2 3
Formed Ti-MWW Formed Re-Ti-MWWf Formed F-Ti-MWWg
63 65 64
236 240 758
Tetra. Tetra. + Hexa. Tetra.
505 511 512
1.78 3.11 2.50
Products distribution (%)e
H2O2 (%)
PO
PG
Conv.
Eff.
99.7 99.5 99.7
0.3 0.5 0.3
13.2 53.7 44.8
92.9 96.2 95.2
a All formed catalysts were ground into powder form for batchwise propylene epoxidation. Reaction conditions: catalyst, 0.03 g; H2O2, 30 mmol; MeCN, 10 g; propylene, 0.4 MPa; temperature, 313 K; time, 1 h. b Determined by ICP analysis. c Evaluated with UV–vis spectroscopy. Tetra., tetrahedral Ti species. Hexa., hexahedral Ti species. d Langmuir specific surface area (SSA) given by N2 adsorption isotherms at 77 K. e PO, propylene oxide; PG, propylene glycol. f Structural rearrangement: hydrothermal treatment under rotation (10 rpm) at PI/SiO2 molar ratio of 0.4 and H2O/SiO2 molar ratio of 10 at 443 K for 1 day, and then calcination in air at 823 K for 6 h. g Fluorination: treatment in a solution of methanol and NH4F at 443 K for 1 day at an NH4F/SiO2 molar ratio of 0.05 and a MeOH/SiO2 molar ratio of 20, and then calcination in air at 823 K for 6 h.
B
A Absorbance (a.u.)
Absorbance (a.u.)
210 260
c b
1608
1448
1490
c
1550
b a
a 200
300
400
Wavelength (nm)
500
1650
1600
1550
1500
1450
1400
-1
Wavenumber (cm )
Fig. 10. UV–visible spectra (A) and pyridine-adsorption FT-IR spectra after evacuation at 423 K for 0.5 h (B) of formed Ti-MWW (a), formed Re-Ti-MWW (b), and formed F-TiMWW (c).
181
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
3720 3745
3676 3510
Absorbance (a.u.)
d c b a 3800
3700
3600
3500
3400
3300
3200
Wavenumber (cm-1) Fig. 11. IR spectra in the hydroxyl stretching region of Ti-MWW powder (a), formed Ti-MWW (b), formed Re-Ti-MWW (c), and formed F-Ti-MWW (d).
intensity (Fig. 11d). In the case of formed Re-Ti-MWW, the band of 3745 cm1 decreased more sharply (Fig. 11c). The change of silanols was further investigated by 29Si MAS NMR spectroscopy. The Ti-MWW-powder showed three obvious resonances at 95, 98, and 102 ppm, attributed to the Q3 site (Fig. 12A). The formed TiMWW catalyst showed a more intensive Q3 resonance than the Ti-MWW powder (Fig. 12B), possibly due to the presence of amorphous silica in the former. After structural rearrangement or fluorination treatment, the intensity of the Q3 resonance decreased (Fig. 12C and D). A calculation of deconvoluted spectra showed that the Q3 percentages were 8.0%, 11.9%, 5.9%, and 7.8% for Ti-MWW
powder, formed Ti-MWW, formed Re-Ti-MWW, and formed F-TiMWW, respectively. In addition, the mechanical strength of the formed catalysts increased in the order formed Ti-MWW, formed F-Ti-MWW. and formed Re-Ti-MWW (Table 2), which was contrary to the intensity of the 3745 cm1 band and the Q3 percentages. These results suggested that the defect sites in the formed Ti-MWW catalyst definitely disappeared through structural rearrangement or fluorination treatment. The Si–O–Si linkages were probably formed as a result of dehydroxylation condensation between amorphous silica binder and Ti-MWW crystals, which contributed to the enhanced mechanical strength. The batchwise epoxidation of propylene was applied to evaluate the reactivity of the modified catalysts. As shown in Table 2, the H2O2 conversion increased in the order formed Ti-MWW, formed F-Ti-MWW, and formed Re-Ti-MWW, which indicated that fluorination treatment and structural arrangement were effective in enhancing the reactivity of Ti-MWW catalyst in the epoxidation of propylene. The positive effects of fluorination treatment and structural arrangement on the reactivity of Ti-MWW catalyst in HPPO process were due to the formation of SiO3/2F units and hexahedral Ti species, respectively, which have been reported in our earlier works [36–39]. However, formed Re-Ti-MWW showed higher PO conversion than formed Ti-MWW in the hydrolysis of PO, while formed F-Ti-MWW showed the lowest PO conversion (Fig. 13). These results were in accordance with the number of acid sites. Moreover, the addition of H2O2 could further accelerate the hydrolysis of PO, suggesting that the hydrolysis ability of Ti–OOH species was greater than that of the Si–OH and Ti–OH groups in the framework. As the catalyst deactivation was mainly due to the deposition of PG or other heavy byproducts inside zeolite micropores, formed F-Ti-MWW, with a relatively high reactivity in the HPPO process and the lowest activity in the hydrolysis of
B
A 3
3
3
4
-80
-90
-100
3
4
Q /(Q +Q ) = 11.9 %
Q /(Q +Q ) = 8.0 %
-110
-120
-130
-140
-80
-90
Chemical shift (ppm)
-100
-110
-120
3
-140
D
C 3
-130
Chemical shift (ppm)
4
Q /(Q +Q ) = 5.9 % 3
3
4
Q /(Q +Q ) = 7.8 %
-80
-90
-100
-110
-120
Chemical shift (ppm) Fig. 12.
29
-130
-140
-80
-90
-100
-110
-120
-130
-140
Chemical shift (ppm)
Si MAS NMR spectra of Ti-MWW powder (A), formed Ti-MWW (B), formed Re-Ti-MWW (C), and formed F-Ti-MWW (D).
182
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
Acknowledgments 91
%
100
80
%
80 60
34
%
44
40
% 37
20
%
(%)
%
PO conversi on
77
The authors gratefully acknowledge financial support from the NSFC of China (21533002, 21373089, 21403069), and China Ministry of Science and Technology under contract of 2016YFA0202804.
Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jcat.2016.07.020.
0
rmed W-fo W M F-Ti rmed W-fo W M ied Re-T form WWM i T
Wi
th
H 2
Wi
th o
ut
O
H
2O
2
Fig. 13. A comparison of PO conversion among formed Ti-MWW, formed Re-TiMWW, and formed F-Ti-MWW in the hydrolysis of PO with H2O2 (magenta) or without H2O2 (green). Reaction conditions: catalyst, 0.1 g; PO, 10 mmol; H2O, 10 g; H2O2 (if added), 10 mmol; temperature, 333 K; time, 1 h.
H2O2 conversion (%)
100
90
a c
b 80
70
60
0
100
200
300
References
2
400
Time on stream (h) Fig. 14. Comparison of the lifetimes of formed Ti-MWW (a), formed Re-Ti-MWW (b), and formed F-Ti-MWW (c). Reaction conditions: WHSV (solvents), 7.2 h1; WHSV (H2O2), 0.3 h1; n(C=3)/n(H2O2) molar ratio, 3; ammonia concentration, 10 ppm; temperature, 313 K.
PO, showed the longest lifetime of 350 h in the HPPO process, in comparison to formed Ti-MWW and formed Re-Ti-MWW (Fig. 14).
4. Conclusions Formed Ti-MWW serves as an active and selective catalyst for the epoxidation of propylene in a fixed-bed reactor. Co-feeding an appropriate concentration of ammonia into the reaction mixture prolongs its lifetime from 74 to 185 h. No structural degradation or Ti leaching occurs during the continuous use of the formed Ti-MWW catalyst. The formation of heavy products such as propylene glycol may deactivate the catalyst, while the deactivated catalyst is regenerated readily after the organic species are removed by calcination. To increase the lifetime of catalysts in the HPPO process, high PO selectivity would be more important than reaction activity. The Fluorination treatment proves to be more effective than structural rearrangement in increasing the lifetime for the HPPO process.
[1] T.A. Nijhuis, M. Makkee, J.A. Moulijn, B.M. Weckhuysen, Ind. Eng. Chem. Res. 45 (2006) 3447. [2] F. Schmidt, M. Bernhard, H. Morell, M. Pascaly, Chim. Oggi 32 (2014) 31. [3] M. Liu, X.X. Ye, Y.Q. Liu, X.Y. Wang, Y.Q. Wen, H.J. Sun, B.J. Li, Ind. Eng. Chem. Res. 54 (2015) 5416. [4] Status and Impact of State MTBE Ban. [27] http://www.dow.com/polyurethane/news/2012/20120104a.htm. [28] F. Song, Y.M. Liu, L.L. Wang, H.J. Zhang, M.Y. He, P. Wu, Stud. Surf. Sci. Catal. 170 (2007) 1236. [29] P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, Chem. Lett. (2000) 774. [30] P. Wu, D. Nuntasri, J.F. Ruan, Y.M. Liu, M.Y. He, W.B. Fan, O. Terasaki, T. Tatsumi, J. Phys. Chem. B 108 (2004) 19126. [31] L.L. Wang, Y. Wang, Y.M. Liu, L. Chen, S.F. Cheng, G.H. Gao, M.Y. He, P. Wu, Micropor. Mesopor. Mater. 113 (2008) 435. [32] S.Y. Kim, H.J. Ban, W.S. Ahn, Catal. Lett. 113 (2007) 160. [33] P. Wu, J.F. Ruan, L.L. Wang, L.L. Wu, Y. Wang, Y.M. Liu, W.B. Fan, M.Y. He, O. Terasaki, T. Tatsumi, J. Am. Chem. Soc. 130 (2008) 8178. [34] L.L. Wang, Y. Wang, Y.M. Liu, H.H. Wu, X.H. Li, M.Y. He, P. Wu, J. Mater. Chem. 19 (2009) 8594. [35] L.L. Wang, Y.M. Liu, W. Xie, H.H. Wu, X.H. Li, M.Y. He, P. Wu, J. Phys. Chem. C 112 (2008) 6132. [36] L. Xu, D.D. Huang, C.G. Li, X.Y. Ji, S.Q. Jin, Z.C. Feng, F. Xia, X.H. Li, F.T. Fan, C. Li, P. Wu, Chem. Commun. 51 (2015) 9010. [37] X.Q. Fang, Q. Wang, A.M. Zheng, Y.M. Liu, Y.N. Wang, X.J. Deng, H.H. Wu, F. Deng, M.Y. He, P. Wu, Catal. Sci. Technol. 2 (2012) 2433. [38] X.Q. Fang, Q. Wang, A.M. Zheng, Y.M. Liu, L.F. Lin, H.H. Wu, F. Deng, M.Y. He, P. Wu, Phys. Chem. Chem. Phys. 15 (2013) 4930. [39] Y.L. Yang, J.H. Ding, B.S. Wang, J. Wu, C. Zhao, G.H. Gao, P. Wu, J. Catal. 320 (2014) 160. [40] B.C. Lippens, J.H. de Boer, J. Catal. 4 (1965) 319. [41] X.Q. Lu, W.J. Zhou, H.H. Wu, A. Liebens, P. Wu, Appl. Catal. A 515 (2016) 51. [42] Y. Zuo, W.C. Song, M.L. Wang, Y.H. Xu, X.S. Wang, X.W. Guo, Acta. Phys. – Chim. Sin. 29 (2013) 183. [43] U. Muller, G. Krug, P. Rudolf, J.H. Teles, H.G. Gobbel, P. Bassler, U.S. Patent 7 608 728, 2010.
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183 [44] X. Han, H.X. Chen, X. Li, J.Y. Wang, C. Li, S.Y. Chen, Y. Yang, J. Mater. Chem. A 4 (2016) 434. [45] P. Cheung, A. Bhan, G.J. Sunley, D.J. Law, E. Iglesia, J. Catal. 245 (2007) 110. [46] G.F. Thiele, E. Roland, J. Mol. Catal. A 117 (1997) 351. [47] Y. Zuo, M.L. Wang, W.C. Song, X.S. Wang, X.W. Guo, Ind. Eng. Chem. Res. 51 (2012) 10586. [48] L.L. Wang, Y.M. Liu, W. Xie, H.J. Zhang, H.H. Wu, Y.W. Jiang, M.Y. He, P. Wu, J. Catal. 246 (2007) 205. [49] L.Z. Wu, S.F. Zhao, L.F. Lin, X.Q. Fang, Y.M. Liu, M.Y. He, J. Catal. 337 (2016) 248.
183
[50] G. Bellussi, A. Carati, M.G. Clerici, G. Maddinelli, R. Millini, J. Catal. 133 (1992) 220. [51] M.G. Clerici, P. Ingallina, J. Catal. 140 (1993) 71. [52] X.J. Deng, S. Zhang, B.S. Wang, Y.N. Wang, H.H. Wu, Y.M. Liu, M.Y. He, Chem. Commun. 49 (2013) 7504. [53] G. Thiele, U.S. Patent 5 675 026, 1997. [54] W.B. Fan, P. Wu, T. Tatsumi, J. Catal. 256 (2008) 62. [55] Y. Wang, Y.M. Liu, X.H. Li, H.H. Wu, M.Y. He, P. Wu, J. Catal. 266 (2009) 258. [56] G. Xiong, Y.Y. Cao, Z.D. Guo, Q.Y. Jia, F.P. Tian, L.P. Liu, Phys. Chem. Chem. Phys. 18 (2016) 190.
Contents lists available at ScienceDirect
Journal of Catalysis journal homepage: www.elsevier.com/locate/jcat
Selective synthesis of propylene oxide through liquid-phase epoxidation of propylene with H2O2 over formed Ti-MWW catalyst Xinqing Lu, Haihong Wu, Jingang Jiang, Mingyuan He, Peng Wu ⇑ Shanghai Key Laboratory of Green Chemistry and Chemical Processes, School of Chemistry and Molecular Engineering, East China Normal University, North Zhongshan Rd. 3663, Shanghai 200062, People’s Republic of China
a r t i c l e
i n f o
Article history: Received 20 June 2016 Revised 25 July 2016 Accepted 26 July 2016
Keywords: HPPO process Ti-MWW titanosilicate Propylene oxide Liquid-phase epoxidation Fixed-bed reactor
a b s t r a c t The liquid-phase epoxidation of propylene to propylene oxide (PO) over formed titanosilicate catalysts was investigated in a fixed-bed reactor. The effects of reaction temperature, n(C=3)/n(H2O2) molar ratio, and weight hourly space velocity (WHSV) of H2O2 or solvent on the catalytic performance of the formed Ti-MWW catalyst have been extensively studied. Adding an appropriate amount of ammonia to the reaction mixture prolonged the catalyst lifetime effectively. The main byproduct of propylene glycol (PG) and other heavy byproducts with high boiling points were deposited inside zeolite micropores, which corresponded to the main reason for the catalyst deactivation. The high-temperature calcination in air recovered readily the reactivity of the deactivated catalyst. Fluorine implantation remarkably enhanced the reactivity and lifetime of the catalyst in the hydrogen peroxide propylene oxide (HPPO) process, exhibiting a PO selectivity of >99%. Ó 2016 Elsevier Inc. All rights reserved.
1. Introduction Propylene oxide (PO) is one of the most important epoxide chemicals, useful for the manufacture of polyether polyols, propylene glycols, and propylene glycol ethers [1]. PO production capacity reached approximately 10 million metric tons in 2012 and continues growing at an annual rate of about 5% [2,3]. The current commercial routes for PO production are the classic chlorohydrin process and variations of the hydroperoxide process, including the propylene oxide/styrene monomer process (PO/SM), the propylene oxide/t-butyl alcohol process (PO/TBA), the cumene hydroperoxide-based process (CHPO), and the titanosilicatebased hydrogen peroxide propylene oxide process (HPPO). The chlorohydrin process was first developed by Wurtz for the synthesis of ethylene oxide (EO) and PO in 1859. This route has played a critical role for many decades and it is even currently widely employed in the production of PO. However, it suffers serious problems of equipment corrosion and environmental pollution. Thus, it has been gradually replaced by more environmentally benign processes using hydroperoxides as oxidants. For co-production processes such as the PO/SM and PO/ TBA techniques, the demand and pricing of the co-products styrene
⇑ Corresponding author. E-mail address: [email protected] (P. Wu). http://dx.doi.org/10.1016/j.jcat.2016.07.020 0021-9517/Ó 2016 Elsevier Inc. All rights reserved.
monomer (SM) or t-butyl alcohol (TBA) may not economically match those of PO well, making the process optimization difficult. Moreover, in the past few years, the fairly stagnant styrene monomer market and the increasing limitation of methyl t-butyl ether (MTBE) in gasoline [4] restricted the further development of such PO processes. In 2006, Sumitomo Chemical Co. Ltd. developed an innovative process free of co-product using cumene hydroperoxide as an oxidant, in which the intermediate cumyl alcohol was reduced to cumene with costly hydrogen for recycling. The CHPO process exhibits an obvious advantage in terms of avoiding a coproduct, but it requires energy-consuming multiple steps including cumene oxidation, propylene epoxidation, and reduction of alcohol intermediate. The HPPO process is a more ecoefficient method for selective production of PO, in which the catalytic epoxidation of propylene hydrogen peroxide is carried out on titanosilicate catalysts under mild reaction conditions (313–323 K and 2–3 MPa), giving water as the main byproduct. The development of the HPPO process dates back to 1983, when ENI patented a titanosilicate (TS-1) as a useful catalyst for this process [5]. Subsequently, there have been many fundamental researches focusing on the synthesis of TS-1 catalysts [6–13] and the investigation of propylene epoxidation with different reaction conditions [14–16], reactors [3,17,18], and additives [19,20], as well as the reaction mechanism and kinetics of this process [1,14,21–23]. Based on deep understanding of the
174
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
HPPO process, Evonik and SKC launched a commercial plant in South Korea in 2008 with a capacity of 100 Kton/yr, and this plant is operated continuously at maximum capacity [24]. The next year BASF and DOW Chemical launched a similar plant based on HPPO technology in Belgium, with a higher capacity of 300 Kton/yr [25]. Later, the HPPO plants were widely used for the production of PO and they have been launched around the world [26,27]. However, all the commercial HPPO processes adopt without exception the catalytic system TS-1/H2O2/methanol. With methanol as the optimized solvent, the separation of PO from the reaction system has high energy consumption due to the zeotropic problem of methanol and PO. Moreover, the solvolysis of PO with the protic methanol solvent would inevitably decrease the PO selectivity. In addition, the manufacturing cost of TS-1 is always an obstacle to its industrial application, because extremely expensive tetrapropylammonium hydroxide (TPAOH) is required as the structuredirecting agent (SDA) for the crystallization of TS-1. Hence, it is desirable to develop a more suitable titanosilicate for the HPPO process. In our earlier work [28], we conducted a HPPO batch reactor process over Ti-MWW, which preferred an aprotic solvent of acetonitrile and showed superior performance to TS-1. Benefiting from the weak basicity of acetonitrile, the solvolysis of PO was effectively suppressed to make the PO selectivity high level 99%. A certainly cheaper SDA of piperidine (PI) or hexamethyleneimine (HMI) is used for the synthesis of Ti-MWW [29]. On the other hand, possessing a special lamellar structure different from that of the conventional three-dimensional zeolites with rigid structures, TiMWW is readily structurally modifiable by swelling [30], full delamination [30], partial delamination [31], pillaring [32], and silylation [33,34]. Moreover, the structural rearrangement [35,36] or fluorination treatment [37,38] was effective to enhance the reactivity of Ti-MWW. In the present study, we investigated the HPPO process in the fixed-bed reactor-formed Ti-MWW catalysts. The effects of reaction temperature, the molar ratio of propylene to H2O2, the concentration of ammonia, and the weight hourly space velocity (WHSV) of H2O2 or solvent on this HPPO process were systematically studied. We have studied the causes of catalyst deactivation as well as the optimized reaction conditions for prolonging the catalyst lifetime. Additionally, the formed Ti-MWW catalyst was chemically modified to enhance its HPPO catalytic performance by structural rearrangement or fluorination.
2. Experimental 2.1. Catalyst preparation 2.1.1. Synthesis of the active component of Ti-MWW zeolite powder The Ti-MWW powder was hydrothermally synthesized using boric acid as a crystallization-supporting agent and PI as a SDA in two steps as follows: the synthetic gels with molar composition 1.0 SiO2:0.05 TiO2:1.4 PI:0.67 B2O3:19 H2O were hydrothermally crystallized at 443 K for 7 days, and then the obtained powder was refluxed in a 2 M HNO3 aqueous solution to remove extraframework Ti species and part of framework boron as well [29]. Finally, the acid-washed product was filtered, washed with deionized water to pH 7, dried at 393 K overnight, and calcined at 823 K for 6 h (denoted as Ti-MWW powder).
added dropwise into a mixture of 80 g Ti-MWW powder and 5 g Sesbania cannabina Pers powder under grinding. Then a certain amount of deionized water was added to the mixture to adjust the hardness of the paste. Finally, the paste was extruded into £ 1.8 mm strips, dried at 393 K overnight, and calcined at 823 K for 6 h (denoted as formed Ti-MWW). 2.1.3. Preparation of chemically modified Ti-MWW catalyst Two different modification methods were employed to enhance the performance of the catalyst. First, the extrudates were subjected to hydrothermal structural rearrangement treatment in an aqueous PI solution according to the procedures adopted previously for the Ti-MWW powder [35,36]. The treatment was carried out under rotation (10 rpm) at 443 K for 1 day at a PI/SiO2 molar ratio of 0.4 and a H2O/SiO2 molar ratio of 10. The treated product was washed with deionized water, dried at 393 K overnight, and calcined at 823 K for 6 h, and was denoted as formed Re-TiMWW. Second, following previously reported fluorination for the titanosilicate powder [39], we conducted the treatment of the extruded catalysts in a solution of methanol and NH4F at 443 K for 1 day at an NH4F/SiO2 molar ratio of 0.05 and a MeOH/SiO2 molar ratio of 20. The product was then dried at 393 K overnight and calcined at 823 K for 6 h (denoted as formed F-Ti-MWW). 2.2. Characterization methods The X-ray diffraction (XRD) pattern was recorded on a Rigaku Ultima IV diffractometer using Ni-filtered CuKa radiation (k = 0.1541 nm) in a scanning range of 5–35° at a scanning rate of 10° min1 to confirm the structure and crystallinity of the catalysts. The voltage and current were 35 kV and 25 mA, respectively. The morphologies and crystal sizes were examined by a Hitachi S4800 scanning electron microscope. The UV–visible spectra were collected on a PerkinElmer UV–vis Lambda 35 spectrophotometer using BaSO4 as a reference. The IR spectra of hydroxyl stretching and pyridine adsorption were collected by a Nicolet Nexus 670 FT-IR spectrometer at a resolution of 2 cm1. In order to avoid the influence of absorbed water, a self-supported wafer that was placed in a quartz IR cell sealed with CaF2 windows was evacuated at 723 K for 2 h; then the spectra in the hydroxyl stretching region were collected at room temperature. For pyridine spectra measurement, the pretreated wafer was exposed to a pyridine vapor at 298 K for 20 min, and then the absorbed pyridine was evacuated at 423 K for 0.5 h. In addition, the IR spectra in the region of organic functional group vibration were measured using a KBr technique. All the spectra were collected at room temperature. 29Si MAS NMR spectra were recorded on a VARIAN VNMRS-400 MB NMR spectrometer. The content of titanium and boron was determined by ICP-AES on an IRIS Intrepid II XSP after the samples were dissolved in HF solution. The textural properties of the catalysts were determined by N2 adsorption at 77 K using a BEL SORP instrument. Prior to the adsorption measurements, the samples were degassed in vacuum at 573 K for 6 h. The microporous volume was derived from the t-plot method [40]. According to the reported method [17], the mechanical strength of the formed catalysts was determined under different weights until the extrudates were crushed. Thermogravimetric (TG) analysis was carried out in a Netzsch Sta 4049 F3 apparatus in air with a heating rate of 10 K min1 to 1073 K. 2.3. Liquid-phase epoxidation of propylene
2.1.2. Preparation of formed Ti-MWW catalyst The catalyst shaping was performed by an extrusion procedure from a homogeneous paste of zeolite active component, silica binder, and pore generator. A solution of 66.7 g silica sol (30 wt.%) was
2.3.1. Continuous reaction As shown in Scheme 1, the continuous epoxidation of propylene was carried out in a fixed-bed reactor (11 cm length, 1 cm internal
175
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
Scheme 1. Schematic diagram of experimental apparatus.
diameter, and with a Teflon inner layer) that was immersed in a water bath to control the reaction temperature. The extrudates were crushed and sieved to be 40–80 mesh size particles, 3.0 g of which were loaded in the reactor. The pressure was maintained at 2 MPa with the assistance of N2, which made the propylene liquefied. After the reaction temperature and pressure both reached the determined values, the propylene liquid and a mixed solution of H2O2/CH3CN/H2O were introduced into the reactor independently by two dual-plunger pumps, each equipped with two 50 mL containers. The solution for the reactor outlet, containing the products, unconverted reactants, and solvent, entered a liquid-storage tank where it was cooled to room temperature. During the reaction, the reaction mixture was sampled at a certain time interval from the tank. 2.3.2. Batchwise propylene epoxidation The batchwise epoxidation of propylene was carried out in a 45 mL autoclave reactor equipped with a Teflon inner layer according to our early works [28,41]. In a typical run, 0.1 g catalyst, 10 g MeCN, and 30 mmol H2O2 (30 wt.%) were fed into the reactor. Propylene was charged into the autoclave to replace the air inside three times, and then the propylene was charged into the autoclave at a constant pressure of 0.4 MPa. After the reaction was carried out under vigorous stirring at 313 K for 2 h, the reactor was cooled with ice water and depressurized slowly before being opened. 2.3.3. Products analysis The reaction products were analyzed on a gas chromatograph (Shimadzu 2014) FID detector that was equipped with an RtxWax capillary column (30 m 0.25 mm 0.25 lm). The quantification was performed using isopropanol as an internal standard substance. The remaining amount of H2O2 after reaction was determined by a conventional titration method with 0.05 M Ce (SO4)2 solution. The products formed were further confirmed by a GC-MS (Agilent 6890 series GC system, 5937 network mass selective detector). The result of the reaction was given using these criteria:
H2 O2 conv: ¼
n0H2 O2 nH2 O2 n0H2 O2
100%;
nPO 100%; n0H2 O2 nPO 100%; PO sel: ¼ nPO þ nPG nPO þ nPG H2 O2 eff: ¼ 0 100%: nH2 O2 nH2 O2 PO yield ¼
H2O2 conv., PO yield, PO sel., and H2O2 eff. stand for the conversion of H2O2, the yield of PO, the selectivity of PO, and the utilization efficiency of H2O2, respectively. nPO and nPG stand for the numbers of moles of PO and PG, respectively. n0H2 O2 and nH2 O2 stand for the initial mole content and the final content of H2O2, respectively. 3. Results and discussion 3.1. A comparison of physicochemical and catalytic propylene epoxidation properties between Ti-MWW powder and formed Ti-MWW Ti-MWW-powder showed the XRD pattern of the corresponding MWW-type zeolite structure with a relatively high crystallinity (Fig. 1A) and exhibited a typical platelet morphology with a uniform thickness of about 50 nm (Fig. 2a). The addition of silica binder during the forming process led to a slight decrease in crystallinity that was in accord with the presence of an amorphous phase on the surface of the crystals (Fig. 2b). Additionally, the forming process made the specific surface area and the micropore volume decrease from 554 m2 g1 and 0.14 cm3 g1 for Ti-MWW powder to 505 m2 g1 and 0.12 cm3 g1 for formed Ti-MWW, respectively (Table 1). However, the mesopore volume increased from 0.28 to 0.41 cm3 g1, possibly due to the stacking of amorphous silica and the removal of the pore generator of Sesbania cannabina Pers powder during the preparation of the formed catalysts. Both Ti-MWW powder and formed Ti-MWW contained only isolated Ti species in the framework, as evidenced by the predominant adsorption at 210 nm in UV–visible spectra, indicating that
176
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
210
d
Intensity (a.u.)
B
Absorbance (a.u.)
A
c b
d c b
a 5
10
15
20
25
30
a 35
200
300
400
500
Wavelength (nm)
2 Theta (deg.)
Fig. 1. XRD patterns (A) and UV–visible spectra (B) of Ti-MWW powder (a), formed Ti-MWW (b), formed Used-Ti-MWW (c), and formed Regenerated-Ti-MWW (d).
a
b
500 nm
500 nm
c
d
500 nm
500 nm
Fig. 2. SEM micrographs of Ti-MWW powder (a), formed Ti-MWW (b), formed Used-Ti-MWW (c), and formed Regenerated-Ti-MWW (d).
Table 1 Physicochemical properties and catalytic performance of various catalysts. No.
1 2 3 4
Catalyst
Ti-MWW powder Formed Ti-MWW Formed used-Ti-MWWe Formed regenerated-Ti-MWWf
Content (mmol g1)a Ti
B
0.39 0.30 0.29 0.30
0.16 0.12 0.02 0.02
SSA (m2 g1)b
554 505 274 498
Pore volume (cm3 g1)
Horizontal strength (MPa)
Microp.d
Mesop.
0.14 0.12 0.05 0.12
0.28 0.41 0.31 0.40
– 1.78 1.79 1.81
PO (%)c
H2O2d (%)
Yield
Sel.
Conv.
Eff.
94.4 77.5 42.0 85.7
99.6 99.4 99.5 99.6
98.7 81.2 43.5 89.2
96.0 96.0 97.0 96.5
a
Determined by ICP analysis. Langmuir specific surface area (SSA) given by N2 adsorption isotherms at 77 K. All formed catalysts were ground into powder form for the batchwise propylene epoxidation. Reaction conditions: catalyst, 0.1 g; H2O2, 30 mmol; MeCN, 10 g; propylene, 0.4 MPa; temperature, 313 K; time, 2 h. d Measured by N2 adsorption using t-plot method. e Formed Ti-MWW after use in continuous propylene epoxidation for 236 h. f Regenerated by calcination in air at 823 K for 6 h. b
c
the coordination environments of the Ti sites were unchanged in the forming process (Fig. 1B). The Ti content decreased from 0.39 to 0.30 mmol g1 as a result of dilution by the silica binder. Rea-
sonably, formed Ti-MWW gave a relatively lower PO yield and H2O2 conversion than pristine Ti-MWW powder in the batchwise epoxidation of propylene (Table 1).
177
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
3.2. Effect of operating conditions on continuous epoxidation of propylene over formed Ti-MWW
3.2.2. Effect of n(C=3)/n(H2O2) molar ratio Generally, the epoxidation of propylene is carried out in an excess of propylene with the purpose of using up H2O2 completely. The amount of propylene thus had a significant influence on the epoxidation performance, in particular the H2O2 conversion [42]. As shown in Fig. 4, the effect of the n(C=3)/n(H2O2) ratio on the performance of propylene epoxidation was investigated in the range from 1 to 4 by changing the WHSV of C=3 under otherwise similar
Conv., yield., sel. & eff. (%)
100
90
80
313 K
308 K
H2O2 conv.
70
60
318 K
323 K
PO yield PO sel. H2O2 eff.
10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40
Time on stream (h) Fig. 3. Effect of reaction temperature on the epoxidation of propylene over formed Ti-MWW. Reaction conditions: WHSV (H2O2), 0.3 h1; WHSV (solvents), 7.2 h1; n(C=3)/n(H2O2) molar ratio, 3.
Solvolysis
Epoxidation [H2O2]
O
H+
Ti4+ Propylene
[H2O]
PO
OH OH PG
Scheme 2. Reaction pathways of Ti-MWW/H2O2-catalyzed propylene epoxidation in the presence of aprotic solvent.
Conv., yield, sel. & eff. (%)
3.2.1. Effect of reaction temperature The effect of the reaction temperature on the performance of propylene epoxidation was investigated in the temperature range 308–323 K (Fig. 3). With the reaction temperature increasing from 308 to 313 K, both H2O2 conversion and PO yield increased and reached a maximum, and they remained almost unchanged from 313 to 318 K. When the reaction temperature was further increased to 323 K, the PO yield and H2O2 utilization efficiency decreased correspondingly, indicating that operating the reaction at high temperatures tended to accelerate nonproductive decomposition of H2O2. MeCN was proved to be the most suitable solvent for alkene epoxidation on Ti-MWW [28]. Benefiting from the weak basicity of this aprotic solvent, the main side reaction, which is the ring-opening of PO to propylene glycol (PG) as a result of acid-sitecatalyzed hydrolysis (Scheme 2), was effectively inhibited. Thus, there was no apparent change in PO selectivity (about 98.0%) with the variation of the temperature.
100
90
=
n(C3 )/n(H2O2)=4
80 =
n(C3 )/n(H2O2)=3 =
n(C3 )/n(H2O2)=2
70 =
n(C3 )/n(H2O2)=1
60
H2O2 conv. PO yield PO sel. H2O2 eff.
10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40
Time on stream (h) n(C=3)/n(H2O2)
Fig. 4. Effect of molar ratio on the epoxidation of propylene over formed Ti-MWW. Reaction conditions: WHSV (H2O2), 0.3 h1; WHSV (solvents), 7.2 h1; temp., 313 K.
reaction conditions. The epoxidation of propylene requires a stoichiometric amount of H2O2. However, when the reaction was performed at an n(C=3)/n(H2O2) molar ratio of 1:1, the H2O2 conversion and PO yield were only 81.9% and 70.9%, respectively. This is because part of H2O2 was decomposed nonproductively. When the n(C=3)/n(H2O2) ratio was increased to 3, the excess of propylene made the H2O2 conversion and PO yield increase significantly to 98.0% and 84.6%, respectively. When the n(C=3)/n(H2O2) ratio increased further to 4, both the H2O2 conversion and the PO yield remained unchanged, as the H2O2 conversion was close to 100%. In consideration of the fact that excess propylene definitely leads to a waste of reactant and energy consumption for the separation and recovery in industrial processes, the n(C=3)/n(H2O2) ratio of 3 is confirmed as the most effective feeding rate for the HPPO process. Nevertheless, the PO selectivity and H2O2 utilization efficiency were maintained at high levels irrespective of the change in the n(C=3)/n(H2O2) ratio, indicating that Ti-MWW is essentially active and selective for developing efficient HPPO processes. 3.2.3. Effect of the WHSV of H2O2 As the residual H2O2 in the reaction mixture cannot be reused and the presence of H2O2 may cause explosions in the purification sections of real processes, the amount of H2O2 charged in has to be exhausted in the epoxidation process. The WHSV of H2O2 determines the production capacity of the HPPO process. Hence, the effect of H2O2 WHSV on the performance of propylene epoxidation was investigated, and the results are shown in Fig. 5. When the H2O2 WHSV was increased from 0.2 to 0.3 h1, the H2O2 conversion and utilization efficiency were nearly invariable; meanwhile the PO yield and selectivity were almost the same. By further increasing the WHSV of H2O2, the H2O2 conversion and utilization efficiency still reached high values of 98.1% and 87.4% at the initial stage of time on stream (TOS, 10 h), respectively. However, the catalyst deactivated rapidly with TOS, resulting in a lowered PO selectivity of 93.7% and a decreased PO yield of 80.4% at TOS = 35 h. Since the aqueous solution of H2O2 is relatively acidic (pH 3.5– 4.5), at a higher WHSV of H2O2, the acidity of the reaction solution would be increased, which would result in aggravated hydrolysis of PO to PG [3]. Moreover, the H2O2 conversion and the PO yield decreased significantly with TOS. The above results showed that the WHSV of H2O2 should be matched well with the capacity of the catalyst in order to maintain high PO selectivity and H2O2 conversion as well as catalyst stability.
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
100
80 -1
0.2h
H2O2 conv.
60
90 80 80 70
70 100
-1
0.3h
70
90
PO yield PO sel. H2O2 eff.
-1
0.4h
90
90 80 80 70
70 0
10 20 30 40 10 20 30 40 10 20 30 40
PO yield (%)
H2O2 conv. (%)
90
PO sel. (%)
Conv., yield, sel. & eff. (%)
100
60
120 180 240 0
60
H2O2 eff. (%)
178
120 180 240
Time on stream (h)
Time on stream (h) Fig. 5. Effect of the WHSV of H2O2 on the epoxidation of propylene over formed TiMWW. Reaction conditions: WHSV (solvents), 7.2 h1; n(C=3)/n(H2O2) molar ratio, 3; temperature, 313 K.
3.2.4. Effect of the WHSV of the solvent It has been reported that MeCN is the best solvent for Ti-MWWcatalyzed HPPO process in terms of PO selectivity and yield [28]. As the H2O2 aqueous solution (30 wt.%) is used as the oxidant, water is definitely brought into or formed as product in the reaction system. The presence of water, in particular its amount relative to that of organic solvent, is presumed to influence the epoxidation activity as well as the PO selectivity. The effect of solvent WHSV was thus investigated, and the results are shown in Fig. 6. A higher PO selectivity and a more stable duration were achieved at higher WHSVs of the solvent (6.0–8.4 h1). However, the low WHSV of the solvent (4.8 h1) lowered the PO selectivity to ca. 95% and made the catalyst deactivate more rapidly with TOS. This means that a low solvent WHSV would lead to a low concentration of MeCN relative to water in the reaction mixture, which is insufficient enough in quantity to inhibit the hydrolysis of PO to PG by water. The formation of a large quantity of heavy byproduct returned to fasten the deactivation of the catalyst. 3.3. Stability and reusability of formed Ti-MWW catalyst The lifetime of a heterogeneous catalyst was the essential factor for actual applications in industrial processes. We conducted a
Conv., yield, sel. & eff. (%)
100
90 -1
8.4 h 80
-1
6.0 h
-1
7.2 h
H2O2 conv.
70 -1
4.8 h 60
PO yield PO sel. H2O2 eff.
10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40
Time on stream (h) Fig. 6. Effect of the WHSV of solvent on the epoxidation of propylene over formed Ti-MWW. Reaction conditions: WHSV (H2O2), 0.3 h1; n(C=3)/n(H2O2) molar ratio, 3; temperature, 313 K.
Fig. 7. Effect of WHSV on the lifetime of the formed Ti-MWW catalyst in the epoxidation of propylene. Reaction conditions: WHSV (solvents), 7.2 h1 (solid) and 3.6 h1 (blank); WHSV (H2O2), 0.3 h1 (solid) and 0.15 h1 (blank); n(C=3)/n(H2O2) molar ratio, 3; temperature, 313 K.
long-term test of propylene epoxidation in this fixed-bed reactor at 313 K while fixing the n(C=3)/n(H2O2) molar ratio at 3 (Fig. 7). When the reaction was carried out at WHSVs of H2O2 and solvent of 0.3 and 7.2 h1, respectively, the H2O2 conversion and the PO yield were maintained stably at >98.0% and 84.0%, respectively, within TOS of 74 h. They then decreased to 86.6% and 74.6%, respectively, when the reaction time was prolonged to TOS of 114 h, indicating a gradual deactivation of the catalyst. Here it is worth noting that the PO selectivity was always maintained at a high level of 99.0%, which had nothing to do with catalytic deactivation. This is different from propylene epoxidation with the TS-1/ MeOH system [18,43]. Stable PO selectivity can be ascribed to the aprotic solvent of MeCN with a weak basicity. Moreover, the lifetime of the catalyst increased from 74 to 201 h when the WHSVs of H2O2 and solvent were decreased from 0.3 and 7.2 h1 to 0.15 and 3.6 h1, respectively. It can be illustrated that the catalyst in the bed is deactivating nonuniformly down the length of the fixed bed and that a small amount of catalyst deactivation at the higher flow rate (lower residence time) is more detrimental to achieving 100% conversion than at the lower flow rate, where only a small portion of the packed bed is needed to achieve 100% conversion. The Used-Ti-MWW formed catalyst was collected after the propylene epoxidation was run under reaction conditions of 313 K, n(C=3)/n(H2O2) ratio of 3, and WHSVs of H2O2 and solvent of 0.15 and 3.6 h1 for 236 h TOS. The spent catalyst was regenerated by calcination in air at 823 K for 6 h, denoted as RegeneratedTi-MWW formed. Propylene epoxidation in a batch reactor was used to evaluate the catalytic performances of formed Ti-MWW, formed Used-Ti-MWW, and formed Regenerated-Ti-MWW. According to the results shown in Table 1, the H2O2 conversion and PO yield decreased from 81.2 and 77.5% for formed Ti-MWW to 43.5 and 42.0% for formed Used-Ti-MWW, respectively, while the Ti content remained unchanged. After regeneration by calcination, formed Regenerated-Ti-MWW showed a totally recovered or even higher reactivity than the fresh formed Ti-MWW-formed catalyst, possibly due to the decrease of the boron species coexisting in the framework (Table 1). Judging from the SEM images shown in Fig. 2b–d, there was no significant difference among the three catalysts, which all showed a platelet-shaped morphology in company with the amorphous silica. As shown in Fig. 1A, the crystallinity of formed Used-Ti-MWW was relatively weaker than that of formed Ti-MWW, due to the deposition of organic byproducts within the pores of zeolite and binder. Formed Regenerated-Ti-MWW showed a crystallinity
179
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
comparable to that of formed Ti-MWW, indicating that the framework remained intact after a long run in the HPPO process. UV–vis spectroscopy was employed as a convenient technique for detecting the coordination state of Ti. There was no significant difference in UV–vis spectra among the three catalysts (Fig. 1B), which were dominated by the isolated Ti-related band around 210 nm. These results verified that the Ti species remained stable in the zeolite framework without change of coordination states during the HPPO process. The FT-IR spectra in the region of the C–H stretching vibration were measured to investigate the organic species (Electronic Supplementary Information, ESI, Fig. S1). Compared with formed TiMWW, two new bands at 2937 and 2978 cm1 were observed in the spectrum of formed Used-Ti-MWW, which are assigned to C– H stretching from CH2 and CH3 groups [44,45]. They should be closely related to the organic species, in particular the heavy byproducts such as PG deposited inside zeolite pores. Calcination in air readily removed these organic species, leading to complete disappearance of these bands (Fig. S1c). The textural properties of the catalysts determined by nitrogen adsorption isotherms are shown in Fig. 8 and Table 1. Compared with those of formed Ti-MWW, the specific surface area and the pore volume (especially the micropore volume) of formed UsedTi-MWW decreased significantly, which can be ascribed to the pore jamming by organic species. This corresponds to the main reason for the catalyst deactivation as well as the lower reactivity of the used catalyst. The specific surface area and pore volume of the catalyst were recovered by reopening the pores after the removal of the organic species by calcination in air. These results indicated that the deactivation of the catalyst could be ascribed to pore blockage by PG or other byproducts with high boiling points, which prevented the H2O2 and propylene molecules from diffusing into the zeolite channels and adsorbing onto the Ti active sites therein. Very similar results have been reported for propylene epoxidation with the TS-1/MeOH catalytic system [22,46,47]. The TG curves of formed Ti-MWW, formed Used-Ti-MWW, and formed Regenerated-Ti-MWW are shown in Fig. S2 in the ESI. The fresh formed Ti-MWW catalyst showed a weight loss of 5% at about 400 K (ESI, Fig. S2a), which can be ascribed to the physically adsorbed water in the zeolite and binder. In the case of formed Used-Ti-MWW, the TG curve showed two stages of weight loss in the temperature range 298–1000 K. The first stage is the removal of the physically adsorbed water, while the second stage
can be ascribed to the decomposition of the occluded organic species. Moreover, the weight loss of 2% at the first stage for formed Used-Ti-MWW was less than that for formed Ti-MWW, indicating that the organic species occluded in the channel of the zeolite lowered the amount of physically adsorbed water. For the formed Regenerated-Ti-MWW, the TG curve turned to be almost the same as that of formed Ti-MWW. It further demonstrated that the occluded organic species could be released from the zeolite channels by calcination, which agreed with the nitrogen adsorption– desorption isotherms. All these results determined that the deactivation of the catalyst was ascribed to the deposition of the byproducts in the micropores. It is possible to regenerate the used Ti-MWW catalyst after removing the byproducts by calcination and to recover its reactivity for propylene epoxidation.
3.4. Effect of ammonia addition on Ti-MWW-catalyzed HPPO process In general, the ring opening of PO closely depends on the concentration of weak acid sites derived from Si–OH, Ti–OH, or Ti– O–O–H in the titanosilicates [48–50]. Moreover, the formed catalyst possessed part of the amorphous silica binder, with a large number of Si–OH groups, which would further increase the acidity amount. Although the solvolysis (or hydrolysis) of PO was inhibited in the aprotic solvent MeCN for the Ti-MWW/CH3CN system [28,41,51], catalyst deactivation still could not be completely avoided due to accumulative production and deposition of PG inside micropores. It has been reported that the presence of ammonia could effectively inhibit the ring opening of PO [49,52]. Hence, the effect of the concentration of ammonia on the performance of propylene epoxidation over formed Ti-MWW was investigated (Fig. 9). The lifetime increased from 74 to 185 h when the concentration of co-fed ammonia was increased from 0 to 10 ppm. The positive effect of the addition of ammonia on the lifetime of the catalyst could be ascribed to the inhibition of the hydrolysis of PO through a neutralization of the acidity of Si–OH, Ti–OH, or Ti– O–O–H with OH. When the concentration of ammonia was further increased to 20 ppm, the H2O2 conversion probably dropped, and it decreased sharply with time on stream, which is due to the poisoning of the active sites by alkalinity [9,52–54]. Hence, the amount of basic additive should be controlled at an appropriate value to achieve a positive effect on PO production. Otherwise, it will have a negative effect on propylene epoxidation.
100
d
600
H2O2 conversion (%)
3
-1
N2 adsorbed (STP) (cm g )
800
+400 400
c
+300
b +100 200
0 0.0
a
0.2
0.4
0.6
0.8
1.0
P/P0 Fig. 8. Nitrogen adsorption desorption isotherms at 77 K of Ti-MWW-powder (a), formed Ti-MWW (b), formed Used-Ti-MWW (c), and formed Regenerated-Ti-MWW (d).
96
92
0 ppm 10 ppm 20 ppm
88
0
40
80
120
160
200
Time on stream (h) Fig. 9. Effect of the concentration of added ammonia on the lifetime of formed TiMWW. Reaction conditions: WHSV (solvents), 7.2 h1; WHSV(H2O2), 0.3 h1; n(C=3)/ n(H2O2) molar ratio, 3; temperature, 313 K.
180
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
pyridine. Fig. 10B showed the FT-IR spectra of formed Ti-MWW and modified catalysts after the physically adsorbed pyridine molecules were removed by evacuation at 423 K. The stretch band at 1599 cm1, related to the hydroxyl groups, was ascribed to hydrogen-bonded adsorbed pyridine [55]. The bands at 1608 and 1550 cm1 were assigned to the vibration of pyridinium ions generated from chemisorbed pyridine on Lewis acid sites (LAS) and Brønsted acid sites (BAS). The presence of BAS could be ascribed to the residual boron in the matrix of silicalite. Additionally, the band at 1490 cm1 associated with both LAS and BAS was also observed in the spectra of all the studied samples. The bands at 1608, 1490, and 1448 cm1 were decreased by fluorination treatment in comparison to the parent sample. However, these bands were significantly increased when the structural rearrangement was performed on formed Ti-MWW, which could be explained by the presence of hexahedral Ti species with stronger acidity than tetrahedral Ti species [56]. The FT-IR spectra in the region of the hydroxyl stretching vibration for Ti-MWW-powder, formed Ti-MWW, formed Re-Ti-MWW, and formed F-Ti-MWW, are shown in Fig. 11. The bands at 3745, 3720, 3676, and 3510 cm1 are attributed to unperturbed external silanols, internal silanols, Ti–OH, and hydrogen-bonded silanol nests, respectively [38]. Compared with Ti-MWW powder, the hydroxyl stretching vibrations of formed Ti-MWW, particularly the 3745 cm1 band, clearly increased in intensity (Fig. 11a and b), due to the presence of amorphous silica in the latter. After fluorination treatment, the 3745 cm1 band decreased in
3.5. Catalyst modification by different methods The formed Ti-MWW catalyst was modified by structural rearrangement and fluorination treatment. When formed Ti-MWW was subjected to hydrothermal treatment in an aqueous solution containing PI at 443 K and then to further calcination at 823 K, it experienced a structural change from 3D MWW to 2D lamellar precursor and then back to 3D MWW, resulting in the formed Re-TiMWW catalyst. This reversible structural change took place as a result of reinsertion of PI molecules into interlayer spaces [35,36]. On the other hand, when formed Ti-MWW was treated with NH4F in methanol to give a fluorinated sample, formed F-TiMWW, the 3D MWW structure was intact but the fluorination of the zeolite framework was expectable [37–39]. As shown in Table 2, the Si/Ti ratio and the specific surface area remained almost the same after different modifications. Moreover, no significant difference was observed in the XRD patterns (ESI, Fig. S3) and the SEM images (ESI, Fig. S4) among formed TiMWW and further modified catalysts. UV–visible spectroscopy was employed to determine the coordination states of the Ti sites in the four catalysts. All the catalysts showed a main absorbance band at approximately 210 nm (Fig. 10A), which was ascribed to the tetrahedral Ti species in the zeolite framework. However, formed Re-Ti-MWW showed a new broad band around 260 nm ascribed to hexahedral Ti species, which was found in our early work [36]. The acidic properties of the catalysts were determined by the FT-IR spectra of adsorbed
Table 2 Effect of post modification on the physicochemical properties and catalytic performance of the formed Ti-MWW catalyst.a No.
Si/Tib
Catalyst
Si/Bb
Ti statec
SSA
Horizontal strength (MPa)
(m2 g1)d 1 2 3
Formed Ti-MWW Formed Re-Ti-MWWf Formed F-Ti-MWWg
63 65 64
236 240 758
Tetra. Tetra. + Hexa. Tetra.
505 511 512
1.78 3.11 2.50
Products distribution (%)e
H2O2 (%)
PO
PG
Conv.
Eff.
99.7 99.5 99.7
0.3 0.5 0.3
13.2 53.7 44.8
92.9 96.2 95.2
a All formed catalysts were ground into powder form for batchwise propylene epoxidation. Reaction conditions: catalyst, 0.03 g; H2O2, 30 mmol; MeCN, 10 g; propylene, 0.4 MPa; temperature, 313 K; time, 1 h. b Determined by ICP analysis. c Evaluated with UV–vis spectroscopy. Tetra., tetrahedral Ti species. Hexa., hexahedral Ti species. d Langmuir specific surface area (SSA) given by N2 adsorption isotherms at 77 K. e PO, propylene oxide; PG, propylene glycol. f Structural rearrangement: hydrothermal treatment under rotation (10 rpm) at PI/SiO2 molar ratio of 0.4 and H2O/SiO2 molar ratio of 10 at 443 K for 1 day, and then calcination in air at 823 K for 6 h. g Fluorination: treatment in a solution of methanol and NH4F at 443 K for 1 day at an NH4F/SiO2 molar ratio of 0.05 and a MeOH/SiO2 molar ratio of 20, and then calcination in air at 823 K for 6 h.
B
A Absorbance (a.u.)
Absorbance (a.u.)
210 260
c b
1608
1448
1490
c
1550
b a
a 200
300
400
Wavelength (nm)
500
1650
1600
1550
1500
1450
1400
-1
Wavenumber (cm )
Fig. 10. UV–visible spectra (A) and pyridine-adsorption FT-IR spectra after evacuation at 423 K for 0.5 h (B) of formed Ti-MWW (a), formed Re-Ti-MWW (b), and formed F-TiMWW (c).
181
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
3720 3745
3676 3510
Absorbance (a.u.)
d c b a 3800
3700
3600
3500
3400
3300
3200
Wavenumber (cm-1) Fig. 11. IR spectra in the hydroxyl stretching region of Ti-MWW powder (a), formed Ti-MWW (b), formed Re-Ti-MWW (c), and formed F-Ti-MWW (d).
intensity (Fig. 11d). In the case of formed Re-Ti-MWW, the band of 3745 cm1 decreased more sharply (Fig. 11c). The change of silanols was further investigated by 29Si MAS NMR spectroscopy. The Ti-MWW-powder showed three obvious resonances at 95, 98, and 102 ppm, attributed to the Q3 site (Fig. 12A). The formed TiMWW catalyst showed a more intensive Q3 resonance than the Ti-MWW powder (Fig. 12B), possibly due to the presence of amorphous silica in the former. After structural rearrangement or fluorination treatment, the intensity of the Q3 resonance decreased (Fig. 12C and D). A calculation of deconvoluted spectra showed that the Q3 percentages were 8.0%, 11.9%, 5.9%, and 7.8% for Ti-MWW
powder, formed Ti-MWW, formed Re-Ti-MWW, and formed F-TiMWW, respectively. In addition, the mechanical strength of the formed catalysts increased in the order formed Ti-MWW, formed F-Ti-MWW. and formed Re-Ti-MWW (Table 2), which was contrary to the intensity of the 3745 cm1 band and the Q3 percentages. These results suggested that the defect sites in the formed Ti-MWW catalyst definitely disappeared through structural rearrangement or fluorination treatment. The Si–O–Si linkages were probably formed as a result of dehydroxylation condensation between amorphous silica binder and Ti-MWW crystals, which contributed to the enhanced mechanical strength. The batchwise epoxidation of propylene was applied to evaluate the reactivity of the modified catalysts. As shown in Table 2, the H2O2 conversion increased in the order formed Ti-MWW, formed F-Ti-MWW, and formed Re-Ti-MWW, which indicated that fluorination treatment and structural arrangement were effective in enhancing the reactivity of Ti-MWW catalyst in the epoxidation of propylene. The positive effects of fluorination treatment and structural arrangement on the reactivity of Ti-MWW catalyst in HPPO process were due to the formation of SiO3/2F units and hexahedral Ti species, respectively, which have been reported in our earlier works [36–39]. However, formed Re-Ti-MWW showed higher PO conversion than formed Ti-MWW in the hydrolysis of PO, while formed F-Ti-MWW showed the lowest PO conversion (Fig. 13). These results were in accordance with the number of acid sites. Moreover, the addition of H2O2 could further accelerate the hydrolysis of PO, suggesting that the hydrolysis ability of Ti–OOH species was greater than that of the Si–OH and Ti–OH groups in the framework. As the catalyst deactivation was mainly due to the deposition of PG or other heavy byproducts inside zeolite micropores, formed F-Ti-MWW, with a relatively high reactivity in the HPPO process and the lowest activity in the hydrolysis of
B
A 3
3
3
4
-80
-90
-100
3
4
Q /(Q +Q ) = 11.9 %
Q /(Q +Q ) = 8.0 %
-110
-120
-130
-140
-80
-90
Chemical shift (ppm)
-100
-110
-120
3
-140
D
C 3
-130
Chemical shift (ppm)
4
Q /(Q +Q ) = 5.9 % 3
3
4
Q /(Q +Q ) = 7.8 %
-80
-90
-100
-110
-120
Chemical shift (ppm) Fig. 12.
29
-130
-140
-80
-90
-100
-110
-120
-130
-140
Chemical shift (ppm)
Si MAS NMR spectra of Ti-MWW powder (A), formed Ti-MWW (B), formed Re-Ti-MWW (C), and formed F-Ti-MWW (D).
182
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183
Acknowledgments 91
%
100
80
%
80 60
34
%
44
40
% 37
20
%
(%)
%
PO conversi on
77
The authors gratefully acknowledge financial support from the NSFC of China (21533002, 21373089, 21403069), and China Ministry of Science and Technology under contract of 2016YFA0202804.
Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jcat.2016.07.020.
0
rmed W-fo W M F-Ti rmed W-fo W M ied Re-T form WWM i T
Wi
th
H 2
Wi
th o
ut
O
H
2O
2
Fig. 13. A comparison of PO conversion among formed Ti-MWW, formed Re-TiMWW, and formed F-Ti-MWW in the hydrolysis of PO with H2O2 (magenta) or without H2O2 (green). Reaction conditions: catalyst, 0.1 g; PO, 10 mmol; H2O, 10 g; H2O2 (if added), 10 mmol; temperature, 333 K; time, 1 h.
H2O2 conversion (%)
100
90
a c
b 80
70
60
0
100
200
300
References
2
400
Time on stream (h) Fig. 14. Comparison of the lifetimes of formed Ti-MWW (a), formed Re-Ti-MWW (b), and formed F-Ti-MWW (c). Reaction conditions: WHSV (solvents), 7.2 h1; WHSV (H2O2), 0.3 h1; n(C=3)/n(H2O2) molar ratio, 3; ammonia concentration, 10 ppm; temperature, 313 K.
PO, showed the longest lifetime of 350 h in the HPPO process, in comparison to formed Ti-MWW and formed Re-Ti-MWW (Fig. 14).
4. Conclusions Formed Ti-MWW serves as an active and selective catalyst for the epoxidation of propylene in a fixed-bed reactor. Co-feeding an appropriate concentration of ammonia into the reaction mixture prolongs its lifetime from 74 to 185 h. No structural degradation or Ti leaching occurs during the continuous use of the formed Ti-MWW catalyst. The formation of heavy products such as propylene glycol may deactivate the catalyst, while the deactivated catalyst is regenerated readily after the organic species are removed by calcination. To increase the lifetime of catalysts in the HPPO process, high PO selectivity would be more important than reaction activity. The Fluorination treatment proves to be more effective than structural rearrangement in increasing the lifetime for the HPPO process.
[1] T.A. Nijhuis, M. Makkee, J.A. Moulijn, B.M. Weckhuysen, Ind. Eng. Chem. Res. 45 (2006) 3447. [2] F. Schmidt, M. Bernhard, H. Morell, M. Pascaly, Chim. Oggi 32 (2014) 31. [3] M. Liu, X.X. Ye, Y.Q. Liu, X.Y. Wang, Y.Q. Wen, H.J. Sun, B.J. Li, Ind. Eng. Chem. Res. 54 (2015) 5416. [4] Status and Impact of State MTBE Ban
X. Lu et al. / Journal of Catalysis 342 (2016) 173–183 [44] X. Han, H.X. Chen, X. Li, J.Y. Wang, C. Li, S.Y. Chen, Y. Yang, J. Mater. Chem. A 4 (2016) 434. [45] P. Cheung, A. Bhan, G.J. Sunley, D.J. Law, E. Iglesia, J. Catal. 245 (2007) 110. [46] G.F. Thiele, E. Roland, J. Mol. Catal. A 117 (1997) 351. [47] Y. Zuo, M.L. Wang, W.C. Song, X.S. Wang, X.W. Guo, Ind. Eng. Chem. Res. 51 (2012) 10586. [48] L.L. Wang, Y.M. Liu, W. Xie, H.J. Zhang, H.H. Wu, Y.W. Jiang, M.Y. He, P. Wu, J. Catal. 246 (2007) 205. [49] L.Z. Wu, S.F. Zhao, L.F. Lin, X.Q. Fang, Y.M. Liu, M.Y. He, J. Catal. 337 (2016) 248.
183
[50] G. Bellussi, A. Carati, M.G. Clerici, G. Maddinelli, R. Millini, J. Catal. 133 (1992) 220. [51] M.G. Clerici, P. Ingallina, J. Catal. 140 (1993) 71. [52] X.J. Deng, S. Zhang, B.S. Wang, Y.N. Wang, H.H. Wu, Y.M. Liu, M.Y. He, Chem. Commun. 49 (2013) 7504. [53] G. Thiele, U.S. Patent 5 675 026, 1997. [54] W.B. Fan, P. Wu, T. Tatsumi, J. Catal. 256 (2008) 62. [55] Y. Wang, Y.M. Liu, X.H. Li, H.H. Wu, M.Y. He, P. Wu, J. Catal. 266 (2009) 258. [56] G. Xiong, Y.Y. Cao, Z.D. Guo, Q.Y. Jia, F.P. Tian, L.P. Liu, Phys. Chem. Chem. Phys. 18 (2016) 190.